Large-scale performance assessment of the BG-Counter 2 used with two different mosquito traps

doi:10.21203/rs.3.rs-4003006/v1

Download PDF

Research Article

Large-scale performance assessment of the BG-Counter 2 used with two different mosquito traps

https://doi.org/10.21203/rs.3.rs-4003006/v1

This work is licensed under a CC BY 4.0 License

You are reading this latest preprint version

Background:

Mosquitoes are important vectors of pathogens. They are usually collected with CO₂-baited traps and subsequently identified by morphology. This procedure is very time-consuming. Automatic counting traps could facilitate timely evaluation of the local risk for mosquito-borne pathogen transmission or decision-making on vector control measures, but the counting accuracy of such devices has rarely been validated in the field.

Methods:

The Biogents (BG)-Counter 2 automatically counts mosquitoes by discriminating the size of captured objects directly in the field and transmits the data to a cloud server. To assess the accuracy of this counting device, 27 traps were placed at 19 sampling sites across Germany and used in daily, weekly or bimonthly intervals from April until October in 2021. The BG Counter 2 was attached to a CO₂-trap (BG-Pro trap = CO₂-Pro) and the same trap converted to also attract gravid mosquitoes (upside down BG-Pro trap with a water container beneath, CO₂-Pro-gravid). All captured mosquitoes were identified by morphology. The number of females (unfed and gravid), mosquito diversity and the number of identified specimens in relation to the counting data of the BG-Counter were compared between both trapping devices to evaluate sampling success and counting accuracy.

Results:

In total 26,714 mosquitoes were collected during 854 trap days. The CO₂-Pro-gravid trap captured significantly more mosquitoes per trap day for all specimens, gravid females and non-gravid females, while there was no difference in the mosquito diversity. The linear model with the captured mosquitoes as response and the counted specimens as a predictor only explained little of the variation within the data (R² = 0.16), but per individual trap the value could reach up to 0.62 (mean R² = 0.23). The counting accuracy for the daily samples had a significant positive correlation with sample size, resulting in a higher accuracy for the CO₂-Pro-gravid trap and higher accuracy for sites and sampling months with high mosquito abundance.

Conclusion:

While the accuracy of the BG-Counter 2 is quite low, the device is able to depict the mosquito phenology and inform about local population dynamics.

mosquito trap

automatic counting

accuracy

Culex

CO2-trap

gravid trap

Mosquitoes are important vectors of pathogens with malaria parasites or dengue virus having the highest relevance on the global scale. It is estimated that these two pathogens alone result in more than 500 million annual cases worldwide, particularly in tropical and subtropical regions [1, 2]. However, especially due to climate warming, there are also emerging mosquito-borne pathogens in Central Europe such as the West Nile virus and the Usutu virus [3, 4]. Since the first emergence of Usutu virus (2011/2012) and West Nile virus (2018) in Germany, both viruses are regularly detected in mosquitoes, horses, birds and humans [5, 6].

Mosquito monitoring and surveillance programs provide the foundation to identify the spatial-temporal risk of mosquito-borne pathogen transmission. Thereby, mosquitoes are usually collected with CO₂-traps in the field and subsequently identified and screened in the laboratory. This process results in a significant delay between the collection and the subsequent identification and pathogen screening of mosquitoes. Automatic traps for counting mosquitoes could significantly reduce the time lag, at least providing mosquito data near real-time directly from the field. This readily available data could, for instance, allow the rapid evaluation of the success of vector control measures, or the estimation of local pathogen transmission risk using mosquito abundance data.

