Black Kites on a flyway between Western Siberia and the Indian Subcontinent

The Black Kite (Milvus migrans) is one of the most widespread raptors in the world. The Palaearctic is populated by two migrating subspecies, Milvus migrans migrans and Milvus migrans lineatus, in the western and eastern part of this realm, respectively. There is a large intergradation zone of M. m. migrans/M. m. lineatus in-between. Although the migration routes of M. m. migrans from Europe to Sub-Saharan Africa and the Middle East are well known, detailed information about migration routes of Black Kites from intergradation zone are missing. Using satellite telemetry we are able to fill this gap in our knowledge of these birds. We tagged with GPS/SMS/GPRS telemetry loggers 13 and 6 Black Kite pulli in lowland around Biysk (Altai Krai, Russia) and in mountains around Kosh-Agach (Altai Republic, Russia), respectively. After fledging, Black Kites from both subpopulations stayed in a small, non-overlapping areas and then migrated to southern Asia through narrow corridors. Black Kites originating from Biysk migrated through the Western Circum-Himalayan Corridor. Black Kites originating from Kosh-Agach used the Trans-Himalayan Corridor crossing the Himalayas in altitudes of up to 6256 m asl. The average total distance travelled of Black Kites from both subpopulations was 9166 km without any significant differences between these subpopulations. Timing of both spring and autumn migration did not vary along different age groups. Black Kites from both subpopulations wintered in low elevations of Pakistan and India. Birds wintered on average for 190 days, and the mean area of individual home ranges in winter was 4704 km2. During the breeding period, birds dwelled in south-western Siberia, where they spent on average 125 days with an average home range size 3537 km2. We found that ontogenetic shifts in migratory behaviour of Black Kites from Eastern Russia differ from those in population/subspecies in Europe. Black Kites crossing the Himalayas fly and, moreover, stay for hours resting at night in the environment of mountains at altitudes over 5000 m.

DNA examination. Contour pin feathers (newly grown feathers, full of blood) were collected from the lower part of a chick body and stored in 96% ethanol. The total DNA was isolated using the ExtractDNA Blood kit (Evrogene, Russia). The sex of tagged birds was determined by a method by Suh et al. (2011). A 699 bp fragment of the mitochondrial cytB gene was analysed to identify haplotypes 36 . The cytB mitochondrial gene fragment (924 bp) was amplified using F3 (5′-CCA CCC CAT CCT CAA AAT AA-3′) and R8 (5′ATT GTG CGC TGT TTG GAC TT-3′). We sequenced PCR fragments in both directions using a 3500 Genetic Analyser capillary sequencer (Applied Biosystems, USA) and aligned resulting sequences using the Vector NTI software (Thermo Fisher Scientific, USA). In order to exclude contamination, operations with genomic DNA and with PCR products were performed in different rooms. In unclear cases, PCR and sequencing were repeated. Table 1. Black Kites from western Siberia, Russia tracked with telemetry devices. All birds were tagged as pulli. F, female; M, male; * birds were tracked also after 30 June 2021 (further data not included in this paper); aa, bb, cc, pairs of siblings. www.nature.com/scientificreports/ Satellite telemetry devices. Black Kites (pulli) were fitted with telemetry loggers in nests in 2018 (subpopulation A, Biysk) and 2019 (subpopulation B, Kosh-Agach). Loggers equipped with solar panels (20 g; Ecotone, Poland, and Ornitela, Lithuania; www. ecoto ne. pl, www. ornit ela. com, respectively) were used to track the birds. Loggers were fitted onto the backs of the birds using harnesses (backpacks) consisting of a 6 mm Teflon ribbon encircling the body by two loops around the bases of the wings and joined in front of the breastbone. Loggers function in GPS (Global Position System)/GSM (Global System for Mobile Communication) systems. The GPS positions of the birds were collected according to individual settings (usually one position fixed per 1-6 h). They were sent as SMS (Short Message Service) text messages by local mobile operators to the Ecotone and Ornitela Centers in Poland and Lithuania, respectively, where they were saved and archived. To analyse the coordinates of bird positions and to create maps of migration we used GIS and the software ArcGIS 10.1 (Esri, Redlands, CA, USA).
