Get 20M+ Full-Text Papers For Less Than $1.50/day. Start a 14-Day Trial for You and Your Team.

Learn More →

Nesting attempts and success of Arctic-breeding geese can be derived with high precision from accelerometry and GPS-tracking

Nesting attempts and success of Arctic-breeding geese can be derived with high precision from... Sensors, such as accelerometers, in tracking devices allow for detailed bio-logging to understand animal behav- iour, even in remote places where direct observation is difficult. To study breeding in birds remotely, one needs to understand how to recognise a breeding event from tracking data, and ideally validate this by direct observation. We tagged 49 adult female pink-footed geese (Anser brachyrhynchus) with transmitter neckbands in Finland in spring of 2018 and 2019, and in Svalbard in summer 2018, and validated inferences from tracking by field observations of nesting sites and family status in 2018–2020 (54 spring–summer tracks). We estimated nesting locations by taking the median coordinates of GPS-fixes at which the goose was motionless (overall dynamic body acceleration, ODBA < 1) on days with a daily median ODBA < 1, which approached the real nesting locations closely (within 1.6–3.7 m, n = 6). The start of nesting was defined as the first day on which the goose spent > 75% of time within 50 m of the nest, because nest site attendances steeply increased within one day to above this threshold. Nesting duration (number of consecutive days with > 75% nest site attendance) ranged between 3 and 44 days (n = 28), but was 30–34 days in confirmed successful nests (n = 9). The prolonged nesting of 39–44 days (n = 3) suggested incubation on unhatch- able egg(s). Nest losses before hatching time occurred mostly in day 3–10 and 23–29 of nesting, periods with an increased frequency of nest site recesses. As alternative method, allowing for non-simultaneous GPS and accelerome- ter data, we show that nesting days were classified with 98.6% success by two general characteristics of breeding: low body motion (daily median ODBA) and low geographic mobility (daily SD of latitude). Median coordinates on nesting days approached real nest sites closely (within 0.8–3.6 m, n = 6). When considering only geographic mobility (allow- ing for GPS data only) nesting locations were similarly accurate, but some short nesting attempts were undetected and non-breeding tracks misclassified. We show that nesting attempts, as short as 3 days, and nesting success can be detected remotely with good precision using GPS-tracking and accelerometry. Our method may be generalised to other (precocial) bird species with similar incubation behaviour. Keywords: Anser brachyrhynchus, ODBA, Incubation, Nesting duration, Brood, Parental care, Recess Background The tracking of individual breeding attempts allows us to estimate population growth parameters and study individual variation in reproductive decisions which *Correspondence: k.schreven@nioo.knaw.nl Department of Animal Ecology, Netherlands Institute of Ecology (NIOO- influence these parameters. In birds, the timing of breed - KNAW ), Droevendaalsesteeg 10, 6708 PB Wageningen, The Netherlands ing, especially relative to the local onset of spring, has Full list of author information is available at the end of the article © The Author(s) 2021. This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http:// creat iveco mmons. org/ licen ses/ by/4. 0/. The Creative Commons Public Domain Dedication waiver (http:// creat iveco mmons. org/ publi cdoma in/ zero/1. 0/) applies to the data made available in this article, unless otherwise stated in a credit line to the data. Schreven et al. Anim Biotelemetry (2021) 9:25 Page 2 of 13 profound effects on breeding propensity [1], clutch size We focus on the migratory pink-footed goose (Anser and hatching success [2–4], chick growth and survival brachyrhynchus), which breeds in remote arctic regions [5–7] and local recruitment [8 reviewed for geese in 9]. and exhibits similar behaviour before, during and after Generally, birds are found to time their breeding so that the breeding season as many other goose species [23, the peak food demand of the offspring during growth 24] (see Fig. 1). During spring staging, pink-footed geese coincides with the peak in food abundance [10, 11] or forage large parts of the day and move tens of kilome- food quality [12]. tres daily between roosting and foraging sites [25]. They Due to climate warming, the onset of spring is advanc- complete the over-sea migration from Norway and Fin- ing [13] and the highest rate of climate change on Earth land to the breeding grounds in 2 days (range 1–9), arriv- occurs in the Arctic [14]. In response, Arctic-breeding ing on the Svalbard breeding grounds mid–late May [26 migratory birds may adjust their arrival to the breed- and own data]. Just after arrival on the breeding grounds, ing grounds and timing of breeding to keep up with this they may roost for the biggest part of a day (pers. comm. environmental change [6], or may adjust their breeding J. Prop), and then forage actively on different sites before location to a place or habitat with a more favourable phe- nesting [26, 27]. We define nesting here as the combina - nology [15]. tion of egg-laying, nest-building, incubation, and hatch- A remote region does not easily allow for direct obser- ing. Pairs may visit prospective nesting sites several times vations of breeding birds. With modern tracking devices, before settling [28]. Once a nest site has been selected, migratory birds can be tagged in a non-breeding location the pair does not leave the area anymore and starts with and followed without direct observation to their remote egg-laying and nest-building immediately, with most breeding areas such as the Arctic [e.g., 16], while addi- nest-building happening up to 6 days after nest initiation tional sensors such as accelerometers allow for detailed [28]. The start of egg-laying occurs usually in late May to bio-logging to understand the birds’ behaviour [e.g., 17, early June, but ranges from mid-May to mid-June [3, 24, 18]. However, to enable the study of breeding biology in 29, 30]. The clutch is completed with commonly 3–5 eggs such birds, one needs to understand how to recognise a (range 1–8) around day 5–9 after nest initiation, when breeding event from tracking data, and ideally validate incubation sharply increases from 40–60% to 90–100% of this with individuals that were observed directly, which is time [3, 24, 28–30]. Incubation is solely carried out by the challenging in remote areas. female, in a sedentary posture with the head forward or Breeding events have been detected in previous stud- leaning on the back, while the male stands nearby. The ies based on GPS data [e.g., 19, 20, 21], but additional incubation period until hatching is 25–28  days, during sensors, especially an accelerometer, are expected to which females usually leave the nest to feed once per day increase the power and precision of nest(ing) detection. [3, 24, 29, 31, 32]. All chicks in one nest hatch within 24 h, Accelerometry enables us to focus on GPS-fixes, or days, and the family leaves the nest within 36–48  h, or even on which the bird displayed an activity level or body ori- within 12–24 h if disturbed [28, 29 pers. obs. J. Madsen). entation of interest [e.g., 22]. Especially for short nest- During this period, the female may forage with the chicks ing attempts, which might otherwise be indiscernible around the nest, but also go back to the nest to brood from daily roosting or foraging routines based on GPS the chicks [28]. After incubation, females spend much patterns, accelerometry could increase the detection time feeding and little time vigilant [33]. Wing moult in probability. Previous studies on wood stork (Mycteria adults starts 2–3 weeks after hatching, 1 week earlier for americana), lesser kestrel (Falco naumanni), Mediterra- non-breeders and renders birds flightless for 25  days in nean gull (Ichthyaetus melanocephalus) [21] concluded July–August [24, 29]. During moult, geese mainly forage, that nesting attempts shorter than 7–14 days can usually preen and roost [34]. Geese depart from Svalbard to Nor- not be detected based on GPS-tracking data. Also, [19] way around mid-September [35, own data]. and [20] defined nests of GPS-tracked barnacle geese We use the above descriptions of breeding behaviour in (Branta leucopsis) and greater white-fronted geese (Anser this species to define characteristics of nesting behaviour albifrons) as locations from which geese stayed within that can be measured quantitatively by GPS-tracking and 2 km for at least 10 days, during a defined breeding sea - accelerometry, and thereby extract nesting locations and son. However, a more precise method which can detect durations for 49 adult females, tagged in Svalbard and short nesting attempts is important to distinguish a low Finland. We ground-truth this by direct observation of breeding propensity from a high breeding failure. Fur- nesting sites on the Svalbard breeding grounds in sum- thermore, nesting success can be determined, by evaluat- mer and family status on subsequent autumn, wintering, ing if the bird incubates the clutch long enough for it to and spring staging sites. We further assess how nesting hatch successfully. failure covaries with nest attendance. We believe that our S chreven et al. Anim Biotelemetry (2021) 9:25 Page 3 of 13 Fig. 1 A pink-footed goose with a transmitter neckband. Incubating female ‘2M’ (solar-powered GPS-GSM transmitter neckband) and guarding male ‘KK9’ (normal plastic neckband) at their nest site in Endalen, Longyearbyen, Svalbard, 29 June 2020. Photo: Christian Stolz, colour-edited by Stijn Schreven method can also be used in other (precocial) bird species and 7 males. Here, we focus on females, although limited with similar incubation behaviour. data on males suggest the method also works when using males. Of the 49 female geese, we obtained 54 tracks dur- ing the breeding season, over 3  years (2018: 4, 2019: 29, Methods 2020: 21). Catching geese Pink-footed geese were tagged in Svalbard and Finland. In Svalbard, geese were caught during wing moult, in GPS‑tracking and accelerometry family groups, on 30 July 2018 (the lake of Isdammen, We used solar-powered GPS-GSM transmitter neck- Longyearbyen, 78°12′12.7′′N, 15°48′10.3′′E) and 1 August bands, type OrniTrack-N38 (Ornitela UAB, Lithuania) 2018 (the coastal plain of Daudmannsøyra, 78°13′16.6′′N, with a weight of 38  g (c. 1.5% of body mass) and an 13°04′10.8′′E). More details on catching and handling inner diameter of 38  mm (see Fig.  1). Tags were white procedure are given in [34]. The capture and tagging and had a black two-digit individual code that could be caused the birds to increase their amount of time spent read from a distance in the field. The tags recorded a preening, but this effect disappeared within 6–7  days GPS-fix every 10 min (when battery voltage is 75–100%, after tagging [34]. In Finland, geese were caught during 4000–4150  mV), 20  min (50–74%, 3850–3995  mV), spring migration, by canon-net on agricultural fields in 30  min (25–49%, 3760–3845  mV) or 60  min (< 25%). In the novel staging area in Tyrnävä, Oulu (64°49′33.1′′N addition, when the voltage was above 85% (4060  mV), 25°33′25.1′′E, see also [36]) on 28 April 2018, 27 April instead of a single GPS-fix a GPS-burst of 10 fixes was 2019 and 1 May 2019. taken at 1 Hz, to increase accuracy of elevation and speed Geese were sexed by cloacal examination in the field, measurements. which was validated molecularly following [37], using Immediately after each GPS-fix, or GPS-burst, a 2-s blood taken from medial metatarsal vein, primer pair accelerometer measurement burst was taken at a fre- 2550F/2718R and the PCR-programme of [38] and run- quency of 20  Hz. Gravitational acceleration (unit g) ning results on a 2% agarose gel. was measured in three dimensions: along the neck In total, 56 geese received a transmitter in Svalbard (y-axis), and around the neck (x and z-axes, which are (n = 35) and Finland (n = 21), of which 49 were females Schreven et al. Anim Biotelemetry (2021) 9:25 Page 4 of 13 interchangeable due to free rotation of the neckband). trampled in the lining of the nest. Remains from depre- Battery voltage level was recorded with each GPS-fix. dated eggs were recognised by an irregular shape, pres- For analysis, a subset of data was taken in which, ence of bite or peck marks, yolk remains and/or a weak besides the single GPS-fixes, only the first GPS-fix of thin membrane. If no eggshells were present, the nest each GPS-burst was kept, to keep data quantity and qual- was assumed to be depredated [3, 24, 29]. ity homogeneous. Further subsetting was not necessary, These field observations resulted in direct observations as the variation in logging interval on the level of interest of two geese incubating on their nests, and checks of 13 (i.e. the within-individual within-day level) was limited: sites of in total seven geese (six geese with one site each, on average only 7% of days contained multiple different one goose with two sites and one goose with five sites). logging intervals and on these days on average only 17% These sites concerned six confirmed nests (one per goose, of fixes had a different logging interval than the most including the two nests where incubating geese were common logging interval length of that individual on that observed), five roosting/foraging areas (all of one goose), day. Overall, 98.3% of fixes were taken at 10-min inter - and one uncertain site, which was excluded (i.e. first site val, 1.3% at 20-min, 0.4% at 30-min, and 0.1% at intervals of the goose with two sites). Of the six confirmed nests, of 1  h or longer. Overall, voltage levels were on average three had hatched and three were depredated. The uncer - 95.6 ± SD 7.8% (range 0–100%). tain site was attended by the goose for > 75% of time The precision of a GPS-fix was evaluated with two tags within 50 m on both 1–2 June 2020, whereas the goose’s placed on a fixed known coordinate, i.e. a wooden pole second site, a confirmed nest 302  m away, was attended at Adventdalen weather station during 4–9 August 2019. for > 75% of time within 50  m on all 11–13 June 2020. This showed that 24% of GPS-fixes was within 5 m of the This uncertain site might also have concerned a nest, as pole, 47% within 10 m, 74% within 20 m and 96% within the time and distance between the sites corresponds to 50 m (n = 1424 fixes). recent findings about replacement clutches of early-failed Barnacle geese (Branta leucopsis, pers. comm. Jouke Prop, Nordenskiöldskysten, Svalbard). However, replace- Field observations ment clutches are not known for pink-footed geese [24, For ground-truthing, nesting and roosting/forag- 29, 30] but are difficult to observe, as early-failed nests ing locations of geese on their breeding grounds were have only accumulated little down [28] that can be blown observed in Adventdalen, Svalbard, during 20 June–11 by wind. At this uncertain site, only goose droppings July 2020. These locations were identified with a rough were found. procedure: selecting