Download as pdf or txt
Download as pdf or txt
You are on page 1of 28

Agricultural and Forest Meteorology 108 (2001) 213–240

Quantification of odors and odorants from swine


operations in North Carolina
Susan S. Schiffman a,∗ , Jeanette L. Bennett a , James H. Raymer b
a Department of Psychiatry, Duke University Medical School, Durham, NC 27710, USA
b Research Triangle Institute, Research Triangle Park (RTP), NC 27709-2194, USA

Received 23 July 2000; received in revised form 5 January 2001; accepted 28 February 2001

Abstract
A total of 331 different VOCs and fixed gases from swine facilities in North Carolina were identified by gas chromatography
and mass spectrometry (GC/MS). Of these, 203 were found in air samples adsorbed onto Tenax® , 112 were found in air samples
adsorbed onto cotton material, and 167 different compounds were identified in the lagoon samples. The compounds identified
were diverse, and included many acids, alcohols, aldehydes, amides, amines, aromatics, esters, ethers, fixed gases, halogenated
hydrocarbons, hydrocarbons, ketones, nitriles, other nitrogen-containing compounds, phenols, sulfur-containing compounds,
steroids, and other compounds. The vast majority of these compounds were present at concentrations below published odor
and irritation thresholds. Yet human assessments indicated that odors (and irritant sensations) in the immediate vicinity of
the swine houses (and even at distances beyond 1000 ft) were strong. Comparison of the findings from chemical and human
assessments points to the importance of the cumulative effects of hundreds of compounds in producing odor and irritation
downwind of swine operations. Many GC peaks from the samples were too small to allow identification of the compounds, but
their presence may also contribute significantly to the odor and irritation. Several methodological difficulties were associated
with the human odor assessments. Odorous air evaluated in the field was simultaneously collected in Tedlar® bags for
evaluation in the laboratory; however, intensity ratings in the field were higher than those in the laboratory. This is due to the
fact that organic dust (dried fecal material and feed) adheres to Tedlar® bags and the tubing of collection/delivery systems;
therefore, only VOCs from the vapor phase (but not the dust) reach the nose of the panelists in sniffing air samples obtained
in Tedlar® bags. Future collection and measurement techniques need to be developed that can evaluate odors from dust and
vapor phases simultaneously in the laboratory. Dispersion models also need to be developed that account accurately for odor
intensities downwind of animal operations. Finally, safety standards for odor exposures need to be determined that consider
the risk of simultaneous exposure to hundreds of low level compounds. © 2001 Elsevier Science B.V. All rights reserved.
Keywords: Odor; CAFOs; Swine; Dust; Olfactometer

1. Introduction

Citizens in communities with odorous air are con-


cerned about potential health effects, long-term degra-
∗ Corresponding author. Tel.: +1-919-660-5657;
dation of their environment, and economic impacts
fax: +1-919-684-8449.
on their property (Miner, 1980; Bundy, 1992; Shus-
E-mail addresses: sss@acpub.duke.edu, schif003@mc.duke.edu terman, 1992; Pate, 1998; Sweeten, 1998; Schiffman
(S.S. Schiffman). et al., 2000). An odor is a sensation that occurs when

0168-1923/01/$ – see front matter © 2001 Elsevier Science B.V. All rights reserved.
PII: S 0 1 6 8 - 1 9 2 3 ( 0 1 ) 0 0 2 3 9 - 8
214 S.S. Schiffman et al. / Agricultural and Forest Meteorology 108 (2001) 213–240

an odorant (the chemical that happens to have an throat, cough, chest tightness, nasal congestion, pal-
odor) impacts the sensory receptors in the nasal cav- pitations, shortness of breath, stress, drowsiness, and
ity. Odorous emissions from agricultural waste have alterations in mood (Shusterman, 1992; Schiffman
raised particular concerns about human health with the et al., 1995; Thu et al., 1997; Schiffman, 1998;
rapid proliferation over the last decade of concentrated Steinheider et al., 1998; Schiffman et al., 2000; Wing
animal feeding operations (CAFOs) (Esteban, 1997). and Wolf, 2000). Typically, these symptoms begin
Odors from animal operations are exempted from reg- upon initial exposure to the odor and lessen after the
ulation, to a large degree, due to the fact that they are odor departs. However, sensitive individuals, such as
covered by agricultural rather than industrial guide- those with asthma and allergies, indicate that health
lines. CAFOs house thousands of animals in a small symptoms induced by odorous air persist for long
area which concentrates potential problems associated periods of time and aggravate their chronic medical
with waste management including odor pollution from conditions.
gaseous emissions into the ambient air as well as pollu- A report from a recent workshop on potential health
tion of surface and ground waters with excess nutrients effects of odors sponsored by the US Environmental
(e.g. nitrogen and phosphorus) and pathogens (GAO, Protection Agency (USEPA) and the National Institute
1995; Esteban, 1997; United States Senate, 1997). on Deafness and Communication Disorders (NIDCD)
The amount of animal waste including manure, urine, concluded that there are three potential ways in which
and carcasses produced by these facilities is enormous odorous emissions from manures and biosolids may
with 130 times more animal waste than human waste produce health symptoms (Schiffman et al., 2000). In
currently produced in the US (United States Senate, the first paradigm, health symptoms are induced by the
1997). In 1997 alone, estimated animal manure pro- irritant properties of the odorants (i.e. the compounds
duction in the US was 1.4 billion tonnes; much of this that induce odor). For a broad range of molecules (or
was stored in open storage structures called “lagoons”. mixture of molecules), irritancy occurs at a concen-
Odor complaints have risen dramatically with the tration somewhat higher (about 3–10 times higher)
increase in the number of CAFOs. Odor sources than the concentration at which odor is first detected
include animal confinement houses, the anaerobic la- (odor threshold). While the concentration of each
goons, and spray fields that are used as the primary individual compound identified in odorous air from
method of waste disposal. The odors are generated agricultural facilities seldom exceeds the concentra-
from microbial degradation of biosolids including tion known to cause irritation, the combined load
manure and feed. Waste from 25,000 animals or more of the mixture of odorants can exceed the irritation
is dumped into the open manure storage pit (lagoon). threshold (Cometto-Muñiz et al., 1997, 1999; Korpi
Anaerobic processing of waste in the lagoons leads et al., 1999). Thus, the irritation induced by the odor-
to the production of many malodorous products in- ous mixture derives from the addition (and sometimes
cluding hydrogen sulfide, acids, ammonia, phenols, synergism) of individual component compounds.
cresol, and indoles (Zahn et al., 1997). Setbacks from In the second paradigm, health symptoms occur at
confinement buildings and lagoons generally range odorant concentrations that are above odor detection
from 750 to 1500 ft on average, and do not take into thresholds but far below the levels that cause irrita-
account the number of animals at the location (NC tion. This typically occurs with odorant classes such
House, 1977). Furthermore, some animal facilities as sulfur-containing compounds and organic amines.
have been sited near dense population areas including The physiological basis by which sulfur gases or
housing developments or shopping centers. organic amines induce health symptoms when odor
Complaints directed to confined animal operations potency far exceeds the irritant potency is not well
generally emphasize the unpleasant quality or char- understood. However, brain imaging studies suggest
acter of the odor; however, a substantial number of a genetic factor may play a role since noxious odors
complaints also involve reports of adverse health stimulate different brain areas than those that process
symptoms. The health symptoms most frequently pleasant odors (Zald and Pardo, 1997).
attributed to odors include eye, nose, and throat irri- In the third paradigm, the odorant is a component
tation, headache, nausea, diarrhea, hoarseness, sore of a mixture that contains a co-pollutant that initially
S.S. Schiffman et al. / Agricultural and Forest Meteorology 108 (2001) 213–240 215