Data processing, migration characteristics. We processed positional data (coordinates) from studied birds for each bird individually. These data were separated into yearlong modules from 01.07.20XY (in the first year from the date of tagging) to 30.06.20XY + 1. The number of modules depends on the lifespan of each bird. We calculated the total distance travelled within the yearlong period and the number of temporary settlements areas (TSA) from these modules. We defined total distance travelled as distances between night roosting places connected chronologically (daily local movements were not calculated within the migration movement). We defined TSA as a preferred place where a bird stayed for > 10 nights within 80 km 2 . This template size was based on roost locations distributed within a 10-km diameter over 10 days, thus, all falling within 80 km 219 . Spring (pre-breeding) and autumn (post-breeding) migrations separate the winter and summer period. We defined the beginning of those migrations as a day when a bird left the winter/summer area and flew north/ south without returning back in consecutive days. The end is defined as a day when a bird reached the summer or winter destination. Bird reached the summer or winter destination when it did not continue on its migration to north or south. During both migrations, birds tend to use stopovers, defined as a day with less than 50 km of a directed flight 26 . The size of the post-fledging area (PFA), winter and summer grounds (home ranges) between migrations were calculated as a Kernel density estimate (KDE) 95%. Before performing KDE estimation, we standardized the data set of each bird to 4 GPS fixes per day (1 each 6 h).
The Himalayas crossing was defined as the period of migration between the first and last day of migration with coordinates recorded by the foothill of the Himalayas with at least one coordinate recorded at over 5000 m asl. For this purpose, we set the loggers to collect the data every 5 min. During the crossing we defined active travelling hours of birds as the time between first and last coordinates with recorded speed over 5 km/h 37 . We classified manually and calculated the length of trajectory segments leading parallel with mountain ridges during the crossing and compared them with the overall distance travelled during the crossing over the Himalayas. We found segments parallel if the bird flew along a mountain slope copyrighting the valley and perpendicular if the bird flew across valleys and ridges not copyrighting the valleys.
We defined checkpoints W1, W2 and W3 as night positions where birds stayed on 31 January of their 2cy (second calendar year), 3cy and 4cy, respectively. It represents where birds were wintering during this date during the first, second and third winter. We defined checkpoints S1, S2 and S3 as positions where birds stayed during the breeding period on 30 June of their 2cy, 3cy and 4cy, respectively. We used the positions during S1, S2, S3, W1, W2 and W3 to compare the latitude of summer and winter areas used by individual birds between years of their life span and among individuals during the first, second and third years of their life.
Meteorological data. Elevation data was downloaded from the mapping and analysing platform www. datab asin. org "30 arc-second DEM of Asia" as a digital elevation model (DEM).
Weather data (wind, temperature and humidity) were obtained from the NCEP/DOE Reanalysis II dataset, using the RNCEP package 38 for the R-software 39 . Weather data of crossing over the Himalayas were extracted for each coordinate in real-time, and pressure level of 700 hPa corresponding to an altitude between 2300 m and 3150 m. Airspeed, flow-assistance and absolute sidewind were calculated by function NCEP.tailwind using RNCEP package, which calculates flow-assistance and forward and sideways movement according to equation Tailwind (Tailwind = wind speed * cos (α), where α is the angle of the wind from the direction of travel). Equation Tailwind considers flow-assistance to be the component of the flow moving parallel to the specified direction (tailwind), with negative values indicating flows against the specified direction (headwind). We have extracted the weather data for coordinates recorded during post-breeding (n = 1790) and pre-breeding (n = 1310) migration over the Himalayas. We excluded coordinates recorded while roosting from the dataset (coordinates with recorded speed lower then 5 km/h).

Statistical analysis.
