Seasonal bat activity related to insect emergence at three temperate lakes

Abstract Knowledge of aquatic food resources entering terrestrial systems is important for food web studies and conservation planning. Bats, among other terrestrial consumers, often profit from aquatic insect emergence and their activity might be closely related to such events. However, there is a lack of studies which monitor bat activity simultaneously with aquatic insect emergence, especially from lakes. Thus, our aim was to understand the relationship between insect emergence and bat activity, and investigate whether there is a general spatial or seasonal pattern at lakeshores. We assessed whole‐night bat activity using acoustic monitoring and caught emerging and aerial flying insects at three different lakes through three seasons. We predicted that insect availability and seasonality explain the variation in bat activity, independent of the lake size and characteristics. Spatial (between lakes) differences of bat activity were stronger than temporal (seasonal) differences. Bat activity did not always correlate to insect emergence, probably because other factors, such as habitat characteristics, or bats’ energy requirements, play an important role as well. Aerial flying insects explained bat activity better than the emerged aquatic insects in the lake with lowest insect emergence. Bats were active throughout the night with some activity peaks, and the pattern of their activity also differed among lakes and seasons. Lakes are important habitats for bats, as they support diverse bat communities and activity throughout the night and the year when bats are active. Our study highlights that there are spatial and temporal differences in bat activity and its hourly nocturnal pattern, that should be considered when investigating aquatic–terrestrial interactions or designing conservation and monitoring plans.


SALVARINA et AL.
Studies that show bat dependence on aquatic resources are more numerous for rivers and streams, than for lakes (36%, 23%, and 17% studies found, respectively, in the total of papers reviewed in Salvarina, 2016); therefore, lake systems and their importance for bats need further investigation.
In Europe, all bat species are insectivorous (except of the Egyptian fruit bat in Cyprus) and many of them include aquatic insects in their diets (Vaughan, 1997). The evidence for this has been mainly from identification of prey remains in feces (reviewed by Vaughan, 1997), and to a lesser degree from stable isotopes (Lam et al., 2013), molecular analyses on feces (e.g., Krüger, Clare, Symondson, Keišs, & Pētersons, 2013), and experiments (Fukui, Murakami, Nakano, & Aoi, 2006). However, the amount of aquatic insects entering the terrestrial systems that is available to terrestrial consumers and whether this food resource fluctuates seasonally is less studied (e.g., Salvarina et al., 2017), particularly with responses to bat activity. The importance of aquatic insects as a food resource for bats may differ seasonally, for example, in early spring when other prey availability is low (Fukui et al., 2006).
Aquatic resources in many areas of the world are degrading, and further deterioration in their quality is predicted (IFRI and Veolia 2015). Additionally, climate change is expected to reduce freshwater in most dry subtropical regions (Jiménez Cisneros et al., 2014).
Numerous water bodies and bat species are under conservation. All bats' dependence on aquatic systems, it is also important to know how much bats rely on aquatic resources. Studying aquatic-terrestrial interactions is an important topic in ecology with increasing interest (e.g., Gratton, Donaldson, & Vander Zanden, 2008;Bartrons, Papes, Diebel, Gratton, & Vander Zanden, 2013) and implications, such as in helping to: (1) investigate and possibly predict the effects of climate change and eutrophication of waters on terrestrial consumers, (2) study food webs, (3) manage the conservation of ecosystems and species effectively, and (4) track transfer of contaminants from aquatic to terrestrial systems (Mogren, Walton, Parker, & Trumble, 2013).
Our general aim was to investigate aquatic subsidies into terrestrial systems and the role they play to explain the activity of bats near lakes. Therefore, we measured aquatic insect emergence from three different types of lakes located in the same region. Earlier, we showed that the total annual biomass of emerging insects from the littoral zone of a low in nutrient content lake (Lake Constance) can reach 1.789 per mg 2 /year (Salvarina & Rothhaupt, 2017). These results confirmed that a considerable insect biomass, even from small or low in production water bodies, can subsidize terrestrial consumers. We also described the pattern of insect emergence every 5 days, during three seasons (Salvarina & Rothhaupt, 2017). Here, we aimed to further investigate how these insect emergence patterns explain bat activity at multi-temporal (nocturnally and seasonally) and spatial scales. Thus, parallel to the insect collections we monitored bat activity, using acoustic monitoring. Our main research question was as follows: Is there a general pattern of bat activity and insect emergence in all study lakes, independently of trophic condition, size and other characteristics? If so, then this pattern can be used as reference for future studies that might aim to predict indirect effects on bats due to climate change or due to modifications on aquatic systems and thus to insect emergence. To restrict other factors, we selected lakes located on same geographical and climatic region. As each lake had a unique pattern of bat activity, we further investigated what factors influence it.
As all eighteen species, reported in our broad study area, are insectivorous with varying degrees of specialization on aquatic or terrestrial insects (Hinweise LUBW 2013, Fledermausschutz Thurgau 2014, see Appendix S1) we predicted that insect availability and season will explain bat activity. We specifically hypothesized that in contrast to lakes with lower insect emergence, at lakes with higher insect emergence, bat activity will correlate stronger with aquatic insect activity than with aerial flying insects (terrestrial insects included). Thus far, studies often acoustically sample bat activity over a limited number of hours after the sunset. This approach might potentially ignore important bat activity displayed later in the night or before sunset. Another objective of our study was therefore to explore the nocturnal (throughout the night) bat activity patterns seasonally and spatially. Activity patterns of bats are suggested as a monitoring tool of animal responses to long-term changes in climate, as it is related to climate and weather conditions (Frick et al., 2012).

