Non-breeding behaviour in the Brown Skua (Stercorarius antarcticus lonnbergi): insights from modelling moulting patterns and stable isotope analyses

ABSTRACT Long-term changes in the life history and behaviour of seabirds during the non-breeding season can reflect shifts in environmental conditions. However, long-term marine studies are scarce, particularly on southern hemisphere seabirds. Here, we used moult scores from 86 Brown Skuas (Stercorarius antarcticus lonnbergi), a large predatory seabird breeding on the Chatham Islands, Aotearoa/New Zealand to model both the timing and duration of primary feather moult. In addition, we analysed stable isotope values (δ13C and δ15N) from 62 modern (2014–16) and ten museum tail feathers. These data provide insights into the non-breeding behaviour of Brown Skua. Interestingly, our results show that the primary feather moult occurred prior to birds departing the colony, starting on average on 2 January ± 5 days (SE). The average start of primary feather moult occurred five days prior to the end of breeding (7 January ± 10 days (SD)) and 42 days before the birds departed the colony (13 February ± 11 days (SD)). The average duration of primary feather moult was 189 ± 14 days (SE). Importantly, low δ13C values in four females suggested that tail feather moult might also occur while skuas are at the colony. There was no difference in tail feather δ13C and δ15N values between any pairwise comparison of modern and museum years. However, values of δ15N from tail feathers sampled in 2014 were different from those sampled in 2015 and 2016. This large annual variation in δ15N values from tail feathers over such a short period makes long-term comparisons difficult to interpret, particularly between years with low sample sizes. While the stable isotope analyses of tail feathers are informative, we recommend future studies of skuas sample the primary coverts rather than tail feathers.