We performed the Mann-Whitney U test for testing the differences in pre-breeding and post-breeding migration and home-range characteristics and the pre-breeding and post-breeding Himalaya's flight-over characteristics. To assess the difference in total distance travelled, number of TSA, and the size of home ranges in summer and winter quarters over the years, we performed Kruskal-Wallis ANOVA test. Before any statistical comparison, we run the Shapiro-Wilk test for normality to assess the data distribution. To assess the effect of weather on bird's movement across the Himalayas, we used linear mixed models (LMMs) in R software using the 'lme4' package 40 to analyse the following dependent variables: bird groundspeed and airspeed, in relation to season, flow-assistance, sidewind, humidity and temperature during the crossing over the Himalayas (Table 2). We used LMM with bird ID as a random effect (as individuals could be tracked over multiple years). Only birds with telemetry loggers Ornitela, which crossed over the Himalayas, were included in the LMM (K14 -K19) due to the high frequency of coordinates recording. The best supported LMM model Migration routes and total distance travelled. Black Kites originating from Biysk migrated through the Western Circum-Himalayan Corridor (Fig. 1). These birds flew through eastern Kazakhstan, Kyrgyzstan, Tajikistan and eastern Afghanistan to winter, mainly in northern and southern Pakistan and western India. After winter, birds flew over the same migration corridor back to Biysk area. Unlike Kites from Biysk, Black Kites originating from Kosh-Agach used a different migration route (Fig. 1). These birds flew over Tian Shan, and the Taklamakan Desert in China, followed by Trans-Karakoram crossing-over through Jammu and Kashmir to winter in northern and western India and eastern Pakistan. After winter, birds flew over the same corridor back to Kosh-Agach area. The average total distance travelled of birds from both subpopulations in the first year was 9191 km (ranging from 6431 to 12,478 km). During the first year, birds used on average 4 TSAs (ranging from 2 to 6). During the second year, birds travelled on an average total distance of 9121 km (ranging from 7422 to 11,268 km) using 5 TSAs (ranging from 4 to 7). The average total distance travelled in the third year was 6839 (ranging from 6594 to 7084) using 5 TSA (ranging from 4 to 5) ( Table 3).
Five tagged birds survived and were tracked for multiple years. For those individuals, we compared the differences in the total distance travelled and the number of used TSAs. We found no significant difference in the total distance travelled (P > 0,05) nor the number of TSA (P > 0,05) used among the years. We also tested the difference in total distance travelled, and the number of TSA used between the two subpopulations without any significant results (P > 0,05).

Timing of autumn and spring migrations.
Timing of autumn migration varied slightly among individuals in departure date (30 August ± 12 days) and noticeable more in arrival date (26 October ± 84 days). The timing of spring migration also varied slightly in departure date (17 April ± 12 days) and arrival date (09 May ± 14 days). Surprisingly, timing of either migration did not vary along different age groups. The tagged kites travelled relatively fast, completing 2535-4842 km journey in 10-94 days, progressing by 62-253 km/day, with significantly faster speeds and lower need to rest in the pre-breeding migration (Table 4). During the pre-breeding migration was the speed and active speed more than 50% and 30% higher in comparison to post-breeding migration. As a result, the pre-breeding migration lasted 10 days less.
Post-fledging area and home ranges in winter and summer. The post-fledging area of tagged Kites varied from 1.7 km 2 to 1567 km 2 with a mean of 396 ± 432 km 2 (Table 5). Some birds left the nest and flew straightforward to the winter quarters. Others birds explored the area around the nest and departed for autumn migration with a slight delay. Black Kites from both subpopulations wintered in Indian Subcontinent in low elevations of areas with high human footprint in Pakistan and India (Figs. 1 and 2). No bird remained in the Indian subcontinent during summer periods (Fig. 2). Birds wintered on average for 190 days, and the mean area of individual home ranges was 4704 km 2 ( Table 5). During the breeding period, birds occupied areas in Table 2. Selecting the best LMMs for the airspeed and groundspeed during the Himalayas crossing. We listed first six models for each dependent variable. Models are ranked according to increasing ΔAIC values, with the best performing model on top. TW-tailwind; SW-sidewind; Seas-season; Temp-temperature; Humhumidity.  Table 5). No bird remained in Siberia during the winter period (Fig. 2). Although the mean area of home ranges was slightly smaller during the breeding season than in the nonbreeding winter period, we found no  www.nature.com/scientificreports/ statistical difference in the spatial use (p > 0.05). Five tagged Black Kites survived and were tracked for multiple years (Fig. 3). For those individuals, we compared the differences in the area size of home ranges in the breeding (summer quarters) and nonbreeding season (winter quarters). We found no difference in spatial use over the years in neither the winter quarters (p > 0.05) or summer quarters (p > 0.05). Birds showed individual changes in the size of winter and summer home-range over the course of time (Fig. 4).