condensed clusters of GPS-fixes We further increased the sample size of confirmed suc - on a map, and taking the average coordinates per clus- cessful geese by seven, based on observations of geese ter. A cluster was defined as a collection of GPS-fixes after nesting on Svalbard and during autumn migration showing that a goose frequently revisited an area with a in Norway, Sweden, Denmark, Netherlands, and Belgium diameter of 20–50 m, for at least 2 days, with relatively entered into the citizen science platform www. geese. org few fixes elsewhere during this period, and occurring [39]. In geese, partners migrate together and with their within the time frame of arrival (mid–late May) until fully fledged brood, enabling an evaluation of breeding the usual hatching period (late June). This procedure success and pair status in autumn and even up to early included all nesting sites as identified by the final proce - spring when juveniles leave their parents [e.g., 40]. The dure below, but also additional roosting/foraging sites. four tracked females in 2018 were seen without chicks In the field, these sites were observed from a distance in autumn. Of the 29 females tracked in 2019, one was (to check presence of a tagged goose, possible partner seen with chicks in summer and/or autumn, 28 without. and chicks), and visited (if no goose was present). In the Of the 21 females tracked in 2020, eight were seen with majority of cases, female geese were not flushed from chicks in summer and/or autumn, and 13 without. Thus, the nest by the observer; only in two cases, the female the total sample of confirmed nesting geese was 13, of left the nest. From a distance it was observed that the which ten were successful and three had failed. female returned to the nest site. During a visit, the sur- We further increased the sample size of roosting sites roundings within 20  m of each site were checked for by including geese roosting just upon arrival at the the presence of a nest, of which the coordinates were breeding grounds, as nesting then is physiologically not then taken with a handheld GPS device (GPSMAP 64 s, ® yet possible [41]. Therefore, if potential ‘nesting’ sites Garmin ). For each nest, the number of eggs and pres- were detected (by the final procedure below) within ence of egg remains in and around the nest was noted. 3  days after arrival, they were regarded as roosting sites Egg remains were evaluated as coming from hatched (n = 4 geese). From the five roosting sites checked in the nests if a thick membrane was present, yolk remains field, described above, only one was detected as potential and predator bite marks were absent, and if they were S chreven et al. Anim Biotelemetry (2021) 9:25 Page 5 of 13 ‘nesting’ site, and therefore the total sample of roosting For our first and main approach, simultaneously sites was five. recorded GPS and ACC data is needed (Fig.  2a). As nesting is characterised by prolonged periods of sitting Deriving nesting attempts still, we expected to measure for breeding geese a time Nesting attempts were derived from GPS tracking and window of consistently low values of body motion. As accelerometry by identifying cut-off values of several proxies for body motion, overall dynamic body accelera- quantifiable characteristics that follow from the breeding tion (ODBA) and vectorial dynamic body acceleration behaviour descriptions in the background. We selected (VeDBA) are in use [42]. Although VeDBA is mathemati- May–July for analysis, i.e. from spring staging and migra- cally better in line with theory of acceleration [42], in our tion until halfway moult and/or chick rearing, and used case, ODBA gave clearer contrasts for a goose within a the tracks of confirmed nesting geese. We developed season, and we therefore used ODBA. We first calcu - two approaches suitable for different data structures (see lated the static acceleration for each of the three dimen- Fig. 2 for a schematic representation of the approaches). sions (x, y, z) within a burst as the average raw measured Fig. 2 Schematic representation of the approaches to extract nesting attempts and success of pink-footed geese based on tracking data. The main approach (a) requires simultaneously recorded GPS and accelerometer (ACC) data, whereas the second approach (b) uses general movement characteristics summarised on a daily level, and can handle non-simultaneous GPS and ACC data, or GPS data only. ODBA overall dynamic body acceleration (see “Methods” Section for calculation). In b, cut-off values were determined using Recursive Partitioning and Regression Trees (see “Methods” section; Fig. 4), and the success of these classifications was determined by comparison with results of approach (a). After identifying possible nests in b, certain nests and successful nests can again be identified with the attendance threshold, period threshold and nesting duration as described in a Schreven et al. Anim Biotelemetry (2021) 9:25 Page 6 of 13 acceleration of each dimension within the burst. We then how well nesting can be predicted by two general char- subtracted for each dimension (x, y, z) the static accelera- acteristics that geese display during nesting: low geo- tion from the raw measured acceleration. This gives the graphic mobility and low body motion, summarised on dynamic acceleration, of which then the absolute value a daily level (see Fig.  2b for a schematic representation). was taken. The absolute values were then summed for all Geographic mobility was measured by the daily stand- 40 points of all three dimensions within a 2-s burst to get ard deviation of latitude of GPS-fixes. Longitude was the ODBA of that burst. We defined an ODBA threshold not analysed, as the daily SD in latitude and daily SD in to identify periods of ‘motionlessness’ based on the direct longitude were strongly correlated (r = 0.92, t = 165.03, observation of geese incubating their clutch (n = 2) and df = 4797, p < 0.0001). We compared daily median other geese that were confirmed to have nested (n = 11, ODBA and daily SD of latitude between different stages see “Field observations” section). As geese attend their of the annual cycle, for each goose based on its GPS and nest almost continuously during nesting, we determined accelerometer data: spring migration (from May 1st until the potential nesting location of each goose by taking the reaching the breeding grounds), pre-nesting (defined median values of the longitude and latitude of a subset as the period between spring migration and nesting), of GPS-fixes. The subset was made by selecting the days nesting (as determined by our first approach above), on which the goose was predominantly motionless (i.e. post-nesting (after nesting until the end of July), and for daily median ODBA below threshold), and then taking non-breeding geese the summering period as one stage from these days all GPS-fixes during which the goose was (i.e. from spring migration until the end of July). Stages motionless (i.e. ODBA below threshold). The standard were compared using Linear mixed models (LMM) using deviation of the coordinates of the selected fixes was used the R-package “lme4” [43] on natural logarithm-trans- to evaluate the precision of the derived potential nest- formed data to approach normality. A random effect of ing site. We assessed nest site attendances by calculating track ID (i.e. each year separately for each individual, n = the distance of each GPS-fix of a goose (also on days on 54) was included and p-values and degrees of freedom which the goose was not mostly motionless) to its poten- were estimated by Satterthwaite’s method (R-package tial nesting location and then reporting the daily amount “lmerTest”) [44]. We then established a procedure to clas- of time that the goose spent within a radius of 5, 10, 20, sify days as either ‘nesting’ or ‘non-nesting’ using Recur- and 50  m of its potential nest site. The radius of 50  m sive Partitioning and Regression Trees (R-package ‘rpart’) gave the clearest contrasts over time and was used in fur- [45]. We split our data randomly into a training set (50%), ther analysis. We then defined an attendance threshold: on which the tree was based, and a testing set (50%), with a daily proportion of time spent within the radius of the which the tree was validated. We tested two different nest that is typical of nesting. We derived this by evalu- classification trees: (1) using only geographic mobility as ating for confirmed nests (n = 13, see “Field observa- a single predictor, or (2) using both geographic mobility tions” section) if there was a steep increase in nest site and body motion as predictors. The comparison of the attendance, which was defined as the start of nesting, and two classifications allowed us to assess the added value of what values characterised attendance before and after the sensor data as opposed to GPS data without accelerome- start. Incubation may be confused with roosting behav- try. Following both classifications, nesting locations were iour (i.e. geese also sitting still, resulting in low ODBA), determined by taking the median coordinates of nesting but roosting is expected to happen in shorter episodes days, and compared ground-truthed nest coordinates and not always on the same location. Therefore, we calcu - (n = 6). After determining possible nesting sites with lated the nesting duration as the number of consecutive this second approach, real nests could be identified by days on which the goose showed above-threshold attend- again using the attendance threshold and period thresh- ance to its potential nesting location. In order to exclude old as explained in the first approach. All statistical tests roosting locations and identify the location as a real nest, were performed in R [46]. the nesting duration should be above a minimum num- ber of days, i.e. the period threshold. To define the period Deriving nesting success threshold, we compared roosting locations (n = 5) and To derive nesting success—binary, i.e. whether any chicks nesting locations (n = 13, see “Field observations” sec- hatched from a clutch or not—we defined a range in nest - tion). If a goose spent fewer days than the period thresh- ing duration typical of successful nests (Fig. 2a). This was old at its potential nesting site, roosting and nesting achieved by comparing confirmed successful geese (n = could not be distinguished. 9) with confirmed unsuccessful geese (n = 3) and breed- Our second and complementary approach does not ing geese with unknown success, i.e. geese that were require that GPS and ACC data were collected simul- breeders according to our first approach, but not seen taneously, and can use GPS only as well. We evaluated with young in autumn (n = 16, see “Field observations” S chreven et al. Anim Biotelemetry (2021) 9:25 Page 7 of 13 section). One successful goose was excluded here typical for nesting (283 out of 377  days, n = 13 nesting because of tag failure from halfway incubation onwards. geese, Figs.  3a, c; 4a), During such time windows, geo- We expected the nesting duration of successful geese to graphic mobility was consistently low as well (Figs.  3b; be 30–38  days, i.e. 5–9  days egg-laying, plus 25–28  days 4a). Our method to extract nesting locations reproduced incubation, plus 0–1  day hatching (see “Background” the field-recorded nesting locations well, i.e. on average section). Any nesting duration outside this range would within 2.7  m (SD = 0.9, range 1.6–3.7  m, n = 6 nests). imply complete hatching failure. Per goose, the SD of latitude in the subset of GPS-fixes To assess whether any nest with unknown nesting suc- from which the nesting location was derived was on cess, but a nesting duration typical of successful nests average 9.7 ± SD 4.9 m (range 5.6–20 m). The subset of could be assumed to be successful, we analysed the geo- one specific goose with a prolonged incubation period graphic mobility and body motion during the post-nest- showed an SD for latitude of 635 m, but the real nest site ing stage. We compared successful breeders, i.e. with was still approximated within 1.7  m. For all identified chicks throughout summer (n = 9), breeders that failed roosting episodes (n = 5 geese), the SD of the subset of during nesting, i.e. without chicks in summer (n = 16) GPS-fixes from which the roosting location was derived and geese with a nesting duration typical of hatched was on average for latitude 90.1 ± SD 153  km (range nests, but with unknown hatching success and without 54.5  m–356  km). The large values were generated by chicks by the time of autumn migration (n = 3). Addi- geese that were motionless for most of the day because tionally, one goose nested successfully but lost its chicks of roosting, but moved between different roost sites, of during summer or autumn. We used LMMs with a ran- which some were several hundred kilometres apart when dom effect of track ID, on natural logarithm-transformed the goose had just arrived in Svalbard. data. For confirmed nesting geese (13 nests), the time a goose spent within 50 m from the nest steeply increased Variation in nest site attendance within one day by on average 70 ± SD 17% (range As we expected to find a range of nesting durations 35–93%), namely from a site attendance of on average (relating to nests that had been depredated, hatched, or 23 ± SD 21% (range 0–65%) to on average 94 ± 7% deserted), we aimed to better understand the moment (range 81–100%, Figs.  3d; 5a). Therefore, the attendance in the nesting cycle at which nests failed. For this, we threshold was set at 75% of the time spent within the evaluated variations in the daily number and duration 50  m radius of the site. The period threshold was set at of nest site recesses, i.e. bouts of time that a goose spent 3 days, as confirmed nesting geese showed nesting dura - beyond 50  m away from its nest. We divided the nest- tions of 3–44 days, i.e. subsequent days with > 75% nest ing cycle before possible hatching into three periods: day site attendance (n = 12 nests), whereas roosting geese 0–10, day 11–19, and day 20–29 after nest initiation. For showed maximally 1 day meeting the attendance thresh- each goose, we selected the period from its first nest site old (n = 5 geese). visit to the day before the end of nesting, and calculated per day the number and average duration of bouts spent General characteristics during breeding: body motion beyond 50 m away from the nest. When a bout spanned and geographic mobility 2 days, we included the bout only on the day on which it During May–July, body motion (daily median ODBA, started. Bout duration was calculated by taking the time log-transformed) of female geese plotted against their difference between the first and last GPS-fix within the geographic mobility (daily SD in latitude, log-trans- bout, plus 10 min. Thereby, bouts with only one GPS-fix formed) showed clear clusters that corresponded to dif- were assumed to have lasted 10 min, as the GPS-logging ferent stages of the annual cycle (Fig.  4). The nesting interval during nesting was nearly always 10 min (median stage was characterised by lower daily median ODBA voltage during