ArticlePDF Available

Odor Plumes and How Insects Use Them

Annu. Rev. Entomol. 1992. 37:505-32
Copyright © 1992 by Annual Reviews Inc. All rights reserved
John Murlis
Department of the Environment, London SW1, United Kingdom
Joseph S. Elkinton and Ring T. Card~
Department of Entomology, University of Massachusetts, Amherst, Massachusetts
KEY WORDS: pheromones, odor dispersion, orientation, anemotaxis
Odor plumes form as the wind disperses odor molecules from their source.
Their structure is complex and is much like that seen in smoke plumes. The
plume as a whole wanders, apparently randomly, over a wide area. Many
species of insects, however, have behavior that enables them to follow odor
plumes to their source and in this way to find mates or distant resources.
Research on odor communication has important practical implications. The
pheromones of many important pest species have been identified, syn-
thesized, and formulated for use in pest-management schemes, and they are
widely used in traps for population monitoring. Several pheromone formula-
tions are marketed, and many more are under development. They are widely
used in crop fields to disrupt insect mating (70, 103).
Despite considerable research, however, pheromones have not been uni-
versally successful in these applications (-32). Pheromone traps have been
used to detect insects at remarkably low population densities, but numbers
caught often do not correlate well with density of these insects. Catch is
affected by features of population ecology (86), and various factors modify
0066-4170/92/0101-0505 $02.00
Annual Reviews
Annu. Rev. Entomol. 1992.37:505-532. Downloaded from
by UNIVERSITY OF CALIFORNIA - RIVERSIDE LIBRARY on 11/11/08. For personal use only.
trap efficiency, including trap shape, placement, and weather (79). Even
more variables influence the success of mating disruption (9, 22).
Entomologists recognize the need for more fundamental understanding of
the factors affecting pheromone communication. They seek answers to sever-
al practical questions concerning, for instance, the effective range of pher-
omone traps or the strength and positioning of sources for mating disruption.
To provide answers, both the odor plume and the insects’ response to it need
study. In this review, we summarize the current knowledge on plumes and
how insects orient to them.
Everyday observation of smoke plumes from chimneys, stacks, and bonfires
provides insight into their structure. Seen from a distance, smoke from an
elevated source appears as a discrete undulating cloud. In a brisk wind, the
cloud seems to widen rapidly close to the source, after which it expands
steadily at a lesser rate. It rises and falls as it leaves the chimney, but, soon
after, the shape becomes frozen into a sinuous pattern that changes little as it
moves with the wind. From a closer vantage, the cloud contains considerable
fine-scale structure, wispy in its center and ragged at the edges. Filaments of
dense smoke intertwine with regions of clean air. The range covered by a
discrete source is indicated, for example, by the deposits of ash on the ground
downwind of a bonfire. These observations suggest three levels of odor plume
1. Large-scale structure--the shape and average odor strength of the body .
of the plume affect the orientation strategy of insects.
2. Small-scale structure--the fluctuating odor concentrations within the
plume body affects the input to insects’ central nervous systems and
hence their instantaneous response to the plume.
3. Time-averaged structure-~determines the probability that an insect will
contact the odor plume at different locations downwind of the source.
Odor-plume structure is determined by the physics of atmospheric disper-
sion. Odor plumes are created when odor molecules are released from their
source and are taken away in the wind. As the cloud of molecules moves away
from the source, it expands and the mean concentration of molecules within it
falls. Two processes are at work: (a) molecular diffusion in which random
motion of the molecules causes them to move gradually apart and (b) turbu-
lent diffusion in which the cloud of molecules is physically torn apart by air
turbulence. The temporal and spatial scales of the two processes are very
Annual Reviews
Annu. Rev. Entomol. 1992.37:505-532. Downloaded from
by UNIVERSITY OF CALIFORNIA - RIVERSIDE LIBRARY on 11/11/08. For personal use only.
different. Molecular diffusion is a slow and small-scale phenomenon. Turbu-
lent diffusion, by contrast, is vigorous and covers a wide range of temporal
and spatial scales. It dominates plume development.
Several authors have summarized the mechanics of turbulent diffusion in
the atmosphere (57, 99, 109). Turbulence arises from the asymmetry of forces
in the earth’s boundary layer, which causes large air masses many hundreds of
meters across to tumble as the wind drives them along. Its production and
maintenance depend on a balance of inertial and buoyancy forces known as
atmospheric stability. When the atmosphere is stable (typically at night under
clear skies), buoyancy forces dampen down turbulence and inhibit its produc-
tion, but in an unstable atmosphere, buoyancy forces encourage the genera-
tion of turbulence. Unstable conditions occur on sunny days when the ground
absorbs heat from the sun’s rays. This heat is transmitted to the air at ground
level and convective updrafts ensue. Under neutral conditions (heavy over-
cast), turbulen~ generated by nonbuoyancy forces is sustained. The energy
contained in the giant vortices or eddies is transferred to smaller vortices that
in turn transfer it to yet smaller vortices. As the vortices become smaller, they
become more randomly oriented until they show no preferred direction. The
result is that the wind speed measured in a given direction fluctuates con-
tinually. The energy injected into the turbulence by the largest eddies is
finally removed by viscous dissipation in the smallest eddies. The size of the
dissipation range of eddies is determined by viscosity of the air and characte-
rized by a parameter known as the Kolmogoroff length. The Kolmogoroff
length is a measure of the size of the smallest turbulent motions. In the
atmosphere, it is roughly a function of wind speed and height (88) and
typically a centimeter or so. Therefore, turbulent eddies have a minimum
size. Instruments used to measure turbulence are stationary and record wind
speed as a function of time. They register the frequency of wind speed
fluctuations or their period rather than the length scales of the eddies that
create them. Conversion from frequency or period to a length scale can be
made using the approximations: L = U/n = Up, where L is the length scale, n
is the frequency, p is the period (time scale), and U is the mean wind speed.
The instantaneous recorded wind speed, U, can be resolved into two
components, a mean component, ti (the overbar denotes a time mean quant-
ity), and a fluctuating component, u’, so that U = a + u’. The strength of
fluctuating quantity is often expressed as the root mean square (rms)
deviations from the mean; for example, the strength of an alternating current
(AC) power supply is given as the rms voltage. The rms of a fluctuating signal
is the equivalent of the standard deviation of samples of a distributed quantity.
The rms of the fluctuating wind speed is a good measure of the intensity of the
turbulence. By convention, the intensity is given as a nondimensional quant-
ity by dividing the rms by the mean wind speed.
Annual Reviews
Annu. Rev. Entomol. 1992.37:505-532. Downloaded from
by UNIVERSITY OF CALIFORNIA - RIVERSIDE LIBRARY on 11/11/08. For personal use only.
The energy contained in a unit volume of a flow is given by a product of
density and half the square of the speed. In the case of turbulence, the
variance of the fluctuating component, fi’ 2, gives a measure of the turbulent
energy. The fluctuations that contribute to the turbulent energy come from a
wide range of different frequencies. The convention is to present measure-
ments of turbulent energy as a spectrum showing how much of the turbulence
energy is contained in each frequency band (84) (Figure 1, top). Currently,
energy spectra are usually derived from anemometer records through digital
processing (12), but the idea is more simply explained by considering the
dominant frequency,
Figure I (Top) Energy spectrum showing the energy contained in the frequency band centered
on n. (Bottom) Normalized energy spectrum, showing the dominant frequency.
Annual Reviews
Annu. Rev. Entomol. 1992.37:505-532. Downloaded from
by UNIVERSITY OF CALIFORNIA - RIVERSIDE LIBRARY on 11/11/08. For personal use only.
analog methods used originally (69). The fluctuating part of the signal
played into a series of band pass filters, each of which allows only frequencies
in a narrow range to pass through. The energy content of each band, derived
from the square of the amplitude of the signal passed (with suitable scaling
factors), is plotted against the center frequency of the band (Figure 1, top).
make spectra measured in different atmospheric conditions comparable, they
can be normalized by dividing by the total variance (Figure 1, bottom).
Spectra contain information about the relative importance of different
turbulence scales. They relate directly to the form taken by dispersion and
hence to the structure of plumes (60). Normalized spectra of turbulence
measured in the open atmosphere have a characteristic shape. The energy
peak, corresponding to the dominant frequency, occurs at time scales of some
hundreds of seconds. Allen (1) measured turbulence spectra in a forest and
reported a progressive change in the spectra from above the forest through the
canopy to ground level. Trees removed energy from the largest and smallest
eddies, leaving a more peaky spectrum dominated heavily by eddies of a size
comparable to the tree spacing. Measurements by Wang (121) show a similar
form with a strongly dominant turbulent scale of a few meters. Turbulence in
the forest is organized into a narrower range of eddy sizes than in the open
In the following sections we summarize what is now known about the
different levels of structure in an odor plume, the influences of habitat and
atmospheric conditions, and how insects respond to them.
If an odor source is smaller than the Kolmogoroff scale, odor molecules
released into the wind form a filament that expands slowly by molecular
diffusion until it reaches the size of the smallest eddies, when the rapid and
vigorous process of turbulence diffusion takes over. During the period of
molecular diffusion, the development of the plume depends on the character-
istics of the odor molecules. Plumes of different materials may behave
The length of this molecular-diffusion stage of plume growth depends on
source size and wind speed. Miksad & Kittredge (88) calculated its length
the case of a small source and concluded that the length could be some meters.
Aylor et al (4), however, argued that for most odor plumes to which insects
respond, for example, pheromone emitted by a female insect perched on a
leaf, the source aerodynamically includes both leaf and insect and compares
in size to the smallest eddies. If so, this first stage would probably not exceed
a centimeter or so.
Beyond this point, the plume structure becomes practically independent of
Annual Reviews
Annu. Rev. Entomol. 1992.37:505-532. Downloaded from
by UNIVERSITY OF CALIFORNIA - RIVERSIDE LIBRARY on 11/11/08. For personal use only.
the properties of the material within it and resembles the familiar form seen in
a smoke plume. Thus a variety of tracers, chosen for visibility or ease of
measurement, may be used to study plume structure. Observations with
tracers show the progressive influence of larger turbulence scales on a de-
veloping odor plume (58). Eddies in the size range from the Kolmogoroff
length up to some hundreds of millimeters determine the small-scale structure:
they stretch and stir the filaments in the plume (66). Those of some meters
size influence the large-scale structure; they cause the fluctuations in the
initial direction of the plume that create the undulating and meandering
patterns. Eddies on the scale of hundreds to thousands of meters give rise to
long-term changes in plume direction.
Pasquill & Smith (99) describe the effect of atmospheric stability on the
large-scale structure of a smoke plume. In unstable conditions, one sees
vigorous vertical and horizontal undulation, which is reduced in neutral and
almost absent in stable conditions. In stable conditions, the continual chang-
ing of wind direction causes the smoke to fan out widely horizontally. A
plume generated in typical daytime conditions will therefore be of a very
different form from one generated at night. Understanding of plume structure
has been advanced by making direct measurements and by the development of
mathematical models based on the physics of diffusion. Recent progress is
summarized below.
