Rangifer Tarandus Platyrhynchus) Females Subject/Study Track Ecology and Evolutionary Biology Level Month and Year Number of Pages Master’S Thesis 12.10.2020 43P

Life History Effects in the Behavior of Svalbard Reindeer (Rangifer tarandus

platyrhynchus) Females

Master’s thesis Laura Puikkonen Faculty of Biological and Environmental Sciences Ecology and Evolutionary Biology University of Helsinki October 2020

Faculty Degree Programme Faculty of Biological and Environmental Sciences Master’s programme in ecology and evolutionary biology Author Laura Puikkonen Title Life history effects in the behavior of Svalbard reindeer (Rangifer tarandus platyrhynchus) females Subject/Study track Ecology and evolutionary biology Level Month and year Number of pages Master’s thesis 12.10.2020 43p. + references and supplements Abstract

Individuals of long-lived animal species can improve their reproductive success through experience. While individual’s resources available for survival and reproduction decrease toward the end of its lifespan through senescence, terminal investment hypothesis predicts the less likely old individuals reproduce again the more they invest in their current offspring. Experience gained through a long lifespan might have an important role in changing behavior to optimize the use of resources and compensate the effects of senescence. In addition, behavioral plasticity allows animals to respond changes in their habitat within much shorter timespan than on an evolutionary timescale.

Svalbard reindeer (Rangifer tarandus platyrhynchus) is a wild subspecies of reindeer. It is only found in Svalbard, a remote archipelago in the Arctic with extreme weather conditions rapidly changing due to climate change. It has been isolated at least 5000 years and adapted to a barren habitat with nearly no hunting, predation or harassment of flying insects. The objective of this study is to investigate the effect of age and a calf at heel in Svalbard reindeer females’ maternal, vigilant, and social behavior and time budget in the light of life history theory and its senescence and terminal investment hypotheses.

I carried out the field work for the study in two periods in summer in Semmeldalen valley and the south-western part of Reindalen valley on the island of Spitsbergen, Svalbard. I collected behavioral data on marked individuals by instant scan sampling and focal watch methods, wrote observations down manually and later fed them into computer. In addition, I have used birth year data collected by the long-term monitoring program by the Norwegian Polar Institute.

I used generalized linear mixed models to analyze the effects of age and calf at heel to the behavior of females. The main results include that young dams maintained shorter distance to their calf in July than in August, and old females were less vigilant. Younger dams and older females without calves were in smaller groups than older dams and younger females without calves. In addition, females with calves spend proportionately less time lying down than females without calves. Dams maintained a longer distance to the nearest neighbor than females without calves. Older dams spend proportionately more time feeding and in groups in August than younger dams.

These results show that the age and calf at heel do play a role in the behavior of Svalbard reindeer females and the effect varies over the course of the short Arctic summer. Experience may make older females more effective mothers by optimize the use of resources for example from vigilance to feeding in a predator-free environment. On the other hand, senescence may affect the amount of energy females can spend on their calves, potentially influencing their survival.

Keywords life history theory, reindeer, age, senescence, terminal investment, maternal behavior, vigilance, time budget Supervisor or supervisors Emma Vitikainen Where deposited HELDA - Digital Repository of the University of Helsinki

Additional information

Field work was conducted in cooperation with University Center in Svalbard (UNIS).

Tiedekunta Koulutusohjelma Bio- ja ympäristötieteellinen tiedekunta Ekologian ja evoluutiobiologian maisteriohjelma Tekijä Laura Puikkonen Työn nimi Elämänhistorian vaikutus Huippuvuorten peuranaaraiden (Rangifer tarandus platyrhynchus) käyttäytymiseen Oppiaine/Opintosuunta Ekologia ja evoluutiobiologia Työn laji Aika Sivumäärä Pro gradu -tutkielma 12.10.2020 43 s. + lähdeluettelo ja liitteet Tiivistelmä

Pitkäikäisten eläinlajien yksilöt voivat parantaa lisääntymismenestystään kokemuksen avulla. Samalla, kun yksilön resurssit selviytymiseen ja lisääntymiseen laskevat senesenssin vuoksi, terminaalisen panostuksen hypoteesi ennustaa vanhan yksilön käyttävän sitä enemmän resursseja jälkikasvunsa hyväksi, mitä epätodennäköisemmin kyseinen yksilö lisääntyy tulevaisuudessa. Pitkän elämänkaaren varrella kerätty kokemus saattaa olla tärkeä elementti käyttäytymisen muutoksessa, jonka avulla yksilö pystyy optimoimaan resurssiensa hyödyntämisen ja kompensoimaan senesenssin vaikutuksia. Sen lisäksi, käyttäytymisen muovautuvuus mahdollistaa eläimen sopeutumisen elinympäristön muutoksiin paljon lyhyemmällä aikavälillä kuin evolutiivinen kehitys.

Huippuvuorten peura (Rangifer tarandus platyrhynchus) on peuran eli poron villi alalaji. Sitä tavataan ainoastaan Huippuvuorilla, joka on syrjäinen saaristo arktisella alueella, ja jonka äärimmäiset sääolot muuttuvat nopeasti ilmastonmuutoksen seurauksena. Alalaji on ollut eristyksissä vähintään 5000 vuoden ajan ja sopeutunut karuun elinympäristöön, sekä metsästyksen, saalistajien ja lentävien hyönteisten aiheuttaman häirinnän puuttumiseen. Tämän tutkielman tavoite on tutkia iän ja vasan vaikutusta Huippuvuorten peuranaaraiden maternaaliseen ja sosiaaliseen käyttäytymiseen, vigilanssiin ja aikabudjettiin elämänhistoriateorian sekä senesenssi ja terminaalinen panostus -hypoteesien puitteissa.

