Adult mosquito trap collection efficacy of selected commercially available mosquito traps and attractants were evaluated in a rice field habitat in a malarious area near the Demilitarized Zone (DMZ) at CP Greaves (US military installation) and Paekyeon-ri (Tongil-Chon), Paju County, in Northern Kyonggi Province, Republic of Korea (ROK). This study is a continuation of trap evaluations conducted during June 2000 (Burkett et al., 2001) but excludes commercially unavailable mosquito traps and those considered unsuitable for routine use for the US military mosquito surveillance program in the ROK. Moreover, only small numbers of Culex tritaeniorhynchus Giles, the primary vector of Japanese encephalitis virus (JEV) in Korea, were captured during the June trials, thus preventing an evaluation of trap efficacy against this medically important mosquito. Populations of Cx. tritaeniorhynchus in this area of Korea historically remain low until mid-July (Kim et al., 1995, 1997, 1999).

Other mosquito trapping studies have been conducted in the ROK (Lee et al., 1984; Lee and Ree, 1991; Joo and Kang, 1992; Yang and Yu, 1992; Shim et al., 1997; Kim et al., 1999, 2000; Strickman et al., 2000) with many focusing almost exclusively on the trapping results and population dynamics of Cx. tritaeniorhynchus (Self et al., 1973; Joo and Wada, 1985; Baik and Joo, 1991; Shim et al., 1990). Most of these investigators used New Jersey (NJ) light traps, black light traps, livestock-baited window traps, and/or human/livestock landing collections. An organized, ROK-wide vector surveillance program was implemented in 1969 and 1974 by the Korean Ministry of Health and Social Affairs and the US Army (Lee et al., 1971; Ree et al., 1973; Lee et al., 1984), respectively.

The poor condition of specimens collected from the NJ or other non-selective light traps often results in inaccurate species identification of morphologically similar malaria or arbovirus vectors. Additionally, the relatively small numbers of mosquitoes captured with these traditional light traps precludes adequate sample sizes for determining malaria and/or arbovirus infection rates. This is the second in a series of reports concerning the effectiveness of selected modern mosquito trap designs and attractants against currently used older techniques for medically important mosquitoes in the ROK. Additionally, hourly host-seeking activity of the commonly captured mosquitoes was recorded to identify periods of greatest human exposure to host-seeking mosquitoes.

A 7 × 7 Latin square design was initially employed to evaluate the effectiveness of selected traps and attractants for mosquito surveillance and/or control. Unfortunately, a malfunctioning Mosquito Magnet™ treatment (without octenol) prevented us from analyzing the data as a Latin Square design. Instead, trap and day effects were evaluated using a two-way ANOVA (SAS Institute, 1995). Trap data were transformed to log₁₀ (x + 1) prior to analysis. Multiple comparisons were made using the Ryan-Einot-Gabriel-Welsh multiple range test (α = 0.05).

The mosquito trap evaluation field trial was conducted during 18-26 September 2000 from 1800 to 0700 h. The position of each trap was changed nightly so that each trap would occupy every position durig each of the test periods. After each trap night, mosquito collections were placed in shipping containers over dry ice and transported to the 5^th Medical Detachment Entomology Laboratory where they were counted and identified using keys specific to Korean mosquitoes (Lee, 1998). As in Burkett et al. (2001), Anopheles species were separated and sent to the Armed Forces Research Institute of Medical Sciences (AFRIMS), Bangkok, Thailand, to determine malaria infection rates [Enzyme Linked Immunosorbent Assay (ELISA)]. Culicine mosquitoes were separated by species, placed in cryovials (30/vial), and then shipped on dry ice to the US Army Medical Research Institute of Infectious Disease (USAMRIID) for virus isolation. Both of the latter studies will be reported separately.

A paucity of reported mosquito-borne diseases among ROK/US military personnel and civilians living in Korea from the 1970s through the mid-1990s provided little impetus for government health, research or military organizations to revamp waning mosquito/vector surveillance that constitute an important part of their respective disease surveillance programs. The antiquity of current mosquito surveillance programs in the ROK prompted an evaluation of newer and potentially more effective surveillance strategies (Burkett et al., 2001). Additional evaluations for selected traps were performed for late season mosquitoes (specifically, Cx. tritaeniorhynchus) to (1) compare their collection efficacy and (2) provide information for there suitability for both detection of malaria and virus infected mosquitoes.

