Results

Koen Hufkens

The swift CC894 is assumed to be at least one year old, but sub-adult, as it did not find a nesting spot during the logging period. Swift CC894 returned to the colony and briefly was established in a nest box (camera box 2) around the 25th of May 2023, as visually confirmed by video feed and sexed as female (copulation, egg laying), before subsequent logger retrieval. The logger on swift CC894 telemetry showed that logging started on the preset start date (2021-07-10) and the battery remained functional and holding charge until retrieval (2023-06-11), running ~25 months, yet exhausting the logger memory for both light level and ancillary measurements. Light and pressure data did not suggest roosting at a nest (box) during the day or night, showing vastly insufficient roosting opportunities to foster a brood. Across the whole logging period only a total of ~48 hours scattered across the whole logging period of >12 000 hours are flagged as potentially inactive, this in contrast to well defined inactive roosting periods for breeding adult swifts (Appendix Figure S1).

Swift CC894 followed a similar flight patterns to adults in the colony when migrating to foraging sites in Eastern Africa, passing briefly through Western Africa and skirting the gulf of Guinea to pass south of the Congo Basin to the coastline of Malawi and Mozambique (Figure 1a, Hufkens et al. 2023). During this trip the swift made several stop-overs. Stop-overs were more common in Eastern Africa than those in Western Africa. Average distances between the sub-adult and adult stop-overs registered to be in the low hundreds of kilometers (303 ± 108 km). Swift CC894 showed great site fidelity between the two observed years, returning to within a hundred kilometres (~129 km) from previously visited final (furthest) foraging locations in eastern Mozambique, year-on-year (see Figure 1a).

A comparison of diurnal flight behaviour during the non-breeding and breeding season showed distinct differences (Figure 1b). During the non-breeding season differences in average daily flight altitude or flight heights are limited. However, during the breeding season twilight flights remain present but are lower in absolute altitude (i.e. due to the terrain) but higher in flight height (i.e. due to a larger diurnal change) than during the non-breeding season. We also note the lower altitudes during daytime for sub-adults than adults (Appendix Figure S3). In short, during the day non-breeding season flights are higher in baseline altitude than those during the breeding season for sub-adults (Appendix Figure S2), while roosting behaviour in adult birds (see Appendix Figure S1) is shown by negative flight heights.

During the non-breeding season the sub-adult swift followed the established diurnal pattern of moonlight induced flight height changes of adults, with high flights in response to moonlit night times (Figure 1b). Increased flight heights were recorded in response to sky brightness values as low as 0.02 - 0.05 lux, translating to a modelled flight height of ~850 m (Figure 2c, with C = 854 (17 Std. Error), k = -65 (5 Std. Error)). Moonlight-driven flight patterns were not observed in sub-adults during the breeding season, with no characteristic asymmetry in flight height, following the moon phases, as observed during the non-breeding season (Figure 1b).

We note that during the breeding season swifts residing around Ghent, Belgium, are exposed to high sky glow values of up to ~0.12 lux (Figure 2d). Night sky brightness data (Falchi et al. 2016) show 31% of the values during the breeding season exceeding the 0.1 lux on cloudy nights, as 55% of the time the mean nighttime cloud cover exceeded 50% (Figure 2d). During ambient illuminance values exceeding ~0.1 lux will trigger a maximum flight height response during moonlit nights (Figure 2b). As such, maximum flight heights under extensive Belgian sky glow (Appendix Figure S3) show correspondence to moonlight responses (Figure 2c). A comparative analysis for a location in Southern Sweden (Lund) during the non-breeding season did not allow us to differentiate sub-adult flight behaviour based on previous radar studies. Although at Lund light pollution and sky glow is lower (at ~0.009 lux and ~0.098 lux, respectively) and high latitude summer nights are brighter with most only reaching nautically twilight with illuminance values varying from 0.008 lux to 2 (Appendix Figure S4). Illuminance values exceeding 0.01 lux are sufficient to trigger an ascending flight, when based on moonlight behaviour during the non-breeding season (Figure 2b).

