Water visitation patterns by Eleonora’s falcon at its breeding grounds: a case study using visual observations and camera traps

Water is a vital resource providing multiple services to many organisms, including birds. Eleonora’s falcon is an iconic, long-distance migrant bird of the Palearctic region. While various aspects of its life history have been thoroughly studied, its dependence on water sources has been poorly documented. In this study, we examine for the first time the species’ visitation patterns at a complex of small, natural ponds of karstic water on a Greek island and provide a detailed description of its bathing and drinking behaviour using visual observations and camera traps. Our results indicate that the species visits the ponds systematically from spring until autumn to bathe and drink. Water visitation patterns suggest communal behaviour, most probably benefitting from increased vigilance in greater numbers. In spring, male falcons prevail during water sessions and the reverse pattern is observed in autumn, whereas morph ratio remains stable, close to the expected 3:1 ratio. Temperature, affecting the falcons’ foraging activity and/or their heat dissipation requirements, together with seasonal differences in labour division between sexes can explain a significant amount of the observed variability in water visitation patterns. Freshwater sources in most islets hosting breeding colonies are scarce, thus natural ponds in nearby larger islands are an important resource many falcons could rely on while at their breeding grounds. The identification and preservation of bathing and drinking spots where falcons gather in numbers across its breeding range is thus highly recommended. Considering the forecasted impact of climate change on karstic water we also encourage an in-depth analysis of the role of water in the adaptation of Eleonora’s falcon to a life in insular environments experiencing global warming. Last but not least, further investigation of the species’ communal behaviour regarding water use can also offer interesting insights in the species’ natural history. HIGHLIGHTS•Eleonora’s falcons visit the water ponds systematically and communally to bathe and drink •Temperature variability and differences in labour division between sexes explain water visitation patterns to a large extent

Water is a vital resource providing multiple services to many organisms, including birds.Eleonora's falcon is an iconic, long-distance migrant bird of the Palearctic region.While various aspects of its life history have been thoroughly studied, its dependence on water sources has been poorly documented.In this study, we examine for the first time the species' visitation patterns at a complex of small, natural ponds of karstic water on a Greek island and provide a detailed description of its bathing and drinking behaviour using visual observations and camera traps.Our results indicate that the species visits the ponds systematically from spring until autumn to bathe and drink.Water visitation patterns suggest communal behaviour, most probably benefitting from increased vigilance in greater numbers.In spring, male falcons prevail during water sessions and the reverse pattern is observed in autumn, whereas morph ratio remains stable, close to the expected 3:1 ratio.Temperature, affecting the falcons' foraging activity and/or their heat dissipation requirements, together with seasonal differences in labour division between sexes can explain a significant amount of the observed variability in water visitation patterns.Freshwater sources in most islets hosting breeding colonies are scarce, thus natural ponds in nearby larger islands are an important resource many falcons could rely on while at their breeding grounds.The identification and preservation of bathing and drinking spots where falcons gather in numbers across its breeding range is thus highly recommended.Considering the forecasted impact of climate change on karstic water we also encourage an in-depth analysis of the role of water in the adaptation of Eleonora's falcon to a life in insular environments experiencing global warming.Last but not least, further investigation of the species'