High-elevation crossing of the Himalayas and influence of the wind on the crossing over the Himalayas. Timing of post-breeding and pre-breeding crossing over Himalayas varied slightly among individuals in departure dates (20 September ± 12 days; 28 April ± 7 days) and arrival dates (21 September ± 12 days; 29 April ± 7 days). Black Kites originating from Koch-Agar travelled relatively fast, crossing over the Himalayas (on average 571 km) in 2 days, progressing with average active speed of 30.2 km per travelling hour, flying from 6 to 10 h per day. Active speed and number of traveling hours were slightly higher during pre-breeding flight-over. During the crossing of the Himalayas birds roosted for one night in average altitude of 4589 m asl, ranging from 1577 to 5171 m asl ( Table 6, Fig. 3).
Wind condition significantly varied during the pre-breeding and post-breeding Himalaya flight-overs (Table 6). Noticeable was the difference in the tailwind speed, sidewind speed and percentage of parallel flight along the mountain ridge. While during the post-breeding flight-over, birds faced mostly a headwind and preferred to fly perpendicularly to mountain ridges and mountain valleys, on their pre-breeding flight-over, birds flew with a tailwind and preferred to fly parallelly along the mountain ridges and mountain valleys (Fig. 5).
Our best LMM model showed that airspeed of birds crossing over the Himalayas were not only positively related to tailwind but also to difference in season. The groundspeed was also positively affected by tailwind and season but negatively affected by sidewind (Table 7). Although the model results showed negative effect of sidewind to groundspeed, plotting the linear regression lines by season showed that the sidewind had a slightly positive effect on groundspeed during autumn migration (Fig. 6B,D). Prevailing tailwind had generally greater positive effect on both the air-and groundspeed of bird during the spring migration (Fig. 6A,C).  www.nature.com/scientificreports/  K1  22  59  ------------K2  38  174  ------------K3  60  273  204  9214  138  5482  201  23,167  125  22,708  ----K4  47  1349  ------------K5  49  501  222  980  44  480  --------K6  60  454  170  6849  103  439  --------K7  68  539  ------------K8  28  17  ------------K9  52  295  211  9278  101  9129  184  1681  121  1791  172  1101  91  694   K10  47  369  205  2002  109  1040  198  7272  132  519  188  4762  145  93   K11  46 552    www.nature.com/scientificreports/ populations chose narrow and non-overlapping migration corridors, where one involved crossing of the Himalayas. Wintering ranges of birds from these two populations where also distanced and non-overlapping. The only interaction between birds from these populations was found in one bird from Biysk which shared the winter area with birds from Kosh-Agach. We assume that the genetic background of the migration behaviour of Black Kites may be strong, forming the uniform behaviour of tracked birds from both studied subpopulations. What is more, our result showing a different genetic history of both populations supports our assumption.  www.nature.com/scientificreports/ On the other hand, studies showed that timing of migration does not always vary between young birds and experienced adults 17,44 . Our results also showed no difference in timing or route selection connected with different age. Under such scenario, young birds may migrate along experienced adults. This offers a social learning opportunity for young birds that can be hard to distinguish from genetic determination of migratory routes. Furthermore, variable routes of raptors (Peregrines, Falco peregrinus) migrating from Siberia to South Asia were demonstrated even if they had the exact gene involved in regulation of the migration distance 45 . Although we provide data showing differences in haplotypes between subpopulations, that might have influenced the uniformity within subpopulations, the role of innate factors and social learning in transmission of routes remain unclear and asks for further research.

Bird (including Black Kite) migration over high altitudes in the Himalayas.
High-altitude flights of birds over the Himalayas are a highly challenging feat of performance underpinned by several specialised physiological traits. Flapping birds like Bar-headed Goose and Ruddy Shelduck (Tadorna ferruginea) can reach high altitudes during their migration across the Himalayas and Tibetian plateau because they can support the metabolic costs of flight as the low-density air becomes extremely hypoxic 35,46 . Like other migrating (soaring) birds, they may occasionally use updraft wind assistance to help offset flight cost 47 . However, they experience periods of intense flapping flight that require extremely high heart rates, wing-beat frequencies, and metabolic power, such as during level flight at high elevation or during climbs that are not assisted by wind 35,48 .