Figure 1. Overview panel plot showing: a) the flight path of the common swifts (Apus apus) from Ghent Voorhaven including the sub-adult CC894 (N21196, top panels) and four adult swifts (top panels) as derived from MDL geolocator data during two consecutive seasons (2021 - 2022 and 2022 - 2023). The general movement of the flight path is indicated by arrows (↑↓) Geolocator based locations were clustered using a breakpoint analysis dividing migration and stationary periods, and only stationary periods were retained and shown as full circles (⏺). Error bars showing quantiles around the median position for every cluster. The originating colony at the Ghent Voorhaven site is marked with a large light blue triangle. Locations outside the mapping domain around the equinoxes were removed. b) The diurnal flight height (normalized altitudes in m, relative to daytime mean values) dynamics, centered on midnight, for both sub-adult and adults swifts according to moon phases (from left to right: new moon, first quarter, full moon and last quarter as indicated by icons). Adult flight dynamics are shown as ribbon plots bounded by the 25th and 75th quantile of the flight height across all individuals. Median flight responses for the sub-adult CC894 are shown as a single full or dotted line, for the breeding and non-breeding season, respectively. Note the overlap between moonlight-driven flight height responses during the non-breeding season for both adults and the sub-adult (see Hufkens et al. 2022). During the breeding season this correspondence in adult swifts is lacking, as well as the typical asymmetrical flight height patterns in response to moon phases.

Figure 1. Overview panel plot showing: a) the flight path of the common swifts (Apus apus) from Ghent Voorhaven including the sub-adult CC894 (N21196, top panels) and four adult swifts (top panels) as derived from MDL geolocator data during two consecutive seasons (2021 - 2022 and 2022 - 2023). The general movement of the flight path is indicated by arrows (↑↓) Geolocator based locations were clustered using a breakpoint analysis dividing migration and stationary periods, and only stationary periods were retained and shown as full circles (⏺). Error bars showing quantiles around the median position for every cluster. The originating colony at the Ghent Voorhaven site is marked with a large light blue triangle. Locations outside the mapping domain around the equinoxes were removed. b) The diurnal flight height (normalized altitudes in m, relative to daytime mean values) dynamics, centered on midnight, for both sub-adult and adults swifts according to moon phases (from left to right: new moon, first quarter, full moon and last quarter as indicated by icons). Adult flight dynamics are shown as ribbon plots bounded by the 25th and 75th quantile of the flight height across all individuals. Median flight responses for the sub-adult CC894 are shown as a single full or dotted line, for the breeding and non-breeding season, respectively. Note the overlap between moonlight-driven flight height responses during the non-breeding season for both adults and the sub-adult (see Hufkens et al. 2022). During the breeding season this correspondence in adult swifts is lacking, as well as the typical asymmetrical flight height patterns in response to moon phases.

Figure 2. Overview plot of the flight dynamics during the 25 month journey of sub-adult common swift (Apus apus) CC894 (ring N21196). We show a) the raw actogram of flight height changes using normalized altitudes (m) relative to daytime mean values with a dashed vertical line marking the logger retrieval date, b) the non-breeding season moonlight driven flight height changes, c) and the changes in sky glow due to cloud cover, relative to a fixed site based light pollution baseline (dotted line) across the breeding season. Nighttime moonlight driven flight responses in (b), for adult swifts and sub-adult swift CC894, are fitted using an inverse exponential fit for months October to May and shown as dashed and full lines, respectively. The expected flight heights from model fits for a reference value of 80% of the asymptote (dashed vertical line), or maximum flight height, is highlighted with dotted vertical line. The sub-adult modelled 80% maximum flight height illuminance level is marked as a horizontal dashed line in (c), with new moon values marked as closed black circles (⏺), linking exceeding this reference value with increased flight heights during the breeding season.

Figure 2. Overview plot of the flight dynamics during the 25 month journey of sub-adult common swift (Apus apus) CC894 (ring N21196). We show a) the raw actogram of flight height changes using normalized altitudes (m) relative to daytime mean values with a dashed vertical line marking the logger retrieval date, b) the non-breeding season moonlight driven flight height changes, c) and the changes in sky glow due to cloud cover, relative to a fixed site based light pollution baseline (dotted line) across the breeding season. Nighttime moonlight driven flight responses in (b), for adult swifts and sub-adult swift CC894, are fitted using an inverse exponential fit for months October to May and shown as dashed and full lines, respectively. The expected flight heights from model fits for a reference value of 80% of the asymptote (dashed vertical line), or maximum flight height, is highlighted with dotted vertical line. The sub-adult modelled 80% maximum flight height illuminance level is marked as a horizontal dashed line in (c), with new moon values marked as closed black circles (⏺), linking exceeding this reference value with increased flight heights during the breeding season.