Tein tutkielmaan kuuluvat kenttätyöt kahtena ajanjaksona kesällä Semmeldalenin laakossa ja Reindalenin laakson lounaisosissa Spitsbergenin saarella Huippuvuorilla. Havainnoin merkittyjen yksilöiden käyttäytymistä instant scan sampling ja focal watch - metodeilla, kirjasin havaintoni ylös käsin, ja siirsin ne myöhemmin tietokoneelle. Lisäksi hyödynsin Norwegian Polar Institute - instituutin pitkäaikaisen seurantaohjelman keräämiä tietoja peurojen syntymävuosista.

Analysoin iän ja alle vieroitusikäisen vasan vaikutusta peuranaaraiden käyttäytymiseen yleistettyjen lineaaristen sekamallien avulla. Tuloksien perusteella nuoret emät ovat lähempänä vasojaan heinä- kuin elokuussa, ja vanhemmat peuranaaraat ovat vähemmän valppaita. Nuoremmat emät ja vanhemmat naaraat ilman vasoja olivat pienemmissä ryhmissä kuin vanhemmat emät ja nuoremmat naaraat. Emien etäisyys lähimpään naapuriin oli keskimäärin pidempi, kuin vasattomien naaraiden. Lisäksi naaraat ilman vasaa viettivät suhteessa enemmän aikaa maaten kuin emät. Vanhemmat emät viettivät suhteessa enemmän aikaa syöden sekä ryhmissä elokuussa, kuin nuoremmat emät.

Tulokset näyttävät, että ikä ja alle vieroitusikäinen vasa vaikuttavat Huippuvuorten peuranaaraiden käyttäytymiseen, ja vaikutus vaihtelee kesän edetessä. Kokemus saattaa tehdä vanhemmista naaraista tehokkaampia emiä auttamalla niitä optimoimaan resurssien käyttöä esimerkiksi valppaudesta syömiseen saalistajista vapaassa ympäristössä. Toisaalta senesenssi saattaa vähentää vasojen hoitoon käytettävissä olevan energian määrään, potentiaalisesti vaikuttaen niiden selviytymiseen.

Avainsanat elämänkaariteoria, peura, ikä, senesenssi, terminaalinen panostus, maternaalinen käyttäytyminen, valppaus, aikabudjetti Ohjaaja tai ohjaajat Emma Vitikainen Säilytyspaikka HELDA - Helsingin yliopiston digitaalinen arkisto / HELDA Muita tietoja

Kenttätyö tehtiin University Center in Svalbard (UNIS) -yliopistokeskuksen avustuksella.

“The differential allocation of energy to either reproduction or survival represents a major conflict with important implications for patterns of life history.” - Loison & Strand (2005)

4

Table of contents

1 Introduction ...... 6 1.1 Life history theory: traits and trade-offs ...... 6 1.2 Senescence and terminal investment in ungulates ...... 7 1.3 Objectives, research questions and hypotheses...... 10 2 Materials and methods ...... 12 2.1 Svalbard reindeer as a study species ...... 12 2.2 Study area ...... 14 2.3 Studied individuals ...... 15 2.4 Observation period ...... 16 2.5 Behavioral observations ...... 17 2.5.1 Instantaneous scan sampling ...... 18 2.5.2 Focal watches ...... 21 2.5.3 Other behavioral observations ...... 22 2.6 Statistical analyses ...... 22 3 Results ...... 25 3.1 The control group and effect of observation period ...... 25 3.2 The effect of age and calf at heel on the behavior of Svalbard reindeer females ...... 26 3.2.1 Maternal behavior ...... 26 3.2.2 Vigilant behavior ...... 28 3.2.3 Social behavior...... 28 3.2.4 Time budget ...... 31 4 Discussion ...... 33 4.1 Effect of age ...... 34 4.1.1 Age and time of the summer affect maternal behavior ...... 34 4.1.2 Experience from predator-free environment tends to decrease vigilance ...... 36 4.1.3 Average group size increases with age in dams but decreases in females without calves...... 38 4.1.4 Effects of senescence are not compensated by the female time budget ...... 39 4.2 Effect of calf at heel ...... 40 4.2.1 Dams decrease their vigilance towards the end of the summer ...... 40 4.2.2 Other reindeer are closer to females without, than females with calves ...... 41 4.2.3 Foraging time of dams increases towards the autumn ...... 43 4.2 Limitations of the study ...... 45 5 Conclusion ...... 47 6 Acknowledgements ...... 48 7 References...... 49 8 Supplementary material ...... 54

5

1 Introduction

1.1 Life history theory: traits and trade-offs

The life history theory is a widely adopted framework in evolutionary ecology focusing on analyzing the evolution and interactions of so-called life history traits (Stearns, 1992). Life history traits include individual qualities such as size at birth, pattern of growth, age and size at maturity, number, size and sex of offspring, reproductive effort, survival rate and lifespan (Stearns, 1992). According to the life history theory, these traits determine an individual’s survival and reproduction rates, being the main components of an individual’s fitness, which is the basis of adaptation by natural selection (Gadgil & Bossert, 1970; McGraw & Caswell, 1996; Stearns, 1992). Life histories of organisms are balancing between extrinsic factors affecting survival and reproduction and intrinsic constraints limiting how life history traits evolve (Stearns, 2000). Trade-offs between different life history traits are the main type of intrinsic constraints (Roff, 1993; Stearns, 1992). A trade-off is defined as a situation where an increase in at least one life history trait is linked to a decrease in at least one other life history trait, and it can occur on two mutually non-exclusive levels: genetic level or the level of resource allocation (physical level) (Flatt & Heyland, 2011; Roff, 1993). The major life history trade-offs are related to the cost of current reproduction to an individual's survival and future reproduction (Crête & Huot, 1993; Stearns, 1989).