INTRODUCTION
Birds are a highly mobile group of organisms that rely on variety of resources encountered in often geographically distinct areas to perform all necessary activities for their existence, such as reproduction, food and water intake, and body maintenance.Water in particular offers multiple services to the avifauna.For example, bathing in water is a widespread behaviour among birds for feather maintenance and body cleaning (Slessers 1970;Rhijn 1977;Spruijt et al. 1992).Bathing in water (Hand et al. 1981) and drinking water (Hazef 1964) is also an effective behaviour for heat dissipation and water supplementation during hot and dry periods.Therefore, assessing the degree to which birds rely on water is urgently required for wellinformed conservation considering the ongoing change in precipitation and temperature regime, especially so in areas mostly impacted by climate change, such as the Mediterranean region (Giorgi 2006).
Eleonora's falcon (Falco eleonorae Géné 1839) is a long-distance migratory raptor of the Palearctic and an emblematic species of the Mediterranean region.Over the years, Eleonora's falcon has received great attention by scientists especially due to its intriguing ecology and life history.The species is an endemic breeder of islands and rocky coastal cliffs in the Mediterranean Sea and Atlantic coast and a winter visitor of forests, rural areas and wetlands primarily in Madagascar, but also in adjacent areas of Africa, and a spring visitor of similar habitats located several kilometres away from its breeding sites in the wider Mediterranean region (Walter 1979;Ristow & Wink 1992-1994;Mellone et al. 2013;Kassara et al. 2021).There have been numerous anecdotal observations of Eleonora's falcons frequenting small, both natural and artificial, ponds, but, to the best of our knowledge, the species' relation to water has not been thoroughly documented to date.The only study referring to the species' bathing behaviour took place in a colony in southern Greece where no water sources were available, thus Eleonora's falcons were observed dust-and sun-bathing and only occasionally bathing in rain for the purpose of plumage and body care, as well as for ectoparasite removal (Ristow et al. 1980).The same study also reports Eleonora's falcons visiting a river in Morocco (Walter 1968cited in Ristow et al. 1980) but without any further details.Therefore, descriptive information on the bathing and drinking behaviour of Eleonora's falcon can furnish valuable information to its natural history.
Even though most breeding colonies are located on islets deprived of water sources, water ponds in nearby larger islands may offer bathing and drinking spots for many Eleonora's falcons, especially during the summer period.Avoidance of thermal stress is particularly important for the species considering that brooding takes place during the hottest period of the year (Wink & Ristow 2000).Moreover, in colour polymorphic birds, like Eleonora's falcons that typically occur in light and dark morphs (Walter 1979;Ristow et al. 1998), physiological adaptation to temperature variability may vary (Galeotti et al. 2003), thus temperature fluctuations may differentially affect water use patterns, even more under extremely hot conditions (e. g.Angelier 2020;Fragueira et al. 2021).Water availability in the vicinity of its breeding locations for thermoregulation, but also for water supplementation, could become a crucial factor in the future considering the anticipated increased severity of droughts, including increased frequency and duration of heat waves, during the summer period in the Mediterranean region (Cramer et al. 2020).Thus, the identification of water sources systematically used by Eleonora's falcons and the exploration of drivers influencing water use patterns can contribute substantially to its conservation considering the dry conditions the species experiences while at its breeding grounds.
In this study we aim to assess the significance of water sources for Eleonora's falcon during both the prebreeding and breeding period by providing a detailed description of water visitation patterns to a complex of small, natural ponds in an insular location, namely the island of Antikythira in southern Greece.To this end, (a) we explore seasonal, daily and diurnal trends as well as traitrelated variability in water visitation patterns, (b) we assess communality in water visits and (c) we describe the species' bathing and drinking behaviour.In the absence of previous knowledge on water use by Eleonora's falcons, we postulate three main hypotheses: (i) falcons use the ponds collectively, (ii) male falcons are more common in spring, and females in autumn because of seasonal differences in labour division between sexes, (iii) more falcons, especially dark ones, frequent the ponds in autumn due to higher temperatures, as well as in hotter periods during the day in both seasons.Our findings are expected to be used as a reference basis for a thorough exploration of the services water sources, including ponds, offer to the species.

Study site
The present study was conducted on the island of Antikythira, which lies at the edge of the Aegean Sea between Crete and Peloponnese in the eastern Mediterranean Sea (Fig. 1).The steep cliffs along the coastline of the island host one of the largest concentrations of breeding pairs in the country, nearly 480 pairs according to the latest population census (Hellenic Ornithological Society unpublished data).The island is a dry place, receiving less than 600 mm of rainfall year-round and with little or no rain between May and October (National Observatory of Athens, Antikythira station unpublished data for the period [2015][2016][2017][2018]. The study site is located at the western side of the island and consists of a complex of small, terraced, natural ponds (Fig. 1) formed by karstic water flushing from a nearby rocky surface.Eleonora's falcons are mainly observed using the three largest ponds for drinking and/ or bathing and these were selected for the purpose of this study.Each pond covers an area of 1-2 m 2 covered by clay sediment.Water depth may fluctuate depending on changes in water flow, but it does not exceed 10 cm during the dry period.At the present, the study site is the only permanent, natural source of freshwater encountered on the island during the dry period.A few drinking troughs for free ranging goats and ephemeral rainwater ponds provide additional drinking and bathing spots.