Time-Averaged Plume Structure
The first models to be developed provided estimates of mean concentration of
odor downwind of a point source based on studies of the dispersion of
battlefield gases following World War I (104). Sutton (112) developed
semiempirical equation for a plume generated from a point source at ground
level. Sutton’s model assumed a Gaussian (normal) distribution of mean
concentration in all directions perpendicular to the plume center line and a
simple power law to describe the fall in the mean concentration along the
plume center line at increasing distances from the source (Figure 2, top). The
equation included constants known as dispersion coefficients that were de-
termined by the level of turbulence or atmospheric stability.
Wright (126) and Bossert & Wilson (15) applied the Sutton model
estimate concentrations of insect pheromones downwind of a source. They
used typical values for the dispersion coefficients suggested by Sutton for
neutral atmospheric conditions. Numerous subsequent applications of the
Sutton and related models to estimate pheromone concentrations are reviewed
elsewhere (41).
The Sutton model lacks flexibility, however, as the dispersion coefficients
do not adequately reflect the physical processes they describe. The Gaussian
plume models currently used to predict mean concentrations for air pollutants
Annual Reviews
Annu. Rev. Entomol. 1992.37:505-532. Downloaded from
by UNIVERSITY OF CALIFORNIA - RIVERSIDE LIBRARY on 11/11/08. For personal use only.
(29, 117) or radioactive materials (63) are more general and empirically based
(Figure 2, top). The basic equatign for a Gaussian plume is:
- Q exp
exp- -~ +
C(x,y,z) 2 7r UrO.yO.
~ff y2
where C~x,y,z) is the mean concentration at any position (x,y,z) with the x axis
aligned with the mean horizontal wind direction, the y axis aligned horizontal-
ly cross-wind, and the z axis representing height above ground (Figure 2,
top). Q is the release rate of the material in the plume, and Ur is the mean wind
speed at a standard reference height (10 m is usual). The coefficients O’y, trz
are standard deviations of the concentration distributions in the respective
horizontal and vertical directions, and h is the height of the source of emission
above ground. The equation also contains a constant, a, the proportion of
material reflected back into the plume when it reaches ground level.
The coefficients, ~ry, Crz, are analogous to the dispersion coefficients of the
Sutton model. They are not constants, however; instead they vary with
distance from the source in addition to atmospheric stability and wind speed.
They are derived from trials such as the US Prairie Grass project (31, 59).
practice, estimating the atmospheric stability is difficult. The Richardson
number, a function of the rates of change of temperature and wind speed with
height, is commonly used (99), but it is not convenient to measure. Pasquill
(98) proposed a classification system for stability based on observations
wind speed and cloud cover. The resulting dispersion coefficients can be
obtained from charts (29, 117). A more quantitative method based on surface
measurements of heat flux and wind speed was recently proposed by Hunt et
al (62).
Elkinton et al (42) tested a range of Gaussian plume models in analysis
gypsy moth response to pheromone. The moths, in field cages, were exposed
to pheromone from a distant point source roughly upwind and the onset and
duration of wing fanning was recorded. The time-averaged plume models
tested did not predict where wing fanning took place in relation to the mean
wind direction. The work revealed two main kinds of difficulty with time-
average plume models:
1. Insects do not respond to mean concentration as calculated by these
models (over an averaging time of 3 min in the case of the Sutton model).
Instead, insects respond to instantaneous concentrations that are frequent-
ly many-fold higher than mean concentrations (91).
2. The averaging times implied by the model are themselves physically
unrealistic in the atmosphere where, because of the influence of continual
shifting in the mean wind direction, stable means do not form.
Annual Reviews
Annu. Rev. Entomol. 1992.37:505-532. Downloaded from
by UNIVERSITY OF CALIFORNIA - RIVERSIDE LIBRARY on 11/11/08. For personal use only.
Time-averaged plume
The filamentous structure of a real plume
Figure 2 (Top) Time-average Gaussian plume model, showing the principle axes and the source
positioned at height h. (Middle) A meandering plume model with concentration in each disc
distributed normally about the meandering center line. (Bottom) The structure of a real plume.
Models of this type cannot provide a general framework for the analysis of
insect odor communication (41), but they may be useful in specific cases, for
example, where there is significant contamination of vegetation by an odor
source (120). Gaussian plume models could be used to map the contaminated
area downwind.
Annual Reviews
Annu. Rev. Entomol. 1992.37:505-532. Downloaded from
by UNIVERSITY OF CALIFORNIA - RIVERSIDE LIBRARY on 11/11/08. For personal use only.
Large-Scale Structure: Meandering Plumes
The most obvious feature of a plume is that it meanders (Figure 2, bottom)
and insects will encounter much higher levels of odor than the time mean.
Predicting concentration in the meandering plume is therefore of great prac-
tical importance, but concentration is difficult to assess. In the absence of
helpful experimental data, the problem was first tackled theoretically.
Gifford (51) suggested a meandering plume model in which Gaussian
functions described the distribution of material across the plume, but the
position of the plume center was free to move. He assumed that the plume was
made up of a series of disks in a plane perpendicular to the mean wind
direction, the centers of which were displaced around the mean wind axis
according to the displacement of the meandering plume (Figure 2, middle).
The overall variance of the concentration fluctuations is a combination of the
variance of the concentration fluctuations with respect to the center of each
disk and the variance of the position of each disk relative to the mean wind
1 is the overall variance,
2 is the variance of concentration across the disk, and
3 is the variance of the position of disk centers.
If the odor is normally distributed along any axis perpendicular to the
plume centerline, the ratio of peak to mean concentration on the mean wind
axis is:
_ cV2i q- CV2M
c ! 21
More general equations for the ratio of the instantaneous concentration to the
local mean concentration can be obtained. Venkatacham (118) proposed
simple model based on the assumption that small parcels of material from an
elevated source are carried downwind at the mean wind speed along straight
lines that deviate vertically and horizontally from the mean wind direction.
The deviations give the parcels a vertical and .horizontal speed component,
and by making assumptions about the distributions of these, one can calculate
the distribution of concentration across the plume downwind. Models of this
kind are known as PDF (probability density function) models.
A plume meandering in roughly sinusoidal form spends most of its time
Annual Reviews
Annu. Rev. Entomol. 1992.37:505-532. Downloaded from
by UNIVERSITY OF CALIFORNIA - RIVERSIDE LIBRARY on 11/11/08. For personal use only.
around the edges of its range. Ride (102) predicted that the cross-wind
distribution of mean concentration from a narrow plume subject to wide
meandering should be bimodal with peaks towards the edges of the plume’s
cross-wind range. Studies with smoke plumes by Hanna (55) illustrate this
bimodal cross-wind distribution. Calculations predict that the degree of
bimodality would depend on the ratio of plume body width to the width of
meandering, r. Bimodality in Ride’s model starts to appear at r = 0.50 and
becomes well developed at r = 0.75. Hanna’s data correspond to r = 0.65.
For pheromone plumes in unstable or neutral conditions, the source is usually
small compared to the size of the eddies responsible for meandering, and r
will be well above 0.5. Ride’s model suggests that the edges of the envelope
in which an odor plume meanders contain higher mean concentrations. These
findings contradict the earlier assumptions of normal distributions of odor
embodied in Gaussian plume models.
Models of these types are simple but powerful. They fit the experimental
data well (46) and have considerable potential for further development (54).
HOgstrOm (61) used Gifford’s model for predicting fluctuating odor con-
centrations, and this model seems an ideal basis for the analysis of trap
interaction (17). These models are still designed to give averaged con-
centrations, however, and do not address the question of the small-scale
structure of a plume. For this, experimental data remain the most reliable
source of information.
Small-Scale Structure
Odors insects use, for example pheromones, cannot at present be measured
with sufficient time resolution to allow the fine-scale structure of odor plumes
to be assessed directly. We rely on data from trials with tracers for this
Concentrations in a smoke plume can be measured over short averaging
periods with photometers [6 s (108), 1 s (55), < I s (37)]. Other studies
53) obtained resolutions better than 30 ms using sulphur dioxide and phos-
phorus pentaflouride tracers with fast flame photometric detectors. Studies
with propylene (93, 94) obtained resolution to better than one second
distances up to 1000 m from source: digital processing improved this figure to
approximately 0.1 s. Fackrell (44, 45) obtained response times of a few
milliseconds in a wind tunnel with methane and propane.
Jones (64) and Jones & Griffiths (67) described a simple and robust system
for generating and detecting unipolar ions in the atmosphere. Ion plumes can
be used to model plumes of odor provided that repulsion between ions and
deflection of the plume (resulting from image forces in the ground) are
minimized or allowances are made for them in analysis (27). Both unipolar
(68) and bipolar (13) ion tracers have also been used in wind-tunnel work
Annual Reviews
Annu. Rev. Entomol. 1992.37:505-532. Downloaded from
by UNIVERSITY OF CALIFORNIA - RIVERSIDE LIBRARY on 11/11/08. For personal use only.
Analysis of Small-Scale Plume Structure
High-resolution tracer detectors produce a signal consisting of a series of short
bursts of activity, separated by periods of zero signal (Figure 3). At the
leading and trailing edges, the signal amplitude rises sharply. Bursts some-
times consist of one clearly defined spike but more often they have consider-
able structure within them. Murlis & Jones (91) defined the burst length as the
time between leading and trailing edges and the burst return period as the time
between the leading edge of one burst and the next.
As with instantanerus recorded wind speed (U), one can resolve the
instantaneous concentration, C, into two components, a mean concentration ~
and a fluctuating component, c’, such that: C = ~ + c’. The intensity of
concentration fluctuations is given by the rms of the fluctuating component
divided by the mean concentration. The variance in the fluctuating component
can be broken down into contributions from the range of frequencies and
formed into a spectrum. Hanna & Insley (58) showed that smoke-plume data
from large-scale field trials produce well-defined spectra but the dominant
scales of concentration fluctuations are considerably shorter than the domi-
nant turbulence scales.
EAG potential
Burst length
Figure 3 Fluctuating signal from an ion concentration sensor showing burst length and burst
return period. Electroanntenogram (EAG) signal recorded from a gypsy moth antenna responding
to pheromone at the same time and in the same position as the ion sensor. Taken from Murlis et al
Annual Reviews
Annu. Rev. Entomol. 1992.37:505-532. Downloaded from
by UNIVERSITY OF CALIFORNIA - RIVERSIDE LIBRARY on 11/11/08. For personal use only.
Time Dependency of Concentration Fluctuations
The few published measurements of the temporal structure of plumes show
that the burst lengths and burst return periods at a fixed position range widely
with an approximately log-normal distribution. Bursts last from less than 10
ms to over a second and burst return periods from 500 ms to several minutes
(66, 90-92). If the turbulence scales responsible for the large-scale structure
of the plume as a whole were distinct from those responsible for the small-
scale structure within the plume (56), then the distributions of burst return
period would be bimodal. Evidence of bimodality, however, can only be
found in distributions of return period measured at great distances from the
source (66) and where the criterion for positioning leading and trailing edges
of bursts was set at higher concentration values (90). The large- and small-
scale structure is not therefore easily separable. A continuum of scales
corresponds to the continuum of eddy sizes. Typical values for burst length
and burst return period (the value of the mode of their distributions) are,
respectively, a few hundreds of milliseconds and a second or so.
As conventionally measured, distributions of neither burst lengths nor burst
return period show significant trends related to source-to-receptor distance
(66). Murlis et al (92), however, took an ensemble mean of both return period
and length measured over short sections of active signal 90 s and 180 s long.