causes the reported health symptom. Odorous airborne of compounds in producing odor and irritation down-
emissions from confined animal housing can contain wind of CAFOs. Dispersion models need to be
other components such as endotoxin and organic dust developed that predict accurately the transport and
that may induce the symptoms. If a person’s initial diffusion of odorous compounds. In addition, further
exposure to odors from swine facilities occurs in the studies must performed to determine the full range
presence of dust or gram-negative endotoxin, he or of health effects associated with specific levels of
she associates the odor with the health effects from odorous emissions from CAFOs.
the co-pollutant. Subsequent exposure to the odor in
the absence of the co-pollutant can then produce the
symptoms via Pavlovian conditioning (e.g. Russell 2. Materials and methods
et al., 1984). Odor associations are readily established
and are difficult to extinguish (Siegel and Kreutzer, 2.1. General overview of methods
1997; Siegel, 1999).
Quantification of odor emissions from CAFOs is Two types of experiments were performed to quan-
necessary in order to determine setback distances and tify the odorants (chemicals) and odors (sensations)
zoning ordinances to reduce exposure of neighbors associated with emissions from swine facilities located
to unpleasant odors with their potential health im- in North Carolina. Odorants were characterized using
pacts. The purpose of this study was to collect and gas chromatography and mass spectrometry (GC/MS),
analyze two types of data including: (1) chemical gas chromatography/flame ionization detection
identification of compounds found in emissions from (GC/FID), and gas chromatography/flame photomet-
North Carolina swine houses (and their concentra- ric detection (GC/FPD). Detectors vary in their sen-
tions) and (2) human assessments of the odors and sitivities and selectivity (Elliott et al., 1978). Human
irritation associated with individual compounds and psychophysical studies were performed to determine
the emissions as a whole. No comprehensive stud- odor intensity, irritation intensity, and odor quality of
ies have been performed previously in the southeast gaseous emissions at swine facilities and downwind.
to measure the concentration of volatile compounds
and perceived odor levels from swine facilities. The
volatile compounds were identified and quantified 2.2. Chemical analysis
by analytical procedures including gas chromatog-
raphy/mass spectrometry. Objective measures of the Volatile organic compounds (VOCs) in air and
odor sensations were obtained using olfactometry, lagoon water at swine operations in North Carolina
which is a measurement technique that uses the hu- were collected and analyzed in order to identify the
man nose as the sensor for odors. Odor and irritation compounds present in odorous samples. The VOCs
thresholds as well as descriptive measurements and from the swine houses were preconcentrated by sorp-
hedonic tone (degree of pleasantness or unpleasant- tion onto two types of adsorbent materials: Tenax®
ness) were obtained. The results were compared to and deodorized cotton. Tenax® is a 2,6-diphenyl-p-
previous studies in other geographical regions. phenylene oxide polymer that has been used exten-
The following points emerged from this study. A sively to adsorb VOCs from air samples (Elliott et al.,
great variety of compounds is found in air (and la- 1978; Krost et al., 1982). Textile fibers can also ad-
goon water) in the vicinity of North Carolina swine sorb odorants and serve as a sink for odorants (Travis
facilities, and the vast majority of these compounds and Elliott, 1977; Miner and Licht, 1980). Pretesting
are present at concentrations below published odor using canisters indicated that a sorbent-based col-
and irritation thresholds. Yet, human assessments in- lection approach was necessary to obtain masses of
dicate that odors and sensations of irritation in the VOCs sufficient for quantitative study. Extracts of
vicinity of the swine houses (and even at distances of VOCs from waste water lagoons were also analyzed
∼1500 ft downwind) can be quite strong. Comparison by GC/MS.
of the chemical and human assessment results points Eleven samples (six air samples from six differ-
to the importance of the cumulative effects of hundred ent swine houses concentrated onto the adsorbent
216 S.S. Schiffman et al. / Agricultural and Forest Meteorology 108 (2001) 213–240

Tenax® , two air samples from two different swine of 15 compounds and the total volume of air sampled.
houses concentrated onto cotton swatches, and three Compounds evaluated to generate response factors
lagoon sample extracts) were analyzed by GC/MS to were thiophene, acetic acid, 2-pentanone, methyl
identify the constituent compounds. Four additional disulfide, propanoic acid, 1-pentanethiol, n-butanoic
Tenax® samples from four houses were analyzed by acid, 2-heptanone, methyl sulfoxide, n-pentanoic
gas chromatography with the effluent split 50:50 be- acid, methyl sulfone, 2-nonanone, 1-nonanethiol, and
tween a flame ionization detector and a flame photo- n-nonanoic acid and were chosen to represent the di-
metric detector (GC/FID/FPD) to compare the peaks versity of chemicals known to be present in the air. No
generated by the two types of detectors. Another four corrections were made for either breakthrough during
Tenax® samples from 4 houses were analyzed by gas sample collection or recovery during thermal desorp-
chromatography with the effluent split 50:50 between tion. Thus, the values calculated represent minimum
a flame ionization detector and an olfactory detector concentrations.
(GC/FID/OLF) to characterize the odor quality as- Gas chromatography with sulfur-specific (FPD)
sociated with each peak. One additional air sample detection was performed on four additional Tenax®
(concentrated onto Tenax® ) from a recently cleaned samples from four different swine facilities to inves-
swine house (floors, walls, ceiling, and fans were tigate the presence of trace-level sulfur-containing
power-washed 1 week prior to testing) was used to compounds from air samples. FPD was necessary
quantify the concentration of VOCs in the ambient because the levels of sulfur-containing compounds
air. This sample was selected for quantification to in the Tenax® and lagoon samples were below that
determine levels associated with best management sufficient to produce a useful mass spectrum (ap-
practices. proximately 5–10 ng). The effluent of the GC column
was split 50:50 between the FID and the FPD, and
the peaks generated by the two different types of
2.2.1. Collection and analysis of Tenax® samples
detectors were compared.
Volatile organic compounds from air (six air sam-
ples from six different swine houses) were collected
onto cartridges packed with Tenax® as described by 2.2.2. Collection and analysis of fabric samples
Krost et al. (1982) and analyzed by GC/MS. A total Two air samples from two different swine houses
volume of approximately 320 l was sampled. Tenax® concentrated onto cotton swatches were also analyzed
cartridges were thermally desorbed at 270◦ C and cryo- by GC/MS to identify the constituent compounds.
genically focused as described by Krost et al. (1982). In the fabric studies, the cotton was cut into squares
Prior to chemical characterization studies, VOCs and extracted with methylene chloride for 48 h using
adsorbed onto Tenax® were desorbed from the cryo- a Soxhlet extraction device to remove any odorants
genic trap of a Fisons Instruments Model 8360 gas associated with the fabric itself. The cotton was then
chromatograph (Fisons Instruments, Danbury, MA) placed in a nitrogen box and purged overnight to re-
into a 100 ml gas-tight syringe. The odor of the syringe move gross amounts of residual solvent. It was then
contents was evaluated by four trained odor panelists placed in a vacuum oven at 50◦ C overnight prior to
for consistency with the odor at the swine facilities. It being packed into a glass tube of the same size as that
was determined that the nature of the odor was largely used for the Tenax® cartridges. The bed length was
preserved. This indicated that the odorants were stable 4.5–5.0 cm. A GilAir-5 (Gilian Instrument Corp., W.
to the adsorption/thermal desorption process. Caldwell, NJ) constant flow pump was used to pull
Mass spectral analysis to aid in the character- air through the tubes at flow rates up to 900 ml/min.
ization of separated constituents was performed Samples were collected at several volumes up to 320 l.
using a Hewlett-Packard GC/MS (Model 5988-A, After collection the cartridges were placed in a freezer.
Hewlett-Packard, Palo Alto, CA) operated in the elec- Due to the high water content found in the cotton mate-
tron impact ionization mode (70 eV). Compounds, rial, the cotton samples were extracted with methylene
identified based on mass spectra, were quantified us- chloride, applied to silanized glass wool from which
ing the average of response factors measured for each the solvent was subsequently removed with a helium
S.S. Schiffman et al. / Agricultural and Forest Meteorology 108 (2001) 213–240 217