In long living iteroparous mammals, probabilities of survival and reproduction vary during the lifespan of an individual, partly due to extrinsic factors and partly due to the trade-offs between these traits. Senescence and terminal investment hypotheses are two non-mutually exclusive hypotheses, which can help us to understand changes in reproductive success during an individual’s lifespan (Weladji et al., 2010). Senescence means the progressing deterioration of vitality which happens as an individual gets older (Curtsinger, 2001). Parental investment is any resource allocation made by a parent increasing the survival of an individual offspring while decreasing the parent’s ability to invest in other offspring (Evans, 1990; Trivers, 1974), whereas parental effort combines the resources and care provided for the offspring regardless of the cost to other offspring (Evans, 1990). Reproductive success is defined as the number of surviving offspring an individual has per breeding event or lifetime, including the reproductive success of individual’s offspring (Howard, 1979). The senescence hypothesis predicts that due to age-related deterioration of cellular and physiological functions, a parent has less resources to be allocated to reproduction with increasing age (Kirkwood & Austad, 2000; Mysterud, Yoccoz, Stenseth, & Langvatn, 2001). Senescence can involve lower survival (Lee et al., 2015), a decline in reproductive success and increased sensitivity to environmental changes

6

(Gaillard, Festa-Bianchet, Yoccoz, Loison, & Toigo, 2000), and it can happen in different pace between the sexes of the same species (Reimers, Holmengen, & Mysterud, 2005). On the other hand, terminal investment hypothesis predicts that as the expected number of future offspring decreases, a parent’s investment in reproduction should increase with age (Evans, 1990; Gadgil & Bossert, 1970; Pianka & Parker, 1975; Trivers, 1974). Terminal investment can be seen in both the use of physiological resources and/or behavior. Generally, longevity has been found to increase fitness and lifetime reproductive success in iteroparous mammals and birds that breed multiple times during their lifespan (Berube, Festa-Bianchet, & Jorgenson, 1999; Kjellander, Gaillard, Hewison, & Liberg, 2004). At the presence of senescence, this requires the reproductive effort to increase with age in respect to age-related decline in reproductive value (Clutton-Brock, 1984), which on the other hand indicates support for terminal investment hypothesis.

1.2 Senescence and terminal investment in ungulates

Ungulates are an iteroparous mammalian group with relatively long lifespans and large body sizes (Clutton-Brock, 1984; McCullough, 1999). Life histories and reproductive success have been studied in multiple ungulate species (Clutton-Brock, 1984; Douhard et al., 2019; Hejcmanová et al., 2010). Reindeer (Rangifer tarandus) is one of the most widely studied species likely because of its economic and cultural importance (Paine, 1988; Vors & Boyce, 2009). Ungulate life histories and trade-offs are in many ways similar between several species, but there are no clear overarching interspecific patterns meaning that comparing results from different species is possible but should be done with caution (Ralls, Lundrigan, & Kranz, 1987). Studies on wild reindeer populations show that besides life history related factors, reproductive success is affected by the population density (Petersson & Danell, 1993; Skogland, 1990), and female fecundity by parasites (Albon et al., 2002; Stien et al., 2002) and winter precipitation (Adams, 2005; Albon et al., 2002). External factors emphasize the consequences of an individual’s differential investments on its survival and reproduction, but trade-offs between these fitness components remain the most prevalent constraints for an individual’s lifetime reproductive success. In reindeer, the trade-off between survival and reproduction shows in the decrease of body fat reserves during lactation (Allaye Chan-Mcleod, White, & Russell, 1999; Crête & Huot, 1993) and body mass decrease when females rear a calf (Reimers et al., 2005; Rönnegård, Forslund, & Danell, 2002). These costs seem to be most prevalent after winters with heavy snow conditions (Adams, 2005) and might increase with age (Weladji, Mysterud, Holand, & Lenvik, 2002). Greater likelihood of calving when previous calf has died before one month of age suggests a cost from rearing a calf also for the future reproduction (Kojola & Eloranta, 1989). No difference in the cost or care between male and female ungulate calves has been found, and this may indicate that a generally high level of

7 maternal cost does not allow biases between sexes (Douhard et al., 2019; Kojola, 1993; Weladji & Holand, 2003b). The costs for overall survival and future reproduction have not always been possible to point out, but this is likely related to supplementary feeding some of the experimental herds receive during winter (Kojola, 1993; Weladji et al., 2008).

Studies on reindeer (Figure 1) show that females in better body condition often give birth to bigger calves (Adamczewski, 1987; Loison & Strand, 2005; Petersson & Danell, 1993; Rönnegård et al., 2002; Skogland, 1990; Taillon, Brodeur, Festa-Bianchet, & Côté, 2012; Weladji et al., 2010). While the female body condition has a major effect on calf condition, age is one of the main traits defining the reproductive success of an individual. Female mass increases with age to a certain point which is defined as the prime age of the animal, before yielding to the effects of Figure 1 Svalbard reindeer female in the front and male on senescence and starting to decrease (Berube et al., the left. Sexes can be told apart by the shape of their heads 1999; Loison & Strand, 2005; Petersson & and antlers. © Laura Puikkonen 2019 Danell, 1993; Reimers et al., 2005; Weladji et al., 2010). Besides body mass, also calf weight has been similarly used to define the prime age (Loison & Strand, 2005; Rönnegård et al., 2002). Timing of the peak body mass and through that the prime age years in ungulates vary between and within species. In reindeer, five years of age has been described as the start of deteriorating body condition of dams in food limited wild populations and the turning point of the positive effect of age on calves’ autumn weight in semi-domestic herds (Lenvik, Bø, & Fjellheim, 1988; Skogland, 1984). Rönnegård et al. (2002) found female mass to increase up to 7-8 years of age and Loison and Strand (2005) defined prime age to be between 3 to 9 years based on multiple earlier studies. Besides physical implications, semi-domestic reindeer female’s rank in the social hierarchy of the herd also tends to increase up to 6-8 years old (Espmark, 1964; Holand et al., 2004). On the other hand, Weladji et al. (2010; 2008) has used 11.5-12 years as the upper limit and 5.5 years as the lower bound for defining prime age females.