Mean burst length and mean burst return period measured in this way in-
creased as source-to-receptor distance increased from 2.5 m to 20 m; mean
burst length increased by a factor of 2, and burst return period increased by a
factor of >7.
Intermittency of Fluctuating Concentration
Intermittency is defined as the proportion of time when the signal is absent.
Because it is also sometimes defined as the proportion of time when the signal
is present (the one being 1.0 minus the other), we have converted where
Intermittency, arising from zero concentrations inside the plume and the
meandering of the plume as a whole, considerably affects concentration
measurements. Zero values can, however, be rejected through a procedure
known as conditional sampling (108). Conditionally sampled means are
formed from ensembles containing only values that meet some set con-
dition--for example, that they are nonzero.
Jones (65) measured intermittency over successive 1-h samples of an ion
detector signal. He found it varied widely and concluded that intermittency is
a highly unreliable signal characteristic because long period events have a
disproportionate effect on samples. The wind may shift away from the
measurement point, for example, for minutes at a time. Murlis et al (92)
Annual Reviews
Annu. Rev. Entomol. 1992.37:505-532. Downloaded from
by UNIVERSITY OF CALIFORNIA - RIVERSIDE LIBRARY on 11/11/08. For personal use only.
estimated intermittency from short samples of active signal (90-s and 180-s
durations) to remove the influence of long periods of zero concentration.
Intermittency estimated in this way increased systematically with increasing
distance from the source, from over 60% at 2.5 m to 90% or more at 20 m.
Much intermittency comes from movements of the plume as a whole, but a
significant part is derived from patches of clean air inside the plume (Figure 1,
bottom). Fackrell & Robins (47) show that intermittency depends on
turbulence structure of the air flow and the source size. Intermittency inside
their wind-tunnel plume decreased as the distance from the source increased,
until at large distances downwind it became close to 0% (no patches of
uncontaminated air left). In the atmosphere, however, where source size is
comparable to that of the smallest eddies, Mylne & Mason (93, 94) found that
the minimum intermittency in the center of a plume reaches an asymptotic
value significantly greater than zero. In the atmosphere, some patches of
odor-free air are apparently always inside a plume from a small source--the
smaller the source relative to the smallest turbulence, the more significant the
patches are.
Concentration Fluctuations in a Plume
Probability distributions of concentration fluctuations are highly skewed
towards high values, low concentration values being disproportionately
numerous. Functions proposed for the distributions include exponential, log
normal, and clipped normal (52, 55, 65, 66, 114). The intensity of concentra-
tion fluctuations decreases with increasing distance from source more rapidly
than mean concentration. The intensity seems eventually to reach some
approximately constant (nonzero) value dependent on source size and the
turbulence scale (46, 47, 93, 94).
Mylne & Mason (93, 94) found that the distribution of mean concentration
across the plume, obtained by conditional sampling, corresponded to a rough-
ly Gaussian form. However, intermittency showed a complementary trend,
and these authors concluded that the lower conditional mean concentrations
found at the edges of the plume were produced by the higher intermittency,
the concentration in odor bearing filaments remaining unchanged.
Aylor (3) recognized the probable behavioral importance of extreme values
in pheromone plumes. Hence, an alternative way of assessing odor concentra-
tion is to focus on peak values in bursts. Storeb0 et al (111) found patches
undilute material 100 m and more from its source. Murlis & Jones (91)
measured probability distributions of peaks of concentration. They were
skewed to higher values, but less so than mean concentrations in bursts.
Distributions of burst-peak concentrations from signals recorded at different
positions overlapped significantly, even when they were as much as 5 m
Annual Reviews
Annu. Rev. Entomol. 1992.37:505-532. Downloaded from
by UNIVERSITY OF CALIFORNIA - RIVERSIDE LIBRARY on 11/11/08. For personal use only.
One can estimate the strength of odor in a plume in’other ways. In
experiments with smell in the atmosphere (10, 61), observers found that smell
arrived in short bursts, both because the plume switched back and forth and
because the human nose is itself a sampling organ, with an averaging period
of a few seconds. Similarly, an insect flying in an odor plume samples
(though over shorter times). Because of its movement through the plume and
because the plume is being driven past it by the wind, the concentration in the
plume has less direct impact on the insect than the flux of material past it (41).
Murlis et al (92) measured the mean flux, the dose (the quantity of ions
contained in each burst), the peak value of flux in each burst, and the
maximum peak recorded in each 90-s or 180-s sample of active dignal using
Langmuir ion probes (13, 14) (Figure 3). Strength as assessed in all these
ways decreased systematically as source-to-receptor distance increased
according to simple power laws, but the mean flux and dose decreased more
rapidly than the peak values.
Effects of Habitat and the Size and Position of an Odor
In an open habitat, dominant turbulent scales are some hundreds of meters and
energy is progressively transferred from large to small scales. There is
considerable energy at small scales, and mixing is vigorous. David et al
showed that, for smoke puffs released from a low source over open ground
(36), each followed (different) linear trajectories for at least 20 m, showing
that individual segments of a plume travel long distances in straight lines
despite differences in original orientation as they leave the source. Meander-
ing is produced as successive plume segments set out in a slightly different
direction from their immediate neighbors.
In a forest, however, where small-scale turbulence is relatively less energe-
tic and the dominant scale is only a few meters, mixing is far more leisurely
and the directional sense of plume segments is less well sustained. The
structure in a forest smoke plume seems almost frozen as it is carried
downwind on a path winding on the scale of the dominant eddy size. Elkinton
et al (43) showed that individual puffs of pheromone beneath a forest canopy
would follow highly nonlinear trajectories even within a few meters of the
source. Under light wind conditions in a forest, tracers (smoke puffs or
bubbles) are frequently observed to change direction by more than 180
over a
distance of 20 m or so. Consecutive puffs follow similar paths to about 20 m
where they frequently become very different, sometimes one in a sequence
doubling back on its trajectory. Brady et al (18) also recorded trajectories
this kind in studies of tsetse response to host odors in African bush (low
Recordings of tracer signals made in a forest (92) had longer bursts than
an equivalent trial in an open field. Flux decreased as source-to-receptor
Annual Reviews
Annu. Rev. Entomol. 1992.37:505-532. Downloaded from
by UNIVERSITY OF CALIFORNIA - RIVERSIDE LIBRARY on 11/11/08. For personal use only.
distance increased less rapidly than in the field. Peak-to-mean ratios were
lower close to the source and increased less rapidly in a forest.
The size of the source has a considerable effect on plume structure (47, 88);
in general, the smaller the source, the higher are the intermittency and the
intensity of concentration fluctuations. An odor source’s size depends on the
size of the insect’s perch rather than the size of the insect or its odor-emitting
gland (4). The size of the perch clearly depends on the habitat. Even in
forest, where the typical source is about the size of the diameter of the tree
trunks, the effective source size is small compared with the energetic turbu-
lent scales and corresponds to the smallest source sizes considered in the wind
tunnel trials of Fackrell & Robins (47).
Fine-scale structure in plumes is also affected by height of the source. From
sources at ground level, it is dominated by intense mixing at the surface (101)
and is independent of source size (47). Hanna & Insley (58) found
systematic data on elevated sources. It is generally assumed (91, 111) that
greater heights than about 1.5 m above ground, the structure of plumes from
sources whose effective size is a few centimeters is affected little by the
surface for distances of tens of meters from the source.
Models for Predicting Concentration Fluctuations
Experimental data of the kind detailed above has enabled modelers to test a
range of different approaches to the prediction of fluctuating concentration.
Hanna (54) reviewed such models and tested them against field data.
concluded that simple analytical formulae derived from such models fit the
data well, but that considerable potential for further development remained.
The most common output from these models is a distribution of concentra-
tion fluctuation intensity, which can be considered as a property to be created,
removed, and transported. For an idealized plume with its center line on the x
axis, the conservation equation for the fluctuating concentration (c’) is given
by Pasquill & Smith (99) as:
1 is an advection term (transport of c’ along the x axis),
2 is a production term,
3 is a diffusion term (transport of
by the cross-wind gradients
of c’), and
4 is a dissipation term (loss of
by molecular diffusion).
The terms v’ and w’ are the fluctuations of wind speed in the y
and z directions, respectively.
Annual Reviews
Annu. Rev. Entomol. 1992.37:505-532. Downloaded from
by UNIVERSITY OF CALIFORNIA - RIVERSIDE LIBRARY on 11/11/08. For personal use only.
The equation can be solved only by making several assumptions and
simplifications. Csanady (33, 34) made the major simplifying assumption that
the cross product terms involving w’, v
, and c’ could be represented by the
product of the mean gradient and a coefficient of diffusivity, K. The conserva-
tion equation then becomes:
where Ta is a time constant for the decay of concentration fluctuations and is a
function of x. Ky and K~ are the coefficients of diffusivity in the y and z
directions, respectively, and are also functions of x.
Although this equation still appears somewhat intractable, all the terms can
be estimated or derived from experiments, and a solution can be found for c’
as a function of downwind position for a specified mean concentration
distribution, C = f(x,y,z). Further simplifications are possible. For example,
far downwind, the production term is small. If it is neglected and Ky = K~ =
K, a constant, the distribution of variance is Gaussian and the equation for
cross-wind distribution of intensity becomes:
~ exp
4~y ~
where y,z=O refers to values on the plume centerline. As a consequence of
this relationship, intensity of concentration fluctuation is greatest at the edges
of the plume body.
Wilson et al (124) proposed an empirical Gaussian model for c’ 2, assuming
it is produced from point sources positioned independently of the source of the
plume. The cross-wind profiles were similar at each position down wind and
were scaled on plume-body widths. The parameters describing position and
strength of the variance sources, the form of cross-wind profiles, and other
input needed were chosen to fit the wind tunnel data of Fackrell & Robins (46,
47). The parameters themselves have a sound physical basis, and the model
therefore has potential for general application. Wilson et al (125) modified
to predict conditionally sampled concentrations and intermittency.
Models of both these kinds predict fluctuations in the plume, but they can
be combined with meandering plume models to predict the overall fluctuation
Two kinds of advanced models have been developed: (a) mathematical
models that solve equations for fluctuation variance directly, such as the
Annual Reviews
Annu. Rev. Entomol. 1992.37:505-532. Downloaded from
by UNIVERSITY OF CALIFORNIA - RIVERSIDE LIBRARY on 11/11/08. For personal use only.
second-order closure models of Sykes and coworkers (113, 114), and (b)
statistical models, such as random walk models (107, 115, 116) or large eddy
simulations (85, 97). Second-order closure methods use more physically
defensible means of calculating the difficult cross-product terms in the con-
tinuity equation for c
Numerical simulations of dispersion calculate large
numbers of trajectories of particles or particle pairs. Single particle models
yield mean concentrations, but by considering particle separation in parti-
cle-pair models, one can also obtain a statistical description of concetra-
tion fluctuations. Once fully developed, numerical simulations should offer
significant new possibilities. By tracing paths of populations of particles,
they could provide a unique realization of the fluctuating concentrations
in a plume. A model of this kind could be used in combination with a
model of insects’ instantaneous response to odor to evaluate theories about
Mason (85) noted that the influence of large-scale motions responsible for
movements of the plume as a whole are often neglected in dispersion model-
ing. He proposed a model in which the movements of large eddies are
simulated by a numerical scheme and the small-scale motions are paramete-
rized. The model produced results that agreed well with experimental data but
was costly in computer resources.