There are a few promising automatic counting devices for mosquitoes. These include different prototypes developed by Chen et al. [7], Lai et al. [8] and Gonzales-Perez et al. [9]. Currently, the BG-Counter (Biogents, Regensburg, Germany) is the only commercially available automatic mosquito counting system. The techniques behind these counting devices are very similar. Infrared light is sent to phototransistors, which measure the obstruction of light caused by a passing insect. These measurements include the object size and wing beat frequency. Through the application of filters and models, underlying information is extracted and subsequently passed to a classification algorithm. Besides the BG-Counter, the prototypes of Lai et al. [8] and Gonzales-Perez et al. [10]are the only ones with published data from the field. For the prototype of Lai et al. mosquitoes were collected at two sites in China from May to August 2021. The average daily counting accuracy was 79.4% on open field and 64.9% near a residence. For the prototype of Gonzales-Perez et al., mosquitoes were collected at two sampling sites in Spain. At site one from July to October, 2021 with an average daily counting accuracy of 89.1% and at site two from June until September 2022 with an average daily counting accuracy of 88.1%. Additionally, this prototype is able to distinguish between the genera Aedes and Culex. The BG-Counter was used for a wider range of applications. For instance, it was used to assess the efficacy of insecticide barrier treatments in remote areas as shown for an inhabited island in a marine bay in Australia [11] or in more urban settings as in Illinois, United States of America [12]. Another example is the evaluation of the impact of a hurricane on the mosquito populations of an inhabited island in the United States of America [13] or to assess the dispersal of mosquitoes from highly productive breeding sites on bay islands to the mainland in Australia [14]. The most comprehensive evaluation of the BG-Counter took place in North Carolina, United States of America [15]. Five BG-Counters were placed in five different counties. The mean daily accuracy ranged from 9.4 to 80.1%. Linear regressions between the BG-Counter and actual mosquito counts resulted in correlation coefficients ranging from 0.0085 to 0.95 depending on the sampling site. However, so far, there are no evaluations of the performance BG-Counters for Europe.

The here presented study is the first large-scale systematic evaluation of the BG-Counter, using 27 BG-Counter deployed over 19 sampling sites in Germany over one complete field season. We aimed to analyse the counting accuracy of the BG-Counter 2, when using it as a standard version with a CO₂-lured BG-Pro or a CO₂-lured BG-Pro combined with a gravid trap. We compared the trapping efficiency of the two trap versions for the total number of mosquitoes, blood-fed and gravid specimens and species diversity. In addition, we analysed the counting accuracy of both trap versions in dependence of sampling site, time period, and sample size.

The BG-Counter uses infrared light emitting diodes and light detectors to discriminate mosquitoes from other objects [16]. The signal detected by the light sensors is dependent on size and wingbeat frequency of the insects. Additionally, the BG-Counter is fitted with sensors for temperature, relative humidity, and ambient light. By using a 4G cellular communication module and a SIM card, it can transmit all data to an online server every 15 minutes. Thereby, the BG-Counter can also be controlled remotely, i.e. the fan can be switched on or off and the CO₂ outlet can be opened or closed.

Two different traps were used with the BG-Counter (Fig. 1). The standard version, as suggested in the BG-Counter user manual, consisting of the BG-Pro, BG-Counter 2 and BG-Trap Station (CO₂-Pro) and a second version converting the standard version into a combination of a CO₂ and gravid trap (CO₂-Pro-gravid). For the gravid version, the entire trap was turned 180°. Beneath the gravid trap, a water container (2.6 l) is placed with approximately 60 g of hay pellets and a tablet containing toxins of Bacillus thuringiensis israelensis (Culinex Tab plus, Becker GmbH, Ludwigshafen am Rhein, Germany), preventing the development of mosquitoes. For the CO₂-Pro-gravid version, the fabric trap body is not attached since the catch bag otherwise would be squashed by its weight (Fig. 1). Both, the standard and CO₂-Pro-gravid version, were used with the adjustable pressure regulator set to 1.5 kg/day. As a technical note, the BG-Counter cannot simply be turned around, since the sensor openings would be exposed to rain allowing the entry of moisture (Additional file 1: Fig. S1). The entering funnel has to be replaced by a thread adapter enabling the connection of the BG-Counter to the BG-Pro trap in upside down position. For this purpose, new holes were drilled into the thread adapter and countersunk since the screw heads would prevent the mounting to the trap.

In 2021, mosquitoes were collected at 19 different sampling sites in Germany (Fig. 2). A total of 27 traps were deployed, comprising of ten CO₂-Pro traps and 17 CO₂-Pro-gravid traps. Eight sampling sites were equipped with both trap versions, nine sites were exclusively equipped with the CO₂-Pro-gravid version and two sites were equipped with the CO₂-Pro trap version only. During each sampling event, the traps were running for approximately 24 hours. A unique label was placed in the capturing net and preserved at -20°C until further analysis. The sampling took place from April until October, and was mainly conducted in private gardens in cooperation with volunteers. Six sampling sites were sampled as often as possible (nearly daily), five sampling sites were sampled weekly and eight on a biweekly basis. The distance between the two trap versions was not standardised but chosen regarding the available space, access to electrical sockets and the convenience of the voluntary helpers. It ranged between 5 and 20 m. At one site, the distance was shorter but the traps were separated by a garden shed. Thus, an interaction between the traps at the sampling sites for the trap comparison cannot be completely excluded. A 12V portable freezer (Dometic CFX3 55, Dometic, Solna, Sweden) was used to maintain the cold chain during transport to the laboratory. All female mosquitoes were identified by morphological analysis, using a taxonomic key [17]. Additionally, the blood-fed status of the caught mosquitoes was assessed using the Sella score [18], categorizing mosquitoes as unfed (Sella score 1) or as blood-fed from freshly engorged to gravid (Sella score 2–7).