Studies on reindeer show that the start of age-related decrease of the female body condition is often related to increasing tooth wear, the effect being most prevalent in food-limited populations (Kojola, Helle, Huhta, & Niva, 1998; Skogland, 1988; Veiberg et al., 2007). The effect of maternal condition

8 on calf is linked to the age and experience of the mother, and the type of linkage and the magnitude of the effect of maternal weight on the calf weight varies between age groups (Côté & Festa-Bianchet, 2001; Lenvik et al., 1988; Rönnegård et al., 2002; Weladji et al., 2010; Weladji et al., 2002). Overall, young ungulate mothers tend to have lower reproductive success and give birth to smaller calves than prime-aged individuals of the same species (Adams & Dale, 1998; Adams, 2005; Bjørkvoll et al., 2016; Festa-Bianchet, 1988; Lenvik et al., 1988; Ropstad, 2000; Rönnegård et al., 2002; Weladji et al., 2006; Weladji & Holand, 2003b). Younger mothers’ fecundity and calf survival seem also more vulnerable to changes in environmental conditions (Adams, 2005; Festa-Bianchet, 1988). Many studies show that fecundity of females as well as body weight of calves at birth and later in the summer decline after the prime age of an individual (Bjørkvoll et al., 2016; Ropstad, 2000; Rönnegård et al., 2002; Verme, 1989; Weladji et al., 2010; Weladji et al., 2002). Old females seem to produce generally lighter offspring especially in food-limited ungulate populations (Clutton-Brock, Price, Albon, & Jewell, 1992; Clutton-Brock, Guinness, & Albon, 1982) and the decrease of productivity accelerates in the late phase of senescence (Adams & Dale, 1998; Weladji et al., 2010). Consequences of reproductive senescence are visible also in multiple ungulate males (Taillon et al., 2012).

Although there is strong evidence of senescence in the way that females are able to allocate resources to their offspring during pregnancy, terminal investment during the rearing period can partly compensate for these negative effects (Festa-Bianchet, 1988; Green, 1990; Weladji et al., 2006; Weladji & Holand, 2003b; Weladji et al., 2010; Weladji et al., 2008). Weladji et al. (2010) showed that the autumn mass of calves born to over 11-year-old reindeer mothers increases with the mother’s age. The mass gain has been shown to be the highest in calves born to old (>11 years) reindeer mothers also in other studies (Weladji et al., 2006; Weladji & Holand, 2003b; Weladji et al., 2010; Weladji et al., 2008). Furthermore, offspring born to older animals seem to have a higher possibility to survive until the autumn (Festa-Bianchet, 1988; Weladji et al., 2010). In white-tailed deer (Odocoileus virginianus), the body mass of newborn fawns is negatively affected by the age of an old mother, but there is no age-related variation in the autumn mass weight, which is influenced by the litter type instead (Verme, 1989). Also studies conflicting with Weladji et al. (2006; 2010) have shown that reindeer calf autumn weight increases with the age of the mother only up to five years with no significant effect of age on the autumn weight after that (Lenvik et al., 1988; Ropstad, 2000). The lack of age-related variation in autumn mass could indicate that if older females give birth to lighter calves (Verme, 1989), they are able to compensate for it during the summer.

Senescence and terminal investment have been well studied in ungulates from a physiological point of view, but the studies on behavioral aspects of these phenomena and their differences between age

9 groups have been marginal. Weladji et al. (2006) found evidence that semi-domestic reindeer improve their fitness through longevity by several simultaneous processes: longer lifespan gives individuals time to breed several times, long-lived individuals generally are of better quality and they accumulate experience in rearing their offspring. Experience may improve the reproductive success of an individual for example through knowledge of the resource locations and quality, advanced ability to rear calves and skills to defend newborns (Côté & Festa-Bianchet, 2001; Mirza & Provenza, 1992; Ortega-Reyes & Provenza, 1993; Weladji et al., 2006). The age is also related to a higher social rank which can bring access to more food (Appleby, 1982; Espmark, 1964; Holand et al., 2004; Kojola & Helle, 1996; Weladji et al., 2006). There are few studies done in other long-living animals than ungulates, but they suggest the same. The age has been shown to affect the behavior of free-ranging rhesus macaques (Macaca mulatta) so that older mothers with lower body mass are less active and spend more time in contact with their young, although their infants still have lower body masses and survival rates than offspring of younger mothers (Hoffman, Higham, Mas-Rivera, Ayala, & Maestripieri, 2010). On the other hand, the infant survival of female Indian gray langur monkeys (Presbytis entellus) seems to increase with age, possibly because older mothers are better at caring for their young (Dolhinow, McKenna, & Laws, 1979). In mice (Mus musculus) which have a much shorter lifespan, the mother’s age has been shown not to influence litter characteristics at birth nor weaning (Girard et al., 2002). These findings in multiple species suggest that experience gained through a long lifespan might have important implications to the maternal behavior and rearing abilities of long-lived iteroparous mammals, affecting an individual’s lifetime reproductive success (Clutton-Brock, 1984).

1.3 Objectives, research questions and hypotheses

This study has two major objectives. First, I investigate the effect of age and having a calf at heel in Svalbard reindeer (Rangifer tarandus platyrhynchus) females’ maternal, vigilant, and social behavior and time budget in the light of life history theory and its senescence and terminal investment hypotheses. Second, as Svalbard reindeer’s habitat and ecology are distinctly different from the reindeer on mainland Norway and other continents, it is important to document its behavior and provide baseline data of behavior for further studies. The baseline information provided here can be used in future studies analyzing the effect of changing climate and increasing human activity on Svalbard reindeer behavior. Based on these objectives, I formulated two research questions and the following a priori hypotheses based on the existing literature on ungulate behavior:

10

1. Does the age of Svalbard reindeer female affect its maternal, vigilant, and social behavior, and time budget?

Based on the terminal investment hypothesis (H1), I predict that the age of a female affects its behavior through experience so that (H1a) older mothers keep closer to their calves to protect them (Green, 1990) and (H1b) are more often involved in the same activity as their calf, for example because of its importance to the development of foraging skills (Kojola, 1990; Mirza & Provenza, 1992; Thórhallsdóttir, Provenza, & Balph, 1990; Weladji et al., 2010). (H1c) In contrast, no differences in suckling bout lengths are expected (Heitor & Vicente, 2008; Hejcmanová et al., 2010). On the other hand, I predict that (H1d) older females are less vigilant due to habituation to the lack of predation (Derocher, Wiig, & Bangjord, 2000; Eide, Eid, Prestrud, & Swenson, 2005; Jepsen, Eide, Prestrud, & Jacobsen, 2002; Prestrud, 1992; Reimers, Lund, & Ergon, 2011) and (H1e) further away from other reindeer than younger females because of the habituation to lack of predators and their higher social rank which enables them to take over feeding patches when animals gather together (Espmark, 1964; Holand et al., 2004; Kojola & Helle, 1996; Kojola & Nieminen, 1988; Weladji et al., 2006). I also predict that (H1f) older and younger females spend proportionately more time feeding and less time lying down and in other activities compared to prime age females to compensate for the cost of growth and senescence (Crête & Huot, 1993; Rönnegård et al., 2002; Weladji et al., 2002).

2. Does having a calf at heel affect the social and vigilant behavior and time budget of Svalbard reindeer females?

Based on earlier studies on reindeer time budget and behavior (H2) I predict that a calf at heel affects female’s social and vigilant behavior as well as its time budget leading to (H2a) females with calves being more vigilant than females without calves (Reimers et al., 2011; Soulliere, 2008) and (H2b) females with calves being socially more segregated from other adults than females without calves (Alendal, de Bie, & van Wieren, 1979; Lent, 1966; Loe et al., 2006). Adapted from Conradt (1998), social segregation from other animals is here considered as spending time in smaller groups and further away from other reindeer. I also predict that (H2c) females with calves spend proportionately more time feeding and less time lying down and in other activities compared to females without calves, to compensate for the cost of reproduction (Allaye Chan-Mcleod et al., 1999; Crête & Huot, 1993; Reimers et al., 2005; Rönnegård et al., 2002).

11

2 Materials and methods

2.1 Svalbard reindeer as a study species

Svalbard reindeer is one of the eight subspecies of reindeer. Reindeer has a circumpolar distribution over large boreal and arctic areas across Northern Europe, North-America and Siberia (Paine, 1988; Vors & Boyce, 2009), but Svalbard reindeer is solely found on the archipelago of Svalbard and its surroundings (33°31.093'E to 10°29.343'E and 74°20.066'N to 80°49.685'N) (Tyler & Øritsland, 1989). It arrived in the archipelago between c. 6700-5000 BP. by migrating over the sea ice from Siberia (Kvie et al., 2016; Van der Knaap, 1989) and since then, the population has been in isolation and slowly adapted on the harsh environmental conditions of the area (Veiberg et al., 2007). Due to adaptation, Svalbard reindeer differs from other reindeer Figure 2 Svalbard reindeer feeding calmly in subspecies in multiple ways. Unlike the barren-ground the front while polar bear is hunting seals in the background. © Laura Puikkonen caribou (Rangifer tarandus groenlandicus) and other migrating subspecies, Svalbard reindeer does not generally move great distances between or within seasons (Côté et al., 2002; Henriksen, Aanes, Sæther, Ringsby, & Tufto, 2003; Lee et al., 2015; Paine, 1988; Tyler & Øritsland, 1989). Stationary strategy combined with decreased metabolic rates is especially prevalent in winter, as it helps the reindeer to cope with limited foraging resources (Cuyler & Øritsland, 1993; Loe et al., 2007; Reimers, 1977; Ringberg, 1979). One reason for the reduced mobility is the virtually predator-free habitat. Although polar bears (Ursus maritimus) (Figure 2), have been occasionally observed to prey or scavenge on them, and arctic foxes (Vulpes lagopus) or glaucous gulls (Larus hyperboreus) may opportunistically feed on weak, starving reindeer, neonatal offspring and carcasses, they are considered to cause only a minor predation pressure to the reindeer populations on Svalbard (Derocher et al., 2000; Eide et al., 2005; Jepsen et al., 2002; Prestrud, 1992). In the lack of predators, the population is mainly regulated by winter weather conditions and population density (Reimers, 1977; Skogland, 1990).

Life histories of Svalbard reindeer are characterized by high survival and low fecundity (Bjørkvoll et al., 2016; Lee et al., 2015; Reimers, 1977). Female reindeer can live at least 16 years (Hansen, Veiberg, & Aanes, 2012), but the male survival decreases rapidly after five years of age and males

12 over ten years old are rare (Reimers, 1977). Females give birth to maximum one calf per year (Bjørkvoll et al., 2016) and the calving season is concentrated to the first half of June (Paulsen, 2018). Regarding the maternal behavior, reindeer belong to so-called “followers”, which means that the offspring follows its mother nearly immediately after its birth and that the mother-young association decreases over time (Ralls et al., 1987). The cows nurse their calves around three months (Ralls et al., 1987). Weather conditions are the major determinant of calf survival as they restrict the food availability in winter affecting both the resources of females and overwinter survival of calves (Reimers, 1977).

Currently, the number of Svalbard reindeer population is estimated to be 22,000 (Norwegian Polar Institute 20.6.2019). Majority of reindeer populations (34) have a declining trend, but Svalbard reindeer is one of the eight populations that are increasing (Vors & Boyce, 2009). Despite the increase, Svalbard reindeer are increasingly vulnerable to negative consequences of climate change (Le Moullec, Pedersen, Stien, Rosvold, & Hansen, 2019; Nedberg, 2012; Tandberg, 2016; Vors & Boyce, 2009). Increasing snowpack, icing and rain-on-snow events (ROS) during winter are causing decreased fecundity and starvation of mainly young and very old animals and changes in the movement patterns of the reindeer (Albon et al., 2002; Hansen, Aanes, & Sæther, 2010; Meland, 2014; Reimers, 1977; Stien et al., 2010). Understanding the effect of life history on the reproductive success and behavior of Svalbard reindeer would help us to understand the consequences of possible demographic changes caused by the climate change to the future of the species.