Therefore, one has several options for modeling the distribution of fluctuat-
ing concentration in a plume. The concept of active space could be revived
with the mean concentration replaced with one or other measure of fluctuating
With an understanding of the generation and structure of odor plumes, we
may now ask what features of the plume are detected and used by insects in
walking and flying toward the odor’s source. This line of inquiry requires
knowledge of the mechanisms of insect orientation to odors (73) and the
sensory inputs that mediate them.
Insects produce and respond to odor plumes that differ markedly. Bark
beetles (Scolytidae) can generate a large plume with an effective source the
size of an entire tree trunk with hundreds of odor-emitting beetles. Arctiid
moths may release pheromone as both a vapor and an aerosol (76), and
because the pheromone gland is rhythmically protruded, the plume is released
as a series of pulses (30). Parasitoid wasps following a plume to a prospective
host may need to navigate a narrow plume of host-released kairomone within
a larger diffuse plume of odor from the plant harboring the host. Such
different plume structures certainly exert different selective pressures. Insects
Annual Reviews
Annu. Rev. Entomol. 1992.37:505-532. Downloaded from
by UNIVERSITY OF CALIFORNIA - RIVERSIDE LIBRARY on 11/11/08. For personal use only.
with differing phylogenies and therefore independently evolved orientation
mechanisms assure multiple solutions to the location of odor sources (21).
Most of our examples are of orientation to odors by flying insects, which
entails the difficult navigational task of detecting wind direction while air-
borne. Most of the evidence on this phenomenon has been accumulated on the
attraction of male moths to a female.
Search Strategies
Which attributes of the plume’s structure and an insect’s pattern of movement
with respect to the wind’s direction are important to the initial contact with an
odor source? Most insects actively move about before entering the plume, and
several modeling studies have asked: are there optimal strategies for contact-
ing the plume?
This approach requires assumptions about plume structure. Some have
suggested that the optimal strategy for the animal is to fly cross-wind if the
plume length exceeds its width and up or downwind otherwise (20, 83).
Sabelis & Schippers (106) provided mathematical expressions of this idea.
Dusenbery (38, 39) developed similar predictive models that incorporated the
cost of moving with or against the air stream. These treatments assumed
simplistic time-average plumes. For instance, plume width has been defined
by the range of the wind direction (106); when the range exceeds
, th e
plume is wider than long. However, the range of the wind direction depends
on the time interval over which it is measured (41). These studies further
assumed that the insect can estimate the mean (as opposed to instantaneous)
wind direction and can fly at some fixed angle with respect to it. They also
assumed that the insect would inevitably contact the plume and locate the
source once within the time-average boundaries. However, much evidence is
to the contrary (e.g. 43).
Few good field data are available on the actual search strategies employed
by flying insects before they contact odor plumes. Some studies suggest a
cross-wind tactic, whereas others do not (e.g. 40). A deficiency in many field
observations is the anecdotal estimation of the instantaneous orientations of
wind and the organism’s path. Realistic models of optimal foraging for a
plume await construction. Such models would incorporate improved repre-
sentations of plume size and shape based on isopleths that indicate the
probability of contacting the instantaneous plume at each point downwind of
the source (54, 102).
Orientation Toward the Source
odor in wind is a simple task because gauging the wind’s instantaneous
direction only requires mechanoreceptive input (11). Upwind flight, however,
Annual Reviews
Annu. Rev. Entomol. 1992.37:505-532. Downloaded from
by UNIVERSITY OF CALIFORNIA - RIVERSIDE LIBRARY on 11/11/08. For personal use only.
requires that the airborne organism determine the wind’s direction by judging
visually how wind has deflected the flyer from its heading. Although an
organism can estimate its airspeed by mechanoreceptive input, the optomotor
reaction (comparing the discrepancy between heading and track over ground
as visually detected drift or side slip) is the only verified mechanism by which
an airborne organism can reckon wind direction (72). When side slip is zero,
the organism is heading due up (or down) wind. The optomotor reaction
wind appears to explain nearly all cases of flying orientation to odor, except
an aim and shoot strategy in which insects determine upwind heading while
perched on a substrate and then fly in that direction (19).
HEADING TOWARD THE PLUME’S SOURCE The instantaneous wind direc-
tion and the plume’s axis are only infrequently coincident, so that simply
heading upwind while in the plume will not routinely lead toward the odor’s
source. In open fields, moving upwind while in a plume should aim the insect
toward the odor’s source, particularly within tens of meters (36), but because
the plume meanders, often an insect heading upwind will exit the plume. In
forests (43) and the African bush (18), the meandering plume can be over-
taken by large-scale eddies traveling in different directions, often producing
highly contorted plume paths. Flight upwind in such plumes often carries an
insect out of the plume. In addition, the upwind direction is only infrequently
aimed toward the odor’s source.
Both of these constraints imply that successful location of an odor over
distances of tens of meters requires a tactic to enable relocation of the plume
when upwind flight carries the insect beyond the plume’s boundary. Casting,
a reiterative zigzag that may progressively widen but does not progress
upwind, enhances the likelihood of directly recontacting the plume as the cast
widens or provides a station-keeping maneuver until the plume shifts back to
the insect’s position (35).
Successful flight toward the source may depend upon sustained (tens of
seconds) intervals when the wind vectors are parallel, such that the directions
along the plume’s axis and upwind are coincident. Such parallel wind fields
within which an insect may successfully follow a plume are more diagnostic
of active spaces than the mere concentration of odor (43). The effective
distance over which insects follow odor sources may be related to the domi-
nant frequency of the large-scale eddies responsible for the major wind shifts.
Models of such large-scale structure might be applicable to insect orientation
induced anemotaxis is the primary mechanism for locating a distant odor
source (74), various forms of chemotaxis have been hypothesized to play
Annual Reviews
Annu. Rev. Entomol. 1992.37:505-532. Downloaded from
by UNIVERSITY OF CALIFORNIA - RIVERSIDE LIBRARY on 11/11/08. For personal use only.
role in orientation. Such a navigational system might rely upon directional
cues extracted from the plume’s structure (such as the position of the bound-
ary) or temporally monitored changes in intensity of the plume’s signal. To
verify that factors other than the optomotor reaction to wind do contribute to
orientation, one experimental scheme is to impose windless conditions. In
their wind tunnel trials, Farkas & Shorey (48) observed that some moths
continued flight along the plume’s axis following a sudden cessation of the
wind. A similar response to a plume in zero wind was verified with Grapholi-
ta molesta (78) and Lymantria dispar (123). Longitudinal klinotaxis (sequen-
tial sampling of concentration) appeared to contribute to flight along the
plume in concert with an initial setting of course polarity, by optomotor
anemotaxis (78). In nature, this maneuver is potentially useful for distances
on the order of meters when wind speed falls below the level needed for the
optomotor reaction.
Wright (126) was the first to point out that the filamentous nature of odor
plumes might provide directional cues. He proposed that flying insects could
measure frequency of bursts to determine source’s direction, but he later
abandoned this idea in favor of anemotaxis (127). Moore & Atema (89)
proposed that lobsters (at relatively close range and in water) extract di-
rectional information from the changes in the rate of increase in concentration
at the leading edges of odor bursts. However, Murlis & Jones (91) concluded
from ion plume trials in the atmosphere that the rates of change in concentra-
tion or other plume characteristics with distance from the source are unreliable
indicators of direction to the source, unless, perhaps, the source was very
Detection of Fine-Scale Structure
Scrutiny of the flight tracks of G. molesta suggests that either contact or loss
of the plume can induce a change in course angle within 0.15 s (5). This
expressed as a surge upwind in pheromone and a tendency toward cross-wind
casting upon loss of the scent. In the silkworm Antheraea polyphemus,
casting latency following odor loss was not as rapid~.3-0.5 s (7). Not only
does the fine-scale structure of plumes alter behavior, it appears to be required
for sustained upwind flight. In wind-tunnel trials with two tortricid moths,
continued upwind progress did not occur in a homogeneous cloud of pher-
omone (75, 122). Rather, a plume or a naturally fenestrated (or
perimentally pulsed) cloud of odor was needed.
Baker et al (8) proposed that cessation of upwind progress, as occurs near
the pheromone source, might be triggered by fusing of the receptor output
when the receptor system could no longer resolve the individual odor pulses.
Annual Reviews
Annu. Rev. Entomol. 1992.37:505-532. Downloaded from
by UNIVERSITY OF CALIFORNIA - RIVERSIDE LIBRARY on 11/11/08. For personal use only.
Field and wind-tunnel electroantennogram (EAG) recordings of G. molesta
suggested that peak-to-mean ratio rather than either the peak or mean con-
centrations induced in-flight arrestment (6).
The ability of an organism to resolve individual bursts of odor and possibly
features of individual bursts rests on the characteristics of the odor-detection
system. In the case of attractant pheromones, odor molecules are thought
(110) to enter pores on the sensilla and bind to receptor proteins on the surface
of the sensory dendrites. The receptor cells themselves, at least in the case of
moth pheromones, appear to have a narrowly tuned specificity. Populations of
such receptors on the antenna respond to a single or perhaps two closely
related compounds in the blend. Other odor receptors many have broad tuning
(2). Graded potentials arising from the dendrite are transduced into trains
action potentials that propagate along the sensory axon to the central nervous
system. The frequency of such action potentials is proportional to the odor
intensity. The odor molecules are rapidly degraded by enzymes (119) so that
receptors are available to respond to a new odor pulse. The latency of axon
response is on the order of 100 ms.
Most receptors exhibit a phasic/tonic response to odor signals of a second
or more duration (71). A strong odor causes an initial high-frequency burst
action potentials (the phasic response), followed by a period of reduced
frequency (the tonic response) that persists for the duration of the odor pulse.
At low odor concentrations, only the tonic phase occurs.
In the receptor cells of A. polyphemus, the two receptor cell types (each
sensitive to a different pheromone component) have different response pat-
terns (105). One cell type can resolve 20-ms pulses at 5 or more stimuli/s,
while another can only distinguish 2 stimuli/s. Thus in A. polyphemus, the
structure of the odor plume might be registered in a different pattern by each
receptor type. At the level of the antenna as a whole, much of the small-scale
plume structure in a plume is reflected in EAG signals (92) (Figure
Central olfactory neurons can detect the pheromone plume’s discontinuous
structure. Christensen & Hildebrand (28) recorded from higher-order in-
terneurones in the deutocerebrum of Manduca sexta and found that pulses
were resolved up to 10/s.
Changes in Behavior as Insects Approach the Source
As flying insects approach an odor source, their flight patterns typically
change, and ultimately landing may ensue. The site of landing may be close to
the odor source, or if the effective source size is large, as occurs for example
when a female gypsy moth calls from a tree trunk, landing may be many
centimeters away from the source. The insect locates the source following
walking orientation (26). Of particular interest are the alterations in the
Annual Reviews
Annu. Rev. Entomol. 1992.37:505-532. Downloaded from
by UNIVERSITY OF CALIFORNIA - RIVERSIDE LIBRARY on 11/11/08. For personal use only.
plume’s signal that influence these behavioral changes. In moths, a typical
change seen in wind-tunnel trials employing lures of varying strength is a
reduction in ground speed and a decrease in the width of the zigzag path, both
of which can be attributed to increases in pheromone concentration because
the plume’s dimensions are ,essentially unaltered (25, 49, 50, 77, 84a). Farkas
et al (50) found that landing was induced by a high concentration of pher-
omone accompanied by an appropriate visual stimulus.