The BG-Counter data were downloaded as CSV file and the same unique ID was generated as used for the capturing nets. The unique ID was then used to align the identification data with the BG-Counter data. All counted and identified mosquitoes were summarized per sampling event. Additionally, the counting accuracy was calculated using the formulas published by Day et al. [15]. In the case the counter undercounted the mosquitoes, the formula was the automatic count divided by the manual count times 100. In case the BG-Counter overcounted the actual number of mosquitoes, the formula was the manual count divided by the automatic count times 100. In order to assess if the BG-Counters transmitted data successfully, we compared the amount of morphologically identified trap days with the trap days delivered by the BG-Counter. To analyse the differences in the trapping efficiency and counting accuracy between CO₂-Pro and CO₂-gravid traps, only the nine sampling sites equipped with both trap versions were considered and only the trap days with corresponding data of both trap versions. Additionally, the mean accuracy, the mean number of caught mosquitoes and the mean number of mosquitoes caught with a Sella score above 1 was calculated and compared between the two trap versions. For each sampling event, the Shannon diversity index was calculated and statistically compared between the two trap versions using a Kruskal-Wallis tests. This statistical analysis only considered specimens that were identified to the lowest taxonomic level possible [17]. Only for sampling sites with more than ten sampling events per season, spearman correlation coefficients and linear models were calculated for each trap and sampling site to statistically analyse the relationship between the manually identified and automatically counted mosquitoes. A categorical factor was generated, dividing all sample sizes in small (0–10 mosquitoes/24 hours) and large (> 10 mosquitoes/24 hours), since higher accuracies were observed during times of higher mosquito abundance. A linear model with all available data points was fitted with the identified mosquitoes as response and BG-Counter counts as predictor to evaluate the general correlation between both. To resolve the cause of differences in accuracies a binomial generalized linear model was fitted with the accuracy as response and the number of identified mosquitoes and the used trap version as predictors. A second binomial generalized linear model was performed using the accuracy as predictor and the month of capture as response. All computational analysis was performed in R (Version: 4.2.2) using the R-Studio IDE (Version:2022.12.0) [19]. Additionally, functions from the following packages were used for data preparation, visualization and analysis: rstatix [20], dplyr [21], terra [22], sp [23], ggmap [24], ggspatial [25], ggplot2 [26], ggdist [27], ggpubr [28], lubridate [29], tidyr [30], vegan [31] and scales [32].

In 2021, 26,714 mosquitoes were captured during a total 854 trap days with both trap versions. The average success rate of data transmission of the 27 BG-Counters was 85.9% (Additional file: Tab. S1). While 17 had a 100.0% success rate, the success rate of the ten remaining BG-Counters ranged from 17.6–93.3%. The mosquitoes belonged to at least 18 species of five genera (Table 1). Culex pipiens s.s./Cx. torrentium was the most abundant mosquito taxon with a total of 5,782 specimens (54.9%) for the CO₂-Pro trap and 12,282 specimens (75.9%) for the CO₂-Pro-gravid trap, followed by Aedes vexans (CO₂-Pro: 2,693 specimens, 25.6%; CO₂-Pro-gravid: 596 specimens, 3.7%). All other taxa were considerably less frequent (maximal 310 specimens, maximal 1.9%).

