The 1998 Brighton Conference - Pests and Diseases

Pheromone dispersion in the canopy trunk space

H W Thistle
USDA Forest Service, Missoula Technology and Development Center, Missoula, Montana, USA

P Shea
USDA Forest Service, Pacific Southwest Research Station, Davis California, USA

E Holsten
USDA Forest Service, Pacific Northwest Research Station, Anchorage, Alask, USA

D Quilici
University of Nevada, Reno Nevada, USA


Optimizing the placement of passive pheromone dispensers in the forest canopy trunk space requires understanding of the air movement in this space. This work concerns western pine beetle (Dendroctonus brevicomis) which typically fly in temperatures greater than 11 ˝C during daylight in the warm seasons. This study utilizes sulphur hexafluoride (SF6 ) as a tracer to study the dispersion patterns in a ponderosa pine (Pinus ponderosa) trunk space in the southern Cascade Range in northern California. Four-hundred and fifty, 30-minute samples of SF6 concentration were collected per test day inside of a circle of 30-m radius, resulting in over 3500 samples over 9 test days. Atmospheric conditions ranging from stable through unstable were sampled. Extensive canopy architecture as well as heat and momentum flux data were also collected. Results showed substantially different mixing regimes as a function of stability.


The primary objective of this project is to provide pest managers with explicit guidance for the placement of passive pheromone releasers in the forest trunk space. Insect pheromones are naturally occurring chemicals and are increasingly being researched for use in pest management strategies. It is difficult to design a tracer experiment to investigate pheromone dispersion in the canopy without knowing the exact nature of the pheromone response mechanism of the insect. There is some evidence that response to a pheromone plume is a signal to noise ratio detection problem and that the insect follows a plume peak centerline concentration to its source. The flight of western pine beetle (Dendroctonus brevicomis) typically occurs during the day. The typical midday dispersion environment under a closed ponderosa pine canopy will be explicitly measured in this experiment. During the day, pheromone may be relatively uniformly mixed under the canopy implying that the beetles may encounter a threshold level of pheromone and use this as a proximity indicator. In this case, the insect could home in by flying upwind. The shaded stem space is stable and may not be well mixed, on the other hand, and the plume-following mechanism might suffice due to the narrower, more concentrated plume. The two approaches might lead to the same result as long as a detectable level of pheromone is encountered by the beetle. These mechanisms will, in all likelihood, be specific to the insect and the specific pheromone being studied. However, the dispersion of any gaseous scalar quantity under this canopy can, at first pass, be treated generically assuming chemical passivity and it is largely determined by the dispersion environment which, in turn, is determined by meteorology, canopy density, and distribution. This discussion is guided by studies in the entomological and meteorological literature (Aylor 1976, Aylor et al. 1976, Murlis & Jones 1981, Elkington et al. 1984, and Farbert et al. 1997).

The degree of coupling of the in-canopy atmosphere and the ‘free’ atmosphere away from the canopy will depend upon canopy density and architecture as well as the state of the atmospheric surface layer. The stem space in a dense, closed canopy can be decoupled from the above-canopy planetary boundary layer (PBL) wind field with exchange between the canopy and the free atmosphere dominated by high-energy, low-frequency downward wind gusts or ‘injection’ events. The crown space converts mean-flow kinetic energy from above the canopy into turbulent kinetic energy (TKE) within the canopy, and much of this TKE is lost to drag due to canopy elements. Mean wind speeds below the crown space are less than those above the canopy. In closed canopies, the crown space receives much more direct-beam solar radiation during daylight hours than the stem space and warms more quickly. Thus an inversion layer is likely to form in the stem space in closed canopies during the daytime which further suppresses momentum transfer and turbulence. Local slope-related flows within the stem space and “chimney effects” due to breaks in the canopy may combine with mean motion above the canopy to drive both transport and dispersion within the stem space. In more open canopies, the mean wind speeds are less in-canopy due to a bulk drag effect of the canopy. However, in these canopies, a large amount of sunlight reaches the forest floor and the atmosphere is closely coupled with the free atmosphere as trees may act as individual obstacles to higher energy flows (higher mean wind speeds) causing large horizontal variability in the flow field. Forest canopies range in density from thick tropical jungle canopies which are completely closed to widely spaced trees in ‘parklands’ typical of much of the high arid western United States.