| Study sites
The study was conducted at three lakes in South Germany that has a temperate seasonal climate ( Figure 1). These three lakes (Lake Constance, Mindelsee and Siechenweiher) were chosen as representative of different nutrient content and size lakes, yet located in the same region and in the same climatic conditions. Lake Constance is a deep (max. depth 254 m), large (500 km²), prealpine, oligotrophic (low in nutrient content) lake, situated in between three countries (Germany, Switzerland, Austria). The sampling location (47 o 41′27.72″N, 9 o 12′08.18″E) was near the city of Konstanz, in Upper Lake Constance, which is less than 10 m deep in this place and is considered as a shallow area (littoral zone) (Baumgärtner, Mortl, & Rothhaupt, 2008). The shore, near our study area, was composed of forest's patches, meadows, small pastures, gardens, and orchard.
Lake Mindelsee (47°45′06.95″N, 9°01′24.80″E) is a shallower (max. depth 12 m), smaller (1.02 km 2 ), mesotrophic to eutrophic lake, included in a nature reserve. We sampled in the southern, steeper littoral zone which is bordered by a hill forested mainly with beech trees (Smukalla & Meyer, 1988). Siechenweiher (47 o 41′47.33″N, 9 o 16′.54.09″E) is a shallow (max. depth 2.5 m), highly eutrophic (Seenprogramm 2010), small (about 0.024 km 2 ) fishing pond at the edge of the town of Meersburg. It is situated between a residential area and a busy road; however, its watershed (227 ha) is composed of forests (10%) and agricultural land (75% of which 22% is meadows, 35% arable land, and 43% orchard). Siechenweiher is about 800 m far from Lake Constance's shores, nevertheless due to its different characteristics as opposed to Lake Constance, it can be correctly considered as an independent point and not just another sampling point along the shores of Lake Constance.
We conducted fieldwork during 2 years (from July through October 2011 and from April through June 2012, plus 2 samplings in May and June 2011) to have seasons covering one "bat year" (spring, summer, autumn), when bats are active.

| Insects
Emerging aquatic insects (hereafter referred to also as aquatic insects) were collected with floating traps (surface: 2,500 cm²: 50 × 50 cm) that looked like pyramids with a bottle of killing solution (either alcohol 80% or 1 alcohol: 1 ethylene glycol: 1 tap water) on the top. The traps were constructed at the University of Konstanz using as model similar traps used in other studies (e.g., Hagen & Sabo, 2012).
Three to five traps were placed in Lake Constance (at about 1, 2, 3, 6, 8 m, the water level varied about 175 cm), and three traps in the other lakes (at 1, 2, 5 m in Mindelsee, respectively, and at 1-1.5 m the two traps and at 2 m the third trap in Siechenweiher). The traps remained on the water from May to October 2011 and from April to June 2012. We sampled insects every 5 days, with some variation due to logistic issues (e.g., bad weather). We also collected separate insect samples the nights (from sunset till sunrise) that we recorded the bat activity. We calculated insect emergence in individuals per hour and per trap. Hereafter, we will refer to the emerged insects (per hr per trap) caught during the 5 days and nights as "total emerged insects" and to the insects emerged during bat recording nights (per hr per trap) as "night emerged insects." Aerial flying insects were caught using one Malaise trap constructed at the Limnological Institute, University of Konstanz (built with model the trap from Bioform ® ; 295 × 175 × 94 cm). The trap was set up at the area of the bat recordings 3 hr before the sunset (this choice was made for practical reasons, as a compromise instead of collecting insects the whole day). The first insect sample was collected at sunset. The emergence rate per hour that corresponds to these 3 hr will be referred as "day aerial insects." The sample from sunset till sunrise per hour corresponded to the "night aerial insects." The Malaise trap was randomly orientated to avoid bias due to wind and a possible corridor of flying insects. The aerial flying insect collection was conducted only when bats were recorded during April-June 2012.
F I G U R E 1 Map of the study area and all the study lakes and the sampling location in each lake: Lake Constance, Mindelsee, and Siechenweiher