Field surveys
Water visitation patterns were recorded by the same observer during the 2018 prebreeding (24 April-25 May) and breeding (3-24 September) period from 7:00 am, i.e. before the falcons arrive at the study area to ensure no observer bias, until 3:00 pm, when most visitation events occur in spring, from a concealed location 25 m away from the ponds.Although in autumn falcons may visit the ponds regularly also in the afternoon, we maintained the same timeframe for the field surveys to obtain comparable results between seasons.
Data collection began 15 min after the observer entered the hide.Water visitation was recorded per session, i.e. for time periods when a falcon (or falcons) landed at the ponds until all falcons took off.For each session the observer recorded as a minimum the time of arrival and departure of the first and last individual respectively, i.e. the timestamp of the session, as well as the total number of falcons landing at the ponds.Additionally, the observer also recorded the morph, sex and, for as many individuals as possible, the time of arrival and departure.The latter was recorded with a stopwatch, while morph and sex identification were based on the coloration pattern of the plumage (Wink et al. 1978) and eye-ring (Walter 1979;Ristow et al. 1998), respectively.The observer also noted whether a session was interrupted due to the presence of goats that forced the falcons to take off.Moreover, the observer recorded the presence and the identity of colour-ringed individuals that have been ringed at the same location in the frame of an ongoing population monitoring scheme.
As a complementary method the presence of the falcons at the ponds was monitored throughout the study period (24 April-24 September) using an automated recording system consisting of two camera traps (Ltl Acorn 5310, Ltl Acorn 5610) positioned at ca 2 m distance from the ponds.The camera traps were set up to take 1 min long videos with no time lag between successive triggers or photos with a trigger interval of 1 min; one camera trap was active from 7:00 am until 1:00 pm and the other one from 1:00 pm until 7:00 pm to ensure energy and memory autonomy.Camera recordings were used to assess whether Eleonora's falcons visited the water ponds throughout the study period, given that visual observations were carried out on specific days and hours in spring and autumn, while not at all during the summer.Camera recordings were also used to describe the bathing and drinking behaviour of Eleonora's falcons in detail.

Weather data
Meteorological data were recorded on 1 min basis at the Antikythira actinometric and meteorological station (1-min averages of 12-sec samples) operated by the Institute for Environmental Research and Sustainable Development (IERSD) of the National Observatory of Athens (NOA).These data were summarized to estimate mean, minimum and maximum airtemperature on an hourly basis.

Statistical analyses
We hereafter applied non-parametric methods as the variables in question did not follow a normal distribution.We assessed seasonal, diurnal and daily variability in a subset of variables describing water visitation patterns by implementing General Additive Models (GAM), namely the session duration, the time interval between successive sessions, the total number of falcons at each session, as well as the percent of male and light morph falcons at each session.GAMs are a generalization of linear models that also permit non-linear relationships between the dependent variable and a set of candidate covariates.For the latter, we used season as a categorical variable, the hour of the day each session started, the maximum temperature recorded the preceding hour, and the day each session occurred since the start of each survey period as continuous variables including their interaction with season.In the model regarding session duration, we also included the total number of falcons as covariate.We built full models and allowed the algorithm to penalize smoothers out of the model using the double penalty approach (Marra & Wood 2011).In essence the algorithm sets the smoothing parameter of non-influential covariates to zero.As a preliminary step we inspected full models for pairs of variables with high concurvity (> 7).This resulted in the exclusion of the hour of the day each session started from the final structure of all models and the day each session occurred from two models.We used (a) the gamma distribution with a log link for the session duration model, (b) the negative binomial distribution with a log link for the number of falcons per session model considering the session duration as weight, (c) the binomial distribution with an identity link in the percentage of males and in the percentage of light morph falcons models, and (d) the tweedie distribution with a log link for the session interval.Finally, we inspected the resulting models using standard diagnostics tests to assess the complexity of non-linear relationship between the response variables and the covariates, as well as the normality, homoscedasticity, as well as lack of autocorrelation in the residuals following Wieling (2018).The component effect of each of the smooth or linear terms (i.e.their partial effect, centred to zero) included in the final models is represented graphically along with the 95% confidence interval.
From the data collected, we also calculated the time difference between the arrival of the first and second falcon of each session, as well as the time difference between the arrival and departure of all falcons for each session to investigate seasonal differences using Wilcoxon rank sum tests.
From all statistical analyses, we excluded sessions that were either disturbed by goats, but for those concerning the timing of the first session of the day, and/or not timed properly.Statistical analyses were implemented in R v.4.2.0 (R Core Team 2022).GAM models were built and tested using the functions "gam" and "gam.check"respectively in the mgcv package (Wood 2011), whereas plots were created using the function "draw" in the gratia package (Simpson 2022).Statistical significance was set at α = 0.05.Mean and standard deviation values are reported, along with the sample size used in each analysis.
We also report the total daily and seasonal number of falcons visiting the study site only as measure of visitation intensity, since individuals may repeatedly visit the study site during the same day or in the same season as it has been the case for colour-ringed individuals in autumn (Table S2 in Supplemental Data), while due to the increased frequency of goat disturbance in autumn the same falcons may take off and return to the ponds shortly after.Thus, we refrained from further analysing these data to avoid spurious results related to pseudoreplication.

