Days to visit an offshore island: effect of weather conditions on arrival fuel load and potential flight range for common blackbirds Turdus merula migrating over the North Sea

Background Crossing open water instead of following the coast(line) is one way for landbirds to continue migration. However, depending on prevailing weather and the birds’ physiological conditions, it is also a risky choice. To date, the question remains as to which interplay between environmental and physiological conditions force landbirds to stop on remote islands. We hypothesise that unfavourable winds affect lean birds with low energy resources, while poor visibility affects all birds regardless of their fuel loads. Methods To test this hypothesis, we caught 1312 common blackbirds Turdus merula stopping over on Helgoland during autumn and spring migration. Arrival fuel load was measured using quantitative magnetic resonance technology. Weather parameters (wind and relative humidity as a proxy for visibility) were interpolated for the night before arrival. Further, we calculated whether caught individuals would have successfully crossed the North Sea instead of landing on Helgoland, depending on wind conditions. Results Both wind and relative humidity the night before arrival were correlated with arrival fuel load. After nights with strong headwinds, birds caught the following day were mostly lean, most of which would not have managed to cross the sea if they had not stopped on Helgoland. In contrast, fat birds that could have successfully travelled on were caught mainly after nights with high relative humidity (≥ 80%). Furthermore, the rate of presumably successful flights was lower due to wind: although only 9% of all blackbirds captured on Helgoland had insufficient fuel loads to allow safe onward migration in still air, real wind conditions would have prevented 30% of birds from successfully crossing the sea during autumn and 21% during spring migration. Conclusions We were able to decipher how physiological condition, wind and relative humidity partially force blackbirds to stop on a remote island. Adverse winds tend to affect lean birds with low energy resources, while poor visibility can affect blackbirds, regardless of whether the arrival fuel load was sufficient for onward flight. Our findings will help to understand different migratory strategies and explain further questions like migration timing. Supplementary Information The online version contains supplementary material available at 10.1186/s40462-021-00290-6.

: Flow chart for overview for better understanding of the method section. With the arrows and flow chart parts we outline the respective temporal and spatial course of our study: 1) take-off of the blackbirds at an unknown position ahead of Helgoland, 2) approach to Helgoland the night before landing (yellow) oriented towards the selected potential flight paths (black), 3) actual stopover on Helgoland and capture the following day (blue) or 4) simulation of a continuation of the flight over Helgoland (including arrival fuel load as energy limitation) oriented towards the selected potential flight paths (including flight distance, grey) towards 5) the selected coastal destinations. Each of the coloured boxes includes point in time and the parameters measured at that point.

Chapter 2: Information on EchoMRI™
For this scanning procedure, we used an appropriately sized ventilated holding tube and the associated 'BIRD' setting of the EchoMRI TM software (accumulation setup: three), cf. Kelsey et al. (2019). Repeatability of the absolute fuel load within the three scans was high (= 0.99).
Required is a weekly calibration of the QMR device using the software 'system test' and the manufacturer's canola oil standard as well as a daily calibration prior to the start of actual scanning. See Guglielmo et al. (2011), Seewagen & Guglielmo (2011 and Kelsey & Bairlein (2019) for more details on the correct use of the EchoMRI TM .

Chapter 3: Overview of weather parameters
Temperature ("air"; K) was not included in the analyses because birds can fly at air temperatures below -15 °C (Bruderer 1971) and above 40 °C (Whitfield et al. 2015), while air temperatures ranged from -13.95 °C (including only two nights with below -10.00 °C) to 19.45 °C during our observation periods (supplementary Table 1). Metabolic rates of flying birds are 5-10 times higher than resting rates, substituting for regular thermogenesis (Dawson & Connor 1996). Therefore, dramatically increased energetic costs of thermoregulation as well as possible ice accumulation on the plumage (Elkins 2004), which could lead to a "forced" landing of the birds, are unlikely. The same applies to possible hyperthermia with heat stress effects (Whitfield et al. 2015), as hyperthermia contributes little to the cost of thermoregulation in birds up to around 100 g (Weathers 1981).
Furthermore, precipitation, cloud cover and cloud altitude were not used because these are not directly assimilated but derived entirely from the predictions of the NCEP reanalysis models and are therefore less reliable (Kalnay et al. 1996;Kistler et al. 2001;Hüppop & Hilgerloh 2012

Chapter 5: Overview of average fuel-dependent flight range
In addition to the measurements of the main text, fat score (levels 0 to 8) was estimated for each blackbird according to Kaiser (1993). In this study, blackbirds only showed fat score levels from zero to five. A linear model (LM) was used to analyse absolute fuel loads (dependent variable) between different fat score levels (six-level explanatory factor: 0 to 5). Since the arrival fuel load (dependent variable) is given in relative proportions, we also used a generalised linear model with a binomial error distribution (GLM; family "quasibinomial") to correlate arrival fuel load (dependent variable) with fat score levels (six-level explanatory factor: 0 to 5).
Absolute and relative arrival fuel loads differed between the fat score levels (Supplementary Table 3), except for the comparison between fat score level "0" and "1". Here, CrI overlapped in term of absolute and relative arrival fuel loads; most likely due to the small sample size for level "0". All of the blackbirds classified in fat score level "5" had sufficient arrival fuel loads enabling them to fly between 446 km and 1,185 km in still air without further refuelling, theoretically allowing them to reach all possible coastal destinations. On the other hand, none of the blackbirds with a fat score of "0" would have been able to reach the next coastal destination in still air. Table 4: Fat score levels, absolute and (relative) arrival fuel load and potential flight range. We have described the estimated mean (in bold) and CrI (in brackets). Corresponding potential flight ranges (km) are given considering still air (groundspeed = 10 m/s).

Chapter 6: Excluded model parameters and additional results
Initially, the models analysing arrival fuel load between migrants caught after "unfavourable weather" nights and migrants caught otherwise (favourable weather") included the interaction between relative humidity and TWC, but as the interaction did not show an effect in any of our models, it was excluded from further analysis.
All models initially included year as a random factor; however, as the variance of this parameter had an effect of (close to) zero, we excluded it from our models. Time of day-effects were not considered as relative fuel load showed high variability during the entire day with coefficients of determination (R²) close to zero in both seasons (spring: R² = 0.0003; autumn: R² = 0.0007), which is in accordance with Dierschke & Bindrich (2001).