Pheromone Blends and the Role of Indi~vidual Components
Most attractant pheromones (indeed most behaviorally relevant odors) consist
of blends. Much debate has focused on the possibility that in moths a portion
of the blend mediates the earliest behaviors in source location (such as locking
on to a plume), whereas the entire mixture (or the natural component ratio)
evokes later behaviors (such as landing or courtship) (24, cf. 82).
alternative system envisages the complete blend evoking all behaviors at a
lower threshold than partial blends (82). Although several behavioral studies
suggest such a nested active-space system (16, 23, 95, 96), the blends tested
have not always been complete (23), and alternative behavioral interpretations
exist. The complete blend has the lowest concentration threshold for eliciting
all behaviors observed in Argyrotaenia velutinana, Tricoplusia ni, and G.
molesta (80, 81) and thus the entire blend governed the active space.
Support for the potential differences in active spaces of partial and com-
plete blends comes from a study of receptors for Antheraea polyphemus and
Antheraea pernyi. The relative number of cells sensitive to each component of
the pheromone, their relative sensitivity, and the release rates of components
from the female suggest that the active space of the major component could
project farther downwind than that of the blend (87). The dynamics of the
receptor’s response to different components in the blend also play a part. The
rapid cells in A. polyphemus antennae (105) are highly responsive at low
concentrations but would soon be overwhelmed by high stimulus levels. The
slower ceils are less sensitive to low concentrations but continue to function at
high stimulus levels. There may be no generic solution to how moths organize
a system of thresholds for blends.
Considerable progress has been made in understanding the fine-scale structure
of odor plumes and consequently in unraveling the information available to
insects. Odor signals consist of short bursts of odor of greatly varying
intensity. Farther from the source, bursts are on average weaker (but not
reliably so). They are also slightly longer and there is a lengthened gap
between them. Overall, the intermittency of the signal is higher at greater
Annual Reviews
Annu. Rev. Entomol. 1992.37:505-532. Downloaded from
by UNIVERSITY OF CALIFORNIA - RIVERSIDE LIBRARY on 11/11/08. For personal use only.
distances from the odor source. Nearer the source, the peak-to-mean ratio is
considerably reduced, but the rate at which the odor concentration rises at the
leading edge of the bursts increases. How much of this rich array of informa-
tion is used in orientation remains to be established.
Insects can resolve odor bursts at least as fast as 10/s in the peripheral
receptors and in interneurons farther up the afferent pathway. The kind of
information derived from these patterns is not yet clear.
Progress in modeling plumes has also been substantial and a variety of
approaches have emerged to deal with fluctuating concentration and its time
dependency. These models have application in the studies of active spaces.
Determinisitic plume models based on numerical simulation of diffusion
could improve our understanding of orientation strategies.
There is promise in applying advanced plume models and field measure-
ments of simulated plumes in different habitats to define the temporal and
spatial cues available in odor plumes. In turn, such information will engender
the behavioral and neurophysiological experiments requisite to explain the
navigational mechanisms used by insects to locate odor sources.
We thank C. Boettner for editorial assistance and D. Hall, C. Jones, L.
Kuenen, K. Mylne, and D. Ride for helpful discussions. Our studies have
been supported by USDA competitive grants.
Literature Cited
Allen, L. H. 1968. Turbulence and
wind-speed spectra within a Japanese
larch plantation. J. Appl. Meteorol.
Atema, J. 1988. Distribution of chemi-
cal stimuli. In Sensory Biology of Aqua-
tic Animals, ed. J. Atema, R. R. Fay, A.
N. Popper, W. N. Tavolga, pp. 29-56.
NY: Springer-Verlag
Aylor, D. E. 1976. Estimating peak con-
centrations of pheromones in the forest.
In Perspectives in Forest Entomology,
ed. J. F. Anderson, M. K. Kaya, pp.
177-88. New York: Academic
4. Aylor, D. E., Parlange, J.-Y., Grannett,
J. 1976. Turbulent dispersion of dis-
parlure in the forest and male gypsy
moth response. Environ. Entomol.
5. Baker, T. C., Haynes, K. F. 1987. Ma-
neuvers used by flying male oriental
fruit moths to relocate a sex-pheromone
plume in an experimentally shifted
wind-field. Physiol. Entornol. 12:263-
6. Baker, T. C., Haynes, K. F. 1989. Field
and laboratory electroantennographic
measurements of pheromone plume
structure correlated with oriental fruit
moth behavior. Physiol. Entomol. 14:1-
7. Baker, T. C., Vogt, R. G. 1988. Mea-
sured behavioral latency in response to
sex-pheromone loss in the large silk
moth Antheraea polyphemus. J. Exp.
Biol. 137:29-38
8. Baker, T. C., Willis, M. A., Haynes, K.
F., Phelan, P. L. 1985. A pulsed cloud
of pheromone elicits upwind flight in"
male moths. Physiol. Entomol. 10:257-
9. Bartell, R. J. 1982. Mechanism of com-
munication disruption by pheromone in
the control of Lepidoptera: a review.
Physiol. Entomol. 7:353-64
10. Barynin, J. A. M., Wilson, M. J. G.
1972. Outdoor experiments on smell_
Atmos. Environ. 6:197-207
11. Bell, W. J. 1984. Chemo-orientation in
walking insects. See Ref. 1 la, pp. 93-
lla. Bell, W. J., Cardr, R. T., eds. 1984.
Annual Reviews
Annu. Rev. Entomol. 1992.37:505-532. Downloaded from
by UNIVERSITY OF CALIFORNIA - RIVERSIDE LIBRARY on 11/11/08. For personal use only.
Chemical Ecology of Insects, London/
New York: Chapman and Hall. 524 pp.
12. Bendat, J. S., Piersol, A. G. 1980.
Engineering Applications of Correlation
and Spectral Analysis. New York: Wiley
Boreham, B. W., Harvey, J. K. 1984.
Dipolar charged particles as markers for
dispersion simulation in wind tunnels. J.
Phys. E: 17:994-98
14. Boreham, B. W., Jones, C. D., Grif-
fiths, R. F. 1985. Field evaluation of a
new Langmuir-type probe for atmo-
spheric dispersion studies using ions as a
tracer. J. Phys. E: 18:218-23
15. Bossert, W. H., Wilson, E. O. 1963.
The analysis of olfactory communication
among animals. J. Theor. Biol. 5:443-
16. Bradshaw, J. W. S., Baker, R., Lisk, J.
C. 1983. Separate orientation and releas-
er components in a sex pheromone. Na-
ture 304:265-67
17. Bradshaw, J. W. S., Ellis, N. W., Hand,
S. C., Stoakley, J. T. 1989. Interaction
between pheromone traps with different
strength lures for the pine beauty moth. J.
Chem. Ecol. 15:2485-94
18. Brady, J., Gibson, G., Packer, M. J.
1989. Odour movement, wind direction,
and the problem of host-finding by tsetse
flies. Physiol. Entomol. 14:369-80
19. Bursell, E. 1984. Observations on the
orientation of tsetse flies (Glossina palli-
dipes) to wind-borne odours. Physiol.
Entomol. 9:133-37
Card6, R. T. 1981. Precopulatory sexual
behavior of the adult gypsy moth. In The
Gypsy Moth: Research Toward Inte-
grated Pest Management. Technical
Bulletin 1584. ed. C. C. Doane, M. L.
McManus, pp. 572-87. Washington,
DC: US Dept. of Agriculture
21. Cardr, R. T. 1986. Epilogue: be-
havioural mechanisms. See Ref. 100,
pp. 175-86
22. Cardd, R. T. 1990. Principlcs of mating
disruption. See Ref. 103, pp. 42-71
23. Cardr, R. T., Baker, T., Roelofs, W.
1975. Ethological function of com-
ponents of a sex attractant system for
Oriental fruit moth males, Grapholitha
molesta. J. Chem. Ecol. 1:475-91
24. Card6, R. T., Charlton, R. E. 1984.
Olfactory sexual communication in
Lepidoptera: strategy, sensitivity, and
selectivity. In Insect Communication,
ed. T. Lewis, pp. 241-64. New York:
25. Card6, R. T., Hagaman, T. E. 1979.
Behavioral responses of the gypsy moth
in a wind-tunnel to air-borne enantiom-
ers of disparlure. Environ. Entomol.
26. Charlton, R. E., Card6, R. T. 1990.
Orientation of male gypsy moths,
Lymantria dispar (L.), to pheromone
sources: the role of olfactory and visual
cues. J. Insect Behav. 3:443-69
27. Chatwin, P. C. 1985. Interaction Be-
tween Turbulent and Electrostatic
Effects in the Atmospheric Dispersion of
Electrically Charged Tracers. Report
PTA/AZ/209. Porton Down: Chemical
Defense Establishment
28. Christensen, T. A., Hildebrand, J. G.
1987. Male-specific, sex pheromone-
selective projection neurons in the an-
tennal lobes of the moth Manduca sexta.
J. Comp. PhysioL A 160:553-69
29. Clarke, R. H. 1979. A model for short
and medium range dispersion of
radionuclides released to the atmos-
phere. Harwell, UK: National Radiolog-
ical Protection Board Report NRPB-R91
30. Conner, W. E., Eisner, T., Vander
Meet, R. K., Gue~rero, A., Ghiringelli,
D., Meinwald, J. 1980. Sex attractant of
an arctiid moth (Utethesia ornatrix):
pulsed signal. Behav. Ecol. Sociobiol.
31. Cramer, H. E., Record, F. A.,
Vaughan, H. C. 1958. The study of the
diffusion of gasses in the lower atmos-
phere. Cambridge, MA: MIT Depart-
ment of Meteorology, Final Report un-
der Contract AF 19(604)-1058
Critchley, B., Campion, D. G.,
McVeigh, L. J. 1989. Pheromone Con-
trol in the Integrated Pest Management
of Cotton. In Pest Management of Cot-
ton, ed. M. V. Green, D. J. deB. Lyon,
pp. 83-92. Chichester: Horwood
Csanady, G. T. 1967. Concentration
fluctuations in turbulent diffusion. J.
Atmos. Sci. 24:21-28
34. Csanady, G. T. 1973. Turbulent Diffu-
sion in the Environment. Dordrecht,
Holland: Reidel
35. David, C. T., Kennedy, J. S., Ludlow,
A. R. 1983. Finding of a sex pheromone
source by gypsy moths released in the
field. Nature 303:804-6
David, C. T., Kennedy, I. S., Ludlow,
A. R., Perry, J. N., Wall, C. 1982. A
reappraisal of insect flight towards a dis-
tant, point source of wind-borne odor. J.
Chem. Ecol. 8:1207-15
37. Dinar, N., Kaplan, H., Kleiman, M.
1988. Characterization of concentration
fluctuations of a surface plume in a neu-
tral boundary layer. Boundary-Layer
Meteorol. 45:157-75
38. Dusenbery, D. B. 1989. Optimal search
direction for an animal flying or swim-
ming in a wind or current. J. Chem.
