Laboratory and Field Performance of Five Cost-Effective Commercial Light Traps for Capturing Mosquitoes In China

Mosquito traps for household use are popular because they are small, cost-effective, user friendly, and environmentally friendly. At present, there are many variations and specications of mosquito traps intended for household use on the market. Their labels claim they are powerful, but research and evaluation of their claims is lacking. This article tested the key parameters, the laboratory capture rate, and the greenhouse eld capture rate of 5 popular mosquito traps intended for household use,and compared them with the BG-trap, used by professionals to monitor mosquitoes in the eld. Results The study found that the wavelength of 395–400 nm had a better capture rate for Culex quinquefasciatus. In the laboratory experiment, the capture rate was between 34.7%-65.0%. The analysis showed that the total radiance, fan speed, and design of the air guide of the traps are important factors that affect the mosquito catch rate. Field tests in the greenhouse found that the 5 mosquito traps had low catch rates for Aedes albopictus. The average percentage of Cx. quinquefasciatus, Ae. albopictus, Anopheles Sinensis, and other ying insects captured every night was 51.76%, 25.29%, 14.12%, and 8.82%. There was no signicant difference in the capture rate of Ae. albopictus by the 5 mosquito traps in the greenhouse, while the mosquito species captured during the same period by the human landing catch method were all Ae. albopictus, suggesting that the dominant species of mosquitoes in the greenhouse was Ae. albopictus. The comparison experiment of mosquito trap 5, with the highest capture rate in the laboratory simulation and greenhouse site, and the BG-trap in the morning, afternoon, and night showed that the capture rate of the BG-trap on Ae. albopictus and Cx. quinquefasciatus was higher than that of mosquito trap 5. Combined with the results of the human landing catch method during the same period, it showed that the BG-trap can more accurately reect the composition of the mosquito community. environment where Ae. albopictus is the dominant species. The mosquito traps intended for household use can be improved by increasing the fan speed and optimizing the air guide. With a higher catch rate, the BG-trap is more suitable for mosquito monitoring than the UV-trap. read the tester data the 4 (points). anemometer the outgoing wind speed of each quadrant of the the the anemometer by the time of the anemometer at to the wind speed (m/s). measuring wind speed should be than 1 control at each and at each Morphological identication of mosquitoes captured, including mosquito species and genders, was performed using an anatomical microscope and the capture performance of the ve mosquito traps was evaluated. All statistical analyses were performed using RStudio (Version 1.2.5001, 64bit) and R (version 3.4.1, 64bit) backends. In terms of statistical signicance level, * means P < 0.05, ** means P < 0.01, *** means P < 0.001.


Introduction
Mosquito-borne diseases such as malaria, dengue fever, Chikungunya fever, and Zika are major threats to global health, especially dengue fever, which has increased 30-fold in the past 50 years (Huang, et al. 2019;Jelinek 2018;Lee, et al. 2018;Wilder-Smith, et al. 2019). Of the more than 50 mosquito-borne diseases known, vaccines are available only for epidemic encephalitis and yellow fever.
The only way to curb other mosquito-borne diseases is mosquito control (Baldacchino, et al. 2015;de Thoisy, et al. 2020;Idoko, et al. 2020;Oliveira, et al. 2019;Trovato, et al. 2020). During mosquito season or a mosquito-borne disease epidemic, it is necessary not only to implement integrated mosquito control in the external environment, but also to control mosquitoes and prevent bites in residents' homes.
In malaria-endemic Africa, indoor control methods for the malaria vector Anopheles mosquitoes are mainly indoor residual spraying (IRS), which usually requires implementation by professional pest control personnel (Bouckenooghe, et al. 2019;Hamre, et al. 2020;Hien, et al. 2020;Kakilla, et al. 2020;Lees, et al. 2020). Mosquito incense and aerosols are traditional mosquito control methods in the home environment in dengue-endemic Southeast Asia and southern China. Due to the social attention to environmental protection and health concern, the number of families who have begun to adopt non-chemical mosquito control methods is increasing. One of these methods is the use of a mosquito trap.
Mosquito traps intended for household use are based on light mosquito trapping techniques for mosquito surveillance, such as the ultraviolet light trap for mosquito surveillance issued by the Centers for Disease Control and Prevention (CDC) (Holderman, et al. 2018;Li, et al. 2016;Silva, et al. 2019;Sriwichai, et al. 2015). Mosquito traps for household use are popular because they are small, costeffective, user friendly, and environmentally friendly. The design principle of this kind of product is the same as that used by professionals. The light source used is an ordinary ultraviolet light or light-emitting diode (LED). The wavelength range of the ultraviolet light is 320 nm-400 nm, and a fan is set to form a guiding air ow, which draws the mosquitoes into the mosquito collection device, where they are trapped.
At present, there are many variations and speci cations of mosquito traps intended for household use on the market. Their labels claim they are powerful, but research and evaluation of their claims is lacking. For that reason, we selected ve popular mosquito traps costing 30 USD or less and evaluated their mosquito control performance in the home environment. We paid attention especially to how well they were made, and their mosquito capturing performance in the laboratory and in the eld to provide valuable information for the control of mosquito-borne diseases by mosquito traps for household use.