pre-nesting and nesting stages 99.0%, than all the other stages combined (Table  1, LMM, t = mean 96.4 ± SD 6.7%, range 33–100). Both the number 86.77, df = 4597, p < 0.0001). Also, the geographic and duration of recesses of all nesting geese were then mobility was lower during nesting than all other stages compared between the three periods in the nesting cycle, combined (Table  1, LMM, t = 52.09, df = 3961, p < using LMMs including a random effect of track ID. 0.0001). Classification success into nesting versus non- nesting days was 96.4% when using only geographic Results mobility as predictor, and 98.6% when using both geo- Extracting nesting locations and nest attendance graphic mobility and body motion as predictors (Fig.  4). The ODBA threshold was defined at 1.0, because the From the 29 breeders determined by the main approach, daily median ODBA was below this value on 75% of 26 were recognised as such by GPS-only (i.e. showing at days during prolonged time windows of motionlessness, Schreven et al. Anim Biotelemetry (2021) 9:25 Page 8 of 13 Fig. 3 Body motion, geographic mobility, and nest attendance of a pink-footed goose during spring and summer. Overall dynamic body acceleration (ODBA) per burst varied largely but showed during nesting a time window of motionlessness (a, c: zoomed in), during which geographic mobility was also notably low (b). GPS-fixes with an ODBA < 1 on days with median ODBA < 1 were selected to extract the nesting site by taking median coordinates (c), from which then nest site attendance was calculated (d). Inferred start and end of nesting are indicated by orange arrows in d, given by a threshold of daily attendance above 75% within 50 m from the nest. This nest was confirmed to have hatched Fig. 4 Body motion and geographic mobility of pink-footed geese during different annual cycle stages (May–July). Per goose, the daily median of the overall dynamic body acceleration (ODBA) was plotted against the daily standard deviation of latitude, both transformed by taking the natural logarithm. Nesting was characterised by days with low body motion and low geographic mobility. Exact cut-off values for the classification of nesting days were determined with Recursive Partitioning and Regression Trees, based on GPS data only (red dotted line) or GPS and accelerometer data combined (green dashed line). a Depicts females that nested (n = 29 tracks), while b depict non-breeding females (n = 25 tracks). The nesting period was defined based on nest attendance patterns (see Fig. 5). For the other periods, see text S chreven et al. Anim Biotelemetry (2021) 9:25 Page 9 of 13 Fig. 5 Nest site attendance and recesses throughout the nesting cycle. Nesting duration, i.e. number of subsequent days with > 75% (dashed horizontal green line) of time spent within 50 m of the nest, varied between 3 and 44 days for all nests, but between 30 and 34 for successful nests (a). Most nest attempts stopped in the first or last 10 days of the incubation cycle (delineated by dashed vertical grey lines). In these periods, geese went more often beyond 50 m from the nest, than during the period in between, while the duration of each of those bouts did not vary significantly (b). Grey shading indicates the start and end of successful nesting. In b, means are depicted ± SE Table 1 Body motion and geographic mobility of pink-footed geese during different annual cycle stages (May–July) Annual cycle stage Female geese included Daily median ODBA Daily SD of latitude (km) Spring migration (from 1 May) Breeders (n = 29) and non-breeders (n = 25) 20.38 ± 9.23 (0.35–54.59) 26.3 ± 59.0 (0.0356–381) Pre-nesting Breeders (n = 29) 30.01 ± 9.73 (0.70–50.57) 6.12 ± 16.7 (0.00857–130) Nesting Breeders (n = 29) 1.08 ± 2.40 (0.49–41.73) 0.0189 ± 0.0434 (0.00278–0.668) Post-nesting (until 31 July) Breeders (n = 29) 17.73 ± 10.80 (0.54–44.41) 0.568 ± 2.67 (0.0145–56.8) Successful, with chicks (n = 9) 15.50 ± 7.24 (1.58–33.95) 0.301 ± 0.223 (0.0156–1.89) Failed, without chicks (n = 16) 19.31 ± 12.02 (0.54–44.41) 0.523 ± 1.78 (0.0145–21.2) Successful, but lost chicks in summer/autumn 22.51 ± 11.19 (0.49–35.53) 3.09 ± 6.41 (0.116–23.3) (n = 1) With unknown hatching success, with 13.94 ± 9.65 (2.95–39.31) 0.323 ± 0.401 (0.0189–2.56) 30–34 days nesting (n = 3) Non-breeding summer (until 31 July) Non-breeders (n = 25) 20.44 ± 12.82 (0.44–49.01) 1.67 ± 7.12 (0.00690–104) For five different annual cycle stages, the daily median overall dynamic body acceleration (ODBA) and daily SD of latitude are summarised here as mean ± SD (range). Nesting was characterised during spring–summer by days with low body motion and low geographic mobility. Post-nesting, breeders with chicks throughout summer showed lower body motion and geographic mobility than breeders without chicks. The nesting period was defined based on nest attendance patterns (see Fig. 1). For the other periods, see text least 3 consecutive nesting days), and 28 by GPS + ACC. detected; GPS + ACC: average 2.2  m, SD 1.0  m, range However, from the 25 non-breeding tracks determined 0.8–3.6 m, n = 6 nests). by the first approach, four were classified as breeding by GPS-only, while GPS + ACC classified them all correctly Deriving nesting success as non-breeding. This resulted in a total classification The nesting durations of all nesting attempts identified success of breeding/non-breeding status for GPS-only with the first approach varied between 3 and 44 days (n = of 87.0%, and for GPS + ACC of 98.1%. The difference 28; Fig.  5a). Of these, nests that were confirmed to have was mainly caused by GPS-only missing short attempts hatched showed a nesting duration of 30–34  days (n = and misjudging some non-breeding tracks. Nevertheless, 9), while nests that were confirmed to have failed showed nesting locations estimated by taking the median coor- nesting durations of 3, 18 and 44  days, respectively. The dinates of classified nesting days were similar based on nesting duration of all geese nesting for shorter than GPS-only or GPS + ACC, and approached the ground- 30 days (including those with unknown success) showed truthed nesting locations within few metres (GPS-only: a bimodal distribution with most nest attempts stopping average 2.5  m, SD 0.8  m, range 1.6–3.6  m, n = 5 nests, after 3–10 days (n = 7) or 23–29 days (n = 5) and only as 1 ground-truthed short nesting attempt was not one after 18  days. Further nesting attempts of unknown success stopped after 32(2 × ), 33, 39 and 41 days. Schreven et al. Anim Biotelemetry (2021) 9:25 Page 10 of 13 During the post-nesting stage, successful breeders of other activities than nesting, e.g., foraging or preen- with chicks had a lower daily body motion level (LMM, ing. This was apparent in our second approach, where t = − 2.301, df = 23.5, p = 0.031; Table 1) and a lower accelerometry increased the likelihood that both short geographic mobility (LMM, for latitude: t = −  2.641, breeding attempts and non-breeding tracks are cor- df = 20.4, p = 0.015; Table 1) than failed breeders with- rectly classified, whereas the goose’s spatial behaviour out chicks. The body motion and geographic mobility per se could be ambiguous (i.e. low geographic mobility of breeders with unknown nesting success but a nesting for a short time). This is useful when one aims to study duration of 30–34  days, seemed more similar to that of the timing of breeding, or to distinguish a low breeding breeders with chicks than failed breeders without chicks propensity from a high breeding failure. We argue that (Table  1), but was not significantly different from either including accelerometer data, if available, can increase group (with chicks: LMM, body motion, t = −  0.279, the power of nest detection. df = 9.7, p = 0.79; LMM, for latitude, t = 0.932, df = By evaluating the proportion of time spent at such 1020, p = 0.35; without chicks: LMM, body motion, t = locations, our method could also distinguish roosting −  1.432, df = 17.5, p = 0.17; LMM, for latitude, t = areas from nesting attempts that lasted at least three − 1.076, df = 15.6, p = 0.30). days. A goose may roost for longer periods, but appar- ently rarely on exactly the same spot during subsequent days. Roosting behaviour may differ for other species, Variation in nest site attendance e.g., raptors, which may have specific roosting trees [47] In the middle of the nesting cycle (day 11–19), geese or gulls, which consistently roost on the same breakwater made 0.87 ± SE 0.18 nest site recesses beyond 50  m [48]. For such species, the body orientation as measured from the nest per day, which tended to be lower, but not by an accelerometer would still be useful to detect nest- significantly, than the 1.1 ± SE 0.14 recesses per day ing sites, e.g., straight up during roosting, versus horizon- in the first period (day 0–10, df = 719, t = 1.882, p = tal during nesting in raptors [49]. Our study provides a 0.060) and was significantly lower than the 1.6 ± SE 0.15 reliable method that enables the study of avian breeding recesses per day in the last period (day 20–29, df = 709, in remote environments, which is often logistically chal- t = 4.537, p < 0.001; Fig. 5b). The duration of these bouts lenging with classic observations. was not different between the middle period (18.5 ± SE The nesting period is unique in its virtual absence of 4.7  min) and the first period (19.5 ± SE 0.73  min, df = movement. Although this would suggest low energy 400, t = 1.342, p = 0.18) or the last period (18.7 ± SE expenditure [42], nesting is energetically a demand- 0.76 min, df = 399, t = 0.222, p = 0.82; Fig. 5b). ing time for geese, resulting in large body mass loss as egg production is costly and foraging opportunities are Discussion traded off against incubation activity and nest defence We have presented a method to use GPS-tracking and [50]. The higher daily activity levels we found in the pre- accelerometry to remotely identify pink-footed goose nesting stage may reflect active foraging for large parts nesting attempts and their locations, the start and end of of the day [27], whereas movement patterns in the post- each nesting attempt, and thus also an indication of nest- nesting stage depended on family status. Geese with ing success given by the duration of nest site attendance. chicks had lower body motion and lower geographic We also showed that > 98% of nesting days could be clas- mobility than geese without chicks, probably the result of sified correctly by two main movement characteristics regularly brooding the chicks in the field and flying less. of the breeding phase of the annual cycle (i.e. low geo- The large geographic mobility that distinguishes spring graphic mobility and low body motion) and that includ- migration results both from daily movements between ing accelerometry especially helps to detect short nesting roost sites and foraging sites within a stop-over, as well as attempts correctly. directional movements between stop-overs [25]. Our main approach acquires a high level of precision The nesting duration of successful nests concentrated, and can detect nest attempts as short as 3 days, which is as expected, at 30–34 days, whereas unsuccessful nests not easily achievable with methods based on only GPS showed both shorter and longer nesting durations. without additional sensor data. For example, [21] sug- For further studies, one may assume that any goose gest for three bird species that nesting attempts below with unknown nesting success but a nesting dura- 7–14  days can usually not be detected based on GPS- tion of 30–34  days is successful, as post-nesting body tracking data, and [19] and [20] use for two goose species motion and geographic mobility of such geese seemed a minimum duration of 10  days. The power of includ - more similar to those with hatched nests than failed ing sensor data in the analysis lies in the fact that one nests. Nesting durations shorter than 30 days are likely can specifically select GPS-fixes (or days) with a certain caused by nest predation or abandonment [1]. Also body motion of interest and exclude those that are part S chreven et al. Anim Biotelemetry (2021) 9:25 Page 11 of 13 human disturbance might indirectly lead to nest loss Combining GPS with accelerometer data can increase the by predation [51]. Nest losses before normal hatching power to detect short nesting attempts and determine the time occurred in our study predominantly in the first timing, location and success of breeding. Studying avian and last ten days of a normal nesting cycle. This pat - breeding in the changing remote arctic environment is of tern may be explained by the slightly higher recorded great interest and will be subject of future research, facili- frequency of nest site recesses in these periods, com- tated by the method presented in this paper. pared to the period in between. Nest recesses, to for- age or drink, expose the nest to predators [3]. Because our GPS-fixes include GPS error, we cannot report Conclusions exactly how far a goose was situated from its nest dur- Nesting attempts of Arctic-breeding pink-footed geese ing a bout. Also, recesses shorter than 10  min can be (as short as 3  days) and their locations can be derived missed due to the logging interval. Still, our analysis is remotely and accurately from GPS-tracking and accel- expected to give a reliable qualitative view on (far and erometry, while nesting success is indicated by the long) nest site recesses. Moreover, the same pattern of nesting duration. We predict that this method can be nest attendance over the incubation cycle was reported generalised to other (precocial) bird species with simi- based on direct observations and showed that females lar incubation behaviour. spend more time foraging in the first week and last days of nesting than in the period in between [28]. In Abbreviations contrast, dark-bellied Brent geese (Branta bernicla ber- ACC : Accelerometry; ODBA: Overall dynamic body acceleration; VeDBA: Vecto- nicla) and greater snow geese (Chen caerulescens atlan- rial dynamic body acceleration; SD: Standard deviation. tica) were found to make fewer and shorter recesses Acknowledgements in the last days before hatching [50, 52]. Differences in We are grateful to people who helped catch geese in Svalbard (Christian incubation constancy between species may result from Sonne, Peter de Vries, Koen Nolet, Ove Martin Gundersen, John Frikke, Cornelia Jaspers, Kevin Kuhlmann Clausen, Cecilia Sandström, Dies Snoodijk and the the interplay of remaining body stores (thus, the need crew of RV “Ulla Rinman”) and in Finland (Jorma Pessa, Tuomas Väyrynen, Esko for foraging trips), foraging opportunities close to the Pasanen, Aija Lehikoinen, Christanne de Vries, Jorma Siira, Matti Tolvanen, nest, and the local risk of nest predation (thus, the need Antti Piironen, Christian Sonne, Lars Haugaard, Jens Peter Hounisen, Michael Schmidt, Niels-Erik Jørgensen, Maël Charbonneaux, Kalle Hiekkanen). Further to stay on the nest). Future studies using our method we thank observers, especially Aija Lehikoinen, Jørgen Peter Kjeldsen, Ole could evaluate how nest attendance and nesting dura- Amstrup, Fred Cottaar, Eckhart