Starting in 1993, autochthonous malaria (Plasmodium vivax) reemerged along the DMZ, and quickly spread through the civilian and military communities to become a primary health threat in Kangwon and Kyonggi Provinces bordering the DMZ (Kho et al., 1999; Ree, 2000). Prior to 1993, few authorities considered the reemergence of malaria in the ROK a reasonable possibility. Similarly, the incidence of human cases of Japanese encephalitis (JE) has largely disappeared from the ROK since the mid 1980's. Historically, JE had been one of the most important diseases in Korea with several thousand cases reported annually up until 1968 and hundreds per year up through the mid 1980's (Sohn, 2000). Through a combination of childhood vaccinations, improved public health and living conditions, insecticides, decreases in vector populations/habitat, vaccination of breeder pigs and improved production methods, the incidence of JE in the ROK decreased from > 18.5/100,000 in 1964 to < 0.02/100,000 over the past 3 decades (Baik and Joo, 1991; Sohn, 2000). In spite of consistent annual JE IgM seroconversions observed in blood from slaughtered pigs throughout the ROK (Self et al., 1973; Sohn, 2000), immunization for JE is not currently required for most US soldiers deployed to Korea. This policy is primarily based on the lack of confirmed JE cases reported for American soldiers deployed to Korea since the 1960's. While transmission may occur, it is estimated that only 1/200 to 1/1,000 infections result in apparent disease. As to whether the lack of JE cases in American soldiers has been due to implemented personal protective measures or whether soldiers have just been lucky is subject to debate. High vector populations (Kim et al., 1995, 1997, 1999) on military installations, extensive outdoor training requirements, low herd immunity for US military personnel, high rates of swine JE seroconversions and incidence of JE virus in mosquito populations should theoretically result in at least periodic symptomatic cases of JE in US soldiers deployed to Korea.

Similar to the problem of distinguishing closely related Anopheles species discussed in Burkett et al. (2001), many of medically important Culex species are also morphologically similar and relatively difficult to accurately identify. Trapping relatively undamaged specimens that preserve key identification characteristics of collected mosquitoes is an important characteristic of trap design. Similar to the previous study by Burkett et al. (2001), the Mosquito Magnet traps captured significantly larger numbers of undamaged and living mosquitoes when compared to those captured in the NJ and even the CDC/ABC style light traps where mosquitoes are often damaged when forced through high-speed fan blades.

With few exceptions (notably An. lesteri and An. yatsushiroensis), other light trap studies in the same Korean province (Sasa and Sabin, 1950; Ree et al., 1969; Ree et al., 1973; Frommer et al., 1979; Lee et al., 1984; Shim et al., 1997; Kim et al., 1995, 1999, 2000; Strickman et al., 2000) found mosquito species composition and relative seasonal abundance consistent with those of our study. Many of the same trends observed in the June trials (Burkett et al., 2001) were again evident in this study where mosquito numbers differed significantly among trap design ranging from < 1 per trap-night for the ABC trap using only light as an attractant to > 900 per trap-night for the Mosquito Magnet trap with octenol. The only mechanical trap that captured significantly more mosquitoes of a particular species than the Mosquito Magnet baited with octenol was the ABC light trap baited with dry ice that captured significantly more Cx. pipiens pallens than the other traps. As noted by Burkett et al. (2001), capture rates for both Cx. pipiens pallens and Cx. orientalis were significantly decreased for traps baited with octenol during the June trials.

During the June trials (Burkett et al., 2001), significantly more An. yatsushiroensis and Ae. vexans were captured in the Shannon trap than the Mosquito Magnet. Conversely, these trials showed the Mosquito Magnet captured significantly more An. sinensis and Cx. tritaeniorhynchus than the other traps including the Shannon trap. Agreeing with Slaff et al. (1983) and Acuff (1976), the ABC light trap baited with dry ice captured significantly more mosquitoes than the NJ or UV light traps. The ABC trap using only light again performed poorly, capturing < 1 mosquito per trap-night. No significant differences in the total number of mosquitoes collected or for any of the species were observed between the UV and the NJ light traps. This is important, as the UV trap is portable, battery operated, and desirable for operation in remote sites. The UV traps could serve as a viable trap substitute for NJ light traps or where CO₂ (dry ice or gas cylinder) or propane, is unavailable.