Description of the Five Light Traps and Quality Check
Source of mosquito trap For this study, we purchased ultraviolet mosquito traps intended for household use from a well-known eCommerce store in China. We selected ve of the most popular variations of mosquito traps with a price not exceeding 30 USD for research and evaluation. The product parameters are shown in Table 1. Detection of radiation wavelength of mosquito traps The radiation wavelength detection of the mosquito traps' UV light was conducted in the Mosquito Trap Quality Monitoring Laboratory of Zhongshan Protostar Optoelectronic Co., Ltd. The test method conformed to the inspection method for appliances with ultraviolet radiation lamps stipulated in Article 32 of the Chinese National Standard Household and Similar Electrical Appliances-Safety-Particular Requirements for Insect Killers (Institute, et al. 2008). The appliances were supplied with rated voltage and operated under normal working conditions. The test equipment used was the PMS-80 ultraviolet (UV)-visible (VIS)-near infrared (NIR) spectroscopy analysis system, which measures radiation at 1 m. The maximum radiation should be recorded when the measuring instrument is placed.
According to the speci c operation process of the spectrometric testing instrument, the ve variations of the mosquito traps were determined in order of number, and the data obtained after the test were analyzed. The total effective radiation was calculated by the following formula. Where: E--Effective Radiation; S λ --Relative Spectral Weight Factor; E λ --Spectral Irradiance(W/m 2 nm) Δ λ --Bandwidth(nm) When measuring spectral irradiance, the radiation required a stable light source. The effective radiation of each wavelength is calculated as the spectrum according to the ultraviolet (UV) of the spectral weight factor of different wavelengths. The total effective radiation (E) should not exceed 1 mW/m 2 .
Ultraviolet mosquito trap air suction fan speed determination The test was carried out following the stipulations of the Chinese National Standard A. C. Fans and Regulators (Institute, et al. 2018), and the test equipment used was the Hima-split anemometer AS8336. During the test, only the anemometer can be placed in front of the outlet of the fan. In the middle of the test, the tester can stay at the inlet side. The tester is only allowed to enter the fan outlet area when they need to control the speed and read the data. The tester should take minimum time to record the data and control the fan speed. The measurement begins about 20 mm from the air outlet side. For a more accurate result, the fan is sectioned into 4 quadrants (points). The anemometer is used to test the outgoing wind speed of each quadrant of the fan. Afterward, the value indicated by the anemometer is divided by the sampling time of the anemometer at that quadrant to measure the wind speed (m/s). The time used in measuring wind speed should not be less than 1 min.

Laboratory Test
Laboratory simulation eld test experiment A simulated eld test was conducted in the Pesticide Evaluation Laboratory of Ningbo Yuying Vector Biocontrol Co., Ltd. in Zhejiang Province from May 25, 2020, to July 10, 2020. The test method followed the stipulations of the Chinese National Standard Laboratory E cacy Test Methods and Criterion of Public Health Equipment-Electronic Trap for Mosquitoes and Flies (Prevention, et al. 2011). The glass test room was 3 m long, 3 m wide, and 3 m high, approximating a square room with a volume of 27 m 3 . The test insect was Cx. quinquefasciatus. For the insecticide-sensitive strain of Cx. quinquefasciatus bred in this laboratory, female adult mosquitoes 3-5 days after emergence without blood suction were selected. The test conditions comprised temperature 26 ℃ ± 1 ℃ and relative humidity 65% ± 10%. The experiment started at 5 p.m. The mosquito trap to be tested was placed in the center of the test room and the light source was set 1.5 m away from the ground. Next, we released 100 mosquitoes into the test room, closed the doors and windows, and turned on the power supply to the mosquito trap after the test insects resumed their normal activities. At 8 a.m. on the second day, we cut off the power supply and wrapped the mosquito trap in a silk yarn cage to prevent the mosquitoes from escaping. Afterward, we extracted the mosquito trap's collection device to check the number of test mosquitoes captured to calculate the capture rate.
Capture rate = number of mosquitoes captured/number of mosquitoes released in the room × 100% The test was repeated three times. The blank control was tested by turning on the fan, but not the light.