| Bats
Bat activity was assessed with acoustic monitoring during three nights (from about 20 min before the sunset till sunrise) per sampling month at each lake from July to October 2011 and April-June 2012 (plus two samplings in May and June 2011). We used an automatic bat recorder, Batcorder (Ecoobs, Nurnberg, Germany), with an omnidirectional microphone, hanging on a 2-m pole, with the microphone parallel to the ground, placed about 3-4 m from lakeshore.
We used the same mode ("Auto+Timer") and the same settings of Ecoobs, Germany). Due to the different sensitivity of the recorder for different species, the recordings are not comparable between species, which was not anyway our aim. Bat activity cannot be compared between species in acoustic studies due to differences in species frequency rates and echolocation intensity (Stahlschmidt & Bruhl, 2012 and references therein). However, the recordings even having a possible bias due to differences between call characteristics of the species, they represent the bat activity situation in each location and the recordings of the same species are comparable between seasons and locations.
We followed these recommendations as much as possible. Bat activity was defined as seconds of recording of bat passes per hour of recording in each night. The time that the recording was stopped due to rain was excluded from the total recording time. We also recorded the wind on a subjective scale from 0 (no wind) to 5 (strong wind) at the time of the sunset till midnight.

| Acoustic analysis
For acoustic analysis, we used software (from Ecoobs) that is specific for recordings made with batcorder: bcAdmin for the management of recorded sessions and sequences; bcDiscriminator that recognizes and takes measurements on bat calls in each sequence; and batIdent that uses those measurements to give a potential identification (on a species or group level) with a probability of this identification to be correct. As the abovementioned programs do not permit listening to the recordings, the sequences that needed to be manually checked were exported to wav files and opened with Raven Pro (Bioacoustics Research Program 2011). All sequences identified only as "Chiroptera" or "nothing" were checked in Raven Pro. Most could be identified to the species, genus, or group level, few remained as Chiroptera and those that were noise were deleted.
The identification was performed by only one of the authors (IS) to avoid bias. For the identification, books (Tupinier, 1997;Barataud, 2002; Koordinationsstellen für Fledermausschutz in Bayern 2009) and papers (Russo & Jones, 2002;Obrist, Boesch, & Fluckiger, 2004) were used. We classified all the calls identified automatically with a probability of ≤70% in the previous lower identification level. The same was true for species, such as M. alcathoe, whose presence in the area is unlikely and has not been confirmed before (pers. comm. Wolfgang Fiedler, Max Planck Institute for Ornithology, Germany).
We grouped Pipistrellus nathusii and P. kuhlii, together, even if they were identified automatically, as due to their similarities in call characteristics, and it is very difficult to distinguish them only from echolocation calls. Nyctaloid species and Myotis species were also grouped, respectively, for further analysis, due to their similarities in call characteristics and usually the low probability that BatIdent identifies them. However, for the species list, we used calls that could be with high confidence identified to species level.
Feeding buzzes are sequences where the pulse duration, interpulse intervals, and frequency decrease (Griffin, Webster, & Michael, 1960). They are produced when a bat is hunting an insect. A number of sequences (3,249 sequences, 25% of the total number), randomly selected, covering all recording sessions were checked manually (visually and acoustically), in Raven Pro, for feeding buzzes.

| Statistical analysis
To explore the data, the bat activity and insect emergence were plotted per lake fitting generalized additive models (gam) and smooths to the data. We searched for direct relationships, between and within insect emergence and bat activity, using linear regression between total emerged insects (from 5 days and nights preceding the recording night) and night emerged insects; between insects and air temperature at the sunset; between total bat activity and activity of each species/group (P. To investigate what influences bat activity on a spatial scale, linear mixed effects models (lmer) were tested with all data together, lake as a random factor and combinations of the same parameters (insects, wind, temperature, Julian day). Then, the data collected in 2012 only were tested in a model that included also the aerial flying insects. The models with the lowest values of AIC were selected (and presented) as those explaining best the variation in the data. All analyses were performed using the statistics package R (R Core 2016) run within R Studio interface, (RStudio 2016).

