Fig 1EVALUATION OF DOG-ASSISTED SEARCHES AND ELECTRONIC ODOR DEVICES FOR DETECTING THE WESTERN SUBTERRAN
TERMITE (ISOPTERA: RHINOTERMITIDAE)
VERNARD R. LEWIS, CALVIN F. FOUCHE and RICHARD L.
LEMASTER
materials and methods | results and discussion | literature cited | footnotes | figures 1, 2 and 3
ABSTRACT
Laboratory blind trails were conducted to evaluate the ability of beagles and an electronic odor device to detect termites in wood. In the 1st test, pine blocks art)ficially infested with either 0, 5, 50, or 200 workers of the western subterranean termite, Reticulitermes hesperus Banks, were randomly presented to five beagles and one electronic odor device. Blocks were presented one-at-a-time to beagles and the electronic odor device. The beagles correctly identified 81% of the blocks while the electronic odor device correctly identified 48%. A 2nd laboratory test comparing five additional electronic odor devices from the same manufacturer resulted in a slightly higher correctly identified value, 62%. Beagles performed best for blocks containing 50 or more termites. However, the percentage of misidentified controls (false positives) for beagles was high, 28%. The electronic odor device did not demonstrate statistically sign)ficant detection ability. Although beagles were almost perfect (49 of 50 blocks) in detecting termites in blocks with 50 or more individuals, neither detection method was reliable with "controls" or low density number of termites. The implications of these findings to use of either of these termite detection methods are discussed.
Termite detection in structures is done almost exclusively by visual searches. In California alone, over 1.4 million inspections are conducted each year (1). Depending solely on visual searches to detect termites often results in missed inspections due to oversights and inability to search inaccessible areas. Some estimates suggest that inaccessible areas in homes, i.e., subareas, attics, and covered walls may exceed 45% of the total area searched during inspection (2).
Acoustic emission (AK) detection has shown promise in laboratory investigations (5, 6,10,12,14, 16, 18, 20). Unfortunately, this technology is only useful for locations within boards. Improvements in sensor probes are needed to minimize damage to wall coverings and for probing areas within boards, deep below the surface. In addition, field verification studies are needed. AE technology is currently not commercially ava~lable. Future detection methods that may allow for the non-destructive searching of entire walls include microwaves, infrared, and laser technologies (11).
Two additional non-visual detection methods are being marketed, dog-assisted termite searches using beagles and electronic odor devices. Both methods exploit gases given offby subterranean termites. The use of dogs for detecting insects is not a new innovation to entomology. Dogs have been used to detect gypsy moth, Porthetria dispar (L.), egg cases (21) and screwworm, Cochliomyia hominivorax (Coquerel), larvae (22).
Although dog-assisted termite searches have been commercially available since the mid-1970s, there is little published information on the efficacy of dogs or other carnivores in detecting termites. The only cited study of any carnivore's ability to detect termites is from field observations of aardwolves (Proteles cristatus) in Africa turning upwinc
of termite foraging parties (Trinervitermes) before approaching them (17). The author suggested that aardwolves used audition and olfaction to locate the termites. The innovation of using beagles to detect termites was first proposed by Robert Outman of Tadd Dog Services (2). Now several nationwide firms train and market beagles to assist in termite inspections. Currently, as many as 150 beagles are in service, although no scientific studies validating the detection capabilities of beagles have been published. The termite detection ability of another breed, bloodhound, has also been mentioned (3, 4). However, the results presented at a national conference reported the detection ability of this breed to be less than 50% (4).
Fifteen gases have been reported from more than 20 species of termites worldwide (8). The most common and abundant gases emitted from termite colonies are carbon dioxide and methane (9). The device used for the current study (Termitect II, Warren Armstrong, Albuquerque, New Mexico) primarily detects methane gas; however other gases may also be detected. The device has three settings for detecting methane: 2 ppm, 4 ppm, and 6 ppm. Additional features include two sensors and a numbered gauge for visual readings of gas levels. A loud beeping sound occurs when above-threshold levels of target gases are detected. There may be as many as 80 of these devices in use in the United States (V. R. Lewis, personal communication).