Kuijken, Christine Verscheure, Leo Schilperoord tion vary with environmental factors and may change and Mirjam Schilperoord-Huisman, Ove M. Gundersen, Kevin K. Clausen, Kent Larsson and Lennart Eriksson, for observing tagged geese in the field, and with arctic warming. Martijn van der Sluijs and Christa Mateman for the molecular sexing. Fieldwork Nesting durations longer than 34  days are longer than in Norway was kindly facilitated by Skogn Folkehøgskole, Skogn. We thank expected, even for large clutches and may indicate that Stijn Schreven for editing figure 1, and two anonymous referees for their constructive comments. the goose was incubating unhatchable eggs. Pink-footed geese may continue to incubate an unhatched egg for Authors’ contributions 3 more days after all other eggs have already hatched, Conceptualisation, KS, BAN; F, KS, CS, JM, BAN; formal analysis, KS; writing— original draft KS; writing—review and editing, KS, CS, JM, BAN; supervision, JM., before they leave with the brood [28]. However, when the BAN. All authors read and approved the final manuscript. whole clutch is unhatchable, many bird species extend their incubation duration by at least 50% [53]. A nest with Funding Our work was supported by grants from the Netherlands Polar Programme solely unhatchable eggs is expected to be rare, but can, of the Dutch Research Council to BAN (ALWPP.2016.024), from the Svalbard for example result from predation of already-hatched Environmental Protection Fund to JM (No. 17/88) and from the Academy Ecol- chicks—in our study one nest was incubated for 44 days, ogy Fund of the Royal Netherlands Academy of Arts and Sciences to KHTS. The funding body had no role in the design of the study and collection, analysis, but contained only one egg on day 43. Egg hatching fail- and interpretation of data or in writing the manuscript. ure can be caused by environmental pollutants result- ing in embryonal death or deformities [e.g., 54, 55]. To Availability of data and materials The datasets generated and/or analysed during the current study are not detect such phenomena, the study of individual breeding publicly available yet, pending publication of another article, but are available attempts by GPS-tracking and accelerometry, as facili- from the corresponding author on reasonable request. tated by this study, is a valuable addition to the extensive monitoring of productivity on the population level [36]. Declarations Our study also highlights the importance of characteris- Ethics approval and consent to participate ing successful nesting attempts not solely by a minimum Permits to catch and tag geese in Svalbard were granted by the Norwegian nesting period, but by a defined range in nesting periods. Food Safety authority (Mattilsynet) (to Aarhus University, 17/210528), by the Technological advances in tracking devices make that Governor of Svalbard (17/01420-4) and Longyearbyen Lokalstyre (2018/347- 5-X70), in Finland by Etelä-Suomen aluehallintovirasto (to Aarhus University, GPS data are increasingly accompanied by sensor data. Schreven et al. Anim Biotelemetry (2021) 9:25 Page 12 of 13 ESAVI/1924/2018 and ESAVI/1880/2018) and Varsinais-Suomen elinkeino-, 16. Davidson SC, Bohrer G, Gurarie E, LaPoint S, Mahoney PJ, Boelman NT, liikenne- ja ympäritsökeskus (to Jorma Pessa, VARELY/551/2018). et al. Ecological insights from three decades of animal movement track- ing across a changing Arctic. Science. 2020;370(6517):712–5. Consent for publication 17. Dokter AM, Fokkema W, Bekker SK, Bouten W, Ebbinge BS, Müskens G, Not applicable. Olff H, van der Jeugd HP, Nolet BA. Body stores persist as fitness correlate in a long-distance migrant released from food constraints. Behav Ecol. Competing interests 2018;29(5):1157–66. https:// doi. org/ 10. 1093/ beheco/ ary080. The authors declare that they have no competing interests. 18. Nuijten RJM, Gerrits T, Shamoun-Baranes J, Nolet BA. Less is more: on- board lossy compression of accelerometer data increases biologging Author details capacity. J Anim Ecol. 2020;89(1):237–47. https:// doi. org/ 10. 1111/ 1365- Department of Animal Ecology, Netherlands Institute of Ecology (NIOO-2656. 13164. KNAW ), Droevendaalsesteeg 10, 6708 PB Wageningen, The Netherlands. 19. de Boer R, Bauer S, van der Jeugd HP, Ens BJ, Griffin L, Cabot D, Exo KM, Norwegian Institute for Nature Research (NINA), 7485 Trondheim, Norway. Nolet BA, Kölzsch A. A comparison of spring migration between three Department of Biology, Norwegian University of Science and Technol- populations of Barnacle Geese Branta leucopsis using GPS satellite trans- ogy (NTNU), 7491 Trondheim, Norway. Department of Bioscience, Aarhus mitters. Limosa. 2014;87:99–106. University, 8410 Rønde, Denmark. Department of Theoretical and Computa- 20. Kölzsch A, Müskens GJDM, Szinai P, Moonen S, Glazov P, Kruckenberg H, tional Ecology, Institute for Biodiversity and Ecosystem Dynamics, University Wikelski M, Nolet BA. Flyway connectivity and exchange primarily driven of Amsterdam, Science Park 904, 1098 XH Amsterdam, The Netherlands. by moult migration in geese. Mov Ecol. 2019;7(1):1–11. 21. Picardi S, Smith BJ, Boone ME, Frederick PC, Cecere JG, Rubolini D, Serra Received: 2 February 2021 Accepted: 21 June 2021 L, Pirrello S, Borkhataria RR, Basille M. Analysis of movement recursions to detect reproductive events and estimate their fate in central place forag- ers. Mov Ecol. 2020;8(1):1–14. 22. Wilson RP, White CR, Quintana F, Halsey LG, Liebsch N, Martin GR, Butler PJ. Moving towards acceleration for estimates of activity-specific meta- bolic rate in free-living animals: the case of the cormorant. J Anim Ecol. References 2006;75(5):1081–90. 1. Anderson HB, Madsen J, Fuglei E, Jensen GH, Woodin SJ, van der Wal 23. Scott DA, Rose PM. Atlas of Anatidae Populations in Africa and Western R. The dilemma of where to nest: influence of spring snow cover, food Eurasia. Wageningen, The Netherlands: Wetlands International; 1996. proximity and predator abundance on reproductive success of an arctic- 24. Cramp S, Simmons KEL. Handbook of the birds of Europe, the Middle breeding migratory herbivore is dependent on nesting habitat choice. East, and North Africa: the birds of the Western Palearctic, vol. I. Oxford: Polar Biol. 2014;38:153–62. Oxford University Press; 1977. 2. Perrins CM. The timing of birds’ breeding seasons. Ibis. 1970;112:242–55. 25. Chudzińska ME, Nabe-Nielsen J, Nolet BA, Madsen J. Foraging behaviour 3. Madsen J, Tamstorf M, Klaassen M, Eide N, Glahder C, Rigét F, Nyegaard and fuel accumulation of capital breeders during spring migration as H, Cottaar F. Eec ff ts of snow cover on timing and success of reproduc- derived from a combination of satellite- and ground-based observations. tion in high-Arctic Pink-footed Geese Anser brachyrhynchus. Polar Biol. J Avian Biol. 2016;47:563–74. 2007;30:1363–72. https:// doi. org/ 10. 1007/ s00300- 007- 0296-9. 26. Glahder CM, Fox AD, Hübner CE, Madsen J, Tombre IM. Pre-nesting site 4. Ross MV, Alisauskas RT, Douglas DC, Kellett DK. Decadal declines in avian use of satellite transmitter tagged Svalbard Pink-footed Geese Anser herbivore reproduction: density-dependent nutrition and phenological brachyrhynchus. Ardea. 2006;94(3):679–90. mismatch in the Arctic. Ecology. 2017;98:1869–83. 27. Fox AD, Francis IS, Bergersen E. Diet and habitat use of Svalbard Pink- 5. Brook RW, Leafloor JO, Abraham KF, Douglas DC. Density dependence footed Geese Anser brachyrhynchus during arrival and pre-breeding and phenological mismatch: consequences for growth and survival of periods in Adventdalen. Ardea. 2006;94(3):691–9. sub-arctic nesting Canada Geese. Avian Conserv Ecol. 2015;10:1. 28. Inglis IR. The breeding behaviour of the Pink-footed Goose: behavioural 6. Lameris TK, van der Jeugd HP, Eichhorn G, Dokter AM, Bouten W, Boom correlates of nesting success. Anim Behav. 1977;25:747–64. MP, Litvin KE, Ens BJ, Nolet BA. Arctic geese tune migration to a warm- 29. Bauer KM, von GlutzBlotzheim UN. Handbuch der Vögel Mitteleuropas, ing climate but still suffer from a phenological mismatch. Curr Biol. vol. 2. Frankfurt: Akademische Verlagsgesellschaft; 1968. 2018;28:2467–73. 30. Bezzel E. Kompendium der Vögel Mitteleuropas: Nonpasseriformes— 7. Ross MV, Alisauskas RT, Douglas DC, Kellett DK, Drake KL. Density- Nichtsingvögel. Wiesbaden: Aula-verlag; 1985. dependent and phenological mismatch effects on growth and survival in 31. Witherby HF, Jourdain FCR, Ticehurst NF, Tucker BW. The handbook of lesser snow and Ross’s goslings. J Avian Biol. 2018;49:e01748. British Birds. Watford Edinburgh: JJ Gray and Sun Printers Ltd; 1944. 8. Reed TE, Grøtan V, Jenouvrier S, Sæther BE, Visser ME. Population growth 32. Del Hoyo J, Elliott A, Sargatal J. Handbook of the Birds of the World, vol. 1. in a wild bird is buffered against phenological mismatch. Science. Barcelona: Lynx Edicions; 1992. 2013;340(6131):488–91. 33. Lazarus J, Inglis IR. The breeding behaviour of the Pink-footed Goose: 9. Nolet BA, Schreven KHT, Boom MP, Lameris TK. Contrasting effects of the parental care and vigilant behaviour during the fledging period. Behav- onset of spring on reproductive success of Arctic-nesting geese. Auk. iour. 1978;65(1/2):62–88. 2020;137(1):ukz063. 34. Clausen KK, Schreven KHT, Madsen J. Eec ff ts of capture and marking 10. Lack D. Ecological adaptations for breeding in birds. London: Methuen; on behaviour of moulting Pink-footed Geese Anser brachyrhynchus on Svalbard. Wildfowl. 2020;70:13–29. 11. Visser ME, van Noordwijk AJ, Tinbergen JM, Lessells CM. Warmer springs 35. Jensen GH, Tombre IM, Madsen J. Environmental factors affecting lead to mis-timed reproduction in Great Tits (Parus major). Proc R Soc numbers of Pink-footed Geese Anser brachyrhynchus utilising an autumn Lond B. 1998;265:1867–70. stopover site. Wildl Biol. 2016;22(5):183–93. 12. Doiron M, Gauthier G, Levesque E. Trophic mismatch and its effects 36. Madsen J, Jensen GH, Cottaar F, Amstrup O, Bak M, Bakken J, Balsby TTJ, on the growth of young in an Arctic herbivore. Glob Change Biol. Kjær Christensen T, Clausen KK, Frikke J, Gundersen OM, Kjeldsen JP, Koffi- 2015;21(12):4364–76. https:// doi. org/ 10. 1111/ gcb. 13057. jberg K, Kuijken E, Månsson J, Nicolaisen PI, Haaning Nielsen H, Nilsson L, 13. Høye TT, Post E, Meltofte H, Schmidt NM, Forchhammer MC. Rapid Reinsborg T, Ødegaard PI, Pessa J, Shimmings P, Tombre I, Verscheure C. advancement of spring in the High Arctic. Curr Biol. 2007;17(12):R449–51. Svalbard Pink-footed Goose Population status report 2017–2018. Bonn, 14. Cohen J, Screen JA, Furtado JC, Barlow M, Whittleston D, Coumou D, Germany: AEWA EGMP Technical Report; 2018. Francis J, Dethloff K, Entekhabi D, Overland J, Jones J. Recent Arctic ampli- 37. Fridolfsson AK, Ellegren H. A simple and universal method for molecular fication and extreme mid-latitude weather. Nat Geosci. 2014;7(9):627. sexing of non-ratite birds. J Avian Biol. 1999;30:116–21. 15. Alzate Valejo A. From local adaptation to range sizes: ecological and 38. Griffiths R, Double MC, Orr K, Dawson RJ. A DNA test to sex most birds. evolutionary consequences of dispersal. PhD thesis, Groningen, The Mol Ecol. 1998;7(8):1071–5. Netherlands: University of Groningen; 2018. S chreven et al. Anim Biotelemetry (2021) 9:25 Page 13 of 13 39. Ebbinge BS, Buij R, de Vries L, Moonen S, van Randen Y, Müskens G, van 48. van Donk SC. A gull’s portrait: On individual foraging strategies and their der Jeugd H, Koffijberg K, Voslamber B, Roosenschoon O, Kramer J. The fitness consequences. PhD Thesis, The Netherlands: Vrije Universiteit website geese.org, an interactive database to report marked waterfowl. Amsterdam; 2020. Goose Bulletin. 2020;25:11–8. 49. Newton I. The sparrowhawk. Calton: A&C Black; 1986. 40. Gupte PR, Koffijberg K, Müskens GJDM, Wikelski M, Kölzsch A. Family size 50. Spaans B, van’t Hoff CA, van der Veer W, Ebbinge BS. The significance of dynamics in wintering geese. J Ornithol. 2019;160:363–75. female body stores for egg laying and incubation in Dark-bellied Brent 41. Klaassen M, Hahn S, Korthals H, Madsen J. Eggs brought in from afar: Geese Branta bernicla. Ardea. 2007;95(1):3–15. Svalbard-breeding Pink-footed Geese can fly their eggs across the Bar - 51. Madsen J, Tombre I, Eide NE. Eec ff ts of disturbance on geese in ents Sea. J Avian Biol. 2017;48:173–9. https:// doi. org/ 10. 1111/ jav. 01364. Svalbard: implications for regulating increasing tourism. Polar Res. 42. Wilson RP, Börger L, Holton MD, Scantlebury DM, Gómez-Laich A, Quin- 2009;28(3):376–89. tana F, Rosell F, Graf PM, Williams H, Gunner R, Hopkins L, Marks N, Geraldi 52. Poussart C, Gauthier G, Larochelle J. Incubation behaviour of NR, Duarte CM, Scott R, Strano MS, Robotka H, Eizaguirre C, Fahlman A, greater snow geese in relation to weather conditions. Can J Zool. Shepard ELC. Estimates for energy expenditure in free-living animals 2001;79(4):671–8. using acceleration proxies: a reappraisal. J Anim Ecol. 2020;89(1):161–72. 53. Skutch AF. The constancy of incubation. Wilson Bull. 1962;74(2):115–52. 43. Bates D, Mächler M, Bolker BM, Walker SC. Fitting linear mixed-effects 54. Skorupa JP, Ohlendorf HM. Contaminants in drainage water and avian risk models using lme4. J Stat Softw. 2015;67(1):1–48. https:// doi. org/ 10. thresholds. In: Dinar A, Zilberman D, editors. The economics and manage- 18637/ jss. v067. i01. ment of water and drainage in agriculture. New York: Springer; 1991. 44. Kuznetsova A, Brockhoff PB, Christensen RHB. lmerTest package: tests in 55. van Oosten HH, van den Burg AB. Dioxines: een niet-onderkend prob- linear mixed effects models. J Stat Softw. 2017;82(13):1–26. https:// doi. leem voor zangvogels? The Netherlands: Report Stichting Bargerveen; org/ 10. 18637/ jss. v082. i13. 2014. 45. Therneau T, Atkinson B, Ripley B. Recursive Partitioning and Regression Trees. 2019. https:// cran.r- proje ct. org/ packa ge= rpart. Accessed 15 May Publisher’s Note Springer Nature remains neutral with regard to jurisdictional claims in pub- 46. R Core Team. R: A language and environment for statistical computing. lished maps and institutional affiliations. R Foundation for Statistical Computing, Vienna, Austria. 2020. https:// www.R- proje ct. org/. Accessed 01 Jan 2020. 47. Deng WH, Wei G, Guang-Mei Z. Nest and roost habitat characteris- tics of the Grey-faced Buzzard in northeastern China. J Raptor Res. 2003;37(3):228–35. Re Read ady y to to submit y submit your our re researc search h ? Choose BMC and benefit fr ? Choose BMC and benefit from om: : fast, convenient online submission thorough peer review by experienced researchers in your field rapid publication on acceptance support for research data, including large and complex data types • gold Open Access which fosters wider collaboration and increased citations maximum visibility for your research: over 100M website views per year At BMC, research is always in progress. Learn more biomedcentral.com/submissions http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Animal Biotelemetry Springer Journals

Nesting attempts and success of Arctic-breeding geese can be derived with high precision from accelerometry and GPS-tracking

Loading next page...
