Artificial resting sites: An alternative sampling method for adult mosquitoes

Mosquito collections are commonly conducted with baited traps predominantly attracting host‐seeking females. In contrast, resting sites are generally colonized by a broader range of the mosquito population, including a higher proportion of males and blood‐engorged females. This study evaluates the sampling success of different artificial resting sites, attached to a deciduous or coniferous tree at different heights. As standard sampling method, carbon dioxide‐baited Biogents Sentinel traps (BG traps) were operated in parallel. BG traps caught a higher number of specimens compared to the resting sites. However, the proportion of blood‐engorged females and males was higher in resting sites. More Culiseta spp. specimens were collected in resting sites compared to BG traps, but less Aedes spp. specimens. In general, fewer specimens and species were recorded in small resting sites and at top height level compared to medium or large resting sites at medium or ground level. The proportion of males was highest at the ground, while the proportion of engorged females was highest at medium and top level. Due to the higher proportion of blood‐engorged females, artificial resting sites are especially useful for studies of host‐feeding patterns or xenosurveillance. Low costs and efforts allow a cost‐effective increase of the number of resting sites per sampling site to collect more mosquitoes.


INTRODUCTION
According to the World Health Organization, vector-borne pathogens are responsible for 17% of infectious diseases worldwide causing more than 700 000 deaths every year (WHO, 2020). Thereby, mosquitoes are the most important vector group, for example, for parasites (e.g. Plasmodium spp., Filaria spp.) or viruses (e.g. dengue virus, chikungunya virus). Knowledge on the distribution and density of mosquitoes is an important information to understand and control the circulation of mosquito-borne pathogens.
Surveillance studies for mosquitoes and associated pathogens are mostly conducted with carbon-dioxide-baited traps (L'Ambert et al., 2012;Lühken et al., 2014). Additional attractants mimicking host cues (e.g. carbon dioxide or olfactory attractants) are commonly added. However, these traps predominantly bait host-seeking females, that is, a high proportion of the trapped specimens can be expected to be non-engorged females (Pezzin et al., 2016). Hence, large numbers of mosquitoes have to be collected to detect human or veterinary pathogens.
Although carbon dioxide-baited traps catch a wide range of mosquito species (Lühken et al., 2014), several alternative trapping methods are applied to collect adult mosquitoes (e.g. gravid traps or human landing catches). These methods vary in their targeted species composition and attract different proportions of the mosquito population (L'Ambert et al., 2012;Panella et al., 2011;Williams & Gingrich, 2007), for example, gravid traps are suitable to collect container-breeding female mosquitoes seeking for breeding sites (Hoel et al., 2011;Reiter et al., 1995;Williams & Gingrich, 2007). Another common method is the sampling of resting mosquitoes. Mosquitoes can be collected from natural (e.g. tree holes) or man-made resting sites (e.g. cellars, barns or pit shelters) (Bhati et al., 1989;Kweka & Mahande, 2009). However, the installation of artificial resting sites allows a standardized sampling within surveillance programmes and is beneficial, when natural resting sites are missing or difficult to find (Edman et al., 1968).
In comparison to carbon-dioxide-baited traps, mosquito collections with artificial resting sites are generally smaller in quantity of specimens and species (L'Ambert et al., 2012;Panella et al., 2011;Williams & Gingrich, 2007). Samples from artificial resting sites are considered more standardized compared to backpack aspiration in the understory vegetation (Brown et al., 2018) and have a higher proportion of gravid and engorged specimens (Brugman, 2016;Sauer et al., 2020), increasing the probability of detection of mosquitoborne pathogens. Furthermore, different studies demonstrated that collections from resting sites allow the trapping of species with a host preference for amphibians, birds or reptiles [e.g. Culex territans (Burkett-Cadena et al., 2008) or Culiseta morsitans (Edman et al., 1968;Sauer et al., 2020)], which are usually underrepresented in samples with other trapping methods. In contrast, species of the genus Aedes rather rest in the understory vegetation and are rarely caught in artificial resting sites (Bidlingmayer, 1971;Burkett-Cadena et al., 2008;Gusciora, 1971;Irby & Apperson, 1992).
Different factors are considered to affect the sampling efficiency from artificial resting sites, while a systematic evaluation of different factors is missing. Herein, we thus studied three different factors, which might influence the sampling efficiency: Size of the resting site, height level and environment of resting site installation (here: Tree type). This study was conducted in rural areas with a high number of mosquito breeding sites (e.g. water drinking troughs, ditches, etc.). One deciduous and one coniferous tree per sampling site were selected for the installation of resting sites. These are known to provide different environmental conditions, for example, microclimate or light regime (Irmak et al., 2018). Although the exact relationship of these factors is mostly unknown, various studies highlight their relevance for the resting site selection of mosquitoes F I G U R E 1 Experimental setup per sampling site. Trapping per sampling site was conducted with one Biogents Sentinel trap and 18 artificial resting sites. Resting sites of three sizes (0.4, 76 and 162 L) were attached to one coniferous and one deciduous tree at three heights (0, 1.5, 4.5 m). Vector graphics for vegetation structures from pixabay.com (Paaijmans & Thomas, 2011;Sauer et al., 2021). Following previous studies (Komar et al., 1995;Morris, 1981;Ritchie et al., 2013), we hypothesize that the number of specimens will increase with size and decreasing height level of the artificial resting sites, because larger resting sites might be better recognized by mosquitoes, and resting sites on the ground are less exposed to harsh climatic conditions (Bidlingmayer, 1971;Edman et al., 1968;Service, 1980).