RESULTS
Visual observations were carried out on 31 days (17 in spring, 14 in autumn) totalling 245 hr of observation during which 470 sessions were recorded.We excluded 158 sessions, most of them recorded in autumn, either for being disturbed by goats and/or for not being timed properly.In the remaining sessions 595 visitation events were recorded in total in spring and nearly the double, namely 1050, in autumn (Fig. S1 in Supplemental Data).During the study period there was no rainfall at all except for 23 May.During the time window of the visual observations, i.e. from 7:00 am until 3:00 pm, the mean air temperature was estimated at 20.8 and 25.2 °C in spring and autumn, respectively, exhibiting greater daily variability in spring but relatively similar diurnal variability in both seasons (Figs S3-S4 in Supplemental Data).

Water visitation patterns
In both seasons the falcons arrived at the study site at approximately 10:00 am each day (Table 1).Sessions lasted 4.9 min on average, being slightly longer in spring compared to autumn by 0.46 min yet not statistically significant (Tables 1-2, Fig. 2).The longest session lasted by far 37 min and was recorded in spring.Sessions lasted  Water visitation patterns by Eleonora's falcon longer when more falcons occurred at the ponds, but in spring this trend was reversed when more than five falcons were present (Table 2, Fig. 2; Fig. S6 in Supplemental Data).In spring the duration of the sessions was negatively affected by temperature until 23 °C while the opposite trend was observed at higher temperatures (Table 2, Fig. 2; Fig. S6 in Supplemental Data).In autumn the effect of temperature resulted in an oscillating pattern without a clear trend (Table 2, Fig. 2; Fig. S6 in Supplemental Data).
At each session there were 5.2 falcons on average, being significantly more numerous in autumn compared to spring by 1.9 individuals (Tables 1-2, Fig. 3).The number of falcons at each session was not affected by temperature nor did it exhibit daily variability, except for a small peak around 12 May (Table 2, Fig. 3; Fig. S7 in Supplemental Data).The time it took for the second falcon to arrive at the ponds in spring was twice that in autumn (64 vs 33 sec, respectively), but it did not vary significantly between seasons (W = 4022, P = 0.082).Conversely, in spring at each session the falcons arrived with 72 sec difference and departed with 66 sec difference on average, whereas in autumn these differences were significantly lower by 26 and 23 sec, respectively (arrival: W = 20479, P < 0.001, departure: W = 15585, P = 0.003, Fig. 4).
At each session ca 76% of the falcons were light morph in both seasons (Tables 1-2).The morph ratio per session varied only slightly with day in spring and temperature in autumn (Table 2, Fig. 5; Fig. S8 in Supplemental Data).On the contrary there was a strong seasonal pattern in sex ratio; males outnumbered females in spring (81.1%) while females outnumbered males in autumn (61.5%) (Tables 1-2, Fig. 6).The same patterns were observed considering the first falcon of each session, which in both seasons it was usually light morph (78%), but male in spring (90.9%) and female in autumn (65.2%).Furthermore, the percentage of male falcons at each session decreased until 23 °C while the opposite trend was observed at higher temperatures (Table 2, Fig. 6; Fig. S9 in Supplemental Data).There was also daily variability with an overall decreasing trend in the percentage of male falcons as the season progressed, yet in autumn this trend was reversed (Table 2, Fig. 6; Fig. S9 in Supplemental Data).
The interval between consecutive sessions was significantly greater in spring as opposed to autumn by 16.4 min on average and varied with temperature in both seasons (Tables 1-2, Fig. 7; Fig. S10 in Supplemental Data).In particular, in spring sessions were less frequent in higher temperatures but between 21 and 23 °C, whereas in autumn the session interval was in general stable with the exception of a decreasing trend between 23 and 25 °C (Fig. 7; Fig. S10 in Supplemental Data).
Moreover, camera traps confirmed the presence of Eleonora's falcons at the water ponds, but also of other bird species, throughout the study period except for a few days (Table S1 in Supplemental Data).