 
/lp/springer-journals/nesting-attempts-and-success-of-arctic-breeding-geese-can-be-derived-qukgX50F5x
Publisher
Springer Journals
Copyright
Copyright © The Author(s) 2021
eISSN
2050-3385
DOI
10.1186/s40317-021-00249-9
Publisher site
See Article on Publisher Site

Abstract

Sensors, such as accelerometers, in tracking devices allow for detailed bio-logging to understand animal behav- iour, even in remote places where direct observation is difficult. To study breeding in birds remotely, one needs to understand how to recognise a breeding event from tracking data, and ideally validate this by direct observation. We tagged 49 adult female pink-footed geese (Anser brachyrhynchus) with transmitter neckbands in Finland in spring of 2018 and 2019, and in Svalbard in summer 2018, and validated inferences from tracking by field observations of nesting sites and family status in 2018–2020 (54 spring–summer tracks). We estimated nesting locations by taking the median coordinates of GPS-fixes at which the goose was motionless (overall dynamic body acceleration, ODBA < 1) on days with a daily median ODBA < 1, which approached the real nesting locations closely (within 1.6–3.7 m, n = 6). The start of nesting was defined as the first day on which the goose spent > 75% of time within 50 m of the nest, because nest site attendances steeply increased within one day to above this threshold. Nesting duration (number of consecutive days with > 75% nest site attendance) ranged between 3 and 44 days (n = 28), but was 30–34 days in confirmed successful nests (n = 9). The prolonged nesting of 39–44 days (n = 3) suggested incubation on unhatch- able egg(s). Nest losses before hatching time occurred mostly in day 3–10 and 23–29 of nesting, periods with an increased frequency of nest site recesses. As alternative method, allowing for non-simultaneous GPS and accelerome- ter data, we show that nesting days were classified with 98.6% success by two general characteristics of breeding: low body motion (daily median ODBA) and low geographic mobility (daily SD of latitude). Median coordinates on nesting days approached real nest sites closely (within 0.8–3.6 m, n = 6). When considering only geographic mobility (allow- ing for GPS data only) nesting locations were similarly accurate, but some short nesting attempts were undetected and non-breeding tracks misclassified. We show that nesting attempts, as short as 3 days, and nesting success can be detected remotely with good precision using GPS-tracking and accelerometry. Our method may be generalised to other (precocial) bird species with similar incubation behaviour. Keywords: Anser brachyrhynchus, ODBA, Incubation, Nesting duration, Brood, Parental care, Recess Background The tracking of individual breeding attempts allows us to estimate population growth parameters and study individual variation in reproductive decisions which *Correspondence: k.schreven@nioo.knaw.nl Department of Animal Ecology, Netherlands Institute of Ecology (NIOO- influence these parameters. In birds, the timing of breed - KNAW ), Droevendaalsesteeg 10, 6708 PB Wageningen, The Netherlands ing, especially relative to the local onset of spring, has Full list of author information is available at the end of the article © The Author(s) 2021. This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http:// creat iveco mmons. org/ licen ses/ by/4. 0/. The Creative Commons Public Domain Dedication waiver (http:// creat iveco mmons. org/ publi cdoma in/ zero/1. 0/) applies to the data made available in this article, unless otherwise stated in a credit line to the data. Schreven et al. Anim Biotelemetry (2021) 9:25 Page 2 of 13 profound effects on breeding propensity [1], clutch size We focus on the migratory pink-footed goose (Anser and hatching success [2–4], chick growth and survival brachyrhynchus), which breeds in remote arctic regions [5–7] and local recruitment [8 reviewed for geese in 9]. and exhibits similar behaviour before, during and after Generally, birds are found to time their breeding so that the breeding season as many other goose species [23, the peak food demand of the offspring during growth 24] (see Fig. 1). During spring staging, pink-footed geese coincides with the peak in food abundance [10, 11] or forage large parts of the day and move tens of kilome- food quality [12]. tres daily between roosting and foraging sites [25]. They Due to climate warming, the onset of spring is advanc- complete the over-sea migration from Norway and Fin- ing [13] and the highest rate of climate change on Earth land to the breeding grounds in 2 days (range 1–9), arriv- occurs in the Arctic [14]. In response, Arctic-breeding ing on the Svalbard breeding grounds mid–late May [26 migratory birds may adjust their arrival to the breed- and own data]. Just after arrival on the breeding grounds, ing grounds and timing of breeding to keep up with this they may roost for the biggest part of a day (pers. comm. environmental change [6], or may adjust their breeding J. Prop), and then forage actively on different sites before location to a place or habitat with a more favourable phe- nesting [26, 27]. We define nesting here as the combina - nology [15]. tion of egg-laying, nest-building, incubation, and hatch- A remote region does not easily allow for direct obser- ing. Pairs may visit prospective nesting sites several times vations of breeding birds. With modern tracking devices, before settling [28]. Once a nest site has been selected, migratory birds can be tagged in a non-breeding location the pair does not leave the area anymore and starts with and followed without direct observation to their remote egg-laying and nest-building immediately, with most breeding areas such as the Arctic [e.g., 16], while addi- nest-building happening up to 6 days after nest initiation tional sensors such as accelerometers allow for detailed [28]. The start of egg-laying occurs usually in late May to bio-logging to understand the birds’ behaviour [e.g., 17, early June, but ranges from mid-May to mid-June [3, 24, 18]. However, to enable the study of breeding biology in 29, 30]. The clutch is completed with commonly 3–5 eggs such birds, one needs to understand how to recognise a (range 1–8) around day 5–9 after nest initiation, when breeding event from tracking data, and ideally validate incubation sharply increases from 40–60% to 90–100% of this with individuals that were observed directly, which is time [3, 24, 28–30]. Incubation is solely carried out by the challenging in remote areas. female, in a sedentary posture with the head forward or Breeding events have been detected in previous stud- leaning on the back, while the male stands nearby. The ies based on GPS data [e.g., 19, 20, 21], but additional incubation period until hatching is 25–28  days, during sensors, especially an accelerometer, are expected to which females usually leave the nest to feed once per day increase the power and precision of nest(ing) detection. [3, 24, 29, 31, 32]. All chicks in one nest hatch within 24 h, Accelerometry enables us to focus on GPS-fixes, or days, and the family leaves the nest within 36–48  h, or even on which the bird displayed an activity level or body ori- within 12–24 h if disturbed [28, 29 pers. obs. J. Madsen). entation of interest [e.g., 22]. Especially for short nest- During this period, the female may forage with the chicks ing attempts, which might otherwise be indiscernible around the nest, but also go back to the nest to brood from daily roosting or foraging routines based on GPS the chicks [28]. After incubation, females spend much patterns, accelerometry could increase the detection time feeding and little time vigilant [33]. Wing moult in probability. Previous studies on wood stork (Mycteria adults starts 2–3 weeks after hatching, 1 week earlier for americana), lesser kestrel (Falco naumanni), Mediterra- non-breeders and renders birds flightless for 25  days in nean gull (Ichthyaetus melanocephalus) [21] concluded July–August [24, 29]. During moult, geese mainly forage, that nesting attempts shorter than 7–14 days can usually preen and roost [34]. Geese depart from Svalbard to Nor- not be detected based on GPS-tracking data. Also, [19] way around mid-September [35, own data]. and [20] defined nests of GPS-tracked barnacle geese We use the above descriptions of breeding behaviour in (Branta leucopsis) and greater white-fronted geese (Anser this species to define characteristics of nesting behaviour albifrons) as locations from which geese stayed within that can be measured quantitatively by GPS-tracking and 2 km for at least 10 days, during a defined breeding sea - accelerometry, and thereby extract nesting locations and son. However, a more precise method which can detect durations for 49 adult females, tagged in Svalbard and short nesting attempts is important to distinguish a low Finland. We ground-truth this by direct observation of breeding propensity from a high breeding failure. Fur- nesting sites on the Svalbard breeding grounds in sum- thermore, nesting success can be determined, by evaluat- mer and family status on subsequent autumn, wintering, ing if the bird incubates the clutch long enough for it to and spring staging sites. We further assess how nesting hatch successfully. failure covaries with nest attendance. We believe that our S chreven et al. Anim Biotelemetry (2021) 9:25 Page 3 of 13 Fig. 1 A pink-footed goose with a transmitter neckband. Incubating female ‘2M’ (solar-powered GPS-GSM transmitter neckband) and guarding male ‘KK9’ (normal plastic neckband) at their nest site in Endalen, Longyearbyen, Svalbard, 29 June 2020. Photo: Christian Stolz, colour-edited by Stijn Schreven method can also be used in other (precocial) bird species and 7 males. Here, we focus on females, although limited with similar incubation behaviour. data on males suggest the method also works when using males. Of the 49 female geese, we obtained 54 tracks dur- ing the breeding season, over 3  years (2018: 4, 2019: 29, Methods 2020: 21). Catching geese Pink-footed geese were tagged in Svalbard and Finland. In Svalbard, geese were caught during wing moult, in GPS‑tracking and accelerometry family groups, on 30 July 2018 (the lake of Isdammen, We used solar-powered GPS-GSM transmitter neck- Longyearbyen, 78°12′12.7′′N, 15°48′10.3′′E) and 1 August bands, type OrniTrack-N38 (Ornitela UAB, Lithuania) 2018 (the coastal plain of Daudmannsøyra, 78°13′16.6′′N, with a weight of 38  g (c. 1.5% of body mass) and an 13°04′10.8′′E). More details on catching and handling inner diameter of 38  mm (see Fig.  1). Tags were white procedure are given in [34]. The capture and tagging and had a black two-digit individual code that could be caused the birds to increase their amount of time spent read from a distance in the field. The tags recorded a preening, but this effect disappeared within 6–7  days GPS-fix every 10 min (when battery voltage is 75–100%, after tagging [34]. In Finland, geese were caught during 4000–4150  mV), 20  min (50–74%, 3850–3995  mV), spring migration, by canon-net on agricultural fields in 30  min (25–49%, 3760–3845  mV) or 60  min (< 25%). In the novel staging area in Tyrnävä, Oulu (64°49′33.1′′N addition, when the voltage was above 85% (4060  mV), 25°33′25.1′′E, see also [36]) on 28 April 2018, 27 April instead of a single GPS-fix a GPS-burst of 10 fixes was 2019 and 1 May 2019. taken at 1 Hz, to increase accuracy of elevation and speed Geese were sexed by cloacal examination in the field, measurements. which was validated molecularly following [37], using Immediately after each GPS-fix, or GPS-burst, a 2-s blood taken from medial metatarsal vein, primer pair accelerometer measurement burst was taken at a fre- 2550F/2718R and the PCR-programme of [38] and run- quency of 20  Hz. Gravitational acceleration (unit g) ning results on a 2% agarose gel. was measured in three dimensions: along the neck In total, 56 geese received a transmitter in Svalbard (y-axis), and around the neck (x and z-axes, which are (n = 35) and Finland (n = 21), of which 49 were females Schreven et al. Anim Biotelemetry (2021) 9:25 Page 4 of 13 interchangeable due to free rotation of the neckband). trampled in the lining of the nest. Remains from depre- Battery voltage level was recorded with each GPS-fix. dated eggs were recognised by an irregular shape, pres- For analysis, a subset of data was taken in which, ence of bite or peck marks, yolk remains and/or a weak besides the single GPS-fixes, only the first GPS-fix of thin membrane. If no eggshells were present, the nest each GPS-burst was kept, to keep data quantity and qual- was assumed to be depredated [3, 24, 29]. ity homogeneous. Further subsetting was not necessary, These field observations resulted in direct observations as the variation in logging interval on the level of interest of two geese incubating on their nests, and checks of 13 (i.e. the within-individual within-day level) was limited: sites of in total seven geese (six geese with one site each, on average only 7% of days contained multiple different one goose with two sites and one goose with five sites). logging intervals and on these days on average only 17% These sites concerned six confirmed nests (one per goose, of fixes had a different logging interval than the most including the two nests where incubating geese were common logging interval length of that individual on that observed), five roosting/foraging areas (all of one goose), day. Overall, 98.3% of fixes were taken at 10-min inter - and one uncertain site, which was excluded (i.e. first site val, 1.3% at 20-min, 0.4% at 30-min, and 0.1% at intervals of the goose with two sites). Of the six confirmed nests, of 1  h or longer. Overall, voltage levels were on average three had hatched and three were depredated. The uncer - 95.6 ± SD 7.8% (range 0–100%). tain site was attended by the goose for > 75% of time The precision of a GPS-fix was evaluated with two tags within 50 m on both 1–2 June 2020, whereas the goose’s placed on a fixed known coordinate, i.e. a wooden pole second site, a confirmed nest 302  m away, was attended at Adventdalen weather station during 4–9 August 2019. for > 75% of time within 50  m on all 11–13 June 2020. This showed that 24% of GPS-fixes was within 5 m of the This uncertain site might also have concerned a nest, as pole, 47% within 10 m, 74% within 20 m and 96% within the time and distance between the sites corresponds to 50 m (n = 1424 fixes). recent findings about replacement clutches of early-failed Barnacle geese (Branta leucopsis, pers. comm. Jouke Prop, Nordenskiöldskysten, Svalbard). However, replace- Field observations ment clutches are not known for pink-footed geese [24, For ground-truthing, nesting and roosting/forag- 29, 30] but are difficult to observe, as early-failed nests ing locations of geese on their breeding grounds were have only accumulated little down [28] that can be blown observed in Adventdalen, Svalbard, during 20 June–11 by wind. At this uncertain site, only goose droppings July 2020. These locations were identified with a rough were found. procedure: selecting condensed clusters of