Table 1

Number of caught mosquito species and corresponding percentages in brackets for both used trap versions: CO2-Pro and CO2-Pro-gravid.
Taxon	CO₂-Pro	CO₂-Pro-gravid
Aedes albopictus	0 (0)	116 (0.7)
Ae. annulipes s.l.	45 (0.4)	11 (0.1)
Ae. caspius	4 (0)	16 (0.1)
Ae. cinereus	7 (0.1)	2 (0)
Ae. communis	1 (0)	0 (0)
Ae. geniculatus	5 (0)	3 (0)
Ae. japonicus	31 (0.3)	45 (0.3)
Ae. punctor	1 (0)	2 (0)
Ae. rusticus	12 (0.1)	10 (0.1)
Ae. spec.	36 (0.3)	97 (0.6)
Ae. sticticus	7 (0.1)	310 (1.9)
Ae. vexans	2693 (25.6)	596 (3.7)
Anopheles claviger s.s./An. petragnani	52 (0.5)	27 (0.2)
An. maculipennis s.l.	36 (0.3)	17 (0.1)
An. plumbeus	7 (0.1)	2 (0)
An. spec.	20 (0.2)	18 (0.1)
Coquillettidia richiardii	56 (0.5)	43 (0.3)
Culiseta annulata	166 (1.6)	242 (1.5)
Cs. morsitans	3 (0)	2 (0)
Cs. spec.	0 (0)	1 (0)
Culicidae	832 (7.9)	740 (4.6)
Culex modestus	8 (0.1)	5 (0)
Cx. pipiens s.s/ Cx. torrentium	5782 (54.9)	12282 (75.9)
Cx. spec.	588 (5.6)	1408 (8.7)
Cx. territans/Cx. hortensis	1 (0)	0 (0)
unidentified males	130 (1.2)	196 (1.2)
Sum	10523	16191

The dataset for the trap comparison contains 283 trap days for each trap version. The BG-Counter generally overestimated the amount of the captured mosquitoes: 90.8% of the trap days for the CO₂-Pro were overestimated and 91.2% for the CO₂-Pro-gravid. For 8.1% of the trap days the number of specimens was underestimated for both traps and accurate for 1.1% of trap days (CO₂-Pro) and 0.7% (CO₂-Pro-gravid). The CO₂-Pro trap captured significantly less mosquitoes per trap day (mean number of specimens = 14.0, 95% confidence interval (CI95) = 10.8–17.3 vs. 32.1, CI95 = 27.0–37.2; Kruskal-Wallis, Chi-squared = 68.3, df = 1, P < 0.0001; Fig. 3A), which was also true for mosquitoes with a Sella-score of one (12.8, CI95 = 9.6–16.0 vs. 21.7, CI95 = 17.3–26.1 Kruskal-Wallis, Chi-squared = 32.9, df = 1, P < 0.0001; Fig. 3B) or larger one (1.7, CI95 = 1.1–2.2 vs. 14.2, CI95 = 12.0–6.4; Kuskal-Wallis, Chi-squared = 210.9, df = 1, P < 0.0001; Fig. 3B). The difference in captured mosquito diversity per trap day was not significant for the Shannon diversity index (Kruskal-Wallis, Chi-squared = 3.2, df = 1, P = 0.0738, Fig. 3C).

A linear model for all available data with the number of captured mosquitoes as predictor variable and counted (BG-Counter) mosquitoes as response variables was significant (R² = 0.16, F _{(1, 852)} = 163.5, P < 0.0001; Fig. 4). For all 18 trap sites with 10 or more trap days, 12 trap sites showed a significant, positive linear relationship between identified and counted mosquitoes, i.e. six trap sites equipped with the CO₂-Pro and five trap sites equipped with the CO₂-Pro-gravid (Table 2). There were no significant differences between the R² values between the two trap versions (CO₂-Pro: mean = 0.29, CI95 = 0.1–0.5; CO₂-Pro-gravid: 0.17, CI95 = 0.1–0.3; Kruskal-Wallis, Chi-squared = 1.2, df = 1, P = 0.270; both traps: mean = 0.23 CI95 = 0.15–0.32; Table 2).