Non-visual alternatives to termite inspections currently are not in large-scale use in the United States. However, it is anticipated that their use will increase due to the increased need to inspect inaccessible areas. Also with the more frequent use of spot treatments as opposed to treatments of entire structures, there will be a greater need for efficacious non-visual detection methods (13). Because of the multibillion dollar value of wooden structures in the United States and the need to protect this financial investment from termite damage, the following study was initiated to determine the ability of trained beagles and electronic odor devices to detect subterranean termites.
MATERIALS AND METHODS
Beagles Versus Odor Device Termites artificially inserted into wooden blocks were used to assess the detection capabilities of the two methods. One hundred ponderosa pine (Pinus ponderosa Dougl. ex Laws.) blocks were randomly chosen from a total of 200 labeled blocks. The size of each block was 115 mm by 87 mm by 37 mm. A 20-mm diameter hole was drilled 103- mm deep into the center of the 87-mm side of each block (Fig. 1). The blocks were then randomly divided into four groups of twenty-five blocks each. The bored area inside each block was rinsed with approximately 25 ml of tap water before termite insertion. Termites placed into wooden blocks were collected from laboratory colonies of the western subterranean termite, Reticulitermes hesperus Banks. We selected healthy and uninjured workers of at least the 3rd instar. Five workers per block were placed into blocks in the first group, 50 workers per block in the second group, and 200 workers per block in third group. The last group consisted of empty control blocks.
To ensure that the status of the blocks, termites or control, remained unknown, blind trials were used. Blind trials were ensured by gluing a 50-mesh galvanized screen with hot paraffin over the bored opening. The screen allowed for gas exchange while concealing the termites. This process was replicated for a total of 100 blocks. Each block, along with 25 ml of tap water, was placed in the bottom of an individual 15-cm diameter plastic container with friction-fitting lid and left overnight.
Five beagles and handlers were provided by Tadd Dog Services, Belmont, California for this study. All beagles were males and ranged in age from 19 months to 11 years. Twenty blocks (five blocks for each termite density used) were randomized, removed from the plastic containers, and presented individually to each dog for inspection. Those blocks evoking a positive response, as reported by the handler after viewing the dog's digging or barking behavior, were scored as positive for termites. Those blocks not eliciting such a response from the dogs were scored as negative. Two separate visits two weeks apart were needed to test all beagles.
When the beagles were finished searching, the same blocks were rerandomized and presented individually to a single electronic odor detection device (Termitect II, Warren Armstrong Co., Albuquerque, New Mexico). The authors conducted all searches of blocks. Our inspection method was developed after consulting with the developer, reading the instruction manual, and viewing an instructional video. A block was removed from its container and the screened end was placed near the extended sensor (Fig. 2). The most sensitive setting, 2 ppm, was used for all tests. Sample blocks were placed near the sensor for 30-s. A loud beeping sound from the device was scored as positive for termites. No beeping after 30-s was scored as negative. After the tests were completed, the sample blocks were opened, and the dead and live termites were counted.
Additional Electronic Odor Device Test
For this study, the variance among five electronic-odor detection devices (same model from the same manufacturer) was compared. Using the same block preparation methods described above, 25 blocks each containing either 5, 50, or 200 R. hesperus workers in addition to 25 empty controls were prepared (n=100). However, instead of covering the bored opening with galvanized screening, a 20-mm diameter cork with a plastic straw 7-mm in diameter and 2.5-cm long inserted through its center, was used to seal the opening in each block (Fig. 3). This mod)fication was requested by the developer to allow for more even gas flow out of the sample blocks. To ensure that termites were not seen during testing and did not crawl out through the open straws, each cork was lightly coated with polytetrafluoroethylene (Fluon AD 1, Northern Products, Woonsocket, RI). Another mod)fication requested and used was that all blocks be aired for several hours before testing.