| Insects
The family Chironomidae accounted for the vast majority (82.5%) of the aquatic insects caught in the floating traps during the whole time of their exposure. The night emerged insects were positively related to the total emerged insects at Lake Constance (R 2 = .771, The most abundant groups of aerial flying insects in all three lakes were Coleoptera (31%) and Chironomidae (16%). Terrestrial origin was attributed to 43% of all aerial flying insects, aquatic (mainly Chironomidae) to 17%, and the rest was not attributed to aquatic or terrestrial origin. This amount might be underestimated as the determination of aquatic or terrestrial origin was rather conservative, and all specimens that were identified only in order or suborder level were characterized as of unknown origin. The day aerial insects (caught during the 3 hr before the sunset) were more in number than the night aerial insects (from sunset to sunrise; p < .021 in all lakes).

| Bats
We recorded 13 bat species with similar numbers of species at each location (Appendix S1), during 63 nights of recording. We recorded most of the expected species in the region (Appendix S1); however, probably some species could have not been recorded due to their low calls or the height of their flight is too high (e.g., Pl. auritus The total bat activity showed seasonal fluctuations ( Figure 2); however, not significant as it was found from the glm models including season or temperature (Table 3). A bimodal pattern of activity was noted in Lake Constance and Siechenweiher with a peak in late spring and early summer respectively and a smaller peak in autumn ( Figure 2). In Mindelsee, the pattern seemed less clear and rather unimodal ( Figure 2). Differences in activity were noted among lakes and seasons (Figure 3). Significant difference, however, was only noted in the activity of Myotis spp. which was greater in summer compared to autumn in Lake Constance (χ 2 (2) = 9.09, p = .01).
The bat activity pattern throughout the night also varied both among the lakes and seasons (Figure 4). In Lake Constance, in spring and summer the greatest activity was recorded about 1 hr after sunset with a smaller peak later in the night before sunrise, while in autumn the activity was more evenly distributed throughout the night.
In Mindelsee, the peak of activity was in the second part of the night for spring and summer, before sunrise, although there was a considerable activity throughout the night. In Siechenweiher, the activity seemed to be also spread through the night, especially in spring.

| Bat activity-insects
Bat activity had weak but positive relationships with the total and night emerged insects ( Figure 5; R 2 = .079, F 1,61 = 5.236, p = .026 and R 2 = .115, F 1,52 = 6.755, p = .012, respectively), and both the day and night aerial flying insects (R 2 = .342, F 1,23 = 11.94, p = .002 and R 2 = .407, F 1,23 = 15.8, p = .001, respectively) when all lakes and  the bat activity correlated with the aquatic insect pattern (Figure 2, Table 1) showing increasing values in spring until they reach a peak in beginning of summer, then they decrease in summer and they increase slightly later in autumn. In spring, in Lake Constance and Siechenweiher the bat activity seemed to increase similarly only with the aerial flying insect numbers ( Figure 2, Table 1).
The highest emergence rate was recorded in Lake Constance, then followed Mindelsee and then Siechenweiher, while bat activity was highest in Siechenweiher than in the other two lakes (Figure 2).
The model with lake as random factor explaining best (selected with the AIC criterion) the bat activity and being biological meaningful was the one including emerged aquatic insects, wind speed, and air temperature ( Table 2). The best model (selected with the AIC criterion) with lake as random factor for the data of 2012 was the one with explanatory parameters emerged aquatic insects, aerial flying insects, wind, and air temperature. This biologically is the same with the full dataset (the one without aerial flying insects) as the aerial flying insects in a sense contain the aquatic emerged insects (that emerged earlier in the night). When the models were applied for each lake separately for the whole dataset, the models that explained the bat activity variation the best were those with insects (emerged aquatic), wind speed, and temperature in Lake Constance and Siechenweiher and the one with insects, season, and wind in Mindelsee (Table 3). But only insects and wind explained significantly the bat activity only in Lake Constance (Table 3). However, the results of the models should be considered with caution as, based on the diagnostic plots, the models did not seem to fit perfectly to the data, possibly implying that the relationships are complex and more replicates are needed or more parameters to be considered.