Raptors use primarily soaring-gliding flight during migration 49 . Soaring flight is an energetically efficient form of flight, and many long-distance migrants are so-called obligate soaring migrants 27 . Updraught necessary for soaring flight includes thermals (pockets of warm rising air) and deflection (orographic) updraughts that occur when horizontal winds strike surface discontinuities, including mountains. The high-altitude terrain of the Himalayas precludes this type of pathway, and hence it is used by raptors 27 . However, some raptors, especially falcons, use flapping flight on their migration across the Himalayas 45 . www.nature.com/scientificreports/ Unfortunately, detailed studies using telemetry devices on raptors crossing the Himalayas are scarce. We can compare our results mainly with a recent study aimed at Black Kites fitted with telemetry loggers in Dehli, India 26 . It seems that Black Kites tagged in Dehli originated, similarly like in our study, to two different population: birds that used Western Circum-Himalayan Corridor may belong to Black kites originating from the intergradation zone between M. m. migrans and M. m. lineatus, birds that used Trans-Himalayan Corridor may belong to M. m. lineatus. Migration routes of these birds were distinct in our study as well as like in a study by Kumar et al. 26 . The birds originating from Upper Altai (Kosh-Agach) crossed the Himalayas over Tian Shan Mts, Taklamakan Desert, and Karakoram Mts like the main portion of Black Kites tagged in Dehli. These birds crossed the Himalayas in extremely high elevation up to 6281 m asl and travelled long periods at elevations above 3500 m. Birds flew across the Himalayas for two days with a single stop to roost at elevations between 1644 to 5448 m asl.
Black Kites crossing the Himalayas may have physiological adaptations that remain to be investigated. They fly and, moreover, stay for hours resting at night in the environment of mountains at altitudes over 5000 m with variable wind speed and direction, where the air density and partial pressure of oxygen is roughly half of that at sea level 35,50 . At the same time, the temperature can be very low, well below freezing year-round, which could require additional metabolic energy for thermogenesis. Maintaining water balance during flight should also be a major challenge in the dry air at high altitudes 35,50 . Ontogenetic shifts in summer areas of immature Black Kites. Contrary to immature Black Kites using the West African-Eurasian flyway 19 , immature Kites in our study returned to the natal area in their first years of life or migrated to even more northerly areas. It seems that high behavioural flexibility is apparent during summer stays of immature Black Kites. Furthermore, such a difference in behaviour brings up an assumption that birds crossing over Himalayas are less constricted by barriers then those wintering in sub-Sahara part of Africa. We believe that the reason for such a difference may be more complex conditioned by many factors such as climate, habitat quality, food abundance, density of populations in breeding areas and with it connected competition and possibly genetic background. Although our results showed that ontogenetic shifts may differ between subspecies of a single species, the causes and consequences of such a variation remain unknown and require further research. Unlike Kumar et al. 26 , we found no difference in the size of the home range during the breeding and nonbreeding seasons.
Environmental influence on migration. Route configuration of Black Kites crossing the Himalayas seemed to be shaped by dominant wind support and barrier avoidance 26 . Black Kites perform circular soaring in areas of higher predicted thermal uplift and linear soaring in areas of higher predicted orographic uplift velocity 51 . During the pre-breeding crossing over Himalayas birds tent to fly parallelly along with the mountain ranges, through the mountain valleys using the up-lifting anabatic winds for soaring up to high altitudes and gliding with the possible strong south valley tailwinds 52 . During the period of pre-breeding migration (from the end of April to the beginning of May, which correspond to the timing of spring migration of tagged Black Kites) with the warmest and driest surface condition, great ascending thermals are forming, creating a great opportunity for soaring birds to glide over Himalayas 53 . While flying north along the mountain ridges, sidewind, that mostly blows from the west 52 , can break over the ridge creating a lee wind perpendicular to bird direction, that may have a negative effect on the birds' groundspeed as the bird has to angle towards the sidewind (as shown by our results). In contrast with the pre-breeding crossing, during the post-breeding Himalayas crossing over bird tent to fly directly across the mountain ranges.
We assume that birds used thermals to stay as high as possible to glid along or against the lee winds to avoid the strong headwinds of the valley breeze 52 . We found that Black Kites increased more their groundspeed and less their airspeed when tailwinds prevailed. For soaring migrants, reducing airspeed under tailwinds allows the birds to attain low sink rate and by that to cover larger distances while decreasing the risk of reaching the ground or switching to energy-expensive flapping flight 54 . However, during pre-breeding Himalayas crossing, birds noticeable increased their airspeed even during stronger tailwind. We believe that this behaviour is partly caused by the abundance of great ascending thermals. Bird can afford to increase its airspeed on the expense of higher sink rate in order to quickly pass the Himalaya barrier. Similar behaviour was observed in Honey Buzzards (Pernis apivorus) that were found to glide at fast airspeeds only in those areas where the best soaring conditions occurred 55 .