Mosquito sampling and identification
In 2016 and 2017, mosquito sampling was conducted at a total of 16 sampling sites (8 per year) in North-West Germany ( Figure S1, Supporting Information). All sampling sites were located in rural regions surrounded by meadows with cattle and horses. Several artificial and temporary water bodies were available at each sampling site, for example, water drinking troughs or water-filled flowerpots.
At each sampling site, resting sites of three different sizes (162, 76, 0.4 L) were attached at three different heights (0, 1.5 and 4.5 m) to one coniferous and one deciduous tree. In addition, one carbon dioxide-baited Biogents Sentinel trap (BG trap; Biogents, Regensburg, Germany) was used per sampling site (Figure 1). At each site, both trees and the BG trap were not more than 20 m apart from each other. This reduced the impact of the surrounding land-use or microclimate conditions and enabled sampling of a closely defined mosquito population. Pop-up garden bags (Relaxdays, Halle, Germany) of two different sizes were used as large-and medium-sized artificial resting sites: 162 L (height: 63 cm, diameter: 56 cm) and 76 L (height: 44 cm, diameter: 46 cm) ( Figure S2). Black plastic cups (0.4 L, height: 9 cm, diameter: 7.5 cm) made from downpipes (Fallrohr NW 80 RAL 8028 anthrazit 2 m, Grüna, Germany) were attached as small resting sites.
In addition, the compass orientation (8 levels: North; North-East; East; North-West; West; South; South-East; South-West) and angle of declination (range: À100 to 12.5) were recorded for each artificial resting site, which differed depending on the natural characteristics of the trees, for example, distribution of branches to attach the resting sites. backpack aspirator in field studies (Maia et al., 2011). BG traps were run biweekly for 24 h in the week before the sampling of the resting sites to reduce direct interference between the two sampling methods. Samples were brought to the laboratory where mosquitoes were killed by freezing À20 C). All mosquito specimens were selected, enumerated, differentiated by sex and identified to lowest taxonomic level possible by using the identification key of Becker et al. (2010).

Statistical analysis
In order to account for sampling site-specific differences, the total number of mosquito specimens and the number of female mosquitoes for each artificial resting site was standardized relative to the samples with the carbon-dioxide-baited BG trap. Accordingly, for each sampling site, specimens per artificial resting site were divided by the sum of specimens per artificial resting site and BG trap. In addition, for each resting site per sampling site, we calculated (1) the proportion of overlapping taxa (collected with both sampling methods) and nonoverlapping taxa (only collected from the resting site) relative to the BG trap, (2) the proportion of blood-fed females and (3) (Hothorn et al., 2008). Data manipulation and visualization were conducted with the R packages magrittr (Bache & Wickham, 2014), plyr (Wickham, 2011) and tidyr (Wickham & Henry, 2017) and ggplot2 (Wickham, 2009).