After the blocks were randomized, inspections with the devices were conducted by the developer. The examination of the sample blocks differed from the methods in the 1st test in that the developer used a medium setting, turned off the beeper, and used the meter-gauged reading to determine whether or not a block contained termites. Blocks positive or negative for termites were reported verbally to one of the authors and recorded. In total, five separate devices were used for the test.
Statistical Analysis
The frequency distributions of blocks correctly identified for subterranean termites by beagles and electronic odor devices were depicted as histogram plots (7). Means and standard deviations for correctly identified blocks for each detection method were calculated using the means procedure (19). Comparisons of survival among blocks of varying termite number were conducted using an F-test (19). Tests of sign)ficance among means for correctly identified blocks for varying termite number were conducted using non-parametric tests; McNemar's test and %2 tests of proportions (23).
RESULTS AND DISCUSSION
Survival of R. hesperus individuals within sample blocks was acceptable. Summing across all blocks, 90.4% (iSD = 16.7%) of the termites survived. This acceptable rate of termite survival suggest that the test design did not overly expose termites to laboratory sources of mortality, e.g., corks dipped in polytetrafluoroethylene (Fluon) during the 2nd odor detection test. However, there were differences in mean percentage survival among termite groups of 5, 50, and 200 individuals, respectively, 83.6 + 26.4%, 95.2 + 6.3%, and 92.2 + 7.5%. These differences were sign)ficant (F = 6.94, df= 2, 149, P < 0.0013). These data suggest that the highest mortality among blocks were those containing 5 termites. The causes of the higher mortality level observed within this group may have been the drying out of blocks during the 2nd odor detection test since this test had the lower survival percentage, 85.5%, compared to 89.6% for the 1st test.
Beagles Versus Electronic Odor Detector
Beagles identified approximately two times more blocks correctly than did the electronic odor device during the 1st test, 81% (SD = + 5.5%) versus 48% (SD = + 11.5%). The difference was significant (X2 = 20.9, df = 1, P < 0.05). Between beagle variability in correct detection was extreme and ranged from 80% to 90% (n = 5 dogs). In the only other study reporting on dog detection of termites, bloodhounds were only 50% effective in finding petri dishes containing live Reticulitermes spp. (4).
Correct identification of blocks with termites by beagles varied with the numbers of termites contained within test blocks. Detection accuracy of beagles was skewed towards those blocks with more termites. This difference was sign)ficant (X:2 = 12.0, df = 3, P < 0.05; Fig. 4). From these data, it appears that the minimum number of termites needed for correct identification by beagles lies between 5 and 50 termites. Misidentified control blocks (false positives) for beagles represented 28% (7 of 25 control blocks). This is a high misidentification value and places doubt on the reliability of beagles in determining the actual presence of subterranean termites in wood. Frequency distributions of the percentage of blocks correctly identified by individual beagles revealed little variability among dogs (Fig. 5). The general pattern revealed was considerable misidentification of controls and blocks containing 5 termites, and almost perfect detection ability for blocks containing 50 or more termites.
During the 1st test, the electronic odor device did not display any sign)ficant subterranean termite detection ability. Even though there was no difference in correct identification among blocks with varying numbers of termites (x2 = 3.3, df = 3, P > 0.05), the accuracy in detection
was 60% or lower for all termite density levels tested (Fig. 4). The range in the percentage of correctly identified blocks among the replicates of the same device for the 1st test ranged from 30% to 60%. The percentage of misidentified controls (false positives) was 40% (10 of 25 blocks). This high level of misidentified controls was even higher than for beagles. There appeared to be little reliability in using the electronic odor device in determining the presence or absence of subterranean termites in wood. For between device variability in correct detection, see results in the Additional Electronic Odor Device Test section that follows.
Additional Electronic Odor Device Test
During the 2nd electronic-odor detection test using five additional devices, the mean percentage of correctly identified blocks increased slightly to 62% (SD = + 6.7%). Similar to the results with beagles, the accuracy of the electronic odor device was skewed towards those blocks with more termites. This difference was sign)ficant (X2 = 7.8, df = 3, P < 0.05; Fig. 4). However, misidentified controls (false positives) were high, 72% (18 of 25 blocks). While the patterns of success for individual beagles were similar, the pattern among electronic odor devices was erratic across varying termite numbers (Fig. 5). With some devices, 100% of the controls were misidentified. Even with developer cooperation in the testing, these data suggest a less than "50-50" chance in determining the presence or absence of termites in wood.