| Bat activity among lakes
We assessed bat activity using acoustic monitoring, an effective and noninvasive method (e.g., Lintott, Fuentes-Montemayor, Goulson, & Park, 2013). Simultaneously, we collected, counted, and identified emerging aquatic insects from the lakes and aerial flying insects from the shores where we recorded bat activity. There was no general pattern of bat activity for the studied region, and activity peaks were not necessarily dependent on insect emergence, as predicted. Interestingly, bat activity showed higher spatial than seasonal variability.
Differences in bat activity among the lakes could be related to factors, such as surrounding habitat, proximity to bat roosts and perches, commuting routes, microclimate, and wind exposure. Wind also played a significant role on bat activity, especially in Lake Constance, although we were avoiding recordings in harsh weather conditions. We speculate that waves in Lake Constance, as this place is open and more affected by wind, might explain the low bat activity there. Bats avoid rough surfaces and wavy waters because they interfere with echolocation (Warren, Waters, Altringham, & Bullock, 2000).
Lake size might also explain bat activity differences among the lakes. Although Siechenweiher had low insect emergence per square meter, the small size might have attracted bats from the surrounding area for drinking or feeding. In contrast, a large lake the size of Lake Constance, which had longer lakeshores, could have had lower bat density at the site of our recording location, compared to Siechenweiher.  Spring Summer Autumn Spring Summer Autumn recorded than in Lake Constance, but this could have been due to the availability of habitats. Nevertheless, regardless of the lake size or characteristics, we were able to continuously record considerable bat activity throughout the three seasons and all recording nights in all three study lakes. This confirms the fact that water bodies are important habitats for bats, no matter if it is for feeding or drinking water.

| Bat and insect activity
Positive relationships between insects and bat activity were found for Siechenweiher and Lake Constance. The relationships between bat activity and aquatic insect emergence were weak in general, and absent in Mindelsee, indicating that the recorded species might feed partly or not at all on aquatic insects. The stronger relationships that were found between bat activity and aerial flying insects, as in spring at Lake Constance, also imply that bats do not depend only on aquatic insects. The species that are known to feed almost exclusively on aquatic insects, M. daubentonii and P. nathusii unfortunately could not be discriminated, from congeneric Myotis and P. kuhlii, respectively, that feed on terrestrial diet. Pipistrellus pipistrellus, which showed highest activity, is considered a generalist, while P. kuhlii is often associated with aquatic habitats (Vaughan, Jones, & Harris, 1997). Both P. pygmaeus and P. kuhlii feed on both terrestrial and aquatic insects. Particularly, in Mindelsee, no relationships were found for any of the species/group and insects, implying that insect availability was a poor predictor of bat activity as has been also found elsewhere (e.g., Wolbert, Zenner, & Whidden, 2014, although sampling place was not close to water at all in this F I G U R E 4 Box-plots of the hourly total bat activity (s of activity per hr of recording) per lake and per season. Hourly intervals are calculated according to the sunset time. 0: 1 hr before the sunset-sunset, 1: sunset-1 hr after sunset, 2: 1-2 hr after sunset, 3: 2-3 hr after sunset and so on. Note the different scales of Y-axis that were kept for better clarity of the hourly pattern although they do not permit easy comparison of the activity. LG, Lake Constance; MI, Mindelsee; SI, Siechenweiher study). Nevertheless, in our study we were able to detect some relationships between the emergence of aquatic insects and bat activity.
Indeed, documenting causal relationships requires a more experimental approach, such as by Fukui et al. (2006) who manipulated emerging insect numbers from a river in Japan and showed the relationship between bat activity and aquatic insects. In particular, in the spring bat foraging activity on emerging insects was higher in the control areas than in the treatment where emergence was prevented. In a field study in Sweden, bat activity was better explained by insect availability (that was also mainly Chironomidae) than in our study (DeJong & Ahlén, 1991). In that study, in early spring (May-June), bats were hunted only in woodlands near lakes where aquatic insects were abundant, while insects elsewhere were scarce. Possibly, aquatic insects are more important resources in cases when terrestrial prey is limited. In Germany, food for bats is almost always available, except during hibernation (Zahn, Rodrigues, Rainho, & Palmeirim, 2007) and probably early spring. However, this might not always be the case in Sweden. Other studies have found that bat activity was influenced not only by insect availability but also habitat structure (e.g., in riverine habitats: Hagen & Sabo, 2011), or air temperature (O'Donnell, 2000. Temperature determined if bats fly at all during one night, while invertebrate abundance determined how long they feed (O'Donnell, 2000).
The absence of strong positive relationships between bat activity and emerged aquatic insects is also probably because total activity is not necessarily feeding activity. Echolocation calls were the majority of the recorded sequences, but we also detected social calls and feeding buzzes. Total activity was positively correlated with feeding buzzes, which is true in similar studies (e.g., Rainho, 2007).
Thus, total activity is considered a good indication of foraging activity. Feeding buzzes show that bats are following insects, and while the result of the hunt might be unknown, we at least know that bat calls were focusing on insects.