The use of octenol as a mosquito attractant has been well documented (Kline, 1994). The malfunction of the "octenol-less" Mosquito Magnet that would occasionally quit operating during the trap night was unfortunate, as the antagonistic/synergistic response to octenol for Cx. pipiens pallens and Cx. tritaeniorhynchus could not be conclusively determined and was not included in the statistical analysis. Even so, 10 times as many Cx. pipiens pallens were collected in the Mosquito Magnet without octenol, agreeing with the results observed in the June field trials. Conversely, far more Cx. tritaeniorhynchus were collected in the Mosquito Magnet with octenol suggesting a potential synergistic octenol affect for this species when used in combination with carbon dioxide. This finding agrees with those of Vythilingam et al. (1992) who captured greater numbers of Cx. tritaeniorhynchus in CDC light traps baited with CO₂ and octenol when compared to CO₂ baited traps without octenol. A synergistic response was not noted for any of the other captured species. Based on surveillance requirements, octenol may not be recommended for use if Cx. pipiens pallens or other repellent species are targeted. Octenol's potential synergistic activity may be beneficial for maximizing collection numbers of Cx. tritaeniorhynchus for JE infection rates and virus isolations. The synergistic/repellent characteristics of octenol-baited traps for mosquito species in Korea should be further investigated, especially since the Mosquito Magnet without octenol malfunctioned, eliminating any possibility of direct comparison during this trial.

Though human landing/biting collections were not used in our study per se, human bait was again used as a primary attractant to gather hourly flight information for host-seeking mosquitoes collected at the Shannon trap. In most studies, Cx. tritaeniorhynchus is known to be only moderately anthropophilic (Self et al., 1973; Yi and Wildie, 1983; Joo and Wada, 1985). In night landing/biting collections, relatively few Cx. tritaeniorhynchus (or Cx. pipiens) were attracted to human bait when compared to An. sinensis, even though concurrent light trap collections captured roughly equal numbers of these species (Joo and Wada, 1985; Baik and Joo, 1991; Strickman et al., 2000). However, in some ROK localities (Self et al., 1973) Cx. tritaeniorhynchus is reported to be the dominant species biting man. If the results of the Shannon trap collection in this study provide some indication of human attractiveness, Culex tritaeniorhynchus was only moderately attracted to humans when compared to the large numbers captured by the Mosquito Magnet. The potential lack of anthropophilic behavior in combination with the low symptomatic/asymptomatic virus infection ratios may be the primary reasons why cases of JE have not been reported in recent years for non-immunized US soldier populations in the ROK.

Time of year, weather conditions and other factors are obvious and well documented factors that play significant roles in mosquito diel flight and biting activities. In this study, host-seeking activity for Cx. tritaeniorhynchus and other common species demonstrated bimodal activity period from 1800 to 2000 and 0500-0600. The lack of activity throughout the night is most likely due to temperatures dropping below 15℃ nightly. While studies show that Cx. tritaeniorhynchus is active throughout the night (Joo and Wada, 1985; Baik and Joo, 1991), none of these studies present data later than August. Though the Shannon trap was useful for estimating host-seeking activity for most species, it did not effectively attract Cx. pipiens pallens or Cx. orientalis. Uniform diel host-seeking activity patterns such as those observed for most species during this late season field trial greatly simplifies the proper timing of ultra-low-volume (ULV) adulticide applications or other management efforts to correspond with those of peak mosquito and human activity.

Based on the seasonal disease risk analysis, the total number of mosquitoes, and/or the number of Cx. tritaeniorhynchus collected in NJ light traps, Yi et al. (1988) developed trap indices for implementing mosquito control measures at US military installations in the ROK. A modified trap index of ≥ 25 females/trap/night (1 May to 31 July) and ≥ 10 females/trap/night (1 August to 15 October) is used by US military pest management/preventive medicine personnel before control measures are implemented. An evaluation of the correlation of human disease risk, biting activity and trap collection indices for more efficient Mosquito Magnets or other traps employing counter flow technology for mosquito/vector surveillance at all or select installations (especially in malarious and JEV areas near the DMZ) are required if these traps are employed by the US military. Changing traps to one of the more efficient Mosquito Magnet traps will not solve the present vector-borne disease threat, but would expand the current mosquito/vector surveillance program into a more comprehensive disease surveillance program. Furthermore, their portability and lack of electrical requirements allows for collections during and prior to local field training exercises to develop trends for estimating population densities, malaria and arboviral infection rates, etc., to be used for estimating risk of human disease and implementing vector and disease control strategies.

The unique mosquito fauna of the ROK make it critical that tests are conducted for local species and not extrapolated from species or populations elsewhere. For example, the anthropophilic behavior of Cx. tritaeniorhynchus appears to vary from one locality to another (Joo and Wada, 1985; Self et al., 1973). The characteristics of these new technologies provide many benefits to the currently used mosquito surveillance system, including live collections for arbovirus or parasite detection and isolation, collecting specimens in better condition for more reliable identification, and collecting large numbers of mosquitoes that may effectively reduce biting populations when traps are employed near human activities. Evaluating new designs and technologies where US soldiers are deployed will increase our knowledge of vectors and assist in the development and implementation of vector and disease control strategies. New technologies that are employed in highly portable, effective, and efficient traps should be considered for inclusion as part of the US military mosquito surveillance/control system.