In some circumstances, during thermally neutral conditions, when the wind speeds above the canopy are greater than 5 to 10 m/s, plumes below the canopy follow the mean PBL wind direction. During stably stratified conditions, however, the plume direction in the canopy can be offset on the order of 45 degrees from the mean wind direction above the canopy. These numbers are specific to the canopy being studied, but they point out that there is a threshold wind event when the air above the canopy mixes with the in-canopy air and that the mean flow direction in-canopy can substantially deviate from the mean flow direction in the free atmosphere.

The degree of coupling will largely depend on the temperature structure of the surface layer or the ‘stability’ of the layer. In stable conditions, cold air is under warm air and vertical turbulence is damped. In these conditions, the stem space may be effectively decoupled from the above canopy atmosphere. A scalar quantity released under the canopy such as a pheromone would tend to stay under the canopy and the pheromone plume would remain concentrated due to the low mixing environment. In the case of an unstable environment, warm air is under cold air. Since the warm air is lighter, it rises through the cold air and mixes the unstable layer. In an unstable atmosphere pheromone would tend to rise and mix and might leave the target area where the insects are active. The canopy complicates this relationship because the stability will change with height in the canopy. Often, the upper leaf surfaces intercept the solar radiation and the shaded stem space will remain stable, but an unstable layer will develop at the canopy top. Generally, western pine beetle (Dendroctonus brevicomis) flight is limited to the stem zone and the mean attack height is around 6 m.

The combination of canopy structure and stability regime will determine optimum spacing of the pheromone assuming the entomological parameters and pheromone elution rate are known. The approach used in this study is to know the canopy structure and micrometeorology in detail and measure the gaseous dispersion field.


The site is in ponderosa pine (Pinus ponderosa) in the southern Cascade Range near the town of McCloud, California. This area offers various canopy densities by way of differing silvicultural treatments. This experiment was sited in a relatively close-crowned canopy on locally flat terrain at 1300 m elevation.


There is evidence that plumes for near-instantaneous releases within a Douglas-fir canopy generally range in width from about 30 degrees to over 120 degrees. Instrument sampling arrays were deployed covering about 240 degrees to capture the plume. Even with 240 degree coverage, one plume edge was off the array during many of the tests. However, the plume centerline and maximum concentration were on the measurement array. The decision of which portion of the array to decorate for each test was based on then-current conditions and wind-direction forecasts for the period of the tracer test. Tests were conducted from June 20 to 28, 1998.

A set of three concentric circles centered on the emission point were marked for sampler deployment. The circles had 5, 10, and 30 m radii. Sampling locations were evenly spaced on each circle and between 47 and 53 samplers were used for each test. The 5-m-radius circle had 12 possible sampling locations, one every 30 degrees. To decorate 240 degrees, the 5-m circle had nine samplers. The 10- and 30-m-radius circles had 24 possible sampling locations each, 1 each 15 degrees. The 10- and 30-m circles had 17 samplers each to decorate 180 degrees. Each sampling location was identified with a circle radius and an azimuth angle. Four sampling towers were deployed, one on each arc as well as the 25-m meteorological tower. The three arc towers were 7 m high, and samplers were placed at 1.5 and 5 m. The meteorological tower had samplers at 1.5 m, 10 m, and 25m.

Krasnec et al., (1984) and Benner and Lamb (1985) provide detailed discussion of the syringe sampler and the analyzer respectively, as well as the general experimental approach. The syringe samplers were programmed to start simultaneously at all sampling locations. The sampling period was set for 30 minutes per sample. Since the samplers contain 9 sampling stations, this means that each test period was 4 hours and 30 minutes. Background concentrations of SF6 could be above detectable levels, so background concentration levels were monitored.