Field Test
The greenhouse eld capture test was performed for the rst time in November 2019 and for the second time in mid-July 2020. The site used for the eld tests was the greenhouse of Guangxi Pastoral Biochemical Co., Ltd. in Nanning City, Guangxi Zhuang Autonomous Region. Geographically, the test site is 108.26° longitude, 22.86° latitude, and 77 m above sea level. It has a humid subtropical monsoon climate, with an annual average temperature of 20 °C -29 °C and annual average precipitation of 1304.2 mm. The greenhouse is 21 m long, 14 m wide, and 5 m high. The average daily temperature and humidity in the greenhouse are 30 ℃ and 80%, respectively. Corn, eggplants, rice, peppers, and other crops are planted all year round.
The test method followed the mosquito trap method and the human landing catch method stipulated in the China National Standard for Vector Biodensity Monitoring Method-Mosquitoes (Department of Analytical Microbiology 2009). In the mosquito trap method, we placed the mosquito traps in a sheltered area away from any interfering light source. The mosquito traps to be tested were placed more than 15 m apart, with plants between them to prevent light from any lamp from interfering with any other lamp. The light source of the mosquito trap was placed 1.5 m away from the ground. One hour before sunset, we turned on the mosquito traps to start the test. The power remained on until 1 hour after sunrise the next day. After turning off the lamp, we wrapped the mosquito traps with a silk yarn cage to prevent the captured mosquitoes from escaping. Then we counted and categorized the number and species of female mosquitoes captured. This test was carried out in two phases, each of which was repeated three times with the trap in a xed position.
The human landing catch test was performed 30 minutes before the test of the mosquito traps. First, the monitors exposed one leg and remained stationary. Second, the species and number of mosquitoes that landed on the leg and were captured by the electric mosquito trap within 30 minutes were recorded. Lastly, the time, location, temperature, humidity, and wind speed at the beginning and end of the human landing catch test were recorded.
Comparative study of capture e cacy of the mosquito trap for household use and the BG-trap The BG-trap (Biogents AG, Regensburg, Germany), used by professionals, is a foldable mosquito monitoring device, 36 cm in diameter and 40 cm in height. It's designed with a black tube in the middle of the plastic surface plate and a small fan under the tube connected to a removable mosquito collection device. Bait called BG-Sweetscent is placed in the container, which releases a mixture of lactic acid, ammonia, and hexanoic acid that mimics the odor on the surface of human skin to attract mosquitoes into the pipeline. Lastly, the fan air ow draws the attracted mosquitoes into the mosquito collection device. Unlike the ve popular mosquito traps tested, it does not use a light to attract mosquitoes. It uses scent instead (Bhalala and Arias 2009;Ponlawat, et al. 2017).
Of the ve mosquito traps, we selected the one with the highest capture rate for comparative testing with the BG-trap. The test was conducted in the same greenhouse as the previous experiment. Both mosquito control devices were placed on a shelf 1.5 m away from the ground. They were placed 15 m apart from each other. The tests were conducted three times per day. The testing periods for each day were as follows: 1) 9:00 A.M. to 11:00 A.M. 2) 2:00 P.M. to 4:00 P.M. 3) 6:00 P.M. to second day morning 8:00 A.M. After each test, the positions of the two mosquito traps were exchanged. Each mosquito control device was tested three times at each location and at each period.

Species Identi cation and Statistical Analysis
Morphological identi cation of mosquitoes captured, including mosquito species and genders, was performed using an anatomical microscope and the capture performance of the ve mosquito traps was evaluated. All statistical analyses were performed using RStudio (Version 1.2.5001, 64bit) and R (version 3.4.1, 64bit) backends. In terms of statistical signi cance level, * means P < 0.05, ** means P < 0.01, *** means P < 0.001.

Product Quality
The ultraviolet wavelengths of the ve mosquito traps were measured by a PMS-80 ultraviolet (UV)-visible (VIS)-near infrared (NIR) spectroscopy analysis system. The results are shown in Table 2. According to the Chinese National Standard Household and Similar Electrical Appliances-Safety-Particular Requirements for Insect Killers (Institute, et al. 2008), mosquito traps exceeding 1 mW/m 2 total effective radiation exceed that which is allowed and are deemed unquali ed. Therefore, mosquito trap 5 is judged to be unquali ed, because its total effective radiation is 2.1980 mW/m 2 , which exceeds the standard allowance. Whereas the remaining 4 traps are quali ed. The fan speed test results of the suction fan of the traps are shown in Table 2. The average fan speed of mosquito trap 5 is 2.10 m/s, which is the highest among all the traps.