Table 2

Summary statistics per sampling site with information on the results of linear models (lm) and spearman correlations between the BG-Counter counts and actually captured mosquitoes for sampling sites with more than 10 trap days
Location	Trap	Trap days	Captured mosquitoes	Mean captured mosquitoes	lm: R-squared	lm: p-value	lm: Estimate	Accuracy (%)	Spearman Correlation	cor: p-value
Buchholz	CO₂-Pro	14	245	17.5	0.58	0.0015	0.58	44.1	0.78	0.0010
Fürth	CO₂-Pro-gravid	74	1233	16.7	0.16	0.0005	0.27	14.1	0.45	< 0.0001
Großheide	CO₂-Pro	12	171	14.3	0.04	0.5426	0.09	20.1	0.18	0.5850
Hamburg 1	CO₂-Pro-gravid	110	3918	35.6	0.20	< 0.0001	0.20	41.7	0.49	< 0.0001
Hamburg 1	CO₂-Pro	104	1020	9.8	0.16	< 0.0001	0.25	36.8	0.50	< 0.0001
Hamburg 2	CO₂-Pro-gravid	33	1373	41.6	0.36	0.0002	0.21	40.0	0.58	0.0004
Hamburg 2	CO₂-Pro	34	250	7.4	0.34	0.0003	0.33	39.2	0.53	0.0013
Hamburg 3	CO₂-Pro	36	530	14.7	0.48	< 0.0001	0.62	52.6	0.83	< 0.0001
Landau	CO₂-Pro-gravid	23	179	7.8	0.04	0.3787	0.11	12.2	0.26	0.2330
Landau	CO₂-Pro	22	568	25.8	0.62	< 0.0001	0.36	26.6	0.66	0.0009
March	CO₂-Pro-gravid	19	234	12.3	0.22	0.0443	0.21	23.3	0.50	0.0309
Neu Wulmsdorf	CO₂-Pro-gravid	19	1418	74.6	0.20	0.0551	0.23	47.3	0.39	0.0978
Neu Wulmsdorf	CO₂-Pro	11	550	50	0.10	0.3518	0.21	37.0	0.55	0.0816
Tübingen	CO₂-Pro-gravid	60	1471	24.5	0.21	0.0003	0.28	32.6	0.55	< 0.0001
Tübingen	CO₂-Pro	50	302	6.0	0.07	0.0683	0.27	14.5	0.25	0.0766
Varel 1	CO₂-Pro-gravid	33	1031	31.2	0.03	0.3061	0.08	33.7	0.10	0.5860
Weinheim	CO₂-Pro-gravid	80	2504	31.3	0.15	0.0005	0.42	35.9	0.26	0.0184
Weinheim	CO₂-Pro	78	1040	13.3	0.27	< 0.0001	0.34	27.1	0.48	< 0.0001
Mean (95% confidence interval)		45.1 (29.3–60.9)	1002.1 (526.5–1477.6)		0.23 (0.15–0.32)			32.2 (26.3–38.1)	0.46 (0.37–0.56)

The mean counting accuracy of the BG-Counter significantly differed between the CO₂-Pro trap (mean accuracy = 29.5, CI95 = 26.2–32.7) and the CO₂-Pro-gravid trap (mean accuracy = 36.8, CI95 = 33.6–40; Kruskal-Wallis, Chi-squared = 11.9, df = 1, P = 0.0006; Fig. 3D). The mosquito sample size has a significant effect on the accuracy of the BG-Counter (Fig. 5). While captures with a small sample size (0–10 mosquitoes/24h) had a mean accuracy of 16.5% (CI95 = 14.6–18.4%), captures with large sample size (> 10 mosquitoes/24h) had a mean accuracy of 50.7% (CI95 = 48.1–53.2%) (Kruskal-Wallis, Chi-squared = 333.3, df = 1, P < 0.0001). The effect of the sample size also at least partly explains the differences in the mean accuracy between the sampling months (Fig. 6). It starts low in April (1.2%) and constantly increases, reaching its peak in August (47.5%). The daily accuracy and the month of capture were significantly correlated (GLM binomial, z = 7.05, P < 0.0001). A joint analysis of both variables indicated that the number of collected mosquito specimens had a significant impact on the accuracy (GLM binomial, z = 4.02, P < 0.0001; Fig. 7), while trap version had no significant impact (GLM binomial, z = -0.27, P = 0.789; Fig. 7).