Obviously, both detection methods demonstrated some level of detection ability but improvements are sorely needed. Reliable results, particularly for determining if termites are present or absent, were not achieved with either detection method. They both also appeared to be highly technique driven. For beagles, more developmental work is needed to minimize the number of false positives and improve detection of low numbers of termites in wood. The physiological capabilities of dogs for detection are high, with olfaction as high as 0.001 parts per billion for some odors (21) and auditory sensitivity at least three octaves higher than humans (15). However, more evaluations are needed to determine detection mode-of-action (audition or olfaction or both), variance within and between breeds, and field validations. With the electronic odor device, more developmental work is needed to discriminate gases that are uniquely produced by termites and determine how background levels of gas produced in the surrounding environment affect detection results. With an increasing number of homes inaccessible to inspection due to construction design and shear wall (plywood) supports for seismic protection and the need for improved detection methods to enhance the efficacy of termite spot-treatment methods, it is imperative that developers and manufacturers begin such additional detection studies. Emphasis should be directed toward comparative field evaluations of all potential termite detection technologies, inclusive of odor, acoustical, microwave, and laser.
Literature Cited
1. Brier, A. N., W. A. Dost, and W. W. Wilcox. 1988. Characteristics of decay and insect attack in California homes. Calif. Agric. 42(5): 2122.
2. Caruba, A. 1981. This beagle sniffs out termites. Pest Control 49(2): 15-16.
. Fincannon, D. W. 1995. The scent of a termite. Pest Control Technology 23(7): 70-72, 74.
'. Fincannon, D. W. and A. L. Fincannon. 1995. Effectiveness of bloodhounds to demarcate specific Reticulitermes fZauipes colonies around residential structures. Submitted paper Section D: Medical & Veterinary, 1995 Entomological Society of America Annual Meeting, Las Vegas, Nevada, December 17-21, 1995.
5. Fujii, Y., M. Noguchi, Y. Imamura, and M. Tokoro. 1989. Detection of termite attack in wood using acoustic emissions. International Research Group on Wood Preservation, Document IRG/WP/2331.
6. Fujii, Y., M. Noguchi, Y. Imamura, and M. Tokoro. 1990. Using acoustic emission monitoring to detect termite activity in wood. Forest Prod. J. 40: 34-36.
7. Jandel Scientific. 1991. SigmaPlot, scientific graph system for the Macintosh, version 4.11 ed. Jandel Scientific, Corte Madera, CA.
8. Khalil, M. A. K., R. A. Rasmussen, J. R. J. French, and J. A. Holt. 1990. The influence of termites on atmospheric trace gases: CH4, CO2, CHCL3, N2O, CO, H2, and light hydrocarbons. J. Geophy. Res. 95(D4): 3619-3634.
9. LaFage, J. P. and W. L. Nutting. 1978. Nutritional dynamics in termites, pp. 165-232. In M. V. Brian [ed.], Production ecology of ants and termites. Cambridge University Press, London.
10. Lemaster, R. L., F. C. Beall, and V. R. Lewis. 1997. Detection of termites with acoustic emission. Forest Prod. J. 47(2): 75-79.
11. Lewis, V. R. 1997. Alternative control strategies for termites. J. Agric. Entomol. In press.
12. Lewis, V. R. and R. L. Lemaster. 1991. The potential of using acoustical emission to detect termites within wood, pp. 34-37. In Proceedings of the symposium on current research on wooddestroying organisms and future prospects for protecting wood in use, 13 September 1989, Bend, OR.
13. Lewis, V. R. and M. I. Haverty. 1996. Evaluation of six techniques for control of the western drywood termite (Isoptera: Kalotermitidae) in structures. J. Econ. Entomol. 89: 922-934.
