Bits, Bytes and Biological Barriers,

“Bits, Bytes and Biological Barriers,”

Spring 1999 Volume 10, Number 1

By Jim Nolen
Jurassic Park taught everyone that mosquitoes have been around for hundreds of millions of years, so it is not surprising that they diversified into species very different from one another. After all, they have successfully adapted to climates from the artic to the equator, and everywhere they have solved fundamental survival problems. They must locate carbohydrates to fuel their flight muscles, blood to supply protein to the egg clutch, and suitable habitats in which to lay their eggs. It is also not surprising that indigenous hosts are as varied as the locale; some mosquitoes are adept at finding frogs, others birds, still others mammals. Specialized adaptations exist within and between these categories. Some mosquitoes prefer birds but are adapted to find mammals. Some are especially well programmed to find small herbivorous mammals but easily detect humans nearby.

Mosquitoes are physiologically well-equipped. They can sense electromagnetic radiation from ultra violet to infrared, detect minute changes in pressure, temperature and humidity, track their prey’s scent like a bloodhound, and they need not stop to smell the flowers as they can do it in flight. Eons of natural selection have focused their instinctive programming on that combination of sensory input that identifies locally abundant hosts, floral nectar and water for oviposition on which their survival depends. We studied three clues mosquitoes use to find a host: scent, sight, and heat. Structures on the antennae and palpae detect the scent of the host’s odor plume at a distance of up to 30 meters. The mosquito follows the plume upwind and makes visual contact at a distance of approximately 10 meters. Three meters away (farther when it is humid) thermal receptors on the tips of the antennae help locate warmer areas where blood is near the surface of the skin. Some species can detect temperature changes as small as 0.1ºC.

Tsetse Fly
Those who understand how sensory inputs affect instinctively programmed behaviors can turn the insect’s evolutionary success against them. In the 1980’s, British scientists succeeded in controlling the Tsetse fly in Africa. Tsetse is a serious health hazard. It feeds on livestock and humans, spreading sleeping sickness. The scientists analyzed ox breath using gas chromatography, and then used electroantennograms to isolate the two most powerful attractants: carbon dioxide (CO₂) and octenol (1-octen-3-ol), a form of alcohol. Octenol and CO₂ are kairomones, chemical components of host odor which mosquitoes can detect. Fermenting vegetation in the ox’s digestive tract produces octenol, which along with CO₂ is expelled with each breath. Today, baited traps are deployed to control several species of Tsetse fly. Aerial pesticide applications stopped in 1991.

Methods the British used in Africa were straightforward, and the sensory structures of mosquitoes and biting flies are similar, so why are success stories like this rare? There are many answers to the question. Tsetse flies are an easy mark compared to mosquitoes. Their hosts were known, their host-seeking behaviors were easily deciphered and vulnerable to attack, and they are not prolific breeders. This is not the case with mosquitoes.

Mosquitoes
There are more than 3,000 species of mosquitoes with many different host-seeking behaviors. Narrow the scope to include only public health pests and hundreds of species remain. Attempt to alter their behavior with attractants, repellents or inhibitors, and the results are influenced by variables we cannot completely control: season of the year, time of day, weather and location. Each influences the behaviors being studied and makes it difficult to isolate experimental results so they can be accurately measured.

The problems are so difficult that an impossibly large number of experimental trials are required to achieve a comprehensive understanding of behavior. Consequently, experimental trials are limited to those that can be completed within the time and budget available. The experimental protocol becomes the embodiment of our priorities; we design it to reveal the knowledge we believe to be most important and postpone the rest of what we want to know.

Repetition of trials also limits what can be done in a given amount of time. Repetition is built into the experimental protocol to average the effect of environmental variables we cannot control over several trials. Repetition builds our confidence when identical trials produce similar results. Equally important, repetition tells us outside influences are at work when identical trials produce different results. Sometimes we get the results we expected. Sometimes we are left wondering how to explain the result we got.

As part of a Cooperative Research and Development Agreement (CRADA) with the United States Department of Agriculture, we demonstrated that accurate numerical predictions of mosquito collections in response to several experimental variables could be developed in as little as 15 days of field trials. Traditional application of the scientific method might take hundreds of trials and years in the field to accomplish the same result. Specifically, we generated 90% accurate models (correlation of predicted vs. actual catch) of each species’ response to any combination of four attractants. The attractants were CO₂, octenol, heat, and visual targets of three different sizes.