The linear model with the captured mosquitoes as response and the counted specimens as predictor only explained little of the variation within the data (R² = 0.16), but per individual trap the value can reach up to 0.62 (mean R² = 0.23). As already emphasized by Day et al. [15], this very low accuracy demonstrates that the BG-Counter does not accurately count the absolute number of collected mosquitoes. However, the R²-value per individual trap shows that the information from the BG-Counter informs us in a predictable manner about the local increase and decrease of mosquito abundances, e.g. for the analysis of phenological patterns. The accuracy of the traps (12–53%) was lower than presented in previous studies: 9.4–80.1% [15] and 62.3–98.7% [33]. The lower number of collected mosquito specimens (mean = 23.5) in our study likely explains the lower accuracy compared to previous studies with much higher numbers: mean = 678.8 [15] and mean = 151.4 [33]. This effect is not so pronounced in the prototype Gonzales-Perez et. al [10], since the daily average caught mosquitoes were 61.2 and the overall average accuracy was 88.1. However, a slight trend can be observed for an increase in accuracy with mosquito abundance. For the BG-Counter this relationship can also be observed for the mean number of caught mosquitoes and counting accuracy per month, where the accuracy was highest during the months of high mosquito abundance, i.e. July-September. In addition, another explanation for the low accuracy in the beginning of the season could be the high abundance of misclassified non-target organisms during springtime, which were not analysed in this study. However, this would also not explain the decreasing accuracy during autumn. The significant correlation of the BG-Counter accuracy with the sample size also result in site-specific dependencies of the BG-Counter accuracy in our dataset. Samples with large sample sizes generally have higher counting accuracies, resulting in higher accuracies for sampling sites with an overall higher mosquito abundance. This was previously also observed for North Carolina, where the three highest mean accuracies were seen at the three sites with the highest numbers of daily mean captures [15].

The modification of the CO₂-Pro trap to a combination of CO₂ and gravid trap was very effective. In line with previous studies comparing gravid traps with CO₂-baited traps, the capture rates of gravid Culex females with the CO₂-gravid traps was significantly higher than for the standard CO₂-only version [34, 35]. Moreover, the CO₂-Pro-gravid also attracted more non-gravid female mosquitoes. Downward suction of a trap as we have done for the CO₂-Pro-gravid has rather shown to decrease the success to collect mosquitoes [36]. However, a higher sampling success was also observed when combining gravid-traps with lures for host-seeking mosquitoes [37]. Therefore, although the exact reasons are unclear, the container with hay-infusion is probably the reason for the increase in host-seeking mosquitoes for the gravid CO₂-Pro-gravid. We found no statistically significant difference between the counting accuracy of the BG-Counter between the two trap versions CO₂-Pro and CO₂-Pro-gravid. Nevertheless, although not quantified in this study, a disadvantage of the BG-Pro-gravid trap might be the collection of a larger proportion of non-target species, which could make sorting costlier. Non-target insects searching for breeding sites or a water source are attracted by the container with hay infusion.

Finally, a few technical notes. The distance between the traps varied from sample site to sample site, but we placed the traps in similar surroundings, e.g. along a hedge. However, in general strong differences can be expected for the number of captured mosquitoes also for traps in close proximity [38]. Unknown microclimatic differences such as wind speed, temperature and humidity have been shown to have large impacts on the capture of mosquitoes in different life stages [39]. The BG-Pro trap has a relative strong fan and is therefore perceived louder than other mosquito traps. The same applies to the sound of the magnet valve of the BG-Counter, which was often recognized as nuisance. For an average of 15% of all sampling events the BG-Counter did not manage to establish an internet connection, resulting in captured mosquitoes without corresponding counter data.

The CO₂-Pro-gravid trap represents a useful modification of the standard CO₂-Pro trap version, colleting a greater number of mosquitoes, particularly gravid Culex species. The accuracy of the BG-Counter was sufficient to capture the phenology of mosquitoes but varied considerably between the sampling sites and month. The counting accuracy seems to be strongly influenced by the number of collected mosquito specimens per sampling event and probably other confounding factors, which requires further research, e.g. weather conditions or bycatch.

Acknowledgements

We greatly acknowledge the voluntary helpers supporting the field work and Lisa J. Winter for her help during laboratory work.

Funding

This project is funded through the Federal Ministry of Education and Research of Germany, with the grant number 01Kl2022.

Availability of data and materials

All data are available in the manuscript and in the supplementary files.

Authors’ contributions

Conceptualization: LR, FGS, RL; data collection: LR, TS, SMMA, HJ, FGS; data analysis: LR, FGS, RL, first drafting: LR, FGS, RL; writing and editing: all authors.

Ethics approval and consent to participate

Not applicable.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.