Ecol. 15:2511-19
Annual Reviews
Annu. Rev. Entomol. 1992.37:505-532. Downloaded from
by UNIVERSITY OF CALIFORNIA - RIVERSIDE LIBRARY on 11/11/08. For personal use only.
Dusenbery, D. B. 1990. Upwind search-
ing for an odor plume is sometimes op-
timal. J. Chem. Ecol. 16:1971-76
40. Elkinton, J. S., Cardr, R. T. 1983.
Appetitive flight behavior of male gypsy
moths (Lepidoptera: Lymantriidae). En-
viron. Entomol. 12:1702-7
41. Elkinton, J. S., Card6 R. T. 1984. Odor
dispersion. See Ref. lla, pp. 73-91
42. Elkinton, J. S., Cardr, R. T., Mason, C.
J. 1984. Evaluation of time-average dis-
persion models for estimating pher-
omone concentration in a deciduous for-
est. J. Chem. Ecol. 10:1081-1108
Elkinton, J. S., Schal, C. S., Cardr, R.
T. 1987. Pheromone puff trajectory and
the upwind flight of male gypsy moths in
a forest. Physiol. Entomol. 12:399-406
44. Fackrell, J. E. 1978. A system for turbu-
lent concentration measurement. J.
Phys. E: 11:1015-22
45. Fackrell, J. E. 1989. A flame ionization
detector for measuring fluctuating con-
centration. J. Phys. E: 13:888-93
46. Fackrell, J. E., Robins, A. G. 1982.
Concentration fluctuations and fluxes in
plumes from point sources in a turbulent
boundary layer. J. Fluid Mech. 117:1-
47. Fackrell, J. E., Robins, A. G. 1982. The
effect of source size on concentration
fluctuations in plumes. Boundary-Layer
Meteorol. 22:335-50
48. Farkas, S. R., Shorey, H. H. 1972.
Chemical trail following by flying in-
sects: a mechanism for orientation to a
distant odor source. Science 178:67-68
49. Farkas, S. R., Shorey, H. H. 1974.
Mechanisms of orientation to a distant
pheromone source. In Pheromones, Part
5, ed. M. C. Birch, pp. 81-95. Amster-
dam: North-Holland Research Mono-
Farkas, S. R., Shorey, H. H., Gaston,
L. K. 1974. Sex pheromones of Lepi-
doptera. Influence of pheromone con-
centration and visual clues on aerial
odour trail following by male of Pecti-
nophora gossypiella. Ann. Entomol.
Soc. Am. 67:633-38
51. Gifford, F. A. 1959. Statistical proper-
ties of a fluctuating plume dispersion
model. In Advances in Geophysics, Vol.
6, Atmospheric Diffusion and Air Pollu-
tion, ed. F. N. Frankiel, R. A. Shep-
pard, p. 117. New York: Academic
52. Gifford, F. A. 1974. The form of the
frequency distribution of air pollution
concentrations. Proc. Symp. on Statis-
tical Aspects of Air Quality Data. Report
650/4-74-038. Washington DC: USA
Environmental Protection Agency
53. Hadjitoffi, A., Wilson, M. J. G. 1979.
Fast response measurement of air pollu-
tion, Atmos. Environ. 13:755-60
Hanna, S. R. 1984. Concentration fluc-
tuations in a smoke plume. Atmos. En-
viron. 18:1091-1106
55. Hanna, S. R. 1984. The exponential
PDF and concentration fluctuations in
smoke plumes. Boundary-Layer
Meteorol. 29:361-75
Hanna, S. R. 1986. Spectra of concen-
tration fluctuations: the two time scales
of a meandering plume. Atrnos. Environ.
57. Hanna, S. R., Briggs, G. A., Hosker,
R. P. 1982. Handbook on Atmospheric
Diffusion. Washington DC: Technical
Information Center. US Dept. of Energy
58. Hanna, S. R., Insley, E. M. 1989. Time
series analyses of concentration and
wind fluctuations. Boundary-Layer
Meteorol. 47;131~47
Haugen, D. A. 1959. Project Prarie
Grass, a Field Programme in Diffusion.
Vol. 3, Geographical Research Paper
No. 59. Bedford, MA: G.R.D.A.F.C.
Hay, J. S., Pasquill, F. 1959. Diffusion
from a continuous source in relation to
the spectrum and scale of turbulence.
Adv. Geophys. 6:345
61. Hrgstrrm, U. 1972. A method for pre-
dicting odour frequencies from a point
source. Atmos. Environ. 6:103-21
62. Hunt, J. C. R., Holroyd, R. J., Carruth-
ers, D. J., Robbins, A. G., Apsley, D.
D., et al. 1990. Developments in mod-
elling air pollution for regulatory uses.
Proc. 18th NATO-CCMS Int. Conf. Air
Pollution Modelling and Its Application,
Vancouver, Canada
63. International Atomic Energy Agency.
1980. Atmopheric dispersion in nuclear
power plant siting, a safety guide. IAEA
Safefy Guides No. 50-SG-$3. Vienna:
Jones, C. D. 1977. Ion concentration
variations at short distances downwind
of continuous and quasi-continuous
point sources. Pestic. Sci. 8:84-95
65. Jones, C. D. 1978. Statistics of the con-
centration fluctuations in short range
atmospheric dispersion. In Proc. Conf.
Mathematical Modelling of Turbulent
Diffusion in the Environment, Liverpool,
ed. C. J. Harris, pp. 277-98. London:
66. Jones, C. D. 1983. On the structure of
instantaneous plumes in the atmosphere.
J. Hazard. Mater. 7:87-112
67. Jones, C. D., Griffiths, R. F. 1984. Full
scale experiments on dispersion round
an isolated building using an ionized air
.tracer technique with very short averag-
ing time. Atmos. Environ. 18:903-16
Annual Reviews
Annu. Rev. Entomol. 1992.37:505-532. Downloaded from
by UNIVERSITY OF CALIFORNIA - RIVERSIDE LIBRARY on 11/11/08. For personal use only.
68. Jones, C. D., Gulliford, N. T. 1979.
Developments in the use of ionized air as
a wind tunnel tracer. J. Phys. E:
69. Jones, J. I. P. 1963. A band-pass filter
technique for recording atmospheric
turbulence. Br. J. Appl. Phys. 14:95-
70. Justum, A. R., Gordon, R. F. S., eds.
1989. Insect Pheromones in Plant Pro-
tection. New York: Wiley
71. Kaissling, K. E. 1986. Chemo-electrical
transduction in insect olfactory recep-
tors. Annu. Rev. Neurosci. 9:121-45
72. Kennedy, J. S. 1940. The visual re-
sponse of flying mosquitoes. Proc.
Zool. Soc. London Ser. A 109:221-42
Kennedy, J. S. 1986. Some current
issues in orientation to odour sources.
See Ref. 100, pp. 11-25
74. Kennedy, J. S. 1977. Olfactory re-
sponses to distant plants and other odor
sources. In Chemical Control of Insect
Behavior: Theory and Application, ed.
H. H. Shorey, J. J. McKelvey, Jr., pp.
76-91. New York: Wiley Interscience.
414 pp.
75. Kennedy, J. S., Ludlow, A. R., Sand-
ers, C. J. 1981 Guidance of flying male
moths by wind-borne sex pheromone.
Physiol. Entomol. 6:395-412
Krasnoff, S. B., Roelofs, W. L. 1988.
Sex pheromone released as an aerosol by
the moth Pyrrharctia isabella. Nature
Kuenen, L. P. S., Baker, T. C. 1982.
Optomotor regulation of ground velocity
in moths during flight to sex pheromone
at different heights. Physiol. Entomol.
78. Kuenen, L. P. S., Baker, T. C. 1983. A
non-anemotactic mechanism used in
pheromone source location by flying
moths. Physiol. Entomol. 8:277-89
79. Lewis, T., Macauley, E. D. M. 1976.
Design and elevation of sex attractant
traps for pea moth Cydia nigricana
(Steph.) and the effect of plume shape
on catches. Ecol. Entomol. 1:175-87
80. Linn, C. E., Campbell, M. G., Roelofs,
W. L. 1986. Male moth sensitivity to
multicomponent pheromones: critical
role of female-released blend in de-
termining the functional role of com-
ponents and active space of the pher-
omone. J. Chem. Ecol. 12:659-68
81. Linn, C. E., Campbell, M. G., Roelofs,
W. L. 1987. Pheromone components
and active spaces: what do moths smell
and where do they smell it? Science
82. Linn, C. E., Roelofs, W. L. 1989.
Response specificity of male moths to
multicomponent pheromones. Chem.
Senses 14:421-37
83. Linsenmair, K. E. 1969. Anemomeno-
taktische orientierung bei Tenebrioniden
und Mistkafem (Insecta, Coleoptera).
Vgl. Physiol. 64:154-211
84. Lumley, J. L., Panofsky, H. A. 1964.
The Structure of Atmospheric Turbu-
lence. New York: Wiley
84a. Marsh, D., Kennedy, J. S., Ludlow,
A. R. 1978. An analysis of anemotactic
zigzagging flight in male moths stimu-
lated by pheromone. Physiol. Entomol.
Mason, P. J. 1989. Large eddy simula-
tion of the convective boundary layer. J.
Atmos. Sci. 46:1492-1516
86. McNeil, J. N. 1991. Behavioral ecology
of pheromone-mediated communication
in moths and its importance in the use of
pheromone traps. Annu. Rev. Entomol.
36:407-30 "
Meng, L. Z., Wu, C. H., Wicklein, M.,
Kaissling, K.-E., Bestmann, H. J. 1989.
Number and sensitivity of three types of
pheromone receptor cells in Antheraea
pernyi and A. p~,plremous. .¢. Com1~
Phvsi,oL A 165:139....46
Mi’ksad, R. W., Kittredge, J. 1979.
Pheromone aerial dispersion: a filament
model. In 14th Conf. Agriculture and
Forest Meteorology, pp. 238...43. Bos-
ton, MA: Am. Meteorol. Soc.
89. Moore, P., Atema, J. 1988. A model of
a temporal filter in chemoreception to
extract directional information from a
turbulent odour plume. Biol. Bull.
90. Murlis, J. 1986. The structure of odour
plumes. See Ref. 100, pp. 27-38
91. Murlis, J., Jones, C. D. 1981. Fine-
scale structure of odour plumes in rela-
tion to insect orientation to distant pher-
omone and other attractant sources.
Physiol Entomol. 6:71-86
Murlis, J., Willis, M. A., Cardr, R. T.
1990. Odour signals: patterns in time
and space. Proc. lOth Int. Syrup. Olfac-
tion and Taste, Oslo, ed. K. B. DCving,
pp. 6-17. Oslo: Univ. Oslo
Mylne, K. R. 1988. Experimental
measurements of concentration fluctua-
tions. See Ref. l17a, 7:555-65
94. Mylne, K. R., Mason, P. J. 1991. Con-
centration fluctuation measurements in
a dispersing plume at a range of up to
100 m. Q. J. R. Meteorol. Soc.
95. Nakamura, K. 1979. Effect of the minor
component of the sex pheromone on the
male orientation to pheromone source in
Annual Reviews
Annu. Rev. Entomol. 1992.37:505-532. Downloaded from
by UNIVERSITY OF CALIFORNIA - RIVERSIDE LIBRARY on 11/11/08. For personal use only.