The level of each attractant was tested over a wide range. For example, CO₂ emissions ranged from zero to 1,000 ml/min, which is approximately equivalent to the respiration of four large men. Octenol ranged from zero to 28 mg/hr, which is equivalent to the emission from several cattle. The visual targets and thermal lures were combined into one device. The visual target consisted of a closed metal cylinder. Inside the metal cylinder were electrical connections for incandescent light bulbs. Because the metal cylinder trapped the light inside it, the energy of the incandescent bulbs (which radiate 10% of their energy as light and 90% as heat) was dissipated as heat through the thin, conductive skin of the cylinder. The outside of the metal cylinder was painted black to radiate energy most quickly. The inside of the metal cylinder was painted a mottled pattern of white and black to produce a non-uniform surface temperature. This effect is intended to simulate the non-uniform thermal emissions of living things. Previous research indicated non-uniform surface temperatures are more attractive than a uniform surface temperature. The size of the smallest visual target was equivalent to the trunk of the body of a small animal such as a rabbit or woodchuck. The next larger size was equivalent to the trunk of an animal such as a deer or goat. The largest size was equivalent to the trunk of a man. Finally, incandescent bulbs of various wattages were used in combination to produce three levels of thermal emissions. The lowest thermal emission was zero (no incandescent bulbs). The next higher thermal emission was 0.016 Watt/cm2 (0.1 Watt/in.2), characteristic of animals with lower body temperatures. The highest thermal emission was 0.031 Watt/cm2 (0.2 Watt/in.2), characteristic of warm-blooded mammals with higher body temperatures.

There are an infinite number of combinations of these four attractants, so how can 15 trials produce an accurate model over the whole range of variables for every species collected? The computer-designed protocol does not test every possible combination of attractants, but specifically selects the fewest combinations from which a statistically valid model may be constructed. Taylor Second Order Expansion Equations are used together with a specially selected fractional factorial design to do so.

The fractional factorial design cleverly uses results (insect collections) of midpoint replicates to minimize the number of trials required to produce a statistically sound model. Just as human researchers gain confidence from identical trials that produce similar results, so to does the computer. The midpoint of every variable is tested repeatedly and the differences in the collections determine the level of confidence we have with the result. Next, the experimental design tests each variable at its extremes, both high and low. Not every combination of high and low extremes is tested. Only trials sufficient to determine whether the effect of a given variable (or its squared value, or its interaction with another variable) is to increase or decrease collections as the variable is increased or decreased.

To validate the predictions, the trials are repeated and the actual collections compared to the predicted values. CRADA research will not be published until later this year, but two examples illustrating very different behaviors are reproduced here: Culex nigripalpus, a St. Louis encephalitis vector, and Culicoides furens, a biting midge.

Figure 1 is the model for Culicoides furens, the infamous biting midge and an aggressive pest in the Southern US and the scourge of tropical beaches. This tiny creature is particularly difficult to control. With a wingspan of 1mm, it can easily pass through physical barriers such as screens or mosquito nets. It deposits its eggs in the inter-tidal zone between the high and low water mark, minimizing their exposure to larvicides. This model indicates that Culicoides furens is strongly attracted to heat and octenol.

The first line of information at the top of the graph contains the key used to identify the curves on the face of the graph. The face of the graph contains three curves representing the predicted catch at three levels of body heat. Low (L) represents no energy radiated as heat. The Mid-point represents 0.0155 Watt/cm2 (0.1 Watt/in.2) energy radiated as heat. High (H) represents 0.031 Watt/cm2 (0.2 Watt/in.c2) radiated as heat. As mentioned previously, the Mid-point and High curves simulate the body heat of living things.

We can graph only two values on a two-dimensional sheet of paper, so other experimental values must be held constant. The second and third line of information at the top of the graph lists the variables that were held constant. In this case, the visual target was held constant at a surface area of 4,580 cm2 (710 in.2), approximately the size of the trunk of a man. This size visual target produced the largest collections of midges. CO₂emissions were held constant at 200 ml/min, equivalent to the respiration of a 90-kg (200 pound) man.

The horizontal axis is the octenol emissions, which range from 0 mg/hr to 28 mg/hr. As mentioned previously, the higher rate is characteristic of herbivorous mammals.

The vertical axis labeled ‘cf’ is the total predicted Culicoides furens collections per night using a CDC trap mounted at the CO₂ discharge point 15.25 cm (6 in.) from the visual target/thermal lure. Collections took place in October 1996 at the University of Florida Medical Entomology Laboratory at Vero Beach Florida. The coefficient of correlation between predicted versus actual collections was 0.97 for this model.

Figure 2 is the model for Culex nigripalpus, the St. Louis encephalitis vector. The left portion of this model indicates that Culex nigripalpus is strongly attracted to heat and CO₂, a profile characteristic of avian hosts. The right portion of this model indicates Culex nigripalpus is also strongly attracted to high levels of octenol, a host profile characteristic of animals such as cattle.