Introduction
Seabirds can be used as indicators of marine ecosystem health (Piatt et al. 2007). Although seabirds must visit land to breed and raise offspring, they spend most of their time at sea, away from the breeding colonies (Brooke 2004). Moult, the replacement of old feathers, is a major part of the annual cycle in birds and is energetically demanding. This usually occurs during the non-breeding period (Hedenström 2006;van Bemmelen et al. 2018), defined as the time from either breeding failure or fledging to egglaying, and includes the wintering period, which is the time from departure until birds return to the colony.
Knowledge of moulting patterns is fundamental for interpreting feather stable isotope data. This is because once grown feathers are metabolically inert and, therefore, their isotopic signatures reflect the environmental conditions during moult (Cherel et al. 2000). While moulting patterns have been described for some seabirds, there is little information for many other species. In addition, there are now a number of useful tools for accurately modelling moult trajectories (Erni et al. 2013).
Stable isotope ratios of carbon ( 13 C/ 12 C ratios, expressed as δ 13 C) from marine phytoplankton vary as a function of latitude, sea surface temperature, and concentrations of aqueous carbon dioxide (CO 2 ) . Moreover, δ 13 C values are influenced by the composition and productivity of phytoplankton communities (Fry and Sherr 1984;) and vary in different marine environments (i.e. benthic versus pelagic, inshore versus offshore), but change little as trophic levels increase (0.4-0.8 ‰) (DeNiro and Epstein 1978;Vander Zanden and Rasmussen 2001;Post 2002). Since baseline δ 13 C values are reflected in the tissues of organisms that feed at high trophic levels, analyses of seabird tissues can be used to infer their foraging areas (Hobson et al. 1994;Jaeger et al. 2010). The isotopic composition of feathers reflects the diet consumed at the time of growth (Cherel et al. 2000). Long-term changes in feather δ 13 C values can thus indicate either shifts in environmental baseline isotope values or changes in foraging areas. For example, Hilton et al. (2006) suggested that lower δ 13 C values in the feathers of Southern Rockhopper Penguins (Eudyptes chrysocome) reflected a recent decline in oceanic primary productivity (and thus increased carbon isotope fractionation) and a reduced carrying capacity of the ecosystem. Furthermore, long-term decreases in feather δ 13 C values can indicate shifts in the non-breeding distributions of seabirds in response to environmental change. For example, this is thought to be the case for both Antarctic Prions (Pachyptila desolata; Grecian et al. 2016) and Slender-billed Prions (P. belcheri; Quillfeldt et al. 2010;Cherel et al. 2014).
Stable isotope ratios of nitrogen ( 15 N/ 14 N ratios, expressed as δ 15 N) from feathers are useful indicators of the trophic position of birds or their respective prey (Hobson et al. 1994). The heavier nitrogen isotope ( 15 N) is preferentially retained in the tissue of a consumer compared to the lighter isotope ( 14 N), producing an enrichment in 15 N of approximately 3.4 ‰ with each trophic level (Mizutani et al. 1992;Post 2002). Longterm trends in δ 15 N values can therefore indicate changes in the trophic position of a consumer (Blight et al. 2015) if corrected for nitrogen baseline values. In seabirds, long-term dietary changes inferred from stable isotope analyses have been reported for Northern Fulmars (Fulmarus glacialis; Thompson et al. 1995), Glaucous-winged Gulls (Larus glaucescens; Blight et al. 2015), Flesh-footed Shearwaters (Ardenna carneipes; Bond and Lavers 2014) and Spotted Shags (Phalacrocorax punctatus; Rayner et al. 2021). Therefore, stable isotope analyses of feathers from museum specimens represent a unique opportunity to investigate past environmental changes. However, such samples are often rare and only a limited number of studies have investigated long-term patterns in the nonbreeding distributions and diets of southern hemisphere seabirds (Hilton et al. 2006;Quillfeldt et al. 2010;Rayner et al. 2011Rayner et al. , 2021Cherel et al. 2014;Grecian et al. 2016).
Brown Skuas (Stercorarius antarcticus lonnbergi) are large predatory seabirds (1804 ± 140 g (SD)) that have a circumpolar breeding distribution and winter between the Antarctic Polar Front and the Subtropical Convergence (Furness 1987;Phillips et al. 2007;Carneiro et al. 2016;Krietsch et al. 2017;Delord et al. 2018;Schultz et al. 2018). Overall, moulting patterns in Brown Skuas are poorly documented and this is complicated by differences between immature and adult birds as well as variation among individuals and populations (Votier et al. 2015). On the Kerguelen Islands, immature Brown Skuas complete primary feather moult between November and February. In contrast, adult Brown Skuas start primary feather moult in February, completing moult before arriving back at the breeding grounds in August/September (Jiguet 2007). This timing is consistent with observations made on Anvers Island, Antarctica, where moult of primary feathers begins in early February followed by nape and head feathers (Higgins and Davies 1996). Furthermore, Graña Grilli and Cherel (2017) found that moult overlapped with the breeding period in some adult Brown Skuas from King George Island. To our knowledge, there is no published information about the timing and duration of tail feather moult in Brown Skuas.
The breeding ecology and behaviour of Brown Skuas on the Chatham Islands, Aotearoa/New Zealand has been studied since the 1970s (reviewed in Young 1999;Schultz et al. 2021). Breeding occurs between October when eggs are laid and January when chicks fledge (Hemmings 1989;Young 1999). This population is characterised by the presence of breeding pairs, a high incidence of communally breeding groups (i.e. three or more birds at a nest) and non-breeders (Young 1978;Hemmings 1994). Of note, the numbers of communal groups in the Chatham Island population have declined markedly during the past three decades (Schultz 2019). This change in the breeding ecology might be linked to non-breeding behaviour, which could reflect changes in the spatial distribution or diet of skuas.
However, until recently, little was known about the non-breeding distribution and diet of Chatham Island Brown Skuas. While earlier reports suggested that Brown Skuas remain resident year-round (Hemmings 1990), more recent findings have shown that, on average, they leave their territories in February for a relatively short period (average wintering period of 146 days ± 20 days (SD)), although some individuals remain in the vicinity of the Chatham Islands (Schultz et al. 2018). Here, we report the patterns of primary feather moult in Chatham Island Brown Skuas and show how this relates to their annual cycle. We also analyse long-term stable isotope data (δ 13 C and δ 15 N) from modern (2014-16;Schultz et al. 2018Schultz et al. ) and museum (1871Schultz et al. -1996 tail feathers to investigate possible changes in the non-breeding distribution and diet of Brown Skuas.

Life history, fieldwork and sample collection
Fieldwork was conducted on South East Island (44°21ʹS, 176°10ʹW) in the Chatham Island archipelago, New Zealand ( Figure S1). Skua nests were monitored during six breeding seasons . Egg-laying dates were recorded in the 1985 breeding season and hatching dates in the 1984, 1986 and 2014-16 breeding seasons. Fledging dates (the date when chicks were capable of sustained flight) were calculated based on hatching dates and the mean age at fledging. The latter was empirically determined based on fledging dates of 12 known-age chicks from eight territories.
Sixty-two adult Brown Skuas were captured at their breeding territories during the incubation and chickrearing periods (October to November, 2014-16) using either a hand net or a remote clap trap. All captured skuas were weighed, marked with numerical metal bands for identification and wings were inspected for primary feather moult. Blood samples (~200 μl) were taken from all individuals for molecular sexing (Griffiths et al. 1998). The tip (~20 mm) of the outermost right tail feather was collected from each individual during the 2014-16 breeding seasons (see Schultz et al. 2018 for details). In addition, one tail feather tip was sampled from each of ten adult Brown Skua study skins that were collected on the Chatham Islands between 1871 and 1996 and are held in the Museum of New Zealand/Te Papa Tongarewa, Canterbury Museum and Auckland War Memorial Museum (Table 1 and  Table S1). In the case of two museum specimens, no tail feathers were available. Instead, the tip of the underwing covert of the eighth primary feather was used. The scientific names used in this article are as detailed in Gill et al. (2022).