Spodoptera litura (F.). Chem. Rev. In-
sect Russia 4:153-56
96. Nakamura, K., Kawasaki, F. 1977. The
active space of Spodoptera litura (F.)
sex pheromone and the pheromone com-
ponent determining this space. Appl. En-
tomol. Zool. 12:162-77
97. Nieuwstadt, T. F. M.,’ van Harem, L.
1988. A large eddy simulation of buoy-
ant and non-buoyant plume dispersion in
the convective atmospheric boundary
layer. See Ref. l17a, 6:525-38
98. Pasquill, F. 1961. The estimation of the
dispersion of windborne material. Mete-
orol. Mag. 90:33-49
Pasquill, F., Smith, F. B. 1983. Atmo-
spheric Diffusion. Chichester: Horwood.
3rd ed.
100. Payne, T. L., Birch, M. C., Kennedy,
C. E. J., eds. 1986. Mechanisms in In-
sect Olfaction. Oxford: Clarendon. 364
Ramsdell, J. V., Hinds, W. T. 1979.
Concentration fluctuations and peak-to-
mean concentration ratios in plumes
from a ground level source. Atmos. En-
viron. 5:483-95
102. Ride, D. J. 1988. A model for the
observed intermittency of a meander-
ing plume. J. Hazard. Mater. 19:131-
Ridgway, R. L., Silverstein, R. M., Ins-
coe, M. N., eds. 1990. Behavior-
Modifying Chemicals for Insect Man-
agement. New York: Dekker. 761 pp.
104. Roberts, O. F. T. 1923. The theoretical
scattering of smoke in turbulent atmo-
sphere. Proc. R. Soc. London Ser. A
105. Rumbo, R., Kaissling, K. E. 1989.
Temporal resolution of odour pulses by
three types of pheromone receptor cells
in Antheraea polyphemus. J. Comp.
Physiol. A 165:281-91
106. Sabelis, M. W., Schippers, P. 1984.
Variable wind directions and anemotac-
tic strategies of searching for an odour
plume. Oecologia 63:225-28
107. Sawford, B. L. 1983. The effect of
Gaussian particle-pair distribution func-
tions in the statistical theory of concen-
tration fluctuations in homogeneous
turbulence. Q. J. R. Meteorol. Soc.
108. Sawford, B. L. 1987. Conditional con-
centration statistics for surface plumes in
the atmospheric boundary layer. Bound-
ary-Layer Meteorol. 38:209-23
109. Smith, F. B. 1990. Proceeding of
Course on Atmospheric Dispersion.
Trieste, Italy: Int. Centre for Theoretical
110. Steinbrecht, R. A. 1987. Functional
morphology of pheromone-sensitive
sensilla. In Pheromone Biochemistry,
ed. G. D. Prestwich, G. L. Blomquist,
pp. 353-84. New York: Academic
111. Storeb~, P. B., Bjorvatten, T., H~n-
nashagen, K., Lillegraven, A., Jones,
C. D., van Buijtenen, C. J. P. 1983.
Tracer experiments with turbulently dis-
persed air ions. Boundary-Layer Meteo-
rol. 26:127-39
112. Sutton, O. G. 1953. Micrometeorology.
New York: McGraw-Hill
113. Sykes, R. I. 1988. Concentration fluc-
tuations in dispersing plumes. In Lec-
tures on Air Pollution Modelling, ed. A.
Venkatram, J. C. Wyngaard, pp. 325-
56. Boston: Am. Meteorol. Soc.
Sykes, R. I., Lewellen, W. S., Parker,
S. F. 1984. A turbulent transport model
for concentration fluctuations and
fluxes. J. Fluid Mech. 139:193-218
115. Thompson, D. J. 1986. A random walk
model of dispersion in turbulent flows
and its application to dispersion in a val-
ley. Q. J. R. Meteorol. Soc. 112:511-
116. Thompson, D. J. 1990. A stochastic
model for the motion of particle pairs in
isotropic high Reynolds number turbu-
lence and its application to the problem
of concentration variance. J. Fluid
Mech. 210:113-53
117. Turner, D. B. 1970. Workbook of Atmo-
spheric Dispersion Estimates. Washing-
ton DC: Office of Air Programs, Pub.
No. AP-26, US Environ. Prot. Agency
l17a. van Dop, H., ed. 1988. Air Pollution
Modelling and Its Applications. Ncw
York: Plenum
118. Venkatacham, A. 1983. On dispersion
in the convective boundary layer.
Atmos. Environ. 17:5529-33
119. Vogt, R. G., Riddiford, L. M. 1986.
Pheromone reception: a kinetic equilibri-
um. See Ref. 100, pp. 201-8
120. Wall, C. Sturgeon, D. M., Greenway,
A. R., Perry, J. N. 1981. Contamination
of vegetation with synthetic sex-attrac-
tant released from traps for the pea moth
Cydia nigricana (F.). Entomol. Exp.
Appl. 30:111-15
121. Wang, Y. 1989. Turbulence structure,
momentum and heat transport in the
edge of broad leaf tree stands. Technical
Report US Forest Service
122. Willis, M. A., Baker, T. C. 1984.
Effects of intermittent and continuous
pheromone stimulation on the flight be-
havior of the oriental fruit moth, Graph-
olita molesta. Physiol. Entomol. 9:341-
Annual Reviews
Annu. Rev. Entomol. 1992.37:505-532. Downloaded from
by UNIVERSITY OF CALIFORNIA - RIVERSIDE LIBRARY on 11/11/08. For personal use only.
123. Willis, M. A., Card6, R. T. 1990. Pher-
omone-modulated optomotor response
in male gypsy moths, Lymantria dispar
L.: upwind flight in a pheromone plume
in different wind velocities. J. Comp.
Physiol. A. 167:699-706
124. Wilson, D. J., Robins, A. G., Fackrell,
J. E. 1982. Predicting the spatial dis-
tribution of concentration fluctuation
from a ground level source. Atmos. En-
viron. 16:497-504
125. Wilson, D. J., Robins, A. G., Fackrell,
J. E. 1985. Intermittency and con-
ditionally averaged concentration fluctu-
ation statistics in plumes. Atmos. En-
viron. 19:1053-64
126. Wright, R.H. 1958. The olfactory guid-
ance of flying insects. Can. Entomol.
127. Wright, R.H. 1964. The Science of
Smell. London: Allen and Unwin
Annual Reviews
Annu. Rev. Entomol. 1992.37:505-532. Downloaded from
by UNIVERSITY OF CALIFORNIA - RIVERSIDE LIBRARY on 11/11/08. For personal use only.
Annu. Rev. Entomol. 1992.37:505-532. Downloaded from
by UNIVERSITY OF CALIFORNIA - RIVERSIDE LIBRARY on 11/11/08. For personal use only.
... To understand the inherent difficulties of robotic gas sensing missions, it is important to become acquainted with the underlying gas plume dispersion. The phenomenon of gas dispersion in the air is characterized by a mixture of the diffusion of the molecules that move away from the source, even when no airflow is present, and the advection of particles due to external airflow [29]. A wind field causes the gas to create a trail, which characterizes the plume dispersion in most real environments. ...
... Laminar wind causes the plume to become wider and less concentrated away from the source, while a turbulent flow does not allow the maintenance of a well-characterized shape [30]. Even though gas dispersion can be modeled using Gaussian distributions [29], this simplification does not capture the patchy nature of the plume filament. ...
Full-text available
Within the scope of the ongoing efforts to fight climate change, the application of multi-robot systems to environmental mapping and monitoring missions is a prominent approach aimed at increasing exploration efficiency. However, the application of such systems to gas sensing missions has yet to be extensively explored and presents some unique challenges, mainly due to the hard-to-sense and expensive-to-model nature of gas dispersion. For this paper, we explored the application of a multi-robot system composed of rotary-winged nano aerial vehicles to a gas sensing mission. We qualitatively and quantitatively analyzed the interference between different robots and the effect on their sensing performance. We then assessed this effect, by deploying several algorithms for 3D gas sensing with increasing levels of coordination in a state-of-the-art wind tunnel facility. The results show that multi-robot gas sensing missions can be robust against documented interference and degradation in their sensing performance. We additionally highlight the competitiveness of multi-robot strategies in gas source location performance with tight mission time constraints.
... The human olfactory system detects various gas molecules in the air, leading us to perceive different odors [1,2]. Generally, what we refer to as 'smelling' is actually the detection of specific gaseous substances associated with an odor. ...
Full-text available
Odor information fills every corner of our lives yet obtaining its spatiotemporal distribution is a difficult challenge. Localized surface plasmon resonance has shown good sensitivity and a high response/recovery speed in odor sensing and converts chemical information such as odor information into optical information, which can be captured by charge-coupled device cameras. This suggests that the utilization of localized surface plasmon resonance has great potential in two-dimensional odor trace visualization. In this study, we developed a two-dimensional imaging system based on backside scattering from a localized surface plasmon resonance substrate to visualize odor traces, providing an intuitive representation of the spatiotemporal distribution of odor, and evaluated the performance of the system. In comparative experiments, we observed distinct differences between odor traces and disturbances caused by environmental factors in differential images. In addition, we noted changes in intensity at positions corresponding to the odor traces. Furthermore, for indoor experiments, we developed a method of finding the optimal capture time by comparing changes in differential images relative to the shape of the original odor trace. This method is expected to assist in the collection of spatial information of unknown odor traces in future research.
... 1992Riffell et al. 2008;Buehlmann et al. 2020). They are detected by insect antennae and extremities of the maxillary or labial palps that contain many chemoreceptors (Schoonhoven 1977). ...
... In still air the odour gradient would be caused by molecular diffusion and, if this mechanism applies to mosquito orientation, it most likely occurs only very near (within decimetres or less) of a host, particularly indoors on a still night when the inhabitants of the dwelling are asleep. Such a gradient also can exist within a wind-formed odour plume, but changes in concentration are steep enough only within decimetres of the source to be of potential value in orientation (Murlis et al. 1992), and this mechanism would be redundant to simply using anemotaxis. ...
... An odour plume consists of a series of filaments, which can be considered as strands of higher concentrations of odour, that are formed by turbulent mixing. Recording VOC concentrations at any one stationary point in a plume as it moves and shifts with air movements would reveal bursts of high concentration as a filament is encountered, in-between absences of VOC, defined as the intermittency of the plume (J Murlis et al., 1992). It is this intermittency of signal that an insect's antennae encounters as it attempts to navigate upwind through to the plume's source. ...
The critical ecological process of animal-mediated pollination is commonly facilitated by odour cues. These odours consist of volatile organic compounds (VOCs), often with short chemical lifetimes, which form the strong concentration gradients necessary for pollinating insects to locate a flower. Atmospheric oxidants, including ozone pollution, may react with and chemically alter these VOCs, impairing the ability of pollinators to locate a flower, and therefore the pollen and nectar on which they feed. However, there is limited mechanistic empirical evidence to explain these processes within an odour plume at temporal and spatial scales relevant to insect navigation and olfaction. We investigated the impact of ozone pollution and turbulent mixing on the fate of four model floral VOCs within odour plumes using a series of controlled experiments in a large wind tunnel. Average rates of chemical degradation of α-terpinene, β-caryophyllene and 6-methyl-5-hepten-2-one were slightly faster than predicted by literature rate constants, but mostly within uncertainty bounds. Mixing reduced reaction rates by 8-10% in the first 2 m following release. Reaction rates also varied across the plumes, being fastest at plume edges where VOCs and ozone mixed most efficiently and slowest at plume centres. Honeybees were trained to learn a four VOC blend equivalent to the plume released at the wind tunnel source. When subsequently presented with an odour blend representative of that observed 6 m from the source at the centre of the plume, 52% of honeybees recognised the odour, decreasing to 38% at 12 m. When presented with the more degraded blend from the plume edge, recognition decreased to 32% and 10% at 6 and 12 m respectively. Our findings highlight a mechanism by which anthropogenic pollutants can disrupt the VOC cues used in plant-pollinator interactions, which likely impacts on other critical odour-mediated behaviours such as mate attraction.