The real-time analyzer used during the tests and for sample assay has a detection range of 10 ppt to 10 ppb (6.07 x 10-7 g/m3). The lower end of the reliable quantification range is probably on the order of 50 ppt (3.04 x 10-5 g/m3) . In order to stay below the 10 ppb limit at the plume centerline on the 5-m-radius circle, the SF6 emission rate was around 5.0 x 10-5 g/s and was adjusted in the field as data were analyzed. The analyzer was calibrated using no less than four audit standards both prior to and after analysis of each 30-minute test series.


Three 7-m towers collected mean meteorological data both under and outside the canopy. These towers had 2 levels of temperature, humidity, wind speed, and direction and one net radiometer. Three three-axis sonic anemometers were used to collect both mean and turbulence statistics. These instruments provide both momentum and heat flux as well as mean wind vectors and turbulent covariance statistics. These sensors were arrayed vertically up to 25 m and were collocated with pheromone samplers. A SoDAR was deployed to characterize the atmospheric boundary layer. It measured wind speed and direction between heights of 100 and 800 m.


Canopy density was measured on a regular grid pattern across the site. Two methods of measurement were used. The first is a LiCor 2000 LAI meter. The second was hemispherical photography. These measurements were taken on a regular grid and LAI was also measured vertically on the main meteorological tower up to 25 m. A detailed stem map of the plot was also drawn.


Two recognized errors entered into the field program. One was due to leakage from the test syringes because of incorrect capping and the second was due to a contaminant volatilizing from an O-ring lubricant in the syringe. The first error was studied in the field (it was corrected near the end of the program) and the affected syringe data can be corrected through a regression equation. The contamination error is not yet completely understood but it was directly evident in the field using the analyzer. It is apparently related to the syringe temperatures and occurs in around 25% of the data. These contaminated data are not discussed here, though it is hoped that after laboratory analysis of the syringes at varying temperatures, the effect of the contaminant can be removed.


Results are preliminary but begin to allow the development of guidance for the placement of passive releasers under this canopy. On still days, the temperature gradient under the canopy controls the vertical spread of the plume. The temperature gradient (or stability) in the trunk space is greatly influenced by the canopy architecture. The leaf area index (LAI, m2m-2) is very difficult to obtain in this type of canopy. Two methods of obtaining this quantity were pursued, as discussed previously. The hemispherical photographs have not been processed yet, so preliminary numbers are only available from the LiCor 2000. This instrument is known to be less accurate in canopies with substantial directional biases in the vegetation which is the case in the canopy here. This canopy extends up to an average height of between 25 and 30 m and is best characterized as individual raised traffic cones. The trunk space is open and easy to move through. A large percentage of the surface is illuminated by direct solar radiation during at least part of the day. The LiCor yields LAI’s in this canopy between 2.5 and 3.

Due to the distribution of the canopy, surface heating commenced early and the stable layer was gone as indicated on the 30-m mast by around 10:30 a.m. The tracer release was at 1.5-m height, and the vertical spread of the plume increased as the stability tended towards unstable. This information will be used to scale the plumes by the source strength to allow a direct indication of effective radius. The test period included 2 days with mean wind speeds of >3ms-1 even at the lower anemometers. These days are neutral in terms of stability with the in-canopy flow highly turbulent. These conditions were unexpected. The nature of the canopy allows the mean velocity to remain high while moving through this field of discrete obstacles. Generally, the canopy should exert a bulk drag and the in-canopy velocity should be much lower than that above the canopy. The highest anemometer was not high enough to evaluate this effect and the SoDAR data were too high. The higher than expected frequency of these relatively high-velocity flows in this canopy require that attention be paid to this neutral stability condition which is highly turbulent due to the canopy obstruction, but quickly transports material long distances and prevents elevated concentrations.