GPS-fixes We further increased the sample size of confirmed suc - on a map, and taking the average coordinates per clus- cessful geese by seven, based on observations of geese ter. A cluster was defined as a collection of GPS-fixes after nesting on Svalbard and during autumn migration showing that a goose frequently revisited an area with a in Norway, Sweden, Denmark, Netherlands, and Belgium diameter of 20–50 m, for at least 2 days, with relatively entered into the citizen science platform www. geese. org few fixes elsewhere during this period, and occurring [39]. In geese, partners migrate together and with their within the time frame of arrival (mid–late May) until fully fledged brood, enabling an evaluation of breeding the usual hatching period (late June). This procedure success and pair status in autumn and even up to early included all nesting sites as identified by the final proce - spring when juveniles leave their parents [e.g., 40]. The dure below, but also additional roosting/foraging sites. four tracked females in 2018 were seen without chicks In the field, these sites were observed from a distance in autumn. Of the 29 females tracked in 2019, one was (to check presence of a tagged goose, possible partner seen with chicks in summer and/or autumn, 28 without. and chicks), and visited (if no goose was present). In the Of the 21 females tracked in 2020, eight were seen with majority of cases, female geese were not flushed from chicks in summer and/or autumn, and 13 without. Thus, the nest by the observer; only in two cases, the female the total sample of confirmed nesting geese was 13, of left the nest. From a distance it was observed that the which ten were successful and three had failed. female returned to the nest site. During a visit, the sur- We further increased the sample size of roosting sites roundings within 20  m of each site were checked for by including geese roosting just upon arrival at the the presence of a nest, of which the coordinates were breeding grounds, as nesting then is physiologically not then taken with a handheld GPS device (GPSMAP 64 s, ® yet possible [41]. Therefore, if potential ‘nesting’ sites Garmin ). For each nest, the number of eggs and pres- were detected (by the final procedure below) within ence of egg remains in and around the nest was noted. 3  days after arrival, they were regarded as roosting sites Egg remains were evaluated as coming from hatched (n = 4 geese). From the five roosting sites checked in the nests if a thick membrane was present, yolk remains field, described above, only one was detected as potential and predator bite marks were absent, and if they were S chreven et al. Anim Biotelemetry (2021) 9:25 Page 5 of 13 ‘nesting’ site, and therefore the total sample of roosting For our first and main approach, simultaneously sites was five. recorded GPS and ACC data is needed (Fig.  2a). As nesting is characterised by prolonged periods of sitting Deriving nesting attempts still, we expected to measure for breeding geese a time Nesting attempts were derived from GPS tracking and window of consistently low values of body motion. As accelerometry by identifying cut-off values of several proxies for body motion, overall dynamic body accelera- quantifiable characteristics that follow from the breeding tion (ODBA) and vectorial dynamic body acceleration behaviour descriptions in the background. We selected (VeDBA) are in use [42]. Although VeDBA is mathemati- May–July for analysis, i.e. from spring staging and migra- cally better in line with theory of acceleration [42], in our tion until halfway moult and/or chick rearing, and used case, ODBA gave clearer contrasts for a goose within a the tracks of confirmed nesting geese. We developed season, and we therefore used ODBA. We first calcu - two approaches suitable for different data structures (see lated the static acceleration for each of the three dimen- Fig. 2 for a schematic representation of the approaches). sions (x, y, z) within a burst as the average raw measured Fig. 2 Schematic representation of the approaches to extract nesting attempts and success of pink-footed geese based on tracking data. The main approach (a) requires simultaneously recorded GPS and accelerometer (ACC) data, whereas the second approach (b) uses general movement characteristics summarised on a daily level, and can handle non-simultaneous GPS and ACC data, or GPS data only. ODBA overall dynamic body acceleration (see “Methods” Section for calculation). In b, cut-off values were determined using Recursive Partitioning and Regression Trees (see “Methods” section; Fig. 4), and the success of these classifications was determined by comparison with results of approach (a). After identifying possible nests in b, certain nests and successful nests can again be identified with the attendance threshold, period threshold and nesting duration as described in a Schreven et al. Anim Biotelemetry (2021) 9:25 Page 6 of 13 acceleration of each dimension within the burst. We then how well nesting can be predicted by two general char- subtracted for each dimension (x, y, z) the static accelera- acteristics that geese display during nesting: low geo- tion from the raw measured acceleration. This gives the graphic mobility and low body motion, summarised on dynamic acceleration, of which then the absolute value a daily level (see Fig.  2b for a schematic representation). was taken. The absolute values were then summed for all Geographic mobility was measured by the daily stand- 40 points of all three dimensions within a 2-s burst to get ard deviation of latitude of GPS-fixes. Longitude was the ODBA of that burst. We defined an ODBA threshold not analysed, as the daily SD in latitude and daily SD in to identify periods of ‘motionlessness’ based on the direct longitude were strongly correlated (r = 0.92, t = 165.03, observation of geese incubating their clutch (n = 2) and df = 4797, p < 0.0001). We compared daily median other geese that were confirmed to have nested (n = 11, ODBA and daily SD of latitude between different stages see “Field observations” section). As geese attend their of the annual cycle, for each goose based on its GPS and nest almost continuously during nesting, we determined accelerometer data: spring migration (from May 1st until the potential nesting location of each goose by taking the reaching the breeding grounds), pre-nesting (defined median values of the longitude and latitude of a subset as the period between spring migration and nesting), of GPS-fixes. The subset was made by selecting the days nesting (as determined by our first approach above), on which the goose was predominantly motionless (i.e. post-nesting (after nesting until the end of July), and for daily median ODBA below threshold), and then taking non-breeding geese the summering period as one stage from these days all GPS-fixes during which the goose was (i.e. from spring migration until the end of July). Stages motionless (i.e. ODBA below threshold). The standard were compared using Linear mixed models (LMM) using deviation of the coordinates of the selected fixes was used the R-package “lme4” [43] on natural logarithm-trans- to evaluate the precision of the derived potential nest- formed data to approach normality. A random effect of ing site. We assessed nest site attendances by calculating track ID (i.e. each year separately for each individual, n = the distance of each GPS-fix of a goose (also on days on 54) was included and p-values and degrees of freedom which the goose was not mostly motionless) to its poten- were estimated by Satterthwaite’s method (R-package tial nesting location and then reporting the daily amount “lmerTest”) [44]. We then established a procedure to clas- of time that the goose spent within a radius of 5, 10, 20, sify days as either ‘nesting’ or ‘non-nesting’ using Recur- and 50  m of its potential nest site. The radius of 50  m sive Partitioning and Regression Trees (R-package ‘rpart’) gave the clearest contrasts over time and was used in fur- [45]. We split our data randomly into a training set (50%), ther analysis. We then defined an attendance threshold: on which the tree was based, and a testing set (50%), with a daily proportion of time spent within the radius of the which the tree was validated. We tested two different nest that is typical of nesting. We derived this by evalu- classification trees: (1) using only geographic mobility as ating for confirmed nests (n = 13, see “Field observa- a single predictor, or (2) using both geographic mobility tions” section) if there was a steep increase in nest site and body motion as predictors. The comparison of the attendance, which was defined as the start of nesting, and two classifications allowed us to assess the added value of what values characterised attendance before and after the sensor data as opposed to GPS data without accelerome- start. Incubation may be confused with roosting behav- try. Following both classifications, nesting locations were iour (i.e. geese also sitting still, resulting in low ODBA), determined by taking the median coordinates of nesting but roosting is expected to happen in shorter episodes days, and compared ground-truthed nest coordinates and not always on the same location. Therefore, we calcu - (n = 6). After determining possible nesting sites with lated the nesting duration as the number of consecutive this second approach, real nests could be identified by days on which the goose showed above-threshold attend- again using the attendance threshold and period thresh- ance to its potential nesting location. In order to exclude old as explained in the first approach. All statistical tests roosting locations and identify the location as a real nest, were performed in R [46]. the nesting duration should be above a minimum num- ber of days, i.e. the period threshold. To define the period Deriving nesting success threshold, we compared roosting locations (n = 5) and To derive nesting success—binary, i.e. whether any chicks nesting locations (n = 13, see “Field observations” sec- hatched from a clutch or not—we defined a range in nest - tion). If a goose spent fewer days than the period thresh- ing duration typical of successful nests (Fig. 2a). This was old at its potential nesting site, roosting and nesting achieved by comparing confirmed successful geese (n = could not be distinguished. 9) with confirmed unsuccessful geese (n = 3) and breed- Our second and complementary approach does not ing geese with unknown success, i.e. geese that were require that GPS and ACC data were collected simul- breeders according to our first approach, but not seen taneously, and can use GPS only as well. We evaluated with young in autumn (n = 16, see “Field observations” S chreven et al. Anim Biotelemetry (2021) 9:25 Page 7 of 13 section). One successful goose was excluded here typical for nesting (283 out of 377  days, n = 13 nesting because of tag failure from halfway incubation onwards. geese, Figs.  3a, c; 4a), During such time windows, geo- We expected the nesting duration of successful geese to graphic mobility was consistently low as well (Figs.  3b; be 30–38  days, i.e. 5–9  days egg-laying, plus 25–28  days 4a). Our method to extract nesting locations reproduced incubation, plus 0–1  day hatching (see “Background” the field-recorded nesting locations well, i.e. on average section). Any nesting duration outside this range would within 2.7  m (SD = 0.9, range 1.6–3.7  m, n = 6 nests). imply complete hatching failure. Per goose, the SD of latitude in the subset of GPS-fixes To assess whether any nest with unknown nesting suc- from which the nesting location was derived was on cess, but a nesting duration typical of successful nests average 9.7 ± SD 4.9 m (range 5.6–20 m). The subset of could be assumed to be successful, we analysed the geo- one specific goose with a prolonged incubation period graphic mobility and body motion during the post-nest- showed an SD for latitude of 635 m, but the real nest site ing stage. We compared successful breeders, i.e. with was still approximated within 1.7  m. For all identified chicks throughout summer (n = 9), breeders that failed roosting episodes (n = 5 geese), the SD of the subset of during nesting, i.e. without chicks in summer (n = 16) GPS-fixes from which the roosting location was derived and geese with a nesting duration typical of hatched was on average for latitude 90.1 ± SD 153  km (range nests, but with unknown hatching success and without 54.5  m–356  km). The large values were generated by chicks by the time of autumn migration (n = 3). Addi- geese that were motionless for most of the day because tionally, one goose nested successfully but lost its chicks of roosting, but moved between different roost sites, of during summer or autumn. We used LMMs with a ran- which some were several hundred kilometres apart when dom effect of track ID, on natural logarithm-transformed the goose had just arrived in Svalbard. data. For confirmed nesting geese (13 nests), the time a goose spent within 50 m from the nest steeply increased Variation in nest site attendance within one day by on average 70 ± SD 17% (range As we expected to find a range of nesting durations 35–93%), namely from a site attendance of on average (relating to nests that had been depredated, hatched, or 23 ± SD 21% (range 0–65%) to on average 94 ± 7% deserted), we aimed to better understand the moment (range 81–100%, Figs.  3d; 5a). Therefore, the attendance in the nesting cycle at which nests failed. For this, we threshold was set at 75% of the time spent within the evaluated variations in the daily number and duration 50  m radius of the site. The period threshold was set at of nest site recesses, i.e. bouts of time that a goose spent 3 days, as confirmed nesting geese showed nesting dura - beyond 50  m away from its nest. We divided the nest- tions of 3–44 days, i.e. subsequent days with > 75% nest ing cycle before possible hatching into three periods: day site attendance (n = 12 nests), whereas roosting geese 0–10, day 11–19, and day 20–29 after nest initiation. For showed maximally 1 day meeting the attendance thresh- each goose, we selected the period from its first nest site old (n = 5 geese). visit to the day before the end of nesting, and calculated per day the number and average duration of bouts spent General characteristics during breeding: body motion beyond 50 m away from the nest. When a bout spanned and geographic mobility 2 days, we included the bout only on the day on which it During May–July, body motion (daily median ODBA, started. Bout duration was calculated by taking the time log-transformed) of female geese plotted against their difference between the first and last GPS-fix within the geographic mobility (daily SD in latitude, log-trans- bout, plus 10 min. Thereby, bouts with only one GPS-fix formed) showed clear clusters that corresponded to dif- were assumed to have lasted 10 min, as the GPS-logging ferent stages of the annual cycle (Fig.  4). The nesting interval during nesting was nearly always 10 min (median stage was characterised by lower daily median ODBA voltage during