RESULTS
In total, 20 592 specimens were collected, 15 950 mosquito speci- and size had a statistically significant impact on the proportional number of collected total specimens and females per resting site relative to the parallel operated BG trap (Figure 2, Tables 1 and 2, D 2 for both glms = 0.32). In contrast, the number of collected mosquitoes was not statistically influenced by type of tree, compass orientation or angle of declination (Table S1) Table S1).
In total, 18 taxa of 5 genera were detected. In both trapping F I G U R E 3 Mean percentage (95% confidence interval) of the total number of overlapping species per resting site and year compared to one parallel running Biogents Sentinel trap per sampling area (n = 16). Each resting sites in three sizes were installed in a coniferous and deciduous tree at three different heights Notably, Aedes species were underrepresented in artificial resting sites (10 specimens, 0.2% of all collected specimens) compared to 335 specimens in the BG traps (2.1%). Moreover, four Aedes taxa were present in the BG traps only: Ae. rusticus, Ae. cinereus/geminus, Ae. annulipes, Ae. geniculatus. In contrast, Cx. territans was exclusively collected by means of artificial resting sites (the mean proportion of these nonoverlapping taxa was 15.5, 95% CI: 10.2-12.9%; Figure S4). Compared to small resting sites, a higher proportion of non-overlapping taxa was collected in the large and medium-sized artificial resting sites