flow, and then subjected to thermal desorption. The fans of six swine facilities. These facilities were the
extract was analyzed using GC/MS. The odor from same ones at which Tenax® samples were obtained;
the cryofocused VOCs was determined to be consis- additional samples were also obtained at the recently
tent with swine odor, as described above for Tenax® . cleaned facility from which compounds were also
quantified (see Section 2.2.1). Samples were taken
2.2.3. Extractions from lagoon waste water throughout the day from 6.00 a.m. to 9.00 p.m. Wind
Three lagoon sample extracts from three differ- speed, wind direction, humidity, and ambient tempe-
ent facilities were analyzed by GC/MS to identify rature were monitored with a Weatherlog® Weather
the constituent compounds. Liquid samples from the Monitoring System (Bar Harbor, ME). Samples were
waste lagoons were collected in clean glass bottles collected in 25 l Tedlar® bags with dual stainless steel
such that no headspace remained. The bottles were fittings (from SKC Inc., Eighty-Four, PA). The bags
capped and refrigerated until analysis. One liter of were cleaned prior to use by sequential rinsing with
each liquid sample was extracted twice with methy- 1 l each of purified water, HPLC grade methanol, and
lene chloride. The extracts were then combined, and HPLC grade pentane to remove any potential water
the volume was reduced to 50 ml using a nitrogen or lipid soluble odorants. The residual solvent and
blow-down at room temperature. A 500 ␮l aliquot of odorants were removed by placing the bag in an oven
this extract was loaded onto glass wool and analyzed at 45◦ C with a continuous purge with house nitrogen
as described above for the cotton sample. GC/FPD until no odor was detected (2–3 days). Subsequent ol-
was also used to investigate the presence of trace-level factory evaluations indicated that the bags had no odor.
sulfur-containing compounds from lagoon samples. A clean bag was placed inside a plastic drum with a
removable top that gave an air tight seal when closed.
2.3. Human assessments The drum had two fittings attached to the lid. One
fitting was used to attach the Tedlar® bag to the inside
Odors were characterized at swine facilities and of the drum, and the other fitting was used to control
downwind by human assessors using six different the pressure in the air space in the drum surrounding
psychophysical techniques: (1) odor threshold and the bag. The bag was filled with ambient air by pulling
sensory assessments of air samples 12 ft from the a vacuum in the air space surrounding the bag. The
exhaust fans and downwind; (2) odor assessments at sampling rate was 5 l/min.
varying delivery rates from Tedlar® bags; (3) scen- A dynamic forced-choice triangle olfactometer
tometer ratings of levels in ambient air; (4) compari- (Dravnieks and Prokop, 1975; Dravnieks, 1980) was
son of ambient odor levels with butanol standards; (5) used to evaluate the gaseous samples collected in
irritation assessments of odorous ambient air; and (6) Tedlar® bags. This forced choice procedure is rec-
human assessment of gas chromatographic effluent. ommended by the American Society for Testing and
All subjects in each of the six human assessments Materials (ASTM, 1991). The bag containing the air
were pretested to assure that they had normal smell sample was placed into a metal drum that was then
thresholds to butanol using the method described by pressurized with a one-third horsepower pump. The
Stevens and Dadarwala (1993). pump pressurized the tank to force air out of the bag
through a pressure regulator that was used to adjust
2.3.1. Odor thresholds and sensory assessment of the flow rate. Once the odorous sample left the drum,
odors collected in Tedlar® bags it was delivered to the olfactometer, which diluted the
The purpose of the odor threshold studies was to sample. The rate of flow was 3 l/min. The odor sam-
determine how many times the odorous source needed ple was diluted by a factor of three at each of the six
to be diluted to reach threshold (D/T or dilutions to dilution stations using deodorized, dehumidified air.
threshold). Ten trained panelists ranging in age from The number of dilutions to threshold (D/T) at each of
26 to 47 years (mean = 31±2.58) participated in these the six stations was: 3, 9, 27, 81, 243 and 729 D/T,
studies. Outdoor air samples were taken 12 ft (3.7 m), respectively. Each of the sniffing stations has three
750 ft (229 m) and 1500 ft (457 m) from the exhaust ports; one port emits diluted odor, and the other two
218 S.S. Schiffman et al. / Agricultural and Forest Meteorology 108 (2001) 213–240

emit odorless air. The panelists were instructed by the with samples from bags filled entirely with deodorized
experimenter to sniff all three ports at each station, (filtered) air (see Fig. 1). Subjects sniffed sequentially
starting with the station with the most diluted sample, from a pair of bags (one containing the odor sample
and to select the port with the strongest odor (forced and the other containing clean air) and were asked
choice method). which of the two bags smelled stronger. They rated
Assessments of odorous air in the field were also the stronger sample for odor and irritation intensity
compared with laboratory assessments of samples ob- as well as pleasantness. The samples were delivered
tained in Tedlar® bags. On 2 days, panelists rated odor at eight different delivery rates: 1, 2, 3, 4, 5, 10, 15
intensity on site at 12 ft from the exhaust fans; air sam- and 20 l/min. In another trial, an undiluted sample ob-
ples were collected simultaneously in Tedlar® bags. tained from 1 m above the lagoon was also evaluated
Subjects wore charcoal masks in the field except dur- at the same delivery rates. The subjects were the same
ing odor evaluations. Three hours later, panelists eval- panelists as described in Section 2.3.1.
uated the air samples in the Tedlar® bags delivered at
full strength at 10 l/min from a metal drum in the labo- 2.3.3. Scentometer ratings
ratory. The samples were rated on three nine-point line The Scentometer (Barnebey and Sutcliffe, Colum-
scales numbered from 0–8 (Schiffman and Williams, bus, OH) is a device used in the field to measure
1999). These included odor intensity, irritation in- the intensity of ambient odors in threshold units D/T
tensity, and hedonic ratings. For odor and irritation (number of dilutions to threshold). It is a hand-held
intensity, the scale was labeled as follows: 0 = none plastic box with two nasal ports on one end and
at all; 1 = very weak; 2 = weak; 3 = moderate weak; six inlets for the odorous air (inlet sizes of 1/32 in.
4 = moderate; 5 = moderate strong; 6 = strong; (0.08 cm), 1/16 in. (0.16 cm), 1/8 in. (0.32 cm), 3/16 in.
7 = very strong; and 8 = maximal. The descriptors (0.48 cm), 1/4 in. (0.64 cm) and 1/2 in. (1.27 cm)) at
for pleasantness/unpleasantness were: 0 = extremely the other end which are connected to a mixing cham-
pleasant; 1 = very pleasant; 2 = moderately pleas- ber. These inlets correspond to 350, 170, 31, 15, 7
ant; 3 = slightly pleasant; 4 = neither pleasant nor and 2 D/T, respectively. Two additional inlets on the
unpleasant; 5 = slightly unpleasant; 6 = moderately top and bottom allow air to be pulled through an acti-
unpleasant; 7 = very unpleasant; and 8 = extremely vated carbon cartridge into the mixing chamber. This
unpleasant. Subjects completed an additional odor provides a clean air source that dilutes the odorous air
character evaluation form that contained 146 odor in the mixing chamber. When a person sniffs, air is
descriptive adjective scales developed by the Amer- pulled simultaneously into the mixing chamber both
ican Society for Testing and Materials (see ASTM, from the odorous air inlets and the clean air source.
1984, 1992). Odor panelists at Duke University have D/T values are measured with all but one of the six
been trained to detect and evaluate each of the 146 odorous air inlets closed, and the size of the open
odor qualities. Panelists selected any of the terms that inlet determines the dilution of the odorous air.
applied to the odor and rated these descriptors on a Four subjects (ranging in age from 18–30 years)
scale from 0 (absent) to 5 (extremely). used the Scentometer to determine odor thresholds
at 12 ft (3.66 m) from the fans and at distances from
750 ft (229 m) to 1250 ft (381 m) at three different
2.3.2. Odor assessments at varying delivery rates swine facilities. The subjects wore carbon masks be-
from Tedlar® bags fore and between trials and were blindfolded during
The second method of testing air samples involved testing to eliminate visual cues. Dilution to thresh-
diluting the air collected 12 ft (3.7 m) from the exhaust old (D/T) assessments were made as follows. Subjects
fans with charcoal-filtered air. The dilutions were pre- were asked to sniff from the Scentometer twice. For
pared directly in the Tedlar® bags themselves. Bags one sniff, the experimenter closed all of the odorous
were partially filled with filtered air prior to sampling air inlets (subject sniffed clean air). For the other sniff,
to obtain dilutions of 1% swine odor, 25% swine odor, the experimenter opened one odorous air inlet. The
50% swine odor, 75% swine odor, and 100% swine subjects were asked which sample was stronger. The
odor. The odor samples from the bags were compared order of presentation of the odorous sample in a pair
S.S. Schiffman et al. / Agricultural and Forest Meteorology 108 (2001) 213–240 219

Fig. 1. Diagram of apparatus used for testing the effect of flow rate on odor judgments.