Scoring and modelling of moult from primary feathers
The progress of primary feather moult in 79 Brown Skuas was scored based on high-resolution photographs taken throughout the Chatham Island archipelago from 2013 to 2021 by the authors and other photographers (see acknowledgement section for details). Furthermore, images of seven Brown Skuas from other New Zealand colonies were included from online sources (ebird.org reference numbers: ML229448791, ML65576961, ML89910481, ML215090721, ML186689841; inaturalist. org reference numbers: 34142943, 15246685). The breeding status (active or failed) and sex of Brown Skuas scored for primary feather moult was not known. Immature and adult birds were distinguished based on plumage characteristics and leg colouration (Jiguet 2007;Howell 2008;Newell et al. 2013;Votier et al. 2015). Birds were identified using the location and time stamp of the images, by individual wing-moult patterns, and in some instances using leg-bands. Each of the ten primary feathers was assigned a value between 0 and 5 where 0 = old; 1 = growing within pin or missing (these feathers were never visible and the status had to be assumed); 2 = pin with tuft to 33% length; 3 = 33-67% length, 4 = 67-99% length, and 5 = new and fully grown (Ashmole 1962 In each photograph, the most visible wing was used for moult scoring (see Figure S2 for examples). Moult scores were then calculated for each individual as the sum of the ten primary feather scores. A moult index, Proportion of Feather Mass Grown (PFMG), was calculated based on the moult scores (Summers et al. 1980). This was achieved using the mean relative mass for each of the ten primary feathers as determined for northern hemisphere Great Skuas (S. skua; van Bemmelen et al. 2018). The moult index accounts for differences in the size of the inner and outer primaries and, therefore, increases linearly with time. This makes it better suited for modelling moult characteristics than raw moult scores (Summers et al. 1980).

Stable isotope analyses from tail feathers
Tail feathers were soaked and cleaned in 70% ethanol, rinsed three times with distilled water, dried in an oven at 60°C and homogenised by cutting into small pieces (<0.5 mm long). Subsequently, 0.75-0.85 mg of homogenised tail feather was transferred into tin capsules for carbon and nitrogen stable isotope analysis at the National Institute for Water and Atmospheric Research in Wellington, New Zealand. Isotopic analysis was performed on a Flash 2000 elemental analyser (Thermo-Fischer Scientific, Bremen, Germany) that was connected to a DELTA V Plus (Thermo-Fischer Scientific, Bremen, Germany) fully automated continuous flow isotope ratio mass spectrometer. Normalisation of isotopic values was performed against a range of National Institute and Technology (NIST) standards (see supplementary material in Schultz et al. (2018) for details). Repeat analysis of NIST standards produced data accurate to within 0.1 ‰ for δ 15 N and 0.2 ‰ for δ 13 C and a precision of better than 0.1 ‰ for N and 0.2 ‰ for C. The increase of anthropogenically derived CO 2 due to burning of fossil fuels has resulted in a reduced abundance of 14 C and 13 C relative to 12 C in the atmosphere leading to decreasing oceanic baseline δ 13 C values, a phenomenon known as the 'Suess effect' (Suess 1955;Keeling 1979). Therefore, all tail feather δ 13 C values were adjusted relative to the youngest sample in the dataset following the method used by Grecian et al. (2016). Furthermore, the combustion of fossil fuels has increased the concentration of aqueous CO 2 in seawater, resulting in an increase in the degree of isotopic fractionation and, consequently, a depletion of 13 C in phytoplankton (see . However, values of δ 13 C were not corrected for concentration or isotopic changes in aqueous CO 2 as the effects are within the margins of the analytical error (0.16 ‰ in 150 years) associated with the stable isotope method (Hilton et al. 2006;Jaeger and Cherel 2011).

