1. Intermittent breeding is an important tactic in long-lived species that trade off survival and reproduction to maximize lifetime reproductive success. When breeding conditions are unfavourable, individuals are expected to skip reproduction to ensure their own survival. 
2. Breeding propensity (i.e. the probability for a mature female to breed in a given year) is an essential parameter in determining reproductive output and population dynamics, but is not often studied in birds because it is difficult to obtain unbiased estimates. Breeding conditions are especially variable at high latitudes, potentially resulting in a large effect on breeding propensity of Arctic-breeding migratory birds, such as geese.
3. With a novel approach, we used GPS-tracking data to determine nest locations, breeding propensity and nesting success of barnacle geese, and studied how these varied with breeding latitude and timing of arrival on the breeding grounds relative to local onset of spring.
4. Onset of spring at the breeding grounds was a better predictor of breeding propensity and nesting success than relative timing of arrival. At Arctic latitudes (> 66°), breeding propensity decreased from 0.89 (95% CI: 0.65-0.97) in early springs to 0.22 (95% CI: 0.06-0.55) in late springs, while at temperate latitudes it varied between 0.75 (95% CI: 0.38-0.93) and 0.89 (95% CI:  0.41-0.99) regardless of spring phenology. Nesting success followed a similar pattern, and was lower in later springs at Arctic latitudes, but not at temperate latitudes.  In early springs, a larger proportion of geese started breeding despite arriving late relative to the onset of spring, possibly because the early spring enabled them to use local resources to fuel egg laying and incubation.
5. While earlier springs due to climate warming are considered to have mostly negative repercussions on reproductive success through phenological mismatches, our results suggest that these effects may partly be offset by higher breeding propensity and nesting success.

We collated tracking data (GPS and accelerometer when available) of 96 adult female barnacle geese. This dataset includes geese with various life history tactics, ranging from long-distance migrants breeding in the Arctic (above 66°N) to short-distance migrants and residents in the temperate zone (between 51°N and 66°N). Birds were caught and equipped with GPS-transmitters on the breeding grounds in the Arctic in 2014 and 2018 (N=6; 68°34' N, 52°18' E), on breeding grounds in the temperate zone (residents) in 2015-2018 (N=7; 51°47' N, 4°08' E), and on the wintering grounds in the North of the Netherlands and North Germany in 2016-2020 (N=72). Additionally we retrieved tracking data from Kölzsch et al. (2015) gathered in 2008-2010 (N=11), which is published on movebank.org (van der Jeugd et al., 2014).  In winter, geese were caught using cannon-nets, while in summer geese were captured either on the nest using clap-nets or by rounding geese up in a catching pen during the wing moult when geese are flightless. Throughout the study period, geese were equipped with different transmitter types: Milsar (24 g, GSMRadioTag, Milsar Technologies S.R.L.; N=2), UvABiTS (19 g, (Bouten et al., 2013); N=11), Madebytheo (27 g, N=24), Ornitela (25 g; N=48) and solar Argos/GPS PTTs (30 g, N=11, Kölzsch et al., 2015). Transmitters were attached using a 16-gram Teflon harness (Lameris et al., 2017), with total mass of transmitter and harness being < 3% of the average body mass of a female barnacle goose. The GPS-tags used by Kölzsch et al. (2015) were attached using a comparable nylon harness. 

We selected data from the period between April 1 and July 31, during which barnacle geese of different breeding populations initiate nesting. IN case of high frequence records, we sub-sampled data to 1 GPS-fix every 15 min to make transmitters with different sampling regimes comparable. All tracks had tracking information for at least 112 days within the 122-day period. ACC data was not available for all tracks, due to differences in transmitter type as well as settings used.  When ACC data were available (Ornitela, UvaBiTS; 78 tracks), we defined a potential nest location as the median coordinates of GPS-fixes at which the goose was motionless, on days when the goose was mostly motionless (see Schreven et al., 2021). We used ACC measurements which were taken simultaneous with GPS positions, or if ACC was not measured simultaneously, we took the nearest ACC-measurement with maximum 10 min time difference. We used VeDBA as measure of activity (Dokter et al., 2018). To deal with the differences in accelerometer types as well as burst length and burst frequency, the threshold in VeDBA below which a goose was categorized as motionless was set per transmitter type, at: 24 (Milsar), 16.5 (Ornitela) and 26.5 (UvABiTS) following methods by Boom et al. (2023). For tracks where ACC-measurements were not available (78 out of 154 tracks), we used the GPS-only method (see Schreven et al., 2021, who showed that this method worked well in comparison with the GPS and ACC based nest detection), defining a potential nest location as the median coordinates of GPS-fixes on days when the standard deviation in latitude was below 50 m. For each potential nest location, we calculated the time spent per day within 50 m of the nest. These locations were classified as nest when at least 4 subsequent days had >75% attendance within 50 m, allowing for differentiation between non-breeding and early nest failure (after first 4 days). Tracks of geese for which no nesting location could be determined (because no nest location was assigned which the goose attended for at least 4 days) were considered non-breeding tracks (N=81).