Bathing and drinking behaviour
Falcons' bathing behaviour based on video footage can be described as follows.First, they opened certain feather tracts on their back, belly, and thighs to expose the apteria while they waded in the water.Then, they squatted to soak the underbody area, while at the same time they could also be observed drinking water.After soaking, they repeatedly dipped their head while rapidly flapping their wings to scatter water over their back and onto their body.This activity was interrupted by pauses during which the birds rotated their head and rubbed it on both sides of their back, a behaviour that Water visitation patterns by Eleonora's falcon possibly facilitates the soaking of the dorsal apteria by water from the head.After bathing and before departing, some individuals remain close to the ponds for a while, either preening or resting, while others depart immediately.While some individuals were observed drinking while bathing, others visited the ponds only to drink.Drinking was performed while the falcons stood at the  verge of the ponds barely touching the water or even entered the ponds, by dipping their beaks to fill their lower mandible and then leaning their heads backwards.

DISCUSSION
In this study, we monitored for the first time the visitation patterns of Eleonora's falcon at a complex of natural ponds on a Greek island, using mainly visual observations and secondarily camera traps.The observed water visitation patterns largely confirm our hypotheses; they are explained to a large extent by season-specific temperature effects and division of labour between sexes, but morph-related differences could not be detected.We also provided for the first time a detailed description of the species' water bathing and drinking behaviour.The appreciation of the multiple roles of small waterbodies in biodiversity and ecosystem functioning is growing, but also of Fig. 6. -Seasonal differences and partial effects of the most influential explanatory variables (i.e.survey day and maximum temperature of the previous hour) for the percentage of male falcons at each session according to the GAM results.Shaded areas represent the 95% confidence interval for the mean shape of the effect.Fig. 7. -Seasonal differences and partial effects of the most influential explanatory variable (i.e.maximum temperature of previous hour) for the time interval between successive sessions according to the GAM results.Shaded areas represent the 95% confidence interval for the mean shape of the effect.the threats they face, especially in arid and semi-arid environments (e.g.Biggs et al. 2017;Zamora-Marín et al. 2022).Thus, our results highlight the significance of natural ponds for a species dwelling in a climate change hot spot for almost half of its annual cycle, delivering novel insight on its natural history and important information for its conservation.
More specifically, Eleonora's falcons visited the water ponds systematically throughout the study period, namely from late April until late September.The first Eleonora's falcons generally arrive in late April and their numbers increase to a maximum by mid-May, as confirmed by our visual observations, then drop until July, becoming numerous again as the breeding season approaches (Antikythira Bird Observatory unpublished data).Moreover, the falcons visited the ponds both for bathing and drinking, exhibiting typical bathing behaviour for raptors exploiting shallow waters (Holthuijzen et al. 1987;Eisermann 2005;Haak & Buchanan 2012;Sazima 2018).
The timing of the first individual observed at our study site each day was consistent between seasons, i.e. water sessions began 3-4 hr after sunrise.Water visits were performed mostly in groups and lasted longer when more falcons frequented the ponds.This communal behaviour was also reflected in the relatively low time difference between the arrival and departure of falcons, especially so in autumn.Eleonora's falcon is well-known as a gregarious breeder; Ristow (2004) proposed that the main reason for its coloniality is insect hunting, which implies locating unpredictable food patches in remote areas.Still, flocking behaviour could also pervade other features of their lives, such as when frequenting water sources, either benefitting from reduced individual vigilance when in larger numbers (Roberts 1996), or as a result of exploiting these scarce resources in insular environments.The observed communal use of water at our study site could be indeed related to increased vigilance, as we have never observed social mobbing while the falcons are at the ponds; on the contrary, when large raptors or goats approach the falcons swiftly fly away.Additionally, in both seasons light morph falcons were more abundant and the morph ratio matched well the expected 3:1 ratio (Wink et al. 1978) reported in various colonies across the species' breeding range (e.g.Ristow et al. 1998;Corso & Gustin 2009;Gangoso et al. 2011;Touati et al. 2017).