Statistical analyses
All statistical tests were conducted using the R computing environment version 3.5.1 (R Core Development Team 2018). Moult index values were used to estimate the mean start date and duration of primary feather moult using the function moult (UZ type 2 model) implemented in the R package 'moult' (Underhill and Zucchini 1988;Erni et al. 2013). Furthermore, the moult model was used to estimate the proportion of Brown Skuas that had started primary feather moult at different times in the annual cycle. Immature individuals were excluded from the analysis.
For modern sampling years only, linear models were fitted to test for differences in δ 13 C and δ 15 N values in relation to the fixed factors sex, sampling year and their interaction. A non-parametric Kruskal-Wallis test and Dunn's test for multiple comparisons were used to examine differences in δ 13 C and δ 15 N values between all possible pairings of the modern (2014-16) and museum  sampling years (α = 0.05). P-values were corrected for multiple comparisons using the false discovery rate (Benjamini and Hochberg 1995).
To compare the isotopic niche width between males and females sampled in modern years (2014-16), the Bayesian Standard Ellipse Area (SEA B ; containing c. 40% of the data) was calculated. This was achieved using a Markov Chain Monte Carlo simulation (chain length = 300,000; burn-in = 200,000) implemented in the R package 'SIBER' (Jackson et al. 2011). The SEA B allows differences between groups to be tested by comparing each pair of posterior draws. The proportion of posterior draws that are either smaller or larger in one group compared to the other is a direct proxy for the probability. The Standard Ellipse Area corrected for small sample sizes (SEA C ) was also calculated for visualisation purposes.

Scoring and modelling of moult from primary feathers
Primary feather moult occurred sequentially, starting from the innermost primary (P1) and progressing outwards. Up to six primaries (P1 to P6) were shed before skuas departed the breeding colony. The estimated mean start date of primary feather moult was 2 January ± 5 days (SE), while the estimated duration was 189 ± 14 days (SE) (Figure 1). According to the moult model, 59% of birds had started primary feather moult by the mean fledging date (7 January ± 10 days (SD)). The model further predicted that 98% of birds had started primary feather moult by the mean date of departure from the colony (13 February ± 11 days (SD))  and that 43% of birds had finished primary feather moult by the mean date of arrival at the colony (7 July ± 19 days (SD); see Schultz et al. 2018). In two instances, moult of the central tail feathers was observed ( Figure S2(b,d)).

Discussion
Moult is an important process in the annual cycle of birds, as individuals need to replace worn feathers to maintain the functionality of their plumage. This process is energetically costly and the timing of moult in relation to the breeding period is key to survival and reproductive success (Hedenström 2006). Furthermore, knowledge of moulting patterns, and particularly timing, is fundamental to accurate interpretation of feather stable isotope data.

Scoring and modelling of moult from primary feathers
We provide the first quantitative analysis of the timing and duration of primary feather moult in Brown Skuas, although some qualitative analyses have been reported (Jiguet 2007;Howell 2008). Although the breeding status of Brown Skuas scored for primary feather moult in this study was unknown, the model indicated that 98% of the birds started primary feather moult before the mean departure date on 13 February. The model also indicated that 59% of birds had started primary feather moult by the mean fledging date on 7 January. This indicates that some birds could have started moult while still breeding. This finding is consistent with studies on Brown Skuas from King George Island (Graña Grilli and Cherel 2017) and the Kerguelen Islands (Jiguet 2007). In contrast, Phillips et al. (2007) did not report any moult in the Bird Island Brown Skua population, but importantly, this study was conducted during the incubation and early to mid chick-rearing periods.
Our estimate of primary feather moult duration in Chatham Island Brown Skuas (body weight 1560-2140 g) was 189 days. This is longer than the estimated moult duration of 150 days in South Polar Skuas (S. maccormicki, body weight 1120-1550 g) (Newell et al. 2013) and the estimated 147 days in northern hemisphere Great Skuas (body weight 1180-1650 g). However, the duration of primary feather moult in Brown Skuas was substantially longer than estimates for the three smaller northern hemisphere species: 122 days in Parasitic Jaegers (S. parasiticus, body weight 335-470 g); 118 days in Pomarine Jaegers (S. pomarinus, body weight 542-917 g) and 82 days in Long-tailed Jaegers (S. longicaudus, body weight 266-336 g) (Furness 1987;van Bemmelen et al. 2018). Given that moult duration is correlated with body size (Rohwer et al. 2009), the longer moult duration in the larger Brown Skua is expected.