pre-nesting and nesting stages 99.0%, than all the other stages combined (Table  1, LMM, t = mean 96.4 ± SD 6.7%, range 33–100). Both the number 86.77, df = 4597, p < 0.0001). Also, the geographic and duration of recesses of all nesting geese were then mobility was lower during nesting than all other stages compared between the three periods in the nesting cycle, combined (Table  1, LMM, t = 52.09, df = 3961, p < using LMMs including a random effect of track ID. 0.0001). Classification success into nesting versus non- nesting days was 96.4% when using only geographic Results mobility as predictor, and 98.6% when using both geo- Extracting nesting locations and nest attendance graphic mobility and body motion as predictors (Fig.  4). The ODBA threshold was defined at 1.0, because the From the 29 breeders determined by the main approach, daily median ODBA was below this value on 75% of 26 were recognised as such by GPS-only (i.e. showing at days during prolonged time windows of motionlessness, Schreven et al. Anim Biotelemetry (2021) 9:25 Page 8 of 13 Fig. 3 Body motion, geographic mobility, and nest attendance of a pink-footed goose during spring and summer. Overall dynamic body acceleration (ODBA) per burst varied largely but showed during nesting a time window of motionlessness (a, c: zoomed in), during which geographic mobility was also notably low (b). GPS-fixes with an ODBA < 1 on days with median ODBA < 1 were selected to extract the nesting site by taking median coordinates (c), from which then nest site attendance was calculated (d). Inferred start and end of nesting are indicated by orange arrows in d, given by a threshold of daily attendance above 75% within 50 m from the nest. This nest was confirmed to have hatched Fig. 4 Body motion and geographic mobility of pink-footed geese during different annual cycle stages (May–July). Per goose, the daily median of the overall dynamic body acceleration (ODBA) was plotted against the daily standard deviation of latitude, both transformed by taking the natural logarithm. Nesting was characterised by days with low body motion and low geographic mobility. Exact cut-off values for the classification of nesting days were determined with Recursive Partitioning and Regression Trees, based on GPS data only (red dotted line) or GPS and accelerometer data combined (green dashed line). a Depicts females that nested (n = 29 tracks), while b depict non-breeding females (n = 25 tracks). The nesting period was defined based on nest attendance patterns (see Fig. 5). For the other periods, see text S chreven et al. Anim Biotelemetry (2021) 9:25 Page 9 of 13 Fig. 5 Nest site attendance and recesses throughout the nesting cycle. Nesting duration, i.e. number of subsequent days with > 75% (dashed horizontal green line) of time spent within 50 m of the nest, varied between 3 and 44 days for all nests, but between 30 and 34 for successful nests (a). Most nest attempts stopped in the first or last 10 days of the incubation cycle (delineated by dashed vertical grey lines). In these periods, geese went more often beyond 50 m from the nest, than during the period in between, while the duration of each of those bouts did not vary significantly (b). Grey shading indicates the start and end of successful nesting. In b, means are depicted ± SE Table 1 Body motion and geographic mobility of pink-footed geese during different annual cycle stages (May–July) Annual cycle stage Female geese included Daily median ODBA Daily SD of latitude (km) Spring migration (from 1 May) Breeders (n = 29) and non-breeders (n = 25) 20.38 ± 9.23 (0.35–54.59) 26.3 ± 59.0 (0.0356–381) Pre-nesting Breeders (n = 29) 30.01 ± 9.73 (0.70–50.57) 6.12 ± 16.7 (0.00857–130) Nesting Breeders (n = 29) 1.08 ± 2.40 (0.49–41.73) 0.0189 ± 0.0434 (0.00278–0.668) Post-nesting (until 31 July) Breeders (n = 29) 17.73 ± 10.80 (0.54–44.41) 0.568 ± 2.67 (0.0145–56.8) Successful, with chicks (n = 9) 15.50 ± 7.24 (1.58–33.95) 0.301 ± 0.223 (0.0156–1.89) Failed, without chicks (n = 16) 19.31 ± 12.02 (0.54–44.41) 0.523 ± 1.78 (0.0145–21.2) Successful, but lost chicks in summer/autumn 22.51 ± 11.19 (0.49–35.53) 3.09 ± 6.41 (0.116–23.3) (n = 1) With unknown hatching success, with 13.94 ± 9.65 (2.95–39.31) 0.323 ± 0.401 (0.0189–2.56) 30–34 days nesting (n = 3) Non-breeding summer (until 31 July) Non-breeders (n = 25) 20.44 ± 12.82 (0.44–49.01) 1.67 ± 7.12 (0.00690–104) For five different annual cycle stages, the daily median overall dynamic body acceleration (ODBA) and daily SD of latitude are summarised here as mean ± SD (range). Nesting was characterised during spring–summer by days with low body motion and low geographic mobility. Post-nesting, breeders with chicks throughout summer showed lower body motion and geographic mobility than breeders without chicks. The nesting period was defined based on nest attendance patterns (see Fig. 1). For the other periods, see text least 3 consecutive nesting days), and 28 by GPS + ACC. detected; GPS + ACC: average 2.2  m, SD 1.0  m, range However, from the 25 non-breeding tracks determined 0.8–3.6 m, n = 6 nests). by the first approach, four were classified as breeding by GPS-only, while GPS + ACC classified them all correctly Deriving nesting success as non-breeding. This resulted in a total classification The nesting durations of all nesting attempts identified success of breeding/non-breeding status for GPS-only with the first approach varied between 3 and 44 days (n = of 87.0%, and for GPS + ACC of 98.1%. The difference 28; Fig.  5a). Of these, nests that were confirmed to have was mainly caused by GPS-only missing short attempts hatched showed a nesting duration of 30–34  days (n = and misjudging some non-breeding tracks. Nevertheless, 9), while nests that were confirmed to have failed showed nesting locations estimated by taking the median coor- nesting durations of 3, 18 and 44  days, respectively. The dinates of classified nesting days were similar based on nesting duration of all geese nesting for shorter than GPS-only or GPS + ACC, and approached the ground- 30 days (including those with unknown success) showed truthed nesting locations within few metres (GPS-only: a bimodal distribution with most nest attempts stopping average 2.5  m, SD 0.8  m, range 1.6–3.6  m, n = 5 nests, after 3–10 days (n = 7) or 23–29 days (n = 5) and only as 1 ground-truthed short nesting attempt was not one after 18  days. Further nesting attempts of unknown success stopped after 32(2 × ), 33, 39 and 41 days. Schreven et al. Anim Biotelemetry (2021) 9:25 Page 10 of 13 During the post-nesting stage, successful breeders of other activities than nesting, e.g., foraging or preen- with chicks had a lower daily body motion level (LMM, ing. This was apparent in our second approach, where t = − 2.301, df = 23.5, p = 0.031; Table 1) and a lower accelerometry increased the likelihood that both short geographic mobility (LMM, for latitude: t = −  2.641, breeding attempts and non-breeding tracks are cor- df = 20.4, p = 0.015; Table 1) than failed breeders with- rectly classified, whereas the goose’s spatial behaviour out chicks. The body motion and geographic mobility per se could be ambiguous (i.e. low geographic mobility of breeders with unknown nesting success but a nesting for a short time). This is useful when one aims to study duration of 30–34  days, seemed more similar to that of the timing of breeding, or to distinguish a low breeding breeders with chicks than failed breeders without chicks propensity from a high breeding failure. We argue that (Table  1), but was not significantly different from either including accelerometer data, if available, can increase group (with chicks: LMM, body motion, t = −  0.279, the power of nest detection. df = 9.7, p = 0.79; LMM, for latitude, t = 0.932, df = By evaluating the proportion of time spent at such 1020, p = 0.35; without chicks: LMM, body motion, t = locations, our method could also distinguish roosting −  1.432, df = 17.5, p = 0.17; LMM, for latitude, t = areas from nesting attempts that lasted at least three − 1.076, df = 15.6, p = 0.30). days. A goose may roost for longer periods, but appar- ently rarely on exactly the same spot during subsequent days. Roosting behaviour may differ for other species, Variation in nest site attendance e.g., raptors, which may have specific roosting trees [47] In the middle of the nesting cycle (day 11–19), geese or gulls, which consistently roost on the same breakwater made 0.87 ± SE 0.18 nest site recesses beyond 50  m [48]. For such species, the body orientation as measured from the nest per day, which tended to be lower, but not by an accelerometer would still be useful to detect nest- significantly, than the 1.1 ± SE 0.14 recesses per day ing sites, e.g., straight up during roosting, versus horizon- in the first period (day 0–10, df = 719, t = 1.882, p = tal during nesting in raptors [49]. Our study provides a 0.060) and was significantly lower than the 1.6 ± SE 0.15 reliable method that enables the study of avian breeding recesses per day in the last period (day 20–29, df = 709, in remote environments, which is often logistically chal- t = 4.537, p < 0.001; Fig. 5b). The duration of these bouts lenging with classic observations. was not different between the middle period (18.5 ± SE The nesting period is unique in its virtual absence of 4.7  min) and the first period (19.5 ± SE 0.73  min, df = movement. Although this would suggest low energy 400, t = 1.342, p = 0.18) or the last period (18.7 ± SE expenditure [42], nesting is energetically a demand- 0.76 min, df = 399, t = 0.222, p = 0.82; Fig. 5b). ing time for geese, resulting in large body mass loss as egg production is costly and foraging opportunities are Discussion traded off against incubation activity and nest defence We have presented a method to use GPS-tracking and [50]. The higher daily activity levels we found in the pre- accelerometry to remotely identify pink-footed goose nesting stage may reflect active foraging for large parts nesting attempts and their locations, the start and end of of the day [27], whereas movement patterns in the post- each nesting attempt, and thus also an indication of nest- nesting stage depended on family status. Geese with ing success given by the duration of nest site attendance. chicks had lower body motion and lower geographic We also showed that > 98% of nesting days could be clas- mobility than geese without chicks, probably the result of sified correctly by two main movement characteristics regularly brooding the chicks in the field and flying less. of the breeding phase of the annual cycle (i.e. low geo- The large geographic mobility that distinguishes spring graphic mobility and low body motion) and that includ- migration results both from daily movements between ing accelerometry especially helps to detect short nesting roost sites and foraging sites within a stop-over, as well as attempts correctly. directional movements between stop-overs [25]. Our main approach acquires a high level of precision The nesting duration of successful nests concentrated, and can detect nest attempts as short as 3 days, which is as expected, at 30–34 days, whereas unsuccessful nests not easily achievable with methods based on only GPS showed both shorter and longer nesting durations. without additional sensor data. For example, [21] sug- For further studies, one may assume that any goose gest for three bird species that nesting attempts below with unknown nesting success but a nesting dura- 7–14  days can usually not be detected based on GPS- tion of 30–34  days is successful, as post-nesting body tracking data, and [19] and [20] use for two goose species motion and geographic mobility of such geese seemed a minimum duration of 10  days. The power of includ - more similar to those with hatched nests than failed ing sensor data in the analysis lies in the fact that one nests. Nesting durations shorter than 30 days are likely can specifically select GPS-fixes (or days) with a certain caused by nest predation or abandonment [1]. Also body motion of interest and exclude those that are part S chreven et al. Anim Biotelemetry (2021) 9:25 Page 11 of 13 human disturbance might indirectly lead to nest loss Combining GPS with accelerometer data can increase the by predation [51]. Nest losses before normal hatching power to detect short nesting attempts and determine the time occurred in our study predominantly in the first timing, location and success of breeding. Studying avian and last ten days of a normal nesting cycle. This pat - breeding in the changing remote arctic environment is of tern may be explained by the slightly higher recorded great interest and will be subject of future research, facili- frequency of nest site recesses in these periods, com- tated by the method presented in this paper. pared to the period in between. Nest recesses, to for- age or drink, expose the nest to predators [3]. Because our GPS-fixes include GPS error, we cannot report Conclusions exactly how far a goose was situated from its nest dur- Nesting attempts of Arctic-breeding pink-footed geese ing a bout. Also, recesses shorter than 10  min can be (as short as 3  days) and their locations can be derived missed due to the logging interval. Still, our analysis is remotely and accurately from GPS-tracking and accel- expected to give a reliable qualitative view on (far and erometry, while nesting success is indicated by the long) nest site recesses. Moreover, the same pattern of nesting duration. We predict that this method can be nest attendance over the incubation cycle was reported generalised to other (precocial) bird species with simi- based on direct observations and showed that females lar incubation behaviour. spend more time foraging in the first week and last days of nesting than in the period in between [28]. In Abbreviations contrast, dark-bellied Brent geese (Branta bernicla ber- ACC : Accelerometry; ODBA: Overall dynamic body acceleration; VeDBA: Vecto- nicla) and greater snow geese (Chen caerulescens atlan- rial dynamic body acceleration; SD: Standard deviation. tica) were found to make fewer and shorter recesses Acknowledgements in the last days before hatching [50, 52]. Differences in We are grateful to people who helped catch geese in Svalbard (Christian incubation constancy between species may result from Sonne, Peter de Vries, Koen Nolet, Ove Martin Gundersen, John Frikke, Cornelia Jaspers, Kevin Kuhlmann Clausen, Cecilia Sandström, Dies Snoodijk and the the interplay of remaining body stores (thus, the need crew of RV “Ulla Rinman”) and in Finland (Jorma Pessa, Tuomas Väyrynen, Esko for foraging trips), foraging opportunities close to the Pasanen, Aija Lehikoinen, Christanne de Vries, Jorma Siira, Matti Tolvanen, nest, and the local risk of nest predation (thus, the need Antti Piironen, Christian Sonne, Lars Haugaard, Jens Peter Hounisen, Michael Schmidt, Niels-Erik Jørgensen, Maël Charbonneaux, Kalle Hiekkanen). Further to stay on the nest). Future studies using our method we thank observers, especially Aija Lehikoinen, Jørgen Peter Kjeldsen, Ole could evaluate how nest attendance and nesting