| Night pattern bat activity
The nocturnal pattern of bat activity differed between lakes, possibly reflecting differences in habitat, microclimate, and proximity of roosts. The Pipistrellus species accounted for most of the activity in all the lakes, and so we do not expect the species composition to be responsible for these activity differences. The nocturnal pattern of bat activity seemed to follow the usual bimodal peaks of insect emergence at dawn and dusk (e.g., Smukalla & Meyer, 1988;Rydell, Entwistle, & Racey, 1996) in the spring/ summer for Lake Constance, and in the summer for Siechenweiher. A possible explanation for the absence of a specific pattern in the autumn could be the low insect availability that might drive bats to search longer for food, or individuals might fly at different hours and places (Swift & Racey, 1983). Seasonal variations in nocturnal activity, foraging time per night, and time of departure and return to the roost have been reported elsewhere as well (Encarnação, Becker, & Ekschmitt, 2010). TA B L E 1 Statistical significant results (p < .05) and R 2 of the linear regressions between bat activity (seconds of activity per hr of recording) and insects (per hr per trap) per lake. Notice that in Mindelsee, there were no significant relationships found. Both bat activity and insect values are log(x + 0.1) transformed

| General conclusions and recommendations
We examined the relationship of aquatic insect emergence to bat activity on a temporal and spatial scale. Our results do not indicate that there is a general pattern applicable for all lakes in the area.
However, our data show that, indeed, even small lakes are important for bats as they support diverse bat communities and bat activity (throughout the night and seasons). The relationship between bat activity and insects is not straight-forward, probably because insect availability is not a limiting factor in the study area. Nevertheless, it is interesting that at the shore of the lake with the lowest aquatic insect emergence, stronger relationship was recorded between the bat activity and the aerial insects than at shores with higher insect emergence, indicating that (probably generalistic) bat species possibly respond to the food availability.
Field experiments, like the one of Fukui et al. (2006) in river, but also in lakes, that control insect availability might provide better insight into what extent aquatic insect resources influence bat activity.
Comparative studies in areas with limited water availability might yield insight into the flexibility and resilience of bat species to changing environmental conditions. We showed that although lakes are exporting important amounts of insect biomass to the adjacent terrestrial systems, but may not predict exclusively the behavior of terrestrial consumers.
Korine, Adams, Russo, Fischer-Phelps, and Jacobs (2016) conclude "studies concerning bats and water are key to better management of water resources." Therefore, our findings may also be of use for the conservation of bat species and lakes, for example, for taking decisions on small conservation actions (e.g., where to install bat boxes) to which restoration actions should be chosen for a specific water body (e.g., decrease in nutrient levels) or fishing strategies. Data like ours can help to predict possible effects of ecosystem restoration actions on bats, for example, increase in benthivorous fishes in a lake, might decrease emerging insects which can lead to lower insect resources available to bats.
By examining hourly nocturnal activity pattern per season, differences among nights could be masked. In our study area, bats were active throughout the entire night, which was consistent with other TA B L E 2 (a) The lmer models (with lake as random factor) that were applied to the full dataset and their Akaike Information Criterion (AIC criterion) value. (b) The results of the best of the above models (the one with the lowest AIC value   (e.g., O'Donnell, 2000). If, in places like Mindelsee, monitoring is performed only few hours after sunset, considerable amount of activity will be missed. Therefore, we recommend when nocturnal pattern of bats is unknown, to conduct a pilot study first with a few nights of full recordings in each season and then decide if only few hours of monitoring are enough and when these hours should be.

ACK N OWLED G M ENTS
We are grateful to Kamran Safi, Hans-Günther Bauer, and Elizabeth Biology (IMPRS), Germany, which she also thanks for her funding.

CO N FLI C T O F I NTE R E S T
None declared.

AUTH O R S CO NTR I B UTI O N
IS and KOR conceived and designed the experiments; IS and DG performed the field and laboratory work and analyzed the data; IS drafted the work and wrote the manuscript; IS, DG, and KOR revised the work critically for important intellectual content.