was randomized across subjects. The odorous air in- series of dilutions of 1-butanol in water. Twelve serial
lets were opened from smallest to largest in a stepwise dilutions of 1-butanol were used starting at 10 ppm
paradigm with repeated presentations as described by butanol with a geometric progression of two: 10 to
Wetherill and Levitt (1965) for determining thresh- 20,480 ppm (10, 20, 40 up to 20,480 ppm). This is a
olds. If a threshold was found with the 1/16 in. inlet standard method (E544-75) established by The Amer-
open, the concentration of the odor was considered to ican Society for Testing and Materials (ASTM, 1997).
be 170 times above threshold. A trained panel also rated these butanol concentrations
for their odor intensity using the nine-point line scale
2.3.4. Comparison to 1-butanol standards described above.
Five subjects aged 41–55 years living an average
of 1520 ± 44.7 ft (463 m) downwind from five swine 2.3.5. Irritation assessments of odorous ambient air
facilities were asked to match the odor intensity level Irritation, unlike olfaction, can be localized to one
of ambient air outside their homes to one of a series nostril or the other (Schneider and Schmidt, 1967;
of bottles containing odor standards. These subjects, Kobal et al., 1989; Roscher et al., 1996). That is,
like those in all of the human studies, were pretested when an odorous stimulus that is not an irritant is
to assure that they had normal smell thresholds to bu- presented to one nostril and a nonodorous stimulus
tanol (see Stevens and Dadarwala, 1993 for methods). is presented to the other nostril, the subject cannot
Subjects were asked to perform the comparison twice determine which nostril received the odor. However,
a day (6.30 a.m. and 7.30 p.m.) for two weeks during if the odorous stimulus is also an irritant, the subject
the summer. Subjects were phoned to remind them to can identify the nostril that received the stimulus.
go outside and sniff the air. The bottles contained a Four subjects ranging in age from 50 to 60 years
220 S.S. Schiffman et al. / Agricultural and Forest Meteorology 108 (2001) 213–240

who lived 1400 ft (427 m) downwind from a swine should be emphasized that many GC peaks were too
facility were presented with odorous ambient air small to allow identification, and these may contribute
to one nostril and charcoal filtered air to the other significantly to the odor and irritation of the samples.
nostril at 4 l/min. Subjects were blindfolded to elim- In column 1 of Table 1, the compound and formula
inate visual cues and any potential ocular irritation. are given. The Chemical Abstracts Service (CAS) reg-
Each subject was presented with five identical pairs istry number is given in column 2. Columns 3, 4 and
of stimuli (odorous ambient air and odorless air) 5, labeled L (lagoon sample), C (cotton sample), and
with at least 3 min between pairs. The nostril that T (Tenax® sample), respectively, list the frequency of
received the ambient air was randomized over trials. occurrence of each compound in the three lagoon sam-
Subjects were asked which nostril received the odor. ples, two cotton samples, and six Tenax® samples,
The five pairs of stimuli were repeated three times respectively. For example, acetic acid in Table 1(a)
with 40 min between sets. Testing began at 7.30 p.m. was found in all three of the lagoon extracts, both of
when the odor of swine waste was perceived in the
the cotton extracts, and all six of the Tenax® samples.
ambient air by the experimenters. Subjects remained
Column 6 is labeled “Y” if the compound has been
indoors until the outdoor tests were performed in
found at swine facilities outside of North Carolina
order to prevent adaptation to the odor.
(see Hammond and Smith, 1981; Hammond et al.,
2.3.6. Comparison of analytic results and human 1989; Ritter, 1989; O’Neill and Phillips, 1992).
assessments Column 7 indicates the concentration of each VOC
Additional Tenax® samples were analyzed by ther- found in the recently cleaned swine house in North
mal desorption (TD)/gas chromatography (GC) with Carolina as determined by GC/MS expressed in
the effluent split 50:50 between a flame ionization bothparts per billion with milligrams per cubic meters.
detector (FID) and an olfactory detector (SGE Inc., Column 8 gives the mean odor threshold in parts per
Austin, TX). Three subjects trained in odor assessment million with the values expressed in milligrams per cu-
evaluated the odor that corresponded to the peaks in bic meters in parentheses. Values with footnote “e” are
terms of their intensity, character, and hedonic tone standardized thresholds from Devos et al. (1990) and
(Schiffman and Williams, 1999). The intensity of are based on multiple studies. Devos et al. (1990) cal-
the odor was rated on a six point scale (0–5) where culated standardized threshold values using a weighted
“0” indicated the absence of odor and “5” indicated average of thresholds from multiple studies. Odor
extremely intense odor. thresholds with footnotes “f” and “g” are from Ruth
(1986) and Van Gemert and Nettenbreijer (1977), re-
spectively; values with footnote “h” are from Billings
3. Results and Jonas (1981). When odor thresholds greater than
one order of magnitude lower than the standardized
3.1. Identity and concentration of VOCs in swine threshold of Devos et al. (1990) were found, they are
houses and lagoons listed in square brackets in column 8. Column 9 gives
the irritation threshold reported by Ruth (1986). The
The compounds from the Tenax® , cotton, and la- odor and other sensory characteristics commonly re-
goon samples were identified by interpretation of the ported in the Merck Index (1996) and a chemical cat-
mass spectra. The compounds were diverse, and the alog (Aldrich Catalog, 2000) are given in column 10.
samples contained many acids, alcohols, aldehydes, An overview of Table 1 indicates that 411 com-
amides, amines, aromatics, esters, ethers, fixed gases, pounds have been associated with swine facilities.
halogenated hydrocarbons, hydrocarbons, ketones, ni- A total of 331 different VOCs and fixed gases were
triles, other nitrogen-containing compounds, phenols, found at swine facilities in North Carolina. Of these,
sulfur-containing compounds, steroids, and other 203 were found in the Tenax® samples, 112 were
compounds (see Table 1). Table 1 give a complete list found on the cotton samples, and 167 different com-
of VOCs identified by GC/MS in this study along with pounds were identified in the lagoon samples. In
data from other studies for comparison. However, it Table 1, the 157 compounds listed were described as
S.S. Schiffman et al. / Agricultural and Forest Meteorology 108 (2001) 213–240 221
222 S.S. Schiffman et al. / Agricultural and Forest Meteorology 108 (2001) 213–240
S.S. Schiffman et al. / Agricultural and Forest Meteorology 108 (2001) 213–240 223
224 S.S. Schiffman et al. / Agricultural and Forest Meteorology 108 (2001) 213–240
S.S. Schiffman et al. / Agricultural and Forest Meteorology 108 (2001) 213–240 225
226 S.S. Schiffman et al. / Agricultural and Forest Meteorology 108 (2001) 213–240
S.S. Schiffman et al. / Agricultural and Forest Meteorology 108 (2001) 213–240 227
228 S.S. Schiffman et al. / Agricultural and Forest Meteorology 108 (2001) 213–240
S.S. Schiffman et al. / Agricultural and Forest Meteorology 108 (2001) 213–240 229
230 S.S. Schiffman et al. / Agricultural and Forest Meteorology 108 (2001) 213–240
S.S. Schiffman et al. / Agricultural and Forest Meteorology 108 (2001) 213–240 231

an “irritant” in the Merck Index (1996) or in the chem- cotton extract. This is likely the result of losses of the
ical supply literature (Aldrich Catalog, 2000). The more volatile VOCs from the cotton extract during
eleven compounds found at the highest concentrations solvent reduction and solvent removal prior to GC
in the sample from the recently cleaned North Car- analysis. Peak identifications of the RIC obtained af-
olina swine facility were: butanoic acid, acetic acid, ter GC/MS analysis of the liquid extract also showed
3-methylbutanoic acid, 4-methylphenol, propanoic many of the same compounds as the Tenax® sample
acid, 2-methylpropanoic acid, 2-methylbutanoic acid, and the cotton extract.
vinyl acetate, 4-ethylphenol, phenol, and acetalde- Fig. 2 shows a comparison of the FID output to the
hyde. The total concentration for all of the measured FPD output for the separation of VOCs from a Tenax®
VOCs in this sample was 0.602 mg/m3 . sample when the effluent of the GC column was split
Table 1 reveals a substantial overlap in many of 50:50 between the FID and the FPD. Many sulfur-
the compounds detected from the Tenax® and the containing compounds were detected, especially in
cotton extract. However, the reconstruction ion chro- the time frame from 0 to 23 min. Pretesting with stan-
matogram (RIC) after GC/MS analysis of the Tenax® dard sulfur-containing compounds (i.e. methylethyl
showed more of the early-eluting VOCs than did the sulfide, thiophene, and tetrahydrothiophene) indicated