 

The format of Figure 2 is identical to Figure 1, with the exception that a different size visual target is used. In this case, the visual target was held constant at a surface area of 516 cm2 (80 in.2) simulating a small animal such as rabbit or woodchuck. This size visual target produced the largest collections of mosquitoes.

The vertical axis labeled ‘cn’ is the total predicted Culex nigripalpus collections per night using the CDC trap described previously. The coefficient of correlation between predicted versus actual collections was 0.98 for this model.

While the scales remain tipped strongly in favor of the mosquito, the computer speeds up the pace of progress. When the insect’s behavior is understood, a multidiscipline team of entomologists, chemists, and engineers can quickly focus on the best opportunities to exploit that behavior. For example:

• Breaths of CO₂ five seconds apart collect more mosquitoes than continuous discharge.
• A very narrow range of temperature, 43ºC ± 8ºC, increases collections. Temperatures less than 35ºC do not increase collections. Temperatures greater than 51ºC reduce collections.
• Irregular infrared patterns (mottled patterns of cooler and warmer areas) produce larger collections than uniform ones.
• The visual image and thermal emission of a small animal produced the greatest mosquito collections, while the visual image and thermal emission of a larger animal produced the greatest midge collections.
• An avian host attractant profile produced larger collections than mammalian host attractant profile for Culex nigripalpus.

These differences profoundly impact the way one designs traps for the poultry industry’s chickens as opposed to the tourist industry’s beach-goers.

The efficiency (target insects collected per ml of CO₂, for example) of traps built to the specifications of the target species (as indicated by the combination of attractants that maximized collections) is very high. Efficient traps maximize collections, lower costs and are harmless to non-target species, making them a viable alternative to insecticides in some cases. For example, Dr. Jonathan Day of the University of Florida has created a midge-free zone 275 meters in length at Boynton Beach, Florida. He used CO₂ and octenol-baited traps spaced at 12 m intervals. While the traps were running they removed flies at distances of at least 6 m in all directions.

Dr. Daniel Kline of the USDA in Gainesville, Florida reports that mud samples taken from this area contain no larva when test line is on and trapping sand flies. Health officials in the Gold Coast of Queensland, Australia are that country’s midge experts. They continue to look for alternative control measures because of insecticide resistance. They occasionally resort to malathion and beach raking to reach eggs buried 6 to 10 cm below the surface of the inter-tidal zone. Considering the difficulties of midge control, Day’s accomplishment is no small feat. It is Day’s opinion that removal trapping may serve as an alternative form of midge control at a cost competitive with present day insecticide strategies. We also are confident that biological attractants can be effective in limited but important applications. Moreover, biological inhibitors that cannot achieve control may, however, reduce pesticide use in important applications such as residential pest control, commercial pest control, and livestock protection.

Our CRADA research originally focused on gaining an understanding of the interaction effects of attractants. As our understanding of the underlying chemistry improved, we soon found ourselves concocting substances that bind more strongly to proteins on the insect’s receptors than do kairomones in the host’s scent. This approach lead to better attractants. For example, one new attractant seems to be particularly effective against Aedes aegypti. In a preliminary trial in the USDA’s 10 by 20 m outdoor cage in Gainesville, Florida, 1,000 Aedes aegypti were released and 750 were recovered. The experimental control, an efficient trap baited with 500 ml/min CO₂, recovered only half as many. This approach also lead to better inhibitors. Because troublesome species of mosquito smell your scent long before they can see you, inhibiting their scent tracking ability seems to be a worthwhile strategy. One new inhibitor reduced landings on humans by 50% compared to landings in an unprotected control location, although there were large individual differences between test subjects, and some species may be less susceptible than others. In preliminary olfactometer trials with Aedes aegypti, this new inhibitor was almost twice as effective as DEET in preventing mosquitoes from locating human scent. Essentially all Aedes aegypti located human scent alone, which was the experimental control.

The technology’s promise has attracted various business partners who wish to develop applications for their markets. What does the future hold? It is early and our crystal ball is very foggy, but some of the possibilities appear below.

• Lower cost alternatives to pyrethrin for animal protection in dairy barns, chicken coops, and other livestock enclosures.
• New methods of releasing high levels of attractants for outdoor spatial barriers.
• New methods of releasing tiny amounts of high-purity attractants or inhibitors for use indoors and in programmable traps.
• Small traps for indoor use in tropical regions.
• Alternatives to pyrethrin for indoor household use in tropical regions.
• Inexpensive devices powered by a 9-Volt battery or solar energy surrounding breeding sites to trap mosquitoes.

Special thanks go to Dr. Jonathan Day of the University of Florida Medical Entomology Laboratory, Dr. Scott Ritchie of the Tropical Public Health Unit, Cairns, Queensland Australia, and Dr. Daniel Kline of the USDA Center for Medical, Agricultural and Veterinary Entomology for their help in the preparation of this article.