The observed seasonal differences in the number and sex ratio of the falcons that visited the ponds at each session agree with the species' biology.More specifically, the prevalence of male falcons in our study site in spring was anticipated given that nesting site selection is performed by males as soon as they arrive at their breeding grounds (Ristow et al. 1980).Apart from provision of breeding sites, large islands like the island of Antikythira also offer foraging opportunities during the early prebreeding period, hence hosting a substantial number of falcons (Kassara et al. 2019).Therefore, the water ponds may also attract individuals foraging on the island in spring.Yet, compared to the breeding period, the ranging activity of both sexes is higher during the prebreeding period according to telemetry data (Mellone et al. 2013;Kassara et al. 2021;authors' unpublished data); thus, in spring falcons could also frequent water sources in other areas, which can explain the relatively lower abundance per session during this time of the year.Conversely, the limited dispersal of females due to brood care and investment during the breeding period in combination with the scarcity (or even lack) of alternative water sources in the close vicinity of the breeding colonies of the wider region, renders our study site a popular place particularly for breeding females in autumn.Although multiple visits performed by the same individuals cannot be ruled out, as suggested by resightings of colour-ringed falcons in our study site (Table S2 in Supplemental Data), the large difference in magnitude between seasons (Fig. S1 in Supplemental Data) is indicative of greater water use by Eleonora's falcons in autumn.September is the month when egg hatching is completed and successful breeding pairs raise their nestlings; more specifically, egg-hatching typically occurs in late August (Walter 1979;Wink et al. 1993) but may be prolonged even until the second half of September (e.g.Hadjikyriakou et al. 2020).Both during the egg incubation and the first days of the nestling period nest attendance is of paramount importance to ensure proper thermoregulation (Wink et al. 1980).This task is mainly performed by the female parent, while its mate is responsible for food provisioning (Walter 1979;Wink et al. 1980).Later in the season breeding females can leave their nestlings unattended for a longer period, as has been confirmed by the increase in ranging activity of GPS-tracked females originating from the wider region (Kassara et al. 2021) that can seek food and water in more remote areas.This fits well the observed increase in the percentage of males per session as the season progressed.Still, since neither the age nor the breeding status of the observed falcons could be determined, future studies are encouraged to provide further insight on the effect of the latter on time investment related to water visits.
Our findings are also indicative of season-specific temperature effects on water visitation patterns.In spring temperature is expected to be an indirect limiting factor of the falcons' foraging activity by exerting a direct effect on insect emergence patterns, i.e. the primary food source of Eleonora's falcon during the prebreeding period (Walter 1979;Ristow 2004;Xirouchakis et al. 2019).A previous study has established a positive relationship between the number of foraging Eleonora's falcons observed on the island of Antikythira during this time of the year and the minimum air temperature (Kassara et al. 2019).Therefore, the observed drop in the session duration and the increase in the session interval until ca 23 °C could thus be attributed to increased time allocation in foraging.When falcons experience higher temperatures they could prioritize heat dissipation over hunting, thus leading to the subsequent increase in the duration and frequency of the sessions in spring.Conversely, in autumn the falcons feed mainly on migrating birds (Walter 1979;Ristow 2004;Xirouchakis et al. 2019), thus during this time of the year water visitation patterns are expected to reflect the relatively increased heat dissipation and/or water supplementation requirements, but knowledge on the species' physiology is currently limited.
The extent to which Eleonora's falcon tolerates hyperthermia and dehydration is currently unknown, but, being a migratory species, it is expected to perform well at a narrower breadth of ambient temperature (Boyles et al. 2011).The only information available to date comes from a study on a rather small sample size indicating that the body temperature of breeding adults is ca 41.5 °C but measurements could have been overestimated as a result of handling-related stress, while ambient temperatures over 43 °C can be lethal for eggs (Wink et al. 1980).During a pilot study we conducted on the island of Antikythira, temperature measurements at two nests under rock, yet unoccupied during the study period, showed that the highest temperatures were recorded around noon depending on nest orientation but did not exceed 37 °C (Fig. S5 in Supplemental Data).However, in general even higher temperatures are likely to be reached depending on sun (Wink et al. 1980) and wind exposure of the nest (Grant 1982), rendering heat dissipation through water bathing a likely explanation for the relative increase in the number of falcons observed at our study site in September.Apart from accommodating heat dissipation needs, water supplementation per se could also result in higher water use during the summer period given the lower water content of birds compared to insects (water content in birds: 60% of body mass, Skadhauge 1981 cited in Mahoney & Jehl 1984 water content in insects: 70-75% of body mass, Bell 1990).