Fig. 2. Comparison of GC/FID to GC/FPD analysis of VOCs from a Tenax® sample. Many sulfur-containing compounds were detected,
especially in the time frame from 0 to 23 min.
232 S.S. Schiffman et al. / Agricultural and Forest Meteorology 108 (2001) 213–240

that FPD was more sensitive and selective for strong) and in the laboratory, 3.2 (moderate weak).
sulfur-containing compounds in this time interval than The mean ratings on 11 adjective scales: sour/acid,
the FID. A large number of sulfur-containing com- like ammonia, sickening, sulfidic, putrid/foul, animal,
pounds were detected, and many of those compounds fecal (like manure), urine-like, cheesy, sharp/pungent,
gave very small or non-detected peaks in the FID chro- and sewer odor, were also significantly higher on site
matogram or co-eluted with other compounds. Each than in the laboratory.
of the four Tenax® samples gave similar GC/FPD
chromatograms, which suggests that many of the 3.2.2. Odor assessments at varying delivery rates
same sulfur compounds were found at each farm. from Tedlar® bags
Fig. 3a–e show the average odor intensity, irritation
intensity, and unpleasantness ratings at differing de-
3.2. Results from human assessments
livery rates (1–20 l/min) for five dilutions of air 12 ft
(3.7 m) from a swine house exhaust fan (odors were
3.2.1. Odor thresholds and sensory assessment of diluted with clean air in the bag). These figures indi-
odors collected in Tedlar® bags cate that the rate of delivery of the odor influences the
The average detection thresholds determined in the magnitude of the intensity and unpleasantness ratings
laboratory for 86 samples obtained 12 ft (3.7 m) from for the most concentrated odors. With no dilution
the fans in Tedlar® bags at the six facilities was 221± (100% odorous air), the average odor intensity rating
26.7 times threshold (i.e. 221 D/T). The mean thresh- was less than 2 (weak) at a delivery rate of 1 l/min.
old at the fans of the recently cleaned swine house was However, when the delivery rate was 20 l/min, the
numerically but not statistically lower than the mean odor intensity for the same sample was rated over 6
of the other samples. There were no trends in threshold (strong). The same trend was seen for irritation inten-
values for air samples taken 12 ft from the fans with sity which varied from an average rating of 1 (very
respect to variation in ambient temperature (0–36◦ C weak) at a delivery rate of 1 l/min to over 3 (mod-
in the field), humidity (up to 90% in the field), wind erate weak) at a delivery rate of 20 l/min. This trend
direction, or wind speed. Thresholds were often diffi- was minimal for highly diluted odor. The ratings of
cult to obtain from samples in Tedlar® bags at 750 ft unpleasantness showed little variation among the dif-
(229 m) and 1500 ft (457 m) in spite of the fact that ferent dilutions, with the exception of undiluted swine
experimenters detected odor at each of the sampling house odor, which was rated more unpleasant than
locations. Only 9.7% of the 93 bags obtained at these diluted odor.
distances contained odorous air that could be detected The delivery rate also affects judgments of the
by subjects using the olfactometer for which the initial undiluted air samples collected above the lagoon as
dilution is represents a D/T = 3. Most of the samples shown in Fig. 3f. The odor intensity increased pro-
at 750 and 1500 ft for which olfactory threshold values gressively with increasing delivery rate from 1 to
could be determined were collected in early morning 20 l/min. The irritation intensity increased at 5 l/min
and late evening with wind speeds less than 1 mile/h and remained elevated up to 20 l/min. The delivery
(little mixing). The mean threshold for samples that rate had little effect on the unpleasantness ratings.
could be detected by panelists at 750 ft (229 m) was
10.1 D/T ± 10.6; the mean threshold for detectable 3.2.3. Scentometer ratings
samples at 1500 ft (457 m) was 9.0 ± 10.1. Scentometer measurements made 12 ft from the ex-
On the 2 days when panelists rated odor intensity on haust fans ranged from 170 to 350 D/T. When an odor
site at 12 ft from the exhaust fans and in the lab from was detected at 750 ft (229 m) to 1250 ft (381 m) from
Tedlar® bags (collected at the time as on site ratings), the swine facilities, scentometer readings ranged from
the mean intensities differed significantly on odor and as high as 170 to as low as 2 D/T. The percentage of the
irritation intensity but not unpleasantness. The mean 78 readings found for 170, 31, 15, 7 and 2 D/T were
odor intensity ratings on site were 7.1 (very strong) 10.3, 25.6, 26.9, 34.6 and 2.6%, respectively. At one
and in the laboratory, 5.1 (moderate strong). The mean facility, the odor level at a distance of 1250 ft (381 m)
irritation intensity ratings on site were 5.0 (moderate corresponded to 170 D/T for all four subjects.
S.S. Schiffman et al. / Agricultural and Forest Meteorology 108 (2001) 213–240 233

3.2.4. Comparison to 1-butanol standards (i.e. subjects made ratings 78.6% of the time at the
The results from comparisons of the intensity of specified times requested during the 2 week test pe-
ambient air at 6.30 a.m. and 7.30 p.m. with butanol riod). The data indicate that odors were perceived
standards are shown in Fig. 4. Compliance was 78.6% 58.3% of the time. The mean intensity ratings on the

Fig. 3. Odor intensity, irritation intensity, and unpleasantness of (a) 1% swine house exhaust; (b) 25% swine house exhaust; (c) 50% swine
house odor; (d) 75% swine house odor; (e) 100% swine house odor; and (f) undiluted air collected above lagoon delivered at 1 to 20 l/min
using the apparatus shown in Fig. 1. For odor and irritation intensity, the scale was labeled as follows: 0 = none at all; 1 = very weak;
2 = weak; 3 = moderate weak; 4 = moderate; 5 = moderate strong; 6 = strong; 7 = very strong; and 8 = maximal. The descriptors for
pleasantness/unpleasantness were: 0 = extremely pleasant; 1 = very pleasant; 2 = moderately pleasant; 3 = slightly pleasant; 4 = neither
pleasant nor unpleasant; 5 = slightly unpleasant; 6 = moderately unpleasant; 7 = very unpleasant; and 8 = extremely unpleasant.
234 S.S. Schiffman et al. / Agricultural and Forest Meteorology 108 (2001) 213–240

Fig. 3 (Continued).

nine-point scale given to the butanol standards by 3.2.5. Irritation assessments of odorous ambient air
the trained panel are given in Table 2 for compari- All four subjects could localize the odor of swine
son. The butanol level most frequently assigned to waste to the nostril that received the ambient air on
the ambient odor near swine facilities was level 12 each of the trials. This indicates that the compounds
(20,480 ppm) which was assigned an intensity rating in the ambient air activated the trigeminal nerve and
of 5.1 (moderate strong) by a trained panel. hence were a sensory irritant.

3.2.6. Olfactory descriptors assigned to components


of swine odor
Results from analysis of four additional Tenax®
samples split 50:50 between the FID and the odor port

Table 2
Intensity ratings by a trained panel of butanol on a nine-point scale
Butanol Butanol concen- Mean rating by
level tration (ppm) a trained panel
1 10 0.8 ± 0.2
2 20 1.1 ± 0.2
3 40 1.3 ± 0.3
4 80 1.4 ± 0.3
5 160 1.9 ± 0.3
6 320 2.2 ± 0.3
7 640 2.6 ± 0.4
Fig. 4. Butanol levels assigned to the perceived intensity of am- 8 1280 2.9 ± 0.4
bient air near swine facilities at 6.30 a.m. and 7.30 p.m. by 9 2560 3.2 ± 0.4
five subjects. The levels correspond to butanol concentrations 10 5120 4.0 ± 0.5
as follows: 1 (10 ppm), 2 (20 ppm), 3 (40 ppm), 4 (80 ppm), 5 11 10240 4.4 ± 0.6
(160 ppm), 6 (320 ppm), 7 (640 ppm), 8 (1280 ppm), 9 (2560 ppm), 12 20480 5.1 ± 0.6
10 (5120 ppm), 11 (10240 ppm), 12 (20,480 ppm).
S.S. Schiffman et al. / Agricultural and Forest Meteorology 108 (2001) 213–240 235

Fig. 5. GC/FID/OLF analysis. Odor descriptors assigned by trained panelists to peaks from a Tenax® sample collected 12 ft from swine
house exhaust fans. The descriptors and intensity judgments (from 0–5 where “0” indicates no odor and “5” indicates extremely intense
odor) are given on the abscissa.