... In the soil matrix, these distinctions apply variably at fine spatial scales, creating a complex chemosensory environment that must be navigated by soil-dwelling organisms (Figure 1). Olfactory semiochemicals aboveground primarily move through air where temperature, atmospheric pressure, and turbulence can affect compound diffusion and mobility, although solid or liquid substrates (plant surfaces and aerosols) can influence movement of these compounds (Murlis et al., 1992;Vickers, 2006). Adapting to this, terrestrial organisms have developed behaviors that enable them to navigate volatile gradients or heterogeneous plumes (Vickers, 2000). ...
Full-text available
Chemical signaling underpins behavioral interactions among organisms in the soil. Understanding chemical communication in the soil requires a paradigm shift in methodology and perspectives compared to aboveground ecosystems because olfaction and gustation, accepted modalities of chemosensation aboveground, may not accurately represent chemical communication in the soil. To fully understand chemical communication in the soil, it is essential to consider how soil properties, such as moisture, pH, and adsorption, affect the transport and perception of semiochemicals. De-anthropomorphizing the study of chemosensation can avoid potential biases, particularly in soil systems, where distinctions between olfaction and gustation are confounded by the heterogeneity of the soil environment and its effects on the mobility of chemical signals. In this perspective, we first explore how soil heterogeneity confounds the dichotomy between olfaction and gustation with hypothetical but ecologically relevant examples. Then we examine how anthropomorphic biases in aboveground chemical ecology have influenced soil chemical ecology. Our examples and discussion are prepared primarily in reference to soil arthropods. We conclude by discussing seven future research directions and outstanding questions. The soil is a premier example of a system where investigators should avoid anthropomorphisms when studying behavioral and chemical ecology. Research in soil chemical ecology should further efforts towards developing a unified view of chemosensation that could apply to all environments where chemical communication occurs.
Any odor cue can be traced to find its release source. So-called “source localization” has been observed in animals in many important tasks including finding food or mates. In particular, the scientific community for a long time focused on unraveling the complex behavior of moths while in pursuit of sex pheromones emitted by their distant female counterpart. These studies have provided many insights including details of the flight paths, sensory organs, and pheromone processing. In turn, this knowledge has provided inspiration to engineers and researchers to devise source-seeking algorithms, whereas sensory organs/-mechanisms led to insect-machine hybrid systems. Therefore, this review revolves around these last two approaches specifically (1) the implementation of moth-inspired algorithms in robotic platforms and the (2) use of biosensors such as antennae or insect-machine hybrid systems.
Full-text available
Most real-world behaviors – such as odor-guided locomotion - are performed with incomplete information. Activity in olfactory receptor neuron (ORN) classes provides information about odor identity but not the location of its source. In this study, we investigate the sensorimotor transformation that relates ORN activation to locomotion changes in Drosophila by optogenetically activating different combinations of ORN classes and measuring the resulting changes in locomotion. Three features describe this sensorimotor transformation: First, locomotion depends on both the instantaneous firing frequency ( f ) and its change ( df ); the two together serve as a short-term memory that allows the fly to adapt its motor program to sensory context automatically. Second, the mapping between ( f, df ) and locomotor parameters such as speed or curvature is distinct for each pattern of activated ORNs. Finally, the sensorimotor mapping changes with time after odor exposure, allowing information integration over a longer timescale.
Full-text available
Motion plays an essential role in sensory acquisition. From changing the position in which information can be acquired to fine scale probing and active sensing, animals actively control the way they interact with the environment. In olfaction, movement impacts the time and location of odour sampling as well as the flow of odour molecules around the olfactory organs. Employing a detailed spatiotemporal analysis, we investigate how the insect antennae interact with the olfactory environment in a species with a well studied olfactory system – the American cockroach. Cockroaches were tested in a wind-tunnel setup during the presentation of odours with different attractivity levels: colony extract, butanol and linalool. Our analysis revealed significant changes in antennal kinematics when odours are presented, including a shift towards the stream position, an increase in vertical movement and high-frequency local oscillations. Nevertheless, the antennal shifting occurred predominantly in a single antenna while the overall range covered by both antennae was maintained throughout. These findings hold true for both static and moving stimuli and are more pronounced for attractive odours. Furthermore, we find that upon odour encounter, there is an increased occurrence of high-frequency antennal sweeps and vertical strokes, which are shown to impact the olfactory environment's statistics directly. Our study lays out a tractable system for exploring the tight coupling between sensing and movement, in which antennal sweeps, in parallel to mammalian sniffing, are actively involved in facilitating odour capture and transport, generating odour intermittency in environments with low air movement where cockroaches dwell.
Current evidence indicates that owl monkeys (Aotus spp.) have species-rich, flexible diets. They can switch resources seasonally, consume ephemeral foods of many life forms, and locate food in small, degraded forests. They focus on fruits (20–87%) yet regularly consume flowers, leaves, and/or insects. Common dietary components are Fabaceae flowers (and leaves in the South American Chaco), Melastomaceae leaves and fruits (tropics), and Moraceae fruits (all), especially figs. They may use relative food brightness and odor as cues, while resource selection seems to be related to availability. Intergroup differences in diet and resource availability of the Azara’s owl monkeys of the South American Chaco are small, while diets diverge in more anthropogenic and species-rich habitats. Owl monkeys experience limited competition with other mammals, perhaps aiding dietary flexibility. Unfortunately, due to a dearth in systematic studies, the most ultimate and proximate determinants of Aotus diet and feeding ecology largely remain a conundrum.
Our objective in compiling a series of chapters on the chemical ecology of insects has been to delineate the major concepts of this discipline. The fine line between presenting a few topics in great detail or many topics in veneer has been carefully drawn, such that the book contains sufficient diversity to cover the field and a few topics in some depth. After the reader has penetrated the crust of what has been learned about chemical ecology of insects, the deficiencies in our understanding of this field should become evident. These deficiencies, to which no chapter topic is immune, indicate the youthful state of chemical ecology and the need for further investigations, especially those with potential for integrating elements that are presently isolated from each other. At the outset of this volume it becomes evident that, although we are beginning to decipher how receptor cells work, virtually nothing is known of how sensory information is coded to become relevant to the insect and to control the behavior of the insect. This problem is exacerbated by the state of our knowledge of how chemicals are distributed in nature, especially in complex habitats. And finally, we have been unable to understand the significance of orientation pathways of insects, in part because of the two previous problems: orientation seems to depend on patterns of distri­ bution of chemicals, the coding of these patterns by the central nervous system, and the generation of motor output based on the resulting motor commands.
With few exceptions, air pollution models are designed to predict dosage or ensemble mean concentrations averaged over time scales of tens of minutes, hours, or more. This is adequate for the study of long-range transport problems and also for short-range dispersion of, for example, radioactive contaminants, for which the time-averaged dosage is the important factor in hazard assessment. The models are also widely used to predict mean concentrations in toxic accidents, for which their time scales are less satisfactory. The toxicity of many gases does not vary linearly with concentration C and exposure time t. For example, the toxicity of Chlorine (Cl 2) varies approximately as C 2.75t (see Griffiths and Megson (1984)). In this case the use of a time averaged dosage could lead to a dangerous underestimation of the hazard over short and medium range. Time averaged concentrations are also unsuitable for the assessment of the inflammability or odours of a gas plume, for which time scales of a few seconds are applicable. While modellers recognise that fluctuations of concentration occur on short time scales, they have not attempted to include them, partly due to a lack of good validation data. Recently technology has advanced to the stage at which tracer experiments may be conducted using continuous chemical analysers capable of measuring concentration time series with a frequency response of up to about 10Hz. Jones (1983) achieves a resolution better than 100Hz using ionized air as a tracer, but is limited to short range experiments and has to account for ionic repulsion in analysis of results. The response required in practise varies according to the application, but 10Hz is adequate for most. purposes. In particular, it is fast relative to the time scale of human breathing.
High resolution concentration measurements were taken in unstable and neutral conditions at 1 m height and up to distances of 15 m from the tracer release point. Values for the intermittency, peak to mean ratios and statistical moments are determined and the probability distribution of the concentration fluctuations examined. (A)
Male adults of Spodoptera litura were released at fixed points and captured in a pheromone trap. By comparing the experimental results with those obtained from a simulation model for the male's attraction process, the active space of the sex pheromone of S. litura was estimated. The maximum range downwind of the active space for one virgin female was estimated as about 80m for wind velocity of 0.50m/sec. This distance decreased with an increase in wind velocity above 0.50m/sec, and decreased greatly with decrease in wind velocity below 0.50m/sec. The reason for the decreased range at low wind velocities was as a result of a deposition effect of the pheromone. Thus, in a windless greenhouse, the active space was limited to a small area close to the attractant. The active space increased proportionally with an increase in the emission rate of the pheromone. It was revealed that the active space was determined by one of two components of litlure (a synthetic sex pheromone of S. litura), viz., cis-9, trans-11-tetradecadienyl acetate. © 1977, JAPANESE SOCIETY OF APPLIED ENTOMOLOGY AND ZOOLOGY. All rights reserved.
In the study of chemoreception, one encounters two principal difficulties that are unique to chemical stimuli. Stimulus qualities (i.e., the spectrum of all stimulatory chemical compounds) are not organized in one linear dimension such as the frequency spectra of mechanical and electromagnetic stimuli; and stimulus quantities (i.e., the distribution of the stimulus in space and time) cannot be expressed easily in mathematical formulations of stimulus dispersal comparable to the wave equation for radiating light and sound sources. Instead, the chemical spectrum of qualities consists of all reactive chemical compounds in the environment organized in what could be perceived as an unknown “n-dimensional hyperspace,” and the distribution of chemical stimuli from their source of release through the environment depends on molecular diffusion and on fluid dynamics of the carrier medium. At the spatiotemporal scale of micro- to millimeters and seconds, diffusion dominates the distribution process; at various larger spatiotemporal scales, chemical stimuli are dispersed by fluid motion. This chapter will focus on the distribution of chemical stimuli and its consequences for chemoreceptive processes, such as receptor physiology and stimulus acquisition behavior.
This section on walking insects, together with the following on flying insects, defines and illustrates mechanisms by which insects utilize chemical information available to them for purposes of locating mates, food and other resources or for avoiding repellents or stress sources. This discussion follows Städler, Chapter 1, and Mustaparta, Chapter 2 on the acquisition and processing of chemical information through peripheral receptors and the central nervous system, Elkinton and Cardé, Chapter 3, on airborne dispersal of chemicals, and is an introduction to the remaining chapters in this volume on ecological implications of resource localization and stress avoidance.