The converse of the neutral situation was the very low velocity situation which existed on three test mornings, making it very difficult to position the array because the mean motion approached zero. Under these conditions, tracer that is released hangs together with little movement of the plume and low mixing, due to stable conditions. This cloud of concentrated material then is available to waft out of the canopy in the vertical motions (thermals) that begin to develop as the surface heats. More detailed analysis of the turbulence data should directly yield heat and momentum fluxes which can then be used to determine scalar (tracer) flux out of the canopy and subsequently be used to quantify the loss term from passive releasers under the canopy.


Conclusions are presented here in two sections. Since this paper largely focuses on methods, the experimental method is evaluated. This is followed by a discussion of preliminary results. Two sources of error exist in the data due to experimental procedures. The QA/QC performed during the experiment focused on the chemical analysis and the source/release system. The weak link from the experimental standpoint turned out to be the syringes themselves. Incorrect capping of the syringes turned out to be a subtle error and therefore was not picked up until well into the test program. Data were plotted and most suspicious outliers in this context were too high because of the frequency of zeroes or background in this type of work. The procedure was to reanalyze these outliers. When the reanalysis returned a lower concentration—which made more sense in the spatial context of the plume—errors were attributed to mislabeling (<1% of the data was deemed suspicious). The incorrect capping would have been spotted immediately if calibration gases had been put into syringes and carried out into the test plot and placed in the data stream. This was an oversight in the QA plan. The second error is more interesting because it may reveal a flaw in recommended equipment that was exactly specified. It was casually suggested that, as a precaution, the syringes be preheated in a ventilated oven to evaporate off any volatiles before the experiment. When this potential problem was discussed with experts, it was stated that the type of contamination seen in this study would not occur. As noted above, this error was immediately obvious at the analyzer when it occurred, and it requires further investigation before the discussion can proceed.

Preliminary results strongly support the idea that the dispersive environment is evolving as the day progresses, at least on the test days that there was not a strong synoptic gradient resulting in high winds at the site. A parallel study was being undertaken at the time this study was performed which indicated there was high beetle activity on the days that a cool, still (stable) morning transitioned into a hot afternoon with a variable breeze (unstable). Therefore, the dispersion of the pheromone will be closely correlated to the time it is released. If the pheromone is released at night, a cloud or 'pool' of the material may accumulate near the infestation which acts as a volume source of material when the air begins moving significantly during the late morning. As the day progresses, a significant amount of a buoyant gas released in the trunk space will be lost upward and out of the canopy top.


Aylor D (1976). Estimating Peak Concentrations of Pheromones in the Forest. In: Perspectives in Forest Entomology, eds Anderson and Kaya, pp.177-188. Academic Press: New York.

Aylor D; Parlange J Granett J (1976). Turbulent dispersion of Disparlure in the forest and male Gypsy moth response. Environmental Entomology 5.

Benner R L; Lamb B (1985). A fast response continuous analyzer for halogenated atmospheric tracers. J. Atmos. Ocean Tech. 2.

Elkington J; Carde R; Mason C (1984). Evaluation of Time-Average Dispersion Models for Estimating Pheromone Concentration in a Deciduous Forest. J. Chemical Ecol. 10.

Farbert P; Koch U; Farbert A; Staten R; Carde R (1997). Pheromone Concentration Measured with Electroantennogram in Cotton Fields Treated for Mating Disruption of Pectinophora gossypiella (Lepidoptera: Gelechiidae). Phys. And Chem. Ecol. 26 5.

Krasnec J; Demaray D; Lamb B; Benner R (1984). Automated Sequential Syringe Sampler for Atmospheric Tracer Studies. J. Atmos. Ocean. Tech. 1.

Murlis J; Jones C (1981). Fine scale structure of odour plumes in relation to insect orientation to distance pheromone and other attractant sources. Physiol. Entomol. 6.

Title page

This page last modified January 21, 2000

Visitor since January 21, 2000