indicated that most of the peaks had unpleasant odors. that many of the same VOCs are present in all of the
An example is given in Fig. 5. The descriptors and samples. Some olfactory descriptors were assigned
intensity judgments (from 0 to 5 where “0” indicates to small peaks in the chromatogram. This indicates
no odor and “5” indicates extremely intense odor) are that intense components of swine odor can be present
given on the abscissa. The chromatograms obtained at very low concentrations in the air. Such com-
from the four samples were similar which suggests pounds would be expected to have very low olfactory
236 S.S. Schiffman et al. / Agricultural and Forest Meteorology 108 (2001) 213–240

thresholds. Compounds present in such small amounts The intensity of the odorous emissions from swine
would not provide mass spectra of sufficient intensity facilities must result not only from individual com-
to permit identification. This problem of trace-level, pounds that are present at concentrations above their
highly odorous components is confounded by the odor thresholds, but from the aggregate effect of
complexity of the chromatograms, especially in the hundreds of odorous chemicals present at subthresh-
region of 30–50 min where the fatty acids elute. old concentrations (Cometto-Muñiz et al., 1997,
1999). There is an additive or synergistic effect by
which low levels (even below known odor or irritant
4. Discussion threshold concentrations) of hundreds of component
compounds lead to odor and even irritant sensations
The main finding of this study is that there are (Cometto-Muñiz et al., 1997, 1999; Schiffman, 1999).
hundreds of compounds associated with swine odor The compounds found in swine houses for which the
in North Carolina, and the vast majority of these concentrations exceeded the standardized odor thresh-
compounds are below published odor and irritation olds reported by Devos et al. (1990) were n-butanoic
thresholds. However, human assessments of emissions acid, isovaleric acid, 2-methylbutanoic acid, indole,
from swine facilities indicate that the odor sensations and p-cresol. The 10 compounds found in emis-
and nasal irritation can be strong. Comparison of the sions from swine facilities with the lowest known
chemical and human assessment results points to the odor thresholds were indole (0.000155 mg/m3 ),
importance of the cumulative load of hundreds of 2,4-decadienal (0.000219 mg/m3 ), 1-pentanethiol
compounds in producing odor and irritation down- (0.000525 mg/m3 ), 2-nonenal (0.000871 mg/m3 ), allyl
wind of CAFOs. Particulates as well as the total mercaptan (0.00126 mg/m3 ), thiophenol (0.00145 mg/
aggregate of VOCs probably contribute to sensory m3 ), methanethiol (0.00209 mg/m3 ), diethyldisulfide
irritation produced by emissions from swine facilities. (0.00219 mg/m3 ), ethanethiol (0.00282 mg/m3 ), and
skatole (0.00309 mg/m3 ) (Devos et al., 1990). Of
4.1. Analytic findings these 10 compounds with the lowest odor thresholds,
six contained sulfur, two were nitrogen-containing,
The results of this study show that over 300 com- and two were aldehydes. The compounds with the
pounds are found in odorous air emissions from lowest odor thresholds also have unpleasant odor
swine facilities in North Carolina. Many more com- qualities. Both indole and skatole have characteristic
pounds are present, but the GC peaks were too small fecal odors. In addition, 1-pentanethiol, allyl mer-
to allow identification. The compounds identified by captan, thiophenol, and ethanethiol all have “stench”
GC/MS from North Carolina swine facilities given in listed as their characteristic odor (Merck Index, 1996;
Table 1 are consistent with those found in other stud- Aldrich Catalog, 2000). Thiophenol is also described
ies (Hammond et al., 1979, 1989; Spoelstra, 1980; as putrid and nauseating.
Hammond and Smith, 1981; Ritter, 1989; O’Neill The magnitude of total VOCs in the air associated
and Phillips, 1992; Hartung and Phillips, 1994; Zahn with a recently cleaned swine house in North Carolina
et al., 1997). The compounds identified were diverse, (0.602 mg/m3 ) was lower than that found in other
and the samples contained many acids, alcohols, alde- studies. This could be due to the cleaning of the house
hydes, amides, amines, aromatics, esters, ethers, fixed or to factors associated with the chemical analysis. No
gases, halogenated hydrocarbons, hydrocarbons, ke- corrections were made here for either breakthrough
tones, nitriles, other nitrogen-containing compounds, during sample collection or recovery during thermal
phenols, sulfur-containing compounds, steroids, and desorption. The average VOC emissions for 21 gases
other compounds. Acids, phenolic compounds, and from swine houses in Germany was 11.72 mg/m3
aldehydes were present in the highest concentra- (Hartung and Phillips, 1994). Higher total analyte
tions. Some of the aldehydes included acetaldehyde, concentrations were also found by Zahn et al. (1997)
n-butanal, 3-methylbutanal, and n-pentanal which as in air near a slurry storage basin in Central Iowa.
a group are a concern as air pollutants because of The total concentrations for 24 VOCs at 0, 25 and
their reactivity and toxicity (Costa and Amdur, 1996). 100 m from the slurry storage basin were 27.70, 19.44
S.S. Schiffman et al. / Agricultural and Forest Meteorology 108 (2001) 213–240 237

and 10.67 mg/m3 , respectively. Even higher levels of person in the field who directly sniffs the ambient air
VOCs (108.7 mg/m3 ) were reported by Hobbs et al. containing both dust and VOCs in the vapor phase. A
(1997) from the headspace of a chamber containing significant amount of odorous VOCs has been shown
slurries produced by weaner pigs. Even though the to be carried on dust (Hammond et al., 1979). Second,
total magnitude of VOCs found here was lower than it is possible that some compounds may be lost be-
in other studies, it was still sufficient to induce strong tween the time of collection and time of testing; it has
odors. been shown that air samples transported from a swine
In the present study, concentration of VOCs onto unit to the laboratory lose H2 S rapidly due to adsorp-
Tenax® and cotton at the source of the odors (swine tion or decomposition (Elliott et al., 1978). Third,
houses and lagoon) provided adequate levels of these delivery rates of the odor may also play a role in the
compounds for identification and quantification by differences between field and laboratory studies.
GC/MS. This method also allowed collection and The average threshold value found for air samples
analysis of VOCs from both the vapor and particulate collected in Tedlar® bags 12 ft from the exhaust
phases. The collection of particulates is important fans using a forced-choice dynamic olfactometer
because dust particles in swine houses contribute sub- (Dravnieks, 1980) according to ASTM standard
stantially to odor intensity as shown by Hammond E679-91 was 221 D/T. This is consistent with Scen-
et al. (1979). Identification and quantification of odors tometer measurements made 12 ft from the exhaust
downwind of swine facilities are more problematic fans which ranged from 170 to 350 D/T. These D/T
because it is often difficult to capture the VOCs at values probably underestimate odor strength at the
adequate levels from the plume. The collection and fans, because both methods filter dust particles. The
concentration of odorous air onto adsorbents off-site D/T values found here using a delivery rate of 3 l/min
needs further development to provide for quantifica- are approximately one-fourth the magnitude of those
tion when odor is intermittent. reported by Jacobson and Guo (2000) using a dif-
ferent olfactometer with a delivery rate of 20 l/min
4.2. Human assessments of odor (St. Croix Sensory Inc., Stillwater, OK). That is,
the intensity of an odor sample with a D/T of 250
using the lower delivery rate with the Dravnieks ol-
Odor assessments by trained panelists in the field
factometer was equivalent to 1000 D/T using the
and laboratory indicate that odors near the exhaust
higher delivery rate with the St. Croix olfactome-
fans of the swine houses are strong. However, there
ter. Cometto-Muñiz et al. (2000) have shown that
were some differences in the magnitude of strength
while each method is internally consistent, olfactory
between field and laboratory assessments of odor
thresholds determined by different stimulus-delivery
samples from identical locations. The mean intensity
systems can differ by a factor of four or more. These
rating in the field at 12 ft from the exhaust fans was 7.1
data suggest that more research is necessary to de-
(very strong) but in the laboratory, this rating of odor-
termine the optimum method/rate of delivering odors
ous air delivered full strength from Tedlar® bags was
to find threshold values in the laboratory that repro-
5.1 (moderate strong). This result raises some ques-
duce the experience of people on and off-site under a
tions about collection and delivery of odor samples.
variety of environmental conditions.
The possible reasons for the lower intensity ratings Assessments in the field suggest that odor down-
in the laboratory are the following. First, fans from wind from swine facilities has the following charac-
swine houses exhaust both dust and gaseous VOCs. teristics. First, ambient air 1400 ft (427 m) downwind
Organic dust (dried fecal material and feed) adheres from a swine facility that had an odor characteristic
to Tedlar® bags as well as the tubing of the collection of swine emissions was found have irritant proper-
and delivery systems; therefore, only VOCs from the ties. The irritancy may be due to the vaporous VOCs
vapor phase reach the nose of the panelists sniffing from the houses and the lagoon, to the dust from
samples collected in Tedlar® bags. The concentrations the swine houses, the particulates formed over the
of the VOCs sniffed from Tedlar® bags in the labora- lagoon (see Hammond et al., 1979; Schiffman et al.,
tory would thus be lower than those experienced by a 2000), or to a combination of the above. VOCs from
238 S.S. Schiffman et al. / Agricultural and Forest Meteorology 108 (2001) 213–240