DISCUSSION
This study systematically compared the number of mosquito specimens and species collected in artificial resting sites and carbon dioxide-baited traps. The study was carried out in North-Western Germany, using resting sites of different sizes attached at different heights in deciduous and in coniferous trees. The number of specimens and species differed according to height level and size of the artificial resting sites only. Further factors included in this study (type of tree, compass orientation and angle of declination) had no statistically significant effect. As reported by previous authors (Morris, 1981;Ritchie et al., 2013), small resting sites and artificial resting sites at the top height level collected fewer specimens and species than larger artificial resting sites at medium or ground level. In addition, the species composition of medium to large artificial resting sites attached at low or medium height was more similar to the carbon dioxide traps.
Dark colour is an effective attractant for mosquitoes (Bidlingmayer, 1971;Edman et al., 1968), which might explain why larger resting sites with a larger dark opening are better recognized by mosquitoes. As hypothesized by Komar et al. (1995), the higher number of mosquito specimens at ground levels could be driven by climatic conditions, which should be harsher for resting sites at higher levels, where they are, for example, exposed to wind (Service, 1980).
Further studies, representing a finer gradation of resting site sizes and height levels should improve our understanding of height preferences of mosquitoes and process guiding species-specific selection.
In general, there is a lack of knowledge on the influence of inclination angles, compass orientation or type of tree on resting site selection, which all did not have a statistically significant effect in this study. Downward directions increase the shaded area within the popup bags and thereby attract higher numbers of mosquito specimens in the resting sites compared to sideways installation (Sauer et al., 2020).
However, the inclination of the resting sites in our study was mostly downwards, that is, did not cover a broad range of inclinations. Few studies detected a positive effect when resting sites were facing east (Edman et al., 1968) or towards swamps (Morris, 1981). In agreement with another study in North-Western Germany (Sauer et al., 2020), the here presented study also did not find a statistically significant impact of compass direction. Trees offer shade and protection from unfavourable weather events like strong wind or rain, but this study did not find any relevance of tree type. The resting sites were not attached in forests, but in single trees or smaller tree groups with relative sparse canopy. Artificial resting sites are considered particularly suitable, when there is a lack of alternative resting sites for mosquitoes (Edman et al., 1968;Pezzin et al., 2016), but the specific impact of land-use on their sampling efficiency is largely unknown.
On average, the number of mosquito specimens collected with an individual artificial resting site was 20 times lower compared to the parallel run BG trap, matching the results of other studies using carbon dioxide as an attractant (Panella et al., 2011;Ritchie et al., 2013;Williams & Gingrich, 2007). Collections with resting sites represent a temporal snapshot only of the resting mosquito population. Mosquitoes can enter and leave the resting site at any time and their resting behaviour is further influenced by different environmental conditions, for example, temperature or light (Panella et al., 2011;Thomson, 1938). Glue (Brown et al., 2018;Degefa et al., 2019), fans (Panella et al., 2011) or lids (Burkett-Cadena, 2011) might be useful to increase the sampling efficiency of artificial resting sites.
The species composition differed strongly between both sampling methods in this study. In particular, only few specimens of the genus Aedes were sampled from the artificial resting sites (Brown et al., 2018;Gusciora, 1971), resulting in a lower number of species compared to collections with the carbon dioxide traps. In contrast, Cs.
Moreover, Cx. territans, which acquires bloodmeals from amphibian hosts (Crans, 1970), was detected exclusively in artificial resting sites, but not in BG traps. This is in concordance with previous studies, comparing different trapping methods for mosquitoes (Pezzin et al., 2016;Sauer et al., 2020;Williams & Gingrich, 2007). Trout et al. (2007 described species of the genus Aedes (Aedes albopictus and Ae. vexans) with resting preferences below 4 m in vegetation.
Irrespective of the differences in the species composition, artificial resting sites collect a different proportion of the mosquito population compared to carbon dioxide-baited traps with a considerable higher proportion of males and engorged females. Male mosquito specimens are generally not in the focus of mosquito sampling campaigns as they have a low relevance for the surveillance of pathogens. Nevertheless, for several taxa only the males allow reliable morphological differentiation of species (e.g. Ae. cinereus and Ae. geminus) (Schaffner et al., 2001). In addition, the collection of both, females and males, give comprehensive insights into mosquito population structures, for example, male mosquitoes are known to provide information on mosquito mass occurrence, allowing targeted control measurements (Fay & Prince, 1968;Sawadogo et al., 2017). As assumed for female mosquitoes, males probably use resting sites to escape unfavourable daytime conditions (Howard et al., 2011). In our study, higher numbers of males were collected in low compared to top positions, which was mainly driven by the occurrence of male Cx. pipiens pipiens. This is consistent with previous studies on Cx.
quinquefasciatus, indicating a preference for low resting site positions (Schreiber et al., 1993). As demonstrated before (Brugman, 2016;Sauer et al., 2020), resting sites also attract a considerable higher proportion of engorged females compared to carbon dioxide-baited traps. Blood-fed females are of specific interest for the xenosurveillance (Drummond et al., 2020;Tomazatos et al., 2019) and to understand the interaction between vectors, hosts and pathogens through the identification of host species. A higher proportion of engorged females for resting sites at upper height levels might be explained by ornithophilic host-seeking females at the top, like it was shown for resting Cx. pipiens s.l. and Culex restuans in proximity to bird roots (Panella et al., 2011). In general, traps at higher levels were found to increase the capture success for ornithophilic mosquitoes (Tempelis et al., 1965).
Artificial resting sites have many operational advantages in comparison to conventional mosquito traps (Burkett-Cadena, 2011;Degefa et al., 2019;Panella et al., 2011;Sauer et al., 2020). The popup garden bags we used are inexpensive, non-insecticidal, easy to store and transport. In the field, they can be easily attached to trees or on the ground, for example, using cable ties, tent pegs or stones.
Due to the weather-resistant material, the pop-up garden bags can remain in the field throughout the complete trapping season. During the term of this biennial study, not a single resting site had to be replaced. Furthermore, resting sites do not require the use of bait (e.g. light or carbon dioxide), which can break down or be used up. In addition, the usage of resting sites as trapping device saves at least 50% of travel time, because the sampling devices do not have to be switched on or off as it is the case for commonly used carbon dioxide traps. This saves both, financial expenses (e.g. for carbon dioxide) and time. To maximize the sampling success, the number of artificial resting sites per sampling site can be easily increased. Our results suggest that approximately 20 resting sites allow the collection of the same number of specimens as one standard carbon dioxide-baited trap.

SUPPORTING INFORMATION
Additional supporting information may be found in the online version of the article at the publisher's website.    Each resting site in three sizes were attached in a coniferous and deciduous tree at three different heights. How to cite this article: Jaworski, L., Sauer, F., Jansen, S., Tannich, E., Schmidt-Chanasit, J., Kiel, E. et al. (2022)