Stable isotope analyses from tail feathers
Carbon and nitrogen isotope values from feathers are often used to interpret the geographic distribution and diet of seabirds during the non-breeding period. Stable isotope signatures are incorporated into feather tissue as feathers grow and remain unchanged once fully formed (Cherel et al. 2000). There is a strong latitudinal gradient in baseline δ 13 C values in the Southern Ocean with lower values occurring at higher latitudes (Jaeger et al. 2010;Magozzi et al. 2017;St John Glew et al. 2021). Although other factors are known to affect δ 13 C values (e.g. carbon source, ocean productivity and phytoplankton community structure), at high latitudes sea-surface temperature is the strongest determinant . Our stable isotope results from the analyses of modern tail feathers suggest that Brown Skuas wintered over subtropical, mixed subtropical-subantarctic, and shelf waters (also see Schultz et al. 2018). This finding is consistent with results from recent geolocation tracking of 27 individuals during the wintering period (also see Schultz et al. 2018 for further details).
Terrestrial diets are generally characterised by low δ 13 C values when compared to marine diets (Hobson 1987). Of the 62 modern tail feathers analysed, four female individuals showed low δ 13 C values indicative of a terrestrial diet (Figure 2). Brown Skuas from the Chatham Islands show a short wintering period of 146 ± 20 days (SD) with females returning slightly earlier than males (Schultz et al. 2018). Assuming that tail feather moult in Brown Skuas starts with the central feathers as reported for Great Skuas (Cramp and Simmons 1983;Schreven and Hammer 2020), the low δ 13 C values in the four females suggests that these birds returned early, moulting the outermost tail feather at the colony. During the breeding season, the diet of female Chatham Island Brown Skuas consists of a high proportion of sheep carrion (~47%), while males feed almost exclusively on marine prey (~94%) (Schultz et al. 2021). Alternatively, the low δ 13 C values could indicate moult in Antarctic waters, which are characterised by relatively low baseline δ 13 C values (Weimerskirch et al. 2015;Graña Grilli and Cherel 2017). However, this seems unlikely given that none of the 27 tracked Brown Skuas migrated south of the Antarctic Polar Front (Schultz et al. 2018).
Our comparison of modern and museum tail feather δ 13 C values suggests that the wintering latitudinal range of Chatham Island Brown Skuas has not changed. However, this suggestion needs to be treated with caution as the number of museum samples was low and the annual and individual variation in stable isotope ratios was high. Despite this caveat our δ 13 C results are consistent with a recent study on Broad-billed Prions (P. vittata) breeding in the Chatham Island archipelago (Grecian et al. 2016), where no changes in feather δ 13 C values were noted over an 87-year period. In contrast, other seabird studies have shown significant changes in feather δ 13 C values over time. For example, feather δ 13 C values from Antarctic Prions breeding on the Auckland Islands, New Zealand have declined over a 60-year period , likely indicating a shift in their wintering areas towards higher latitudes (Grecian et al. 2016). Similarly, comparisons of δ 13 C values in modern and museum feathers of Slender-billed Prions from the Falkland and Kerguelen Islands suggest a recent poleward shift in their wintering distributions, likely as a consequence of temperature-mediated changes in the distribution of prey (Quillfeldt et al. 2010;Cherel et al. 2014). Furthermore, δ 13 C values in feathers of Australian Pied Cormorants (P. varius), Spotted Shags and Little Penguins (Eudyptula minor) from New Zealand colonies have declined over a 141year period (1878-2019), likely indicating a shift in foraging locations from inshore to offshore environments (Rayner et al. 2021).
Of interest was both a reduced variance and a significantly higher mean tail feather δ 15 N value in 2014 when compared to 2015 and 2016. This difference could be due to the 2014 individuals feeding in areas with higher baseline δ 15 N values (e.g. shelf waters), at higher trophic levels or due to annual differences in the timing of tail feather moult (Fromant et al. 2020). Furthermore, the SEA B of males was smaller than that of females, suggesting some differences in the non-breeding behaviour between the sexes, which is consistent with geolocation tracking data during the wintering period (Schultz et al. 2018).
Finally, our results underline the importance of a detailed understanding of species-specific moulting patterns when inferring both non-breeding and wintering behaviour from stable isotope data. In our study, moult was quantified based on primary feathers while stable isotope data were derived from tail feathers. Ideally, moult and stable isotope analyses should be based on the same type of feather. Because the moult sequence is better known and easier to observe, we recommend using primary feathers or the corresponding coverts for stable isotope analyses in Brown Skuas. Furthermore, these types of feathers may reduce the annual and individual variation in stable isotope values observed here in tail feathers. Given that some inner primaries are moulted while still at the colony, we recommend using one of the outer primary feathers or coverts (ideally P8). We suggest sampling the primary coverts, which we believe moult simultaneously with the primary feathers (author's observation). The coverts are less important for flight and therefore sampling is likely to have less impact on the individual. While we provide the first quantitative analysis of moult in Brown Skuas, we suggest that additional research is needed on other southern hemisphere skuas to better understand variation within and among populations and species.