trees, grass, and motor vehicle emissions may also offsite (Zhu et al., 2000; Zhu, 1999; Schiffman, 1999).
contribute to the total load of VOCs and particu- There are several possible reasons why accurate dis-
lates in the air shed. Irritants can induce respiratory persion models for odor have not yet been developed.
symptoms in healthy persons and may be of special First, agricultural facilities have multiple sources of
concern for sensitive individuals such as asthmatic pa- odor that are difficult to model for a variety of at-
tients (Schiffman et al., 2000). A second finding from mospheric conditions. Second, surrogate compounds
off-site measurements is that threshold values (D/T) have not been identified for which there is a one-to-one
at a distance of 1250 ft (381 m) can be as high as relationship between compound concentration and
170 D/T. These D/T values obtained from NC swine smell intensity. Third, compounds in an odor plume
facilities are in the same range as those reported by are additive and even synergistic with one another and
Sweeten (1998) from swine facilities in Nebraska and with dust; yet, algorithms for addition and synergism
Iowa. A third finding is that the majority of intensity of odors and/or dust have not been developed. Fourth,
evaluations by residents living 1500 ft from swine odors can be detected for compounds at very low con-
facilities were equivalent to 20,480 ppm 1-butanol centrations (e.g. in the parts per trillion range). Further
in water. Nicolai et al. (2000) have indicated that research is necessary to develop air dispersion models
odors equivalent in magnitude to 750 ppm 1-butanol for agricultural facilities that perform accurately both
in water are “a little annoying”; 2250 ppm 1-butanol, on and offsite, especially beyond 500 ft.
“annoying”; 6750 ppm 1-butanol, “very annoying”; Finally, medical studies must be performed to evalu-
and 20,250 ppm 1-butanol, “extremely annoying”. ate the health effects of swine emissions. At the present
Using this scale, most of the odors recorded by per- time, safety standards (OSHA, NIOSH) for VOCs are
sons downwind in this study (see Fig. 4) were rated based on exposure to a single compound rather than
strong enough to be very or extremely annoying. simultaneous exposure to 300 or more compounds.
Future studies are needed to refine the method-
ological issues of olfactometry to reproduce the ex-
perience of persons exposed to swine odors in the 5. Conclusions
field. These experimental methods can then be em-
ployed to develop dispersion models that estimate A total of 411 compounds have been found in odor-
the impact of odorous emissions from agricultural ous emissions from swine facilities, with 331 VOCs
facilities on neighbors. Dispersion models have been and fixed gases found in North Carolina, and the
proposed and/or utilized in an attempt to predict odor remainder found elsewhere in other studies. Although
levels and setback distances that reduce odor impacts each individual compound was found at low concen-
(Welch et al., 1999; Cavalini, 1994; Van den Hazel trations, the additive and/or synergistic effect of these
and Waegemaekers, 1991; Smith, 1993; Zhu, 1999; hundreds of compounds in aggregate produces strong
Zhu et al., 2000; Leonardos, 1999; Jiang and Sands, odor intensities. Odor levels off-site were as high as
1999; Liu et al., 1995). Computer models and pro- 170 dilutions to threshold (D/T). Emissions down-
grams are currently under development to predict wind from swine facilities are not only odorous but
odor levels and setback distances to reduce these cause nasal irritation as well. The nasal irritation may
odor impacts (Jacobson and Guo, 2000; Petersen and be induced by the vaporous VOCs from the houses
Lavdas, 1986). However, dispersion modeling to date and the lagoon, by the dust from the swine houses, by
has not yet provided predictions that reach a high particulates formed over the lagoon or a combination
enough level of certainty to estimate odor intensity of these. Current methods for sampling odor in the
offsite of animal operations. While concentrations field for laboratory assessment need improvement to
of odorants (chemicals) and odors (sensations) can capture dust as well as vaporous VOCs. Dispersion
be measured at the source using gas chromatogra- models need to be developed that account accurately
phy/mass spectometry and olfactometry, respectively, for odor intensities downwind of animal operations.
dispersion models tested by our laboratory and others Medical studies must also be performed to assess the
(e.g. ISCST3, INPUFF-2) lack the capacity to predict health effects of simultaneous exposure to 300 or more
with adequate certainty whether an odor is present compounds.
S.S. Schiffman et al. / Agricultural and Forest Meteorology 108 (2001) 213–240 239

References Hammond, E.G., Fedler, C., Junk, G., 1979. Identification of


dust-borne odors in swine confinement facilities. T ASAE 22,
Aldrich Catalog, 2000–2001. Aldrich Handbook of Fine Chemicals 1186–1189, 1192.
and Laboratory Equipment. Milwaukee, WI. Hammond, E.G., Heppner, C., Smith, R., 1989. Odors of swine
American Society for Testing and Materials (ASTM), 1984. waste lagoons. Agric. Ecosyst. Environ. 25, 103–110.
Correlation of subjective–objective methods in the study of Hartung, J., Phillips, V.R., 1994. Control of gaseous emissions
odors and taste. STP 440, ASTM, Philadelphia. from livestock buildings and manure stores. J. Agric. Eng. Res.
American Society for Testing and Materials (ASTM), 1991. 57, 173–189.
Hobbs, P.J., Misselbrook, T.H., Pain, B.F., 1997. Characterisation
Standard practice for determination of odor and taste thresholds
of odorous compounds and emissions from slurries produced
by a forced-choice ascending concentration series method of
from weaner pigs fed dry feed and liquid diets. J. Sci. Food
limits. E679-91, ASTM, Philadelphia.
Agric. 73, 437–445.
American Society for Testing and Materials (ASTM), 1992. Atlas
Jacobson, L.D., Guo, H., 2000. Odor from feedlots setback
of Odor Character Profiles. DS 61, ASTM, Philadelphia.
estimation tool (OFFSET). In: Livestock Odor Workshop I
American Society for Testing and Materials (ASTM), 1997.
Handbook. February 2000.
Standard practices for referencing suprathreshold odor intensity.
Jiang, J.K., Sands, J.R., 1999. Controlling noxious animal odours:
E544-75, ASTM, Philadelphia.
an imperative at the rural–urban interface — Review. Asian
Billings, C., Jonas, L., 1981. Odor thresholds in air as compared
Aust. J. Anim. 12 (4), 633–641.
to threshold limit values. Am. Ind. Hyg. Assoc. J. 42, 479–480.
Kobal, G., Van Toller, S., Hummel, T., 1989. Is there directional
Bundy, D.S., 1992. Odor issues with wastes. In: National Livestock
smelling. Experientia 45, 130–132.
Poultry and Aquaculture Waste Management. ASAE Publ.
Korpi, A., Kasanen, J.P., Alarie, Y., Kosma, V.M., Pasanen, A.L.,
03-92, American Society of Agricultural Engineers, St. Joseph,
1999. Sensory irritating potency of some microbial volatile
MI, pp. 288–292.
organic compounds (MVOCs) and a mixture of five MVOCs.
Cavalini, P.M., 1994. Industrial odorants: the relationship between
Archiv. Environ. Health 54, 347–352.
modeled exposure concentrations and annoyance. Arch. Krost, K.J., Pellizzari, E.D., Walburn, S.G., Hubbard, S.A., 1982.
Environ. Health 49, 344–351. Collection and analysis of hazardous emissions. Anal. Chem.
Cometto-Muñiz, J.E., Cain, W.S., Hudnell, H.K., 1997. Agonistic 54 (4), 810–817.
sensory effects of airborne chemicals in mixtures: odor, nasal Leonardos, G., 1999. Beyond detectability. J. Air Waste Manage.
pungency, and eye irritation. Percept. Psychophys. 59, 665–674. 49, 582–587.
Cometto-Muñiz, J.E., Cain, W.S., Abraham, M.H., Gola, J.M.R., Liu, Q., Bundy, D.S., Hoff, S.J., 1995. A study on the air flow
1999. Chemosensory detectability of 1-butanol and 2-heptanone and odor emission rate from a simplified open manure storage
singly and in binary mixtures. Physiol. Behav. 67, 269–276. tank. T ASAE 38, 1881–1886.
Cometto-Muñiz, J.E., Cain, W.S., Hiraishi, T., Abraham, M.H., Merck Index, 1996. Whitehouse Station, NJ Merck Research
Gola, J.M.R., 2000. Comparison of two stimulus-delivery Laboratories, 12th edition.
systems for measurement of nasal pungency thresholds. Chem. Miner, J.R., 1980. Controlling odors from livestock production
Senses 25, 285–291. facilities: state-of-the-art. In: Livestock Waste: A Renewable
Costa, D.L., Amdur, M.O., 1996. Air pollution. In: Klaasen, Resource. American Society of Agricultural Engineers, St.
C.D, Amdur, M.O and Doull, J. (Eds.), Casarett and Doull’s Joseph, MI, pp. 297–301.
Toxicology, The Basic Science of Poisons, 5th Edition. Miner, J.R., Licht, L.A., 1980. Fabric swatches as an aid in
McGraw-Hill, New York, Chapter 28, pp. 857–882. livestock odor evaluations. In: Livestock Waste: A Renewable
Devos, M., Patte, F., Rouault, J., Laffort, P, Van Gemert, Resource, Proceedings of the 4th International Symposium
L.J. (Eds.), 1990. Standardized Human Olfactory Thresholds. on Livestock Wastes, Amarillo, TX, 15–17April 1980.
Oxford University Press, New York. American Association of Agricultural Engineers, St. Joseph,
Dravnieks, A., 1980. Odor threshold measurement by dynamic MI, pp. 302–305.
olfactometry: significant operational variables. J. Air Pollut. Nicolai, R.E., Clanton, C.J., Guo, H., 2000. Modeling the
Control Assn. 30, 1284–1289. relationship between threshold and intensity of swine odors. In:
Dravnieks, A., Prokop, W.H., 1975. Source emission odor Proceedings of the 2nd International Conference, Des Moines,
measurement by a dynamic forced choice triangle olfactometer. IA, 9–11 October 2000. American Association of Agricultural
J. Air Pollut. Control Assn. 25, 28–35. Engineers, St. Joseph, MI, pp. 296–304.
Elliott, L.F., Doran, J.W., Travis, T.A., 1978. A review of analytical NC House, 1977. North Carolina House Bill 515, August 1977.
methods for detecting and measuring malodors from animal Clean Water Responsibility Act/Environmentally Sound Policy
wastes. T ASAE 21 (1), 130–135. Act.
Esteban, E., 1997. The confinement animal feeding operation O’Neill, D.H., Phillips, V.R., 1992. A review of the control of
workshop. Centers for Disease Control and Prevention. National odour nuisance from livestock buildings. Part 3. Properties of
Center for Environmental Health. Atlanta, GA. the odorous substances which have been identified in livestock
Hammond, E.G., Smith, R.J., 1981. Survey of some molecularly wastes or in the air around them. J. Agric. Eng. Res. 53, 23–50.
dispersed odorous constituents in swine-house air. Iowa State Pate W., 1998. Odor: a public health practice viewpoint. In:
J. Res. 55 (4), 393–399. Schiffman, S.S., Walker, J.M., Small, R., Millner, P. (Organizing
240 S.S. Schiffman et al. / Agricultural and Forest Meteorology 108 (2001) 213–240