Svalbard reindeer has been studied relatively widely. Most of the studies about Svalbard reindeer focus on its population dynamics (Aanes et al., 2003), interaction with plant communities (Van der Wal, Brooker, Cooper, & Langvatn, 2001) and behavior. Behavioral studies focus on foraging (Beumer, Varpe, & Hansen, 2017; Henriksen et al., 2003; Karbø, 2019; Walter et al., 2018), vigilance and disturbance responses (Baskin & Skogland, 1997; Hansen & Aanes, 2015; Reimers & Eftestøl, 2012; Reimers et al., 2011), spatial distribution of the reindeer (Hansen et al., 2010; Meland, 2014; Tyler & Øritsland, 1989) and responses to extreme weather events and climate change (Hansen, Aanes, Herfindal, Kohler, & Sæther, 2011; Hansen, Lorentzen, et al., 2019; Loe et al., 2016; Stien et al., 2010). However, to this date, there has been only a marginal number of studies on the reproductive behavior of the Svalbard reindeer (Heatta, 2009) and there are no previous studies on the social and maternal behavior of Svalbard reindeer in the summertime. This knowledge gap combined with a habitat lacking several external factors such as predators, insect harassment and competitors, and a tendency of not moving great distances in regular weather conditions, makes Svalbard reindeer an

13 interesting model species for an observational field study analyzing the intrinsic effect of life histories on the long-living iteroparous mammalian behavior.

2.2 Study area

The behavioral observations of reindeer were done in Semmeldalen valley and the south-western part of Reindalen valley on the island of Spitsbergen, in the archipelago of Svalbard, approximately 34 kilometers away from Longyearbyen. Semmeldalen-Reindalen area including the adjacent valleys of Kalvdalen and Istjørndalen are common calving grounds of Svalbard reindeer (Hansen et al., 2010; Paulsen, 2018). The extent of the whole study area was about 21km2 but most of the observations were made in the southern parts of the Semmeldalen valley (Figure 3). Although the visibility is generally good, there are numerous small gullies and hills which can cause interruptions to the scan sampling. The study area was chosen because it belongs to the long-term mark–recapture program run since 1994 as a part of long-term monitoring of Svalbard reindeer. From 1994 to 2018, a total of 1282 reindeer have been marked (personal communication, Loe 16.11.2019) and 5-15% of the population is captured yearly in April for marking and estimating the population size. In addition to marking and recapturing, summer censuses have been done each year since 1996. In the censuses the

Figure 3 Valleys of Semmeldalen and Reindalen on Spitsbergen, with the study area marked with red. Map edited from TopoSvalbard (Norwegian Polar Institute, n.d.).

14 numbers of unmarked animals and the identity of observed marked individuals have been recorded, including information about their reproductive status. (Lee et al., 2015)

Human activity in Reindalen is classified as medium-low while the scientific activity is high (Reimers et al., 2011). Semmeldalen and the south western corner of Reindalen are outside of the hunting areas for the Svalbard reindeer, but 285 females were culled between 1994 and 2008 for scientific purposes and in the surrounding areas 531 female reindeer were hunted between 1994 and 2011 (Lee et al., 2015). In total there are four cabins in Reindalen and three cabins in Semmeldalen. Norwegian Polar Institute’s Tarandus cabin was used for accommodation during the observation periods in the field.

2.3 Studied individuals

The population in Reindalen-Semmeldalen-Colesdalen area consists of approximately 1313 animals (Norwegian Polar Institute 20.6.2019). The marked females have 5 cm wide ear tags in both ears, as well as 3.5 cm wide plastic neck collars while males have only ear tags (Lee et al., 2015) (Figure 4). All tags and collars are color-coded by the year of birth and have individual identification numbers (Lee et al., 2015). In total I observed 41 marked individuals including 12 females with calves, 9 females without calves, two males and 14 unmarked females with calves (Supplementary Table 1).

I chose the individuals for this study based on random encounter and same animals were observed on several days whenever possible. The study area has mostly good visibility up to several kilometers in distance, which made finding marked animals and observing several individuals at once relatively easy. However, reindeer move around in the valley changing their place up to a couple of kilometers per day, which made recurring scanning of same individuals during multiple days challenging. I prioritized marked females with calves, and instant scan sampling was only started if one of the

Figure 4 Svalbard reindeer in Semmeldalen. Marked reindeer mother 203B with its calf (left picture) and female 344R without a calf. Both are photographed in July - note the slightly taller antlers of females without calves. Images © Sanne Moedt 2019

15 encountered animals was a female with a calf. In total, one to five animals were scanned simultaneously for instant scan sampling and one animal for focal watches. During the first observation period, I sampled all marked females whenever possible. Several long continuous sampling sessions with same individuals proved to be difficult to obtain due to the movement of the animals and gullies in the valley. For this reason, I sampled only females with calves during the second observation period and ensured a sufficient amount of data from at least the mothers and a larger number of scans of the same individuals. Scanned calves were identified by nursing events and association with their mother. Allosuckling, defined as suckling of other than calf’s own mother, has been identified in semi-domestic reindeer (Engelhardt et al., 2014; Engelhardt, Weladji, Holand, Røed, & Nieminen, 2015) but it has not been documented in Svalbard reindeer. Mother-calf association is generally an accurate measure of kinship in follower ungulates (Djakovic et al., 2012; Ralls et al., 1987) and would have corrected for failed identification caused by possible allosuckling.

No evidence of a lower calving success in animals captured in the previous winter has been found (Omsjoe et al., 2009). However, in order to analyze the effects of ear tags and neck collar on individual behavior, I established a control group sampled by both instant scans and focal watches. Only females with calves were included in the control group due to the difficulty to distinguish between sexes from a distance by other means than observing a calf at heel. I chose control animals randomly from a position where it was possible to scan marked individuals at the same time. When animals moved, the priority was given to getting longer scan samples of marked individuals, which caused control animals sometimes to be observed for very short periods of time (< 6 scans). I made drawings of control animals and distinguished them from each other by their antlers and color, which made sure that no individual was sampled as two different animals. I observed seven control animals in July and seven in August.