Brady OJ, Gething PW, Bhatt S, Messina JP, Brownstein JS, Hoen AG, et al. Refining the global spatial limits of dengue virus transmission by evidence-based consensus. PLoS Negl Trop Dis. 2012; 6(8):e1760. https://doi.org/10.1371/journal.pntd.0001760
WHO: World malaria report 2022. Geneva: World Health Organization; 2022. https://www.who.int/teams/global-malaria-programme/reports/world-malaria-report-2022. Accessed 29.02.2024.
Cadar D, Lühken R, van der Jeugd H, Garigliany M, Ziegler U, Keller M, et al. Widespread activity of multiple lineages of Usutu virus, western Europe, 2016. Euro Surveill. 2017;22(4):30452. https://doi.org/10.2807/1560-7917.ES.2017.22.4.30452
Ziegler U, Lühken R, Keller M, Cadar D, Van Der Grinten E, Michel F, et al. West Nile virus epizootic in Germany, 2018. Antivir. Res. 2019;162:39–43. https://doi.org/10.1016/j.antiviral.2018.12.005
Becker N, Jöst H, Ziegler U, Eiden M, Höper D, Emmerich P, et al. Epizootic emergence of Usutu virus in wild and captive birds in Germany. PloS one. 2012;7(2):e32604.
Ziegler U, Santos PD, Groschup MH, Hattendorf C, Eiden M, Höper D, et al. West Nile virus epidemic in Germany triggered by epizootic emergence, 2019. Viruses. 2020;12 (4):448.
Chen YP, Why A, Batista G, Mafra-Neto A, Keogh E. Flying insect detection and classification with inexpensive sensors. JoVE. 2014;92:e52111.
Lai Z, Wu J, Xiao X, Xie L, Liu T, Zhou J, et al. Development and evaluation of an efficient and real-time monitoring system for the vector mosquitoes, Aedes albopictus and Culex quinquefasciatus. PLoS Negl Trop Dis 2022;16(9):e0010701. https://doi.org/10.1371/journal.pntd.0010701
Gonzalez-Perez MI, Faulhaber B, Williams M, Brosa J, Aranda C, Pujol N, et al. A novel optical sensor system for the automatic classification of mosquitoes by genus and sex with high levels of accuracy. Parasit Vectors 2022;15(1):1–11. https://doi.org/10.1186/s13071-022-05324-5
González-Pérez MI, Faulhaber B, Aranda C, Williams M, Villalonga P, Silva M, et al. Field evaluation of an automated mosquito surveillance system which classifies Aedes and Culex mosquitoes by genus and sex. Parasit Vectors 2024;17 (1):97. https://doi.org/10.1186/s13071-024-06177-w.
Johnson BJ, Manby R, Devine GJ. The use of automated traps to assess the efficacy of insecticide barrier treatments against abundant mosquitoes in remote environments. J. Med. Entomol. 2022;59 (1):384–389. https://doi.org/10.1093/jme/tjab178
Clifton ME, Xamplas CP, Nasci RS, Harbison J. Gravid Culex pipiens exhibit a reduced susceptibility to ultra–low volume adult control treatments under field conditions. J Am Mosq Control Assoc 2019;35(4):267–278
Lucas KJ, Watkins A, Phillips N, Appazato DJ, Linn P. The impact of hurricane Irma on population density of the black salt-marsh mosquito, Aedes taeniorhynchus, in Collier County, Florida. J Am Mosq Control Assoc. 2019;35(1):71–74.
Johnson BJ, Manby R, Devine GJ. Further evidence that development and buffer zones do little to reduce mosquito nuisance from neighboring habitat. J Am Mosq Control Assoc. 2020;36(3):204–207.
Day CA, Richards SL, Reiskind MH, Doyle MS, Byrd BD. Context-dependent accuracy of the BG-Counter remote mosquito surveillance device in North Carolina. J Am Mosq Control Assoc. 2020;36(2):74–80.
Geier M, Weber M, Rose A, Obermayr U, Abadam C, Kiser J, et al: The BG-Counter: A smart Internet of Things (IoT) device for monitoring mosquito trap counts in the field while drinking coffee at your desk. In: American Mosquito Control Association Conference 2016:1–2.
Becker N, Petrić D, Zgomba M, Boase C, Madon MB, Dahl C, et al. Mosquitoes: identification, ecology and control. Springer 2020.
Detinova TS, Bertram DS, Organization WH. Age-grouping methods in Diptera of medical importance, with special reference to some vectors of malaria. Monogr Ser World Health Organ. 1962;47:13–191.