dura- Amstrup, Fred Cottaar, Eckhart Kuijken, Christine Verscheure, Leo Schilperoord tion vary with environmental factors and may change and Mirjam Schilperoord-Huisman, Ove M. Gundersen, Kevin K. Clausen, Kent Larsson and Lennart Eriksson, for observing tagged geese in the field, and with arctic warming. Martijn van der Sluijs and Christa Mateman for the molecular sexing. Fieldwork Nesting durations longer than 34  days are longer than in Norway was kindly facilitated by Skogn Folkehøgskole, Skogn. We thank expected, even for large clutches and may indicate that Stijn Schreven for editing figure 1, and two anonymous referees for their constructive comments. the goose was incubating unhatchable eggs. Pink-footed geese may continue to incubate an unhatched egg for Authors’ contributions 3 more days after all other eggs have already hatched, Conceptualisation, KS, BAN; F, KS, CS, JM, BAN; formal analysis, KS; writing— original draft KS; writing—review and editing, KS, CS, JM, BAN; supervision, JM., before they leave with the brood [28]. However, when the BAN. All authors read and approved the final manuscript. whole clutch is unhatchable, many bird species extend their incubation duration by at least 50% [53]. A nest with Funding Our work was supported by grants from the Netherlands Polar Programme solely unhatchable eggs is expected to be rare, but can, of the Dutch Research Council to BAN (ALWPP.2016.024), from the Svalbard for example result from predation of already-hatched Environmental Protection Fund to JM (No. 17/88) and from the Academy Ecol- chicks—in our study one nest was incubated for 44 days, ogy Fund of the Royal Netherlands Academy of Arts and Sciences to KHTS. The funding body had no role in the design of the study and collection, analysis, but contained only one egg on day 43. Egg hatching fail- and interpretation of data or in writing the manuscript. ure can be caused by environmental pollutants result- ing in embryonal death or deformities [e.g., 54, 55]. To Availability of data and materials The datasets generated and/or analysed during the current study are not detect such phenomena, the study of individual breeding publicly available yet, pending publication of another article, but are available attempts by GPS-tracking and accelerometry, as facili- from the corresponding author on reasonable request. tated by this study, is a valuable addition to the extensive monitoring of productivity on the population level [36]. Declarations Our study also highlights the importance of characteris- Ethics approval and consent to participate ing successful nesting attempts not solely by a minimum Permits to catch and tag geese in Svalbard were granted by the Norwegian nesting period, but by a defined range in nesting periods. Food Safety authority (Mattilsynet) (to Aarhus University, 17/210528), by the Technological advances in tracking devices make that Governor of Svalbard (17/01420-4) and Longyearbyen Lokalstyre (2018/347- 5-X70), in Finland by Etelä-Suomen aluehallintovirasto (to Aarhus University, GPS data are increasingly accompanied by sensor data. Schreven et al. Anim Biotelemetry (2021) 9:25 Page 12 of 13 ESAVI/1924/2018 and ESAVI/1880/2018) and Varsinais-Suomen elinkeino-, 16. Davidson SC, Bohrer G, Gurarie E, LaPoint S, Mahoney PJ, Boelman NT, liikenne- ja ympäritsökeskus (to Jorma Pessa, VARELY/551/2018). et al. Ecological insights from three decades of animal movement track- ing across a changing Arctic. Science. 2020;370(6517):712–5. Consent for publication 17. Dokter AM, Fokkema W, Bekker SK, Bouten W, Ebbinge BS, Müskens G, Not applicable. Olff H, van der Jeugd HP, Nolet BA. Body stores persist as fitness correlate in a long-distance migrant released from food constraints. Behav Ecol. Competing interests 2018;29(5):1157–66. https:// doi. org/ 10. 1093/ beheco/ ary080. The authors declare that they have no competing interests. 18. Nuijten RJM, Gerrits T, Shamoun-Baranes J, Nolet BA. Less is more: on- board lossy compression of accelerometer data increases biologging Author details capacity. J Anim Ecol. 2020;89(1):237–47. https:// doi. org/ 10. 1111/ 1365- Department of Animal Ecology, Netherlands Institute of Ecology (NIOO-2656. 13164. KNAW ), Droevendaalsesteeg 10, 6708 PB Wageningen, The Netherlands. 19. de Boer R, Bauer S, van der Jeugd HP, Ens BJ, Griffin L, Cabot D, Exo KM, Norwegian Institute for Nature Research (NINA), 7485 Trondheim, Norway. Nolet BA, Kölzsch A. A comparison of spring migration between three Department of Biology, Norwegian University of Science and Technol- populations of Barnacle Geese Branta leucopsis using GPS satellite trans- ogy (NTNU), 7491 Trondheim, Norway. Department of Bioscience, Aarhus mitters. Limosa. 2014;87:99–106. University, 8410 Rønde, Denmark. Department of Theoretical and Computa- 20. Kölzsch A, Müskens GJDM, Szinai P, Moonen S, Glazov P, Kruckenberg H, tional Ecology, Institute for Biodiversity and Ecosystem Dynamics, University Wikelski M, Nolet BA. Flyway connectivity and exchange primarily driven of Amsterdam, Science Park 904, 1098 XH Amsterdam, The Netherlands. by moult migration in geese. Mov Ecol. 2019;7(1):1–11. 21. Picardi S, Smith BJ, Boone ME, Frederick PC, Cecere JG, Rubolini D, Serra Received: 2 February 2021 Accepted: 21 June 2021 L, Pirrello S, Borkhataria RR, Basille M. Analysis of movement recursions to detect reproductive events and estimate their fate in central place forag- ers. Mov Ecol. 2020;8(1):1–14. 22. Wilson RP, White CR, Quintana F, Halsey LG, Liebsch N, Martin GR, Butler PJ. Moving towards acceleration for estimates of activity-specific meta- bolic rate in free-living animals: the case of the cormorant. J Anim Ecol. References 2006;75(5):1081–90. 1. Anderson HB, Madsen J, Fuglei E, Jensen GH, Woodin SJ, van der Wal 23. Scott DA, Rose PM. Atlas of Anatidae Populations in Africa and Western R. The dilemma of where to nest: influence of spring snow cover, food Eurasia. Wageningen, The Netherlands: Wetlands International; 1996. proximity and predator abundance on reproductive success of an arctic- 24. Cramp S, Simmons KEL. Handbook of the birds of Europe, the Middle breeding migratory herbivore is dependent on nesting habitat choice. East, and North Africa: the birds of the Western Palearctic, vol. I. Oxford: Polar Biol. 2014;38:153–62. Oxford University Press; 1977. 2. Perrins CM. The timing of birds’ breeding seasons. Ibis. 1970;112:242–55. 25. Chudzińska ME, Nabe-Nielsen J, Nolet BA, Madsen J. Foraging behaviour 3. Madsen J, Tamstorf M, Klaassen M, Eide N, Glahder C, Rigét F, Nyegaard and fuel accumulation of capital breeders during spring migration as H, Cottaar F. Eec ff ts of snow cover on timing and success of reproduc- derived from a combination of satellite- and ground-based observations. tion in high-Arctic Pink-footed Geese Anser brachyrhynchus. Polar Biol. J Avian Biol. 2016;47:563–74. 2007;30:1363–72. https:// doi. org/ 10. 1007/ s00300- 007- 0296-9. 26. Glahder CM, Fox AD, Hübner CE, Madsen J, Tombre IM. Pre-nesting site 4. Ross MV, Alisauskas RT, Douglas DC, Kellett DK. Decadal declines in avian use of satellite transmitter tagged Svalbard Pink-footed Geese Anser herbivore reproduction: density-dependent nutrition and phenological brachyrhynchus. Ardea. 2006;94(3):679–90. mismatch in the Arctic. Ecology. 2017;98:1869–83. 27. Fox AD, Francis IS, Bergersen E. Diet and habitat use of Svalbard Pink- 5. Brook RW, Leafloor JO, Abraham KF, Douglas DC. Density dependence footed Geese Anser brachyrhynchus during arrival and pre-breeding and phenological mismatch: consequences for growth and survival of periods in Adventdalen. Ardea. 2006;94(3):691–9. sub-arctic nesting Canada Geese. Avian Conserv Ecol. 2015;10:1. 28. Inglis IR. The breeding behaviour of the Pink-footed Goose: behavioural 6. Lameris TK, van der Jeugd HP, Eichhorn G, Dokter AM, Bouten W, Boom correlates of nesting success. Anim Behav. 1977;25:747–64. MP, Litvin KE, Ens BJ, Nolet BA. Arctic geese tune migration to a warm- 29. Bauer KM, von GlutzBlotzheim UN. Handbuch der Vögel Mitteleuropas, ing climate but still suffer from a phenological mismatch. Curr Biol. vol. 2. Frankfurt: Akademische Verlagsgesellschaft; 1968. 2018;28:2467–73. 30. Bezzel E. Kompendium der Vögel Mitteleuropas: Nonpasseriformes— 7. Ross MV, Alisauskas RT, Douglas DC, Kellett DK, Drake KL. Density- Nichtsingvögel. Wiesbaden: Aula-verlag; 1985. dependent and phenological mismatch effects on growth and survival in 31. Witherby HF, Jourdain FCR, Ticehurst NF, Tucker BW. The handbook of lesser snow and Ross’s goslings. J Avian Biol. 2018;49:e01748. British Birds. Watford Edinburgh: JJ Gray and Sun Printers Ltd; 1944. 8. Reed TE, Grøtan V, Jenouvrier S, Sæther BE, Visser ME. Population growth 32. Del Hoyo J, Elliott A, Sargatal J. Handbook of the Birds of the World, vol. 1. in a wild bird is buffered against phenological mismatch. Science. Barcelona: Lynx Edicions; 1992. 2013;340(6131):488–91. 33. Lazarus J, Inglis IR. The breeding behaviour of the Pink-footed Goose: 9. Nolet BA, Schreven KHT, Boom MP, Lameris TK. Contrasting effects of the parental care and vigilant behaviour during the fledging period. Behav- onset of spring on reproductive success of Arctic-nesting geese. Auk. iour. 1978;65(1/2):62–88. 2020;137(1):ukz063. 34. Clausen KK, Schreven KHT, Madsen J. Eec ff ts of capture and marking 10. Lack D. Ecological adaptations for breeding in birds. London: Methuen; on behaviour of moulting Pink-footed Geese Anser brachyrhynchus on Svalbard. Wildfowl. 2020;70:13–29. 11. Visser ME, van Noordwijk AJ, Tinbergen JM, Lessells CM. Warmer springs 35. Jensen GH, Tombre IM, Madsen J. Environmental factors affecting lead to mis-timed reproduction in Great Tits (Parus major). Proc R Soc numbers of Pink-footed Geese Anser brachyrhynchus utilising an autumn Lond B. 1998;265:1867–70. stopover site. Wildl Biol. 2016;22(5):183–93. 12. Doiron M, Gauthier G, Levesque E. Trophic mismatch and its effects 36. Madsen J, Jensen GH, Cottaar F, Amstrup O, Bak M, Bakken J, Balsby TTJ, on the growth of young in an Arctic herbivore. Glob Change Biol. Kjær Christensen T, Clausen KK, Frikke J, Gundersen OM, Kjeldsen JP, Koffi- 2015;21(12):4364–76. https:// doi. org/ 10. 1111/ gcb. 13057. jberg K, Kuijken E, Månsson J, Nicolaisen PI, Haaning Nielsen H, Nilsson L, 13. Høye TT, Post E, Meltofte H, Schmidt NM, Forchhammer MC. Rapid Reinsborg T, Ødegaard PI, Pessa J, Shimmings P, Tombre I, Verscheure C. advancement of spring in the High Arctic. Curr Biol. 2007;17(12):R449–51. Svalbard Pink-footed Goose Population status report 2017–2018. Bonn, 14. Cohen J, Screen JA, Furtado JC, Barlow M, Whittleston D, Coumou D, Germany: AEWA EGMP Technical Report; 2018. Francis J, Dethloff K, Entekhabi D, Overland J, Jones J. Recent Arctic ampli- 37. Fridolfsson AK, Ellegren H. A simple and universal method for molecular fication and extreme mid-latitude weather. Nat Geosci. 2014;7(9):627. sexing of non-ratite birds. J Avian Biol. 1999;30:116–21. 15. Alzate Valejo A. From local adaptation to range sizes: ecological and 38. Griffiths R, Double MC, Orr K, Dawson RJ. A DNA test to sex most birds. evolutionary consequences of dispersal. PhD thesis, Groningen, The Mol Ecol. 1998;7(8):1071–5. Netherlands: University of Groningen; 2018. S chreven et al. Anim Biotelemetry (2021) 9:25 Page 13 of 13 39. Ebbinge BS, Buij R, de Vries L, Moonen S, van Randen Y, Müskens G, van 48. van Donk SC. A gull’s portrait: On individual foraging strategies and their der Jeugd H, Koffijberg K, Voslamber B, Roosenschoon O, Kramer J. The fitness consequences. PhD Thesis, The Netherlands: Vrije Universiteit website geese.org, an interactive database to report marked waterfowl. Amsterdam; 2020. Goose Bulletin. 2020;25:11–8. 49. Newton I. The sparrowhawk. Calton: A&C Black; 1986. 40. Gupte PR, Koffijberg K, Müskens GJDM, Wikelski M, Kölzsch A. Family size 50. Spaans B, van’t Hoff CA, van der Veer W, Ebbinge BS. The significance of dynamics in wintering geese. J Ornithol. 2019;160:363–75. female body stores for egg laying and incubation in Dark-bellied Brent 41. Klaassen M, Hahn S, Korthals H, Madsen J. Eggs brought in from afar: Geese Branta bernicla. Ardea. 2007;95(1):3–15. Svalbard-breeding Pink-footed Geese can fly their eggs across the Bar - 51. Madsen J, Tombre I, Eide NE. Eec ff ts of disturbance on geese in ents Sea. J Avian Biol. 2017;48:173–9. https:// doi. org/ 10. 1111/ jav. 01364. Svalbard: implications for regulating increasing tourism. Polar Res. 42. Wilson RP, Börger L, Holton MD, Scantlebury DM, Gómez-Laich A, Quin- 2009;28(3):376–89. tana F, Rosell F, Graf PM, Williams H, Gunner R, Hopkins L, Marks N, Geraldi 52. Poussart C, Gauthier G, Larochelle J. Incubation behaviour of NR, Duarte CM, Scott R, Strano MS, Robotka H, Eizaguirre C, Fahlman A, greater snow geese in relation to weather conditions. Can J Zool. Shepard ELC. Estimates for energy expenditure in free-living animals 2001;79(4):671–8. using acceleration proxies: a reappraisal. J Anim Ecol. 2020;89(1):161–72. 53. Skutch AF. The constancy of incubation. Wilson Bull. 1962;74(2):115–52. 43. Bates D, Mächler M, Bolker BM, Walker SC. Fitting linear mixed-effects 54. Skorupa JP, Ohlendorf HM. Contaminants in drainage water and avian risk models using lme4. J Stat Softw. 2015;67(1):1–48. https:// doi. org/ 10. thresholds. In: Dinar A, Zilberman D, editors. The economics and manage- 18637/ jss. v067. i01. ment of water and drainage in agriculture. New York: Springer; 1991. 44. Kuznetsova A, Brockhoff PB, Christensen RHB. lmerTest package: tests in 55. van Oosten HH, van den Burg AB. Dioxines: een niet-onderkend prob- linear mixed effects models. J Stat Softw. 2017;82(13):1–26. https:// doi. leem voor zangvogels? The Netherlands: Report Stichting Bargerveen; org/ 10. 18637/ jss. v082. i13. 2014. 45. Therneau T, Atkinson B, Ripley B. Recursive Partitioning and Regression Trees. 2019. https:// cran.r- proje ct. org/ packa ge= rpart. Accessed 15 May Publisher’s Note Springer Nature remains neutral with regard to jurisdictional claims in pub- 46. R Core Team. R: A language and environment for statistical computing. lished maps and institutional affiliations. R Foundation for Statistical Computing, Vienna, Austria. 2020. https:// www.R- proje ct. org/. Accessed 01 Jan 2020. 47. Deng WH, Wei G, Guang-Mei Z. Nest and roost habitat characteris- tics of the Grey-faced Buzzard in northeastern China. J Raptor Res. 2003;37(3):228–35. Re Read ady y to to submit y submit your our re researc search h ? Choose BMC and benefit fr ? Choose BMC and benefit from om: : fast, convenient online submission thorough peer review by experienced researchers in your field rapid publication on acceptance support for research data, including large and complex data types • gold Open Access which fosters wider collaboration and increased citations maximum visibility for your research: over 100M website views per year At BMC, research is always in progress. Learn more biomedcentral.com/submissions

Journal

Animal BiotelemetrySpringer Journals

Published: Jul 13, 2021

Keywords: Anser brachyrhynchus; ODBA; Incubation; Nesting duration; Brood; Parental care; Recess

References