To retrieve information on location (latitude), timing of arrival and spring phenology for non-breeding bird tracks (N=81), we needed to estimate the most likely potential nest location of non-breeding birds (hereafter referred to as “potential nest location”). To determine the potential nest locations, we calculated for each non-breeding bird the similarity between its track and the tracks of breeding birds (i.e. those with an assigned nest, N = 73). Tracks were subset to the period April-June and mean locations per day were taken. We used Dynamic Time Warping (DTW), using the R package “dtw” (Giorgino, 2009) to calculate trajectory similarities (Janoska, 2014). The potential nest location of non-breeding birds was determined as the nest location of the most similar breeding bird (track of the bird with the lowest “distance” as determined by the DTW-analysis).

For every track, arrival date was determined as the first day a goose was within 35 km of its assigned nest location (for breeders) or potential nest location (for non-breeders). Not all non-breeding birds came within 35 km of the potential nest location. We concluded that for these birds determination of a potential nest location was not reliable and therefore these birds were excluded from the  statistical analysis.

We derived information on the local onset of spring from temperature data for all years between 2008-2020. We used the R package RNCEP (Kemp et al., 2012) to retrieve temperature data from January 1 till September 30 for all assigned (breeders) and potential (non-breeders) nest locations. Air temperature (2 m above the surface) was retrieved from a 2.5°x2.5° gridded dataset with 6 h temporal resolution, and interpolated using planar interpolation. We calculated the Growing Degree Days (GDD) for every nest location. Due to the wide latitudinal range in our study, the threshold used to determine the GDD was latitude-dependent, following a linear relationship (Tthreshold = -0.25 x Latitude + 13; see van Wijk et al. 2012). Subsequently, we fitted a sigmoid curve, and derived the GDD jerk (third derivative) following van Wijk et al. (2012). By solving the derivative of the GDD jerk to zero, we determined the date of peak spring temperature acceleration. To correct for variation in timing of onset of spring with latitude, local onset of spring was standardized.

Boom, M. P., Lameris, T. K., Schreven, K. H., Buitendijk, N. H., Moonen, S., de Vries, P. P., Zaynagutdinova, E., Nolet, B.A., van der Jeugd, H.P. & Eichhorn, G. (2023). ‘Year-round activity levels reveal diurnal foraging constraints in the annual cycle of migratory and non-migratory barnacle geese’. Oecologia, pp. 287–298.  doi: 10.1007/s00442-023-05386-x.

Bouten, W.,Baaij, E.W.,Shamoun-Baranes, J. & Camphuysen, K.C.J. (2013) ‘A flexible GPS tracking system for studying bird behaviour at multiple scales’, Journal of Ornithology, 154(2), pp. 571–580. doi: 10.1007/s10336-012-0908-1

Giorgino, T. (2009) ‘Computing and visualizing dynamic time warping alignments in R: The dtw package’, Journal of Statistical Software, 31(7), pp. 1–24. doi: 10.18637/jss.v031.i07.

Janoska, Z. (2014) Trajectory similarity calculation using Dynamic Time Warping. https://rpubs.com/janoskaz/10351. Accesed on 2022-01-24

Kemp, M.U.,Emiel van Loon, E.,Shamoun-Baranes, J. & Bouten, W. (2012) ‘RNCEP: Global weather and climate data at your fingertips’, Methods in Ecology and Evolution, 3(1), pp. 65–70. doi: 10.1111/j.2041-210X.2011.00138.x.

Kölzsch, A.,Bauer, S.,de Boer, R.,Griffin, L.,Cabot, D.,Exo, K.-M.,van der Jeugd, H.P. & Nolet, B.A. (2015) ‘Forecasting spring from afar? Timing of migration and predictability of phenology along different migration routes of an avian herbivore.’, The Journal of animal ecology. Edited by S. Bearhop, 84(1), pp. 272–83. doi: 10.1111/1365-2656.12281.

Lameris, T.K.,Kölzsch, A.,Dokter, A.M.,Nolet, B.A. & Müskens, G.J.D.M. (2017) ‘A novel harness for attaching tracking devices to migratory geese’, Goose Bulletin, (22), pp. 25–30.

Schreven, K.H.T.,Stolz, C.,Madsen, J. & Nolet, B.A. (2021) ‘Nesting attempts and success of Arctic-breeding geese can be derived with high precision from accelerometry and GPS-tracking’, Animal Biotelemetry, 9(1), pp. 1–13. doi: 10.1186/s40317-021-00249-9.

van der Jeugd, H.P.,Oosterbeek, K.,Ens, B.J.,Shamoun-Baranes, J. & Exo, K. (2014) ‘Data from: Forecasting spring from afar? Timing of migration and predictability of phenology along different migration routes of an avian herbivore [Barents Sea data].’, Movebank Data Repository. doi: 10.5441/001/1.ps244r11

van Wijk, R.E.,Kölzsch, A.,Kruckenberg, H.,Ebbinge, B.S.,Müskens, G.J.D.M. & Nolet, B. a. (2012) ‘Individually tracked geese follow peaks of temperature acceleration during spring migration’, Oikos, 121(5), pp. 655–664. doi: 10.1111/j.1600-0706.2011.20083.x.