R Core Team A, Team RC: R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. 2012. 2022.
Kassambara A. rstatix: Pipe-friendly framework for basic statistical tests. R package version 06 0. 2020.
Wickham H, François R, Henry L, Müller K and Vaughan D. dplyr: a grammar of data manipulation. 2023. https://CRAN.R-project.org/package=dplyr.
Hijmans RJ. terra: Spatial Data Analysis. 2023. https://CRAN.R-project.org/package=terra.
Bivand RS, Gómez-Rubio V. Applied spatial data analysis with R. vol. 747248717: Springer; 2013.
Kahle DJ, Wickham H. ggmap: spatial visualization with ggplot2. R J. 2013;5 1:144.
Dunnington D. ggspatial: Spatial data framework for ggplot2. 2023. https://CRAN.R-project.org/package=ggspatial.
Wickham H. ggplot2: Elegant graphics for data analysis. Springer, New York; 2016.
Kay M. ggdist: Visualizations of distributions and uncertainty in the grammar of graphics. IEEE Transactions on Visualization and Computer Graphics. 2023.
Kassambara A. ggpubr: 'ggplot2' Based Publication Ready Plots. 2023. https://CRAN.R-project.org/package=ggpubr.
Wickham GGaH. Dates and times made easy with [29]. Journal of Statistical Software. 2011;40:1–25. https://www.jstatsoft.org/v40/i03/.
Hadley Wickham DVaMG. tidyr: tidy messy data. 2023. https://CRAN.R-project.org/package=tidyr.
Oksanen J, Simpson G, Blanchet F, Kindt R, Legendre P, Minchin P, et al: Vegan: community ecology package, R Package Version 2.6-4. 2022.
Seidel HWaD. scales: scale functions for visualization. 2022. https://CRAN.R-project.org/package=scales.
Johnson BJ, Manby R, Devine GJ. What happens on islands, doesn’t stay on islands: patterns of synchronicity in mosquito nuisance and host-seeking activity between a mangrove island and adjacent coastal development. Urban Ecosys. 2020;23(6):1321–33.
Pezzin A, Sy V, Puggioli A, Veronesi R, Carrieri M, Maccagnani B, et al. Comparative study on the effectiveness of different mosquito traps in arbovirus surveillance with a focus on WNV detection. Acta trop. 2016;153:93–100.
Williams GM, Gingrich JB. Comparison of light traps, gravid traps, and resting boxes for West Nile virus surveillance. J. Vector Ecol. 2007;32(2):285–91.
Guindo A, Epopa PS, Doumbia S, Millogo A-A, Diallo B, Yao FA, et al. Improved BioGents® Sentinel trap with heat (BGSH) for outdoor collections of Anopheline species in Burkina Faso and Mali, West Africa. Parasite Vectors. 2021;14(1):1–14.
Liu H, Dixon D, Bibbs CS, Xue R-D. Autocidal gravid ovitrap incorporation with attractants for control of gravid and host-seeking Aedes aegypti (Diptera: Culicidae). J. Med. Entomol. 2019;56(2):576–578.
Landau KI, van Leeuwen WJ. Fine scale spatial urban land cover factors associated with adult mosquito abundance and risk in Tucson, Arizona. J. Vector Ecol. 2012;37(2):407–18.
Karki S, Hamer GL, Anderson TK, Goldberg TL, Kitron UD, Krebs BL, et al. Effect of trapping methods, weather, and landscape on estimates of the Culex vector mosquito abundance. Environ. Health Insights. 2016;10:EHI. S33384.

No competing interests reported.

Addditionalfile1.docx

Download PDF

Editorial decision: Revision requested
02 May, 2024
Reviews received at journal
30 Apr, 2024
Reviewers agreed at journal
30 Apr, 2024
Reviews received at journal
28 Apr, 2024
Reviewers agreed at journal
17 Apr, 2024
Reviewers agreed at journal
19 Mar, 2024
Reviewers invited by journal
14 Mar, 2024
Editor assigned by journal
02 Mar, 2024
Submission checks completed at journal
02 Mar, 2024
First submitted to journal
01 Mar, 2024

You are reading this latest preprint version

Large-scale performance assessment of the BG-Counter 2 used with two different mosquito traps

Status:

Version 1

Abstract

Background:

Methods:

Results:

Conclusion:

Figures

Background

Methods

Results

Discussion

Conclusion

Declarations

References

Additional Declarations

Supplementary Files

Status:

Version 1