Increased water use has been already demonstrated for breeding birds (e.g.Hand et al. 1981;Oswald et al. 2008) performing regular visits to water for bathing and drinking as a means to thermoregulate in a thermally stressful environment.Other behaviours, such as belly-soaking in water, offer a cooling function as exhibited by Charadriiformes during egg incubation (Grant 1982).The same study also suggests that belly-soaking can serve as a cooling technique both for the parents and nestlings, as well as a means to regulate nest humidity.Another function of plumage wetting by breeding birds, including belly soaking, is transportation of drinking water to their nestlings (e.g.Cade & MacLean 1967).The review by Commans (2018) is enlightening on the mechanisms of passive water collection and transportation by a variety of animal taxa, including birds, as an adaptation to limited water supply, such as when living in dry environments like the island of Antikythira.Whether such a wide range of behaviours is also exhibited by Eleonora's falcon remains to be investigated by future studies.Moreover, time budget analysis at the individual level can provide a better understanding on heat stress responses of Eleonora's falcons in the context of resource allocation in their daily routines (Oswald et al. 2008).Additionally, plumage polymorphism could be affecting Eleonora's falcon thermoregulation as has been shown for Rock pigeons (Columba livia) in an experimental setting where darker pigeons reached higher plumage and cloacal temperature when exposed to the sun compared to light ones (Angelier 2020).Although we did not detect variability in the morph ratio of falcons during the water sessions, dark Eleonora's falcons could exhibit a variety of counterbalancing behaviours, like prolonged duration of water bathing and higher frequency of gular flattering, that are worth investigating in the future.
The significance of freshwater bodies to the breeding ecology of the species has been highlighted in a previous study exploring the environmental factors explaining the distribution pattern of Eleonora's falcon breeding colonies in the Aegean Sea (Kassara et al. 2013).In this study we show that even smaller water sources, namely ponds, play an important role in the species' biology as indicated by the attraction of a substantial number of Eleonora's falcons and their consistent presence from late April until at least late September at our study site.The identification and preservation of water bodies where Eleonora's falcons gather in numbers for bathing and drinking across the species' breeding range is thus highly recommended.Moreover, like in other Mediterranean regions, these karstic ponds are the only available, permanent water source on the island of Antikythira during the dry period.The forecasted reduction of karstic flow due to climate change (e.g.Nerantzaki & Nikolaidis 2020) could be an issue of major concern for Eleonora's falcon, but also for other bird species that systematically use water ponds.Future studies are prompted to build on from our findings and bring light on the role of water in the behavioural and physiological adaptation of Eleonora's falcons to a life in insular environments experiencing global warming, as well as explore further the communal behaviour of the species in relation to water use.ACKNOWLEDGMENTS This is contribution No. 35 from Antikythira Bird Observatory -Hellenic Ornithological Society/BirdLife Greece.We would like to thank two anonymous reviewers for their fruitful comments on a previous version of the manuscript.Special thanks to the volunteers of the Antikythira Bird Observatory for field assistance, to Dr George Iliopoulos for providing information on the geomorphology of the study site and to Alexios Lazaris for assistance in model building.

Fig. 1 .
Fig.1.-The study site consists of three small, terraced, natural ponds located on the island of Antikythira in southern Greece.Two camera traps were placed ca 2 m away from the ponds operating from 7:00 am to 7:00 pm from late April until late September 2018.

Fig. 2 .
Fig. 2. -Partial effects of the most influential explanatory variables (i.e.number of falcons at each session and maximum temperature of previous hour) for the session duration according to the GAM results.Shaded areas represent the 95% confidence interval for the mean shape of the effect.

Fig. 3 .
Fig. 3. -Seasonal differences and partial effects of the most influential explanatory variable (i.e.survey day) for the number of falcons at each session.

Table 2 .
GAM results depicting seasonal and daily trends, as well as temperature effects on water visitation patterns.The estimate of the intercept (β), effective degrees of freedom (edf) for the smooth term of the explanatory variables, their associated P-values, and the deviance explained by the model are reported.All models were built using 252 observations but for the session interval where 195 observations were used.