Laboratory Tests
The results of the mosquito capture rate test in the laboratory are shown in Table 3. The Shapiro-Wilk test was used to verify that the data were normally distributed. The results of variance analysis and the Tukey test showed that there were signi cant differences in the trapping rate of the ve mosquito traps (P < 0.001). The capture rate of mosquito trap 5 and 1 exceeded 50%, which was signi cantly higher than that of the other three mosquito traps.  After the Shapiro-Wilk test was performed to con rm that the data were not a normal distribution, the Kruskal-Wallis test was performed. The results are shown in Fig. 1 The results showed that mosquito trap 5 caught the most Ae. albopictus on average every night(2.67 ± 0.61). However, there was no signi cant difference in the average number of mosquitoes captured per night among the ve traps (P = 0.17). The results are shown in Fig. 1A.
The results of the number of Cx. quinquefasciatus captured showed that there were signi cant differences among the ve mosquito traps (P < 0.001). The average number of mosquitoes captured per night of trap 5 was about 8.7 times that of trap 4 (P < 0.01), about 3.9 times that of trap 3 (P = 0.031), and about 7.2 times that of trap 1 (P = 0.0065). The results are shown in Fig. 1B.
In terms of the number of An. sinensis captured, only trap 5 captured more than 1 of this species, on average, and there was no signi cant difference among the ve mosquito traps (P = 0.315). The results are shown in Fig. 1C.
The Kruskal-Wallis test result showed that there were signi cant differences among the ve mosquito traps in the average total number of mosquitoes captured per night (P < 0.001). The average number of mosquito captured per night of trap 5 was about 3.7 times that of trap 4 (P < 0.001), about 3.0 times that of trap 3 (P < 0.01), about 2.66 times that of trap 2 (P = 0.023), and about 1.8 times that of trap 1 (P = 0.039). The mosquito capture performance of trap 5 was the best among the ve traps, followed by trap 1. The results are shown in Fig. 1D.
In the rst phase of the test in the greenhouse, the capture performance of trap 5 was the best. Thus, we decided to compare the performance of trap 5 with that of the BG-trap. We renamed trap 5 as UV-trap and compared it with the BG-trap in the greenhouse18 times. After the test, a total of 457 specimens were collected.
The second phase of the test was divided into three periods: morning, afternoon, and evening. A total of 457 mosquitoes were captured, 378 of which were female (82.7%). The largest ratio of mosquito species trapped was Ae. albopictus, of which 333 were captured, accounting for more than half of the total number (72.9%), followed by 91 Cx. quinquefasciatus (19.9%) and 33 An. sinensis (7.2%). Next, we analysed the conditions under which the mosquitoes were captured in different periods. The results are shown in Fig. 2. As for the result of the human landing catch method, the average number of captured mosquitoes per time (30 min) in the morning, afternoon, and evening were 1.67 ± 0.33, 1.83 ± 0.17, 3.00 ± 0.37, respectively. All the mosquitoes captured in the three periods were Ae. albopictus.
In the morning ( Fig. 2A), the performance of the BG-trap in capturing Ae. albopictus and Cx. quinquefasciatus was signi cantly higher than that of the UV-trap. The number of Ae. albopictus captured by the BG-trap (4.83 ± 0.91) was 7.2 times higher than that of the UV-trap (0.67 ± 0.33) (P < 0.01). The number of Cx. quinquefasciatus captured by the BG-trap (3.00 ± 1.18) was 17.6 times higher than that of the UV-trap (0.17 ± 0.17) (P < 0.05).
In the afternoon (Fig. 2B), the performance of the BG-trap in capturing Ae. albopictus and Cx. quinquefasciatus was signi cantly higher than that of the UV-trap. The number of Ae. albopictus captured by the BG-trap (5.17 ± 0.75) was 30.4 times higher than that captured by the UV-trap (0.17 ± 0.17) (P < 0.001). The number of Cx. quinquefasciatus captured by the BG-trap (2.00 ± 0.37) was 6.1 times higher than that captured by the UV-trap (0.33 ± 0.21) (P < 0.01).
In the evening (Fig. 2C), the performance of the BG-trap in capturing the three species of mosquitoes was signi cantly higher than that of the UV-trap. The number of Ae. albopictus captured by the BG-trap (37.67 ± 1.67) was 5.2 times higher than that captured by the UVtrap (7.00 ± 1.15) (P < 0.001), while the number of Cx. quinquefasciatus captured by the BG-trap (6.33 ± 0.82) was 1.9 times higher than that captured by the UV-trap (3.33 ± 0.92) (P < 0.05). Lastly, the number of An. sinensis captured by the BG-trap (2.83 ± 0.60) was 5.7 times higher than that captured by the UV-trap (0.50 ± 0.34) (P < 0.01).
Further, we compared the total number of mosquitoes captured per hour in each period and obtained the result shown in Fig. 2D. The average number of mosquitoes captured per hour by the BG-trap (3.96 ± 1.03) was higher than that captured by the UV-trap (0.57 ± 0.54), and the difference was signi cant (t = 12.381, df = 25.78, P < 0.001).