2.4 Observation period

The mean calving date on Nordenskiöld Land is on the 7th of June and varies between the beginning and middle of June (Paulsen, 2018). The first observation period was conducted in the latter half of July and the second one in late August, when calves are slightly over a month and two months old, respectively. The first 18-day field trip was taken 12.-29.7.2019 and the second 10-day field trip took place 15.-24.8.2019. Based on the development of their digestive organs and the ruminal content, one-month old reindeer still get their energy intake from milk, although they have started to express foraging-like behavior (Knott, Barboza, & Bowyer, 2005). At about two months of age the diet of the calves starts to resemble the diet of an adult reindeer (Knott et al., 2005). Studies made on sheep (Ovis

16 aries) have shown that lambs up to two or even three months old learn foraging preferences from their mothers (Mirza & Provenza, 1992; Thórhallsdóttir et al., 1990). During this time reindeer calves also gain most of their weight increase (Petersson & Danell, 1993). These phases in calf development make July and August suitable months for observing essential age-related differences in the maternal behavior of reindeer females.

I made observations on twelve days during the first field trip and on six days on the second one. Working hours in the field were on average ten hours a day, of which approximately two hours was used for looking for reindeer to sample and hiking to and from the observation spots and eight hours for observations. The rest of the time was used for hiking in and out from the field base and for general upkeep and rest. During summer, moving around in Svalbard is only possible by foot, boat, or helicopter. This combined to the limited budget and challenging hiking conditions in the early summer due to flooding of rivers also affected the schedule of the field work.

The observation periods took place during the polar day, which starts on the 19th of April and ends on the 24th of August. Mean temperature during the field work period was 7.6°C with minimum 4.7°C and maximum 12.6°C in July, and 4.3°C with minimum -0.1°C and maximum 6.6°C in August. The average wind in July was 4.3 m/s with gusts up to 10.8 m/s and 3.5 m/s with gusts up to 8.7 m/s in August. The mean precipitation was 0.19 mm with median of zero and maximum 1.9 per day in July and 0.21 mm with median of zero and maximum 1.3 per day in August. Weather data is collected at Adventdalen observation station, which is 28 kilometers from Semmeldalen and the closest weather observation station to the area. (Norwegian Meteorological Institute and the Norwegian Broadcasting Corporation 2007-2020).

2.5 Behavioral observations

Both focal and instantaneous scans were done during both July and August observation periods. I searched the animals for the scans walking by foot but sat still while scanning (Figure 5). I made all observations for focal and instant scan samples by myself to avoid interpretation biases. I was accompanied by one polar bear guard, who was the same person throughout the field season. The observation distance of a minimum of 200 meters was pre-determined based on a study of reindeer

17 reaction distances in Reindalen from Colman et al. (2001) and Reimers et al. (2011). This distance was adjusted after the first days in the field, depending on weather conditions, terrain and the habituation of animals to the observer (Reimers, Loe, Eftestøl, Colman, & Dahle, 2009). However, no observations were made closer than 150 meters away from the animal, the average observation distance was 440 meters, and the maximum distance was one kilometer.

Reindeer were spotted and observations were made by using Nikon 10x42 6.7 binoculars and Nikon Spotting Scope 80 with 20X-60X zoom. The focal watches and scanning were interrupted if the Figure 5 Sampling in Semmeldalen © Sanne Moedt 2019 animal presented disturbance behavior including staring towards observers, approaching observers intentionally or animal taking an alarm pose, urinating or moving away from observers because it had noticed the observers. This happened relatively often, as reindeer tended to ‘check us out’ and come to stare at us within 10-50 meters distance for up to half hour if they happened to notice me and the polar bear guard. To make sure that no behavioral effects introduced by the observer took place, I only continued instantaneous scans after the animal had left us and returned to its undisturbed behavior for at least ten minutes. I did not continue focal watches on the same day if they were interrupted because of disturbance of the animal. After finishing scanning of a certain individual for the day, I approached it until a distance where it was possible to read the number of its ear tags.

2.5.1 Instantaneous scan sampling

I observed reindeer females by instantaneous scan sampling method (Altmann, 1974) in one to ten- hour blocks and scanned the focal animals every 10 minutes. Each scan took approximately ten seconds and I scanned one to five animals ‘simultaneously’. I completed all simultaneous scans within one minute of the start of the scan. Scan sampling was continued for as long as possible, but due to animals moving in terrain with restricted visibility, a number of scans had to be interrupted. Scans were only included in the statistical analysis if scans continued uninterrupted for at least one hour.

Preliminary behavioral categories for scan sampling were constructed based on previously applied categories in existing literature before the first field work period (Supplementary Table 2). As studies about Svalbard reindeer behavior are limited, also studies on reindeer and caribou behavior using the instantaneous scan sampling method were used as the basis of the selection. The criteria for chosen

18 categories was that a category was used in at least two peer-reviewed research articles. However, as the study focuses on the maternal, social, and vigilant behavior of reindeer mothers, final, more detailed categories were constructed during the first day in the field based on preliminary observations and literature (Table 1). I supplemented the main categories with behaviors described in a detailed analysis of reindeer and caribou behavior by Thomson (1977) and Pruitt (1960), definition of play behavior from Meyer‐Holzapfel and Frisch (1956), definition of head-hind leg contact by Espmark (1977), the detailed descriptions of caribou maternal behaviors presented by Soulliere (2008) in her MSc thesis and by Jingfors, Miller, Glaholt, and Gunn (1983) on vigilant behavior of caribou. In addition, the article on techniques for analyzing vertebrate social structure presented by Whitehead and Dufault (1999) was used as a support to construct the final categories.

Table 1 Final behavioral categories, states, events, and contacts States and events marked with * are classified as vigilant behaviors in this study.