14. Lewis, V. R., R. L. Lemaster, F. C. Beall, and D. L. Wood. 1991. Using AE monitoring for detecting economically important species of termites in California. International Research Group on Wood Preservation, Document IRG/WP/2375.
15. Lipman, E. A. and J. R. Grassi. 1942. Comparative auditory sensitivity of man and dog. Amer. J. Psych. 55: 84-89.
16. Noguchi, M., Y. Fujii, M. Owada, Y. Imamura, M. Tokoro, and R. Tooya. 1991. AE monitoring to detect termite attack on wood of commercial dimension and posts. Forest Prod. J. 41: 32-36.
17. Richardson, P. R. K. 1985. The social behavior and ecology of the aardwolf, Proteles cristatus (Sparrman, 1783) in relation to its food resources. Unpublished Ph.D. dissert., University of Oxford, Oxford, 289 pp.
18. Robbins, W. P., R. K. Mueller, T. Schaal, and T. Ebeling. 1991. Characteristics of acoustic emission signals generated by termite activity in wood, pp. 1047-1051. In Proceedings IEEE Ultrasonics Symposium, 8-11 December 1990, Orlando, FL.
19. SAS Institute. 1985. SAS/STAT guide for personal computers. Version 6 ed. SAS Institute, Cary, NC.
20. Scheffrahn, R. H., W. P. Robbins, P. Busey, N.-Y. Su, and R. K. Mueller. 1993. Evaluation of a novel, hand-held, acoustic emissions detector to monitor termites (Isoptera: Kalotermitidae, Rhinotermitidae) in wood. J. Econ. Entomol. 86(6): 1720- 1729.
21. Wallner, W. E., and T. L. Ellis. 1976. Olfactory detection of gypsy moth pheromone and egg masses by domestic canines. Environ. Entomol. 5: 183-186.
22. Welch, J. B. 1990. A detector dog for screwworms (Diptera: Calliphoridae). J. Econ. Entomol. 83: 1932-1934.
23. Zar, J. H. 1984. Biostatistical analysis. Prentice-Hall, Des Moines, IA.
Footnotes
The authors are, respectively, Assistant Extension Specialist, Insect Biology, Dept. of Environmental Sciences, Policy and Management, Univ. of California, Berkeley, CA 94720; Research Associate, Dept. Wood and Paper Science, North Carolina State Univ., Raleigh, NC 27502; and Farm Advisor, Cooperative Extension, San Joaquin County, Stockton, CA 95205. The authors wish to thank Robert Outman and his staff at Tadd Services Corporation, and Warren Armstrong for their cooperation. This research was supported in part by the Univ. of California Statewide IPM Project, Univ. of California, Davis.
Figure Captions
Figure 1. Wooden block used to test detection ability of beagles and termite odor device. The size of the block is 115 mm by 87 mm by 37 mm. A 20-mm diameter drilled hole was covered with a 50-mesh galvanized screen. Blocks containing termites were stored in 15-cm diameter clear plastic dishes.
Figure 2. Placement of electronic odor device sensor near screened opening in center of wooden block. The sensor is approximately 5-cm long and 2-cm in diameter.
Figure 3. Cork and straw assemblage used during the 2nd electronic odor detection test to improve gas flow and restrict termite escape from blocks.
Figure 4. Percentage of blocks containing 4 different densities of subterranean termites correctly identified by beagles or two different tests representing six electronic odor devices from the same manufacturer. Columns represent averages correctly identified for 25 inspected blocks. Five beagles inspected 20 blocks each. During the 1st odor device test, one device was used to inspect 100 blocks. For the 2nd test, five devices were used to inspect a total of 100 blocks: 20 per device. Vertical bars on columns represent standard deviations.
Figure 5. Frequency distribution of the percentage of blocks correctly identified by each beagle (Dog 1 to Dog 5) in the 1st test, and electronic odor devices (Device 1 to Device 5) in a 2nd test. A repeated measures design was used to present 20 blocks, containing varying numbers of subterranean termites plus controls, to each dog or device. Total number of blocks tested in both tests was 200.
Fig 1
Fig 2
Fig 3
Fig 4
Fig 5