Committee), Proceedings of the Workshop on Health Effects Stevens, J.C., Dadarwala, A.D., 1993. Variability of olfactory thres-
of Odors: Participant Reviews and Opinions, Duke University. hold and its role in assessment of aging. Percept. Psychophys.
Petersen, B.W., Lavdas, L.G., 1986. Inpuff 2.0-A multiple 54 (3), 296–302.
source Gaussian puff dispersion algorithm — User’s guide. Sweeten, J.M., 1998. Separation distances for swine odor control
Atmospheric Sciences Research Laboratory, Office of Research in relation to manure nutrient balances. Appl. Eng. Agric. 14,
and Development. US EPA, Research Triangle Park, NC. 543–549.
Ritter, W.F., 1989. Odour control of livestock wastes: state-of- Thu, K., Donham, K., Ziegenhorn, R., Reynolds, S., Thorne, P.S.,
the-art in North America. J. Agric. Eng. Res. 42, 51–62. et al., 1997. A control study of the physical and mental health
Roscher, S., Glaser, C., Hummel, T., Kobal, G., 1996. An easy of residents living near a large-scale swine operation. J. Agric.
method for separating olfactory from trigeminal stimulation. Safety Health 3, 13–26.
Chem. Senses. 21, 492. Travis, T.M., Elliott, L.F., 1977. Quantitation of indole and skatole
Russell, M., Dark, K.A., Cummins, R.W., Ellman, G., Callaway, in a housed swine unit. J. Environ. Qual. 6, 407–410.
E., Peeke, H.V.S., 1984. Learned histamine release. Science United States General Accounting Office (GAO), 1995. Animal
225, 733–744. agriculture: information on waste management and water quality
Ruth, J.H., 1986. Odor thresholds and irritation levels of several issues. Briefing report to the United States Senate Committee
chemical substances: a review. Am. Ind. Hyg. Assoc. J. 47, on Agriculture, Nutrition, and Forestry.
A142–A151. United States Senate, 1997. Animal waste pollution in America:
Schiffman, S.S., 1998. Livestock odors — implications for human An emerging national problem. Environmental risks of livestock
health and well-being. J. Anim. Sci. 76, 1343–1355. and poultry production. Report compiled by the minority staff of
Schiffman, S.S., 1999. Unpublished data. the United States Senate Committee on Agriculture, Nutrition,
Schiffman, S.S., Williams, C.M., 1999. Evaluation of swine odor and Forestry for Senator Tom Harkin.
control products using human odor panels. In: Proceedings of Van den Hazel, P.J., Waegemaekers, C.H., 1991. Odour annoyance
the Animal Waste Management Symposium, Raleigh, NC State in a residential area near a papermill. Public Health Rev. 19,
University, pp. 110–118. 251–262.
Schiffman, S.S., Sattely-Miller, E.A., Suggs, M.S., Graham, B.G., Van Gemert, L.J., Nettenbreijer, A.H. (Eds.), 1977. Compilation of
1995. The effect of environmental odors emanating from odour threshold values in air and water. National Institute for
commercial swine operations on the mood of nearby residents. Water Supply, Voorburg, The Netherlands and Central Institute
Brain Res. Bull. 37, 369–375. for Nutrition and Food Research TNO, Zeist, The Netherlands.
Schiffman, S.S., Walker, J.M., Dalton, P., Lorig, T.S., Raymer, Welch, F., Murray, V.S., Robins, A.G., Derwent, R.G., Ryall, D.B.,
J.H., Shusterman, D., Williams, C.M., 2000. Potential health Williams, M.L., Elliott, A.J., 1999. Analysis of a petrol plume
effects of odor from animal operations, wastewater treatment, over England: 18–19 January 1997. Occup. Environ. Med. 56,
and recycling of byproducts. J. Agromed. 7, 7–81. 649–656.
Schneider, R.A., Schmidt, C.E., 1967. Dependency of olfactory Wetherill, G.B., Levitt, H., 1965. Sequential estimation of points
localization on non-olfactory cues. Physiol. Behav. 2, 305–309. on a psychometric function. Br. J. Math. Stat. Psychol. 18,
Shusterman, D., 1992. Critical review: the health significance of 1–10.
environmental odor pollution. Arch. Environ Health. 47, 76–87. Wing, S., Wolf, S., 2000. Intensive livestock operations, health,
Siegel, S., Kreutzer, R., 1997. Pavlovian conditioning and multiple and quality of life among eastern North Carolina residents.
chemical sensitivity. Environ. Health Perspect. 105 (Suppl. 2), Environ. Health Perspect. 108 (3), 233–238.
521–526. Zahn, J.A., Hatfield, Y.S.D., DiSpirito, A.A., Laird, D.A., Pfeiffer,
Siegel, S., 1999. Multiple chemical sensitivity as a conditional R.L., 1997. Characterization of volatile organic emissions and
response. Toxicol. Ind. Health. 15, 323–330. wastes from a swine production facility. J. Environ. Qual. 26,
Smith, R.J., 1993. Dispersion of odours from ground level 1687–1696.
agricultural sources. J. Agric. Eng. Res. 54, 187–200. Zald, D.H., Pardo, J.V., 1997. Emotion, olfaction and the
Spoelstra, S.F., 1980. Origin of objectionable odorous components human amygdala: amygdala activation during aversive olfactory
in piggery wastes and the possibility of applying indicator stimulation. Proc. Natl. Acad. Sci. U.S.A. 94, 4119–4124.
components for studying odor development. Agric. Environ. Zhu, J., 1999. The influence of stability class on downwind odor
5 (3), 241–260. concentration predicted by air dispersion models. J. Environ.
Steinheider, B., Both, R., Winneke, G., 1998. Field studies on Sci. Health A 34, 1919–1931.
environmental odors inducing annoyance as well as gastric and Zhu, J., Jacobson, L.D., Schmidt, D.R., Nicolai, R., 2000. Evalu-
general health-related symptoms. J. Psychophysiol. 12 (Suppl. ation of INPUFF-2 model for predicting downwind odors from
1), 64–79. animal production facilities. Appl. Eng. Agric. 16, 159–164.

You might also like