What we found interesting is the effect of different season on air and groundspeed of migrating birds. Tagged Kites kept increasing their airspeed even with prevailing tailwind, which shows on birds own motivation to increase its overall speed during the spring crossing on the expense of energy that they could have saved with lowering the airspeed in tailwind and gliding with low sink ratio. Many studies of avian migration showed that birds tend to migrate faster during spring migration than autumn migration 56 . Migration theory predicts that migrants minimalize the duration of spring migration to arrive in breeding area as soon as possible. Birds that arrive sooner start to breed earlier which can positively affect the reproductive performance 56,57 . Additionally, they will have more time for raising better quality offsprings that have better chance to survive their first migration 58 . Although there are cases when the spring migration took approximately the same time or longer 59,60 . We found the spring migration to be significantly shorter in comparison with autumn (post-breeding) migration, although the duration of the Himalayan crossing was found to be the same. As we mentioned earlier, for aerial migrants, wind represent a major support that can considerably reduce both energy and time cost of migration 61 . A stronger tailwind prevailing during spring increased the birds' speed and eased the Himalaya crossing. Birds were less exhausted from the Himalayas crossing over and arrived at summer destination much faster. Based on all that, we suggest that birds in our study migrated faster during the spring migration due to both favourable wind conditions and inner motivation.  63,64 . Novel observations of the communal roosting of Black Kites during the winter months have been reported in southeastern Europe, Egypt, and Turkey; however, their taxonomic subspecies status was not mainly investigated [65][66][67][68][69][70] .
Increasing number of Black Kites spotted in the Middle East seems to be related to a consistent increase in Black Kites numbers migrating along eastern part of Black Sea from 2011 43 . Now, the Black Kite is the most common wintering raptor in Israel, and a proportion of kites wintering in Israel showed morphological characteristics of M. m. lineatus, likely representing the western outpost of wintering M. m. lineatus 71 . Alternatively, these individuals may comprise birds from the broad intergradation zone between M. m. migrans and M. m. lineatus 72 .
It now appears that Black Kites with M. m. lineatus features supposedly originated from a large intergradation zone between M. m. migrans and M. m. lineatus can be found anywhere in Europe west of Russia 16 . Recent data on numerous wintering of Black Kites in Georgia in an area of the Black Sea Basin correspond well with these data 73 . Moreover, Black Kites with M. m. lineatus features can be found migrating from southern and eastern Africa as documented in South Africa in November 1972 and Ethiopia in November 2011 74,75 .

Conclusion
By telemetry research and DNA analyse of Black Kites from Western Siberia we found differences in subpopulations of Black Kites from Upper Altai close to Kosh-Agach and Black Kites from Biysk, pointing at the intergradation zone between M. m. migrans and M. m. lineatus and revealing their migration routes. Black Kites M. m. lineatus migrating to winter in Indian Subcontinent were challenged by the crossing of the main Himalayan ridge. They flew and roosted in the environment of mountains at altitudes over 5000 m in unfavourable weather conditions. During crossing, birds showed a response to the wind direction which helped them to overcome the environmental obstacle. Remarkable behavioural flexibility of Black Kites to surmount various environmental obstacles on their migration routes may be one reason that the species has been able to colonize such a large breeding range and may also elucidate the ongoing rapid establishment of novel wintering areas by Black Kites. What is more, Black Kites crossing the Himalayas may have physiological adaptations that remain to be investigated.
Ethics statement. Black Kite trapping and tagging were done in accordance with Art. 44 of the Federal Law No. 52-FZ "On the Animals"-the use of the animals for scientific, cultural, educational, recreational and aesthetic purposes through various forms of observation, tagging, photographing and other research methods without removing the animals from the habitat. In Russian Federation, the Black kite is not classified as protected species and no permits are required for any manipulations with it. Trapping and tagging of birds was performed by trained and experienced person. We performed all methods in accordance with the relevant guidelines and regulations with respect to our study animals. We confirm that the study is reported in accordance with ARRIVE guidelines 76 .

Availability of data and materials
The datasets used and/or analysed during the current study are available from the corresponding author on reasonable request.