Discussion
In this study, the ultraviolet light of all ve household mosquito traps had a wavelength range of 390 nm-400 nm, which conformed to the standard ultraviolet light range. According to the Pearson correlation coe cient, the correlation coe cient between the ultraviolet light wavelength of the ve mosquito traps and the mosquito capture rate in the laboratory was subtle (P > 0.05). Therefore, the ultraviolet light wavelength was not a signi cant factor in uencing the difference in the mosquito traps' capture rate.
According to Pearson correlation coe cient analysis on the air suction e ciency of the mosquito traps, the fan speed and capture rate exhibited a linear relationship, which showed that fan speed might be a crucial factor in uencing the mosquito capture performance of the traps (P < 0.05). Other researchers tested the effect of different fan speeds against the performance of the trap in capturing mosquitoes. The result showed that 1.7 m/s was the ideal suction rate to obtain a higher capture rate and lower damage to the captured mosquitoes' bodies (ZHANG Hong-xiang 2002). In our study, mosquito trap 5 had the highest capture rate with an air suction rate of 2 m/s. And the mosquitoes captured didn't show critical damage to their bodies. Therefore, we believe that whether it is a mosquito trap for household use or a mosquito monitoring trap, the mosquito capture performance can be enhanced by increasing the air suction rate.
Mosquito trap 5 and trap 1 had the highest mosquito capture rates during the laboratory eld test. This may be due to their shape and structural design, which were different from the other three mosquito traps. Trap 5 and trap 1 have inclined upward-opening entries, which means they can capture mosquitoes from 360° around the top, whereas the entries of the other three traps are located at the middle, where the air ow into the entries is parallel and thus there is a smaller capture area. The capture area might be one factor in uencing the mosquito capture rate.
In the rst phase of the greenhouse eld test, the largest ratio of one species of mosquito species captured by the ve mosquito traps was 51.76% and the species captured was Cx. quinquefasciatus, while the main mosquito species captured by the human landing catch method was that of Ae. albopictus. However, Ae. albopictus was not the main species of mosquito captured by the ve mosquito traps and did not re ect the mosquito species composition of the mosquito community in the eld. This experiment shows that the ultraviolet trap with a wavelength of 390 nm-400 nm is ine cient in capturing Ae. albopictus and is not suitable for use as a light source for mosquito traps for household use in areas where Ae. albopictus is the dominant species, such as China. It has been reported that the capture rate of a light source with 520 nm wavelength had a higher capture rate of Ae. albopictus, which deserves further study (Costa-Neta, et al. 2017).
Mosquito trap 5 and trap 1 had a relatively high capture rate of mosquitoes, which was signi cantly higher than that of the other three traps. Further, the difference was particularly signi cant in the capturing of Cx. quinquefasciatus. The total effective radiation of mosquito trap 5 exceeded the standard quite a bit, and its air suction rate was also the largest, which may be the reason for its high capture rate. Mosquito trap 1 achieved high capture e ciency under the premise of product compliance and should be an excellent choice among the ve mosquito traps for household use that we evaluated.
The published comparative test studies of the BG-trap and the UV-trap were conducted in the eld, and this experiment evaluated both in the laboratory (Hoshi, et al. 2019;Ponlawat, et al. 2017). The experiment showed that in the greenhouse, the capture rate of Ae. albopictus and Cx. quinquefasciatus by the BG-trap (used by professionals) in the morning, afternoon, and evening was signi cantly higher than that by the UV-trap (mosquito trap 5). The capture rate of the BG-trap of Ae. albopictus was higher, which was corroborated by the human landing catch method and re ected the composition of the mosquito community more accurately. In the eld of mosquito surveillance, the BG-trap is more objective and accurate than a lamp trap, and it can replace the human landing catch method.
There are few studies on the capture rate of mosquito traps for household use. This study tested the product parameters of ve popular mosquito traps, the capture rate in the laboratory and the capture rate in a greenhouse, and obtained preliminary data, which provided research and development ideas for improving the performance of mosquito traps marketed for household use in China.