Behavioral categories States Events Contacts

Feeding Feeding still Animal stance* Nursing, successful Lying Feeding walking Alarm pose* Nursing, attempted Standing Lying head down Urinating Sniffing Walking Lying head up* Urinating (vigilant)* Grooming Running Stand head down Head-hind leg contact Vocal contact Other Stand head up* Groom self Playing Walking head down Stretch Other bonding interaction Walking head up* Yawn Front or back leg kick Trotting (play or social) Body shake Antler threat Trotting (vigilant)* Excitation jump* Head swing Galloping (play) Rush towards another Galloping (vigilant)* Other agonistic interaction

The categories were divided into individual animal’s behavioral states and events and contacts between at least two animals (Table 1). This division was adapted from Jingfors et al. (1983) and followed the same principles. States cannot occur simultaneously and when one state ends, another one begins, while events on the other hand can occur during states (Jingfors et al., 1983). All events described by Jingfors et al. (1983) were included in the behaviors listed here except for head bobbing, which has not been observed to occur considerably in other reindeer subspecies than caribou (Pruitt, 1960). When scanning animals, I noted down the major behavioral state during a ten-second-long scan and made additional notes on events occurring at the time of the scan. In contacts between animals, I defined the initiator and recipient, while in agonistic interactions these were described as the winner and the loser (Engelhardt et al., 2014). I grouped these contact events into three groups: A. Bonding interactions between mother and calf, B. Bonding interactions, excluding own calf and C. Agonistic interactions. The latter includes interactions resulting in increased distance between

19 individuals, as presented by Müller-Schwarze, Källquist, and Mossing (1979). Here, I considered bonding interactions as the counterpart of agonistic interactions by intentionally reducing the distance between animals and/or otherwise socially bonding the animals involved. As expected, contact events were happened rarely during the scans and no statistical analysis was made on them.

I defined vigilant behavior (Table 1) broadly based on the previous studies on reindeer as a state occurring with animal’s head above its shoulders while the animal is observing its surroundings (Bøving & Post, 1997; Frid, 1997; Reimers, Røed, & Colman, 2012). I also interpreted trotting and galloping in the context of a clear reason to be vigilant, and scans including any events defined as a vigilant behavior, as vigilant scans. The broad definition was used due to occasionally long observation distances and the 10-minute interval of instantaneous scan sampling, which made it impossible to confidently identify more fine scale behaviors.

In addition to the scanned animals’ behavioral activity, several other parameters were observed and noted down during the scan. These included group size, group structure (number of males, females, calves, yearlings, adults of unknown sex, reindeer of unknown age group, other marked reindeer and all animals excluding calves), the behavioral state of the majority of animals (excluding calves) in the group, distance to the nearest neighbor (Müller-Schwarze et al., 1979), the sex, antler size and reproductive status of the nearest neighbor, distance to the own calf (if applicable), observation distance, antler size, relative calf size (if applicable), approximate location and vegetation in the immediate surroundings of the animal. A group was defined as animals which would gather in case of disturbance. In practice, this meant a maximum 50 body lengths between animals and case-specific evaluation of the effect of terrain and state of activity. In nine occasions, several focal animals were in the same large group while scanned. The Svalbard reindeer are mostly solitary or stay in small groups of two to three animals, and larger aggregations are loose and temporary (Alendal et al., 1979). Based on my field observations the compositions of larger groups often change and/or they disperse within two hours and animals are often involved in different activities within the group. Colman et al. (2010) showed that in captivity, semi-domestic reindeer synchronized their behavior within small, enclosed groups of three, but not between groups. In addition, studies in wild red deer (Cervus elaphus) have shown that the behavior of individual deer is influenced mainly by its immediate neighbors rather than group members further away (Rands, Muir, & Terry, 2014). Therefore, there was no reason to expect severe non-independence issue of sampling reindeer in the same group and possible effect of the sampling situation was accounted with random effect of date in the statistical models. Scanning multiple animals in a same large group made it possible to notably increase the sample size.

20

2.5.2 Focal watches

In focal watches, the focal animal was observed continuously for 30 minutes during its active bout. The 30-minute length of the focal watches has been regularly used in behavioral studies of reindeer (Colman, Pedersen, Hjermann, & Holand, 2003; Kojola, 1989; Witter, Johnson, Croft, Gunn, & Gillingham, 2012). The focal watches were conducted when the focal reindeer and the group it was in had not been paying attention to the observer for at least 30 minutes after arrival to the observation spot. Focal watches, in which the focal animal lied down or noticed the observer before half of the 30-minute watch had elapsed, were discarded. If the focal individual moved out of sight for less than five minutes, the watch was paused and continued after the animal was in sight again. The focal watches were weighted proportionally according to the final duration of the watch as in two cases the focal animal laid down or noticed the observer after 15 minutes but before 30 minutes had passed. Whenever possible, a focal watch was started within the first 15 minutes after the end of a resting bout, as it was more likely that the animal did not start a new resting bout within this time frame. There were no autocorrelation issues related to the focal watches, as there was at least three hours between the focal watches with same animal, and the focal watches were always done during feeding bouts with at least one resting bout in between the watches. Therefore, the animals changed their behavior and moved between the focal watches for such a long time spans that one focal observation was not depending on the former, as explained by Flydal, Korslund, E, Frode Berntin, and Colman (2009).

During the focal watches, two different main variables were continuously observed: 1. Vigilant time of the focal female. 2. Distance between the focal female and its calf. Vigilance was defined based on the previous studies on reindeer as an interruption in feeding activity to lift head above the shoulders and observing surroundings (Frid, 1997; Reimers et al., 2011) (Figure 6). These vigilant bouts were measured in seconds. Unlike Reimers et al. (2011) and Bøving and Post (1997), also vigilance bouts lasting more than 10 seconds were included in the analysis, as they seemed not to end up in non-vigilance behavior as Reimers et al. (2011) described. Altogether, 15 focal watches were done with marked females with calves, three in marked females without calves, two in marked males Figure 6 Svalbard reindeer female and calf in August in and eight with unmarked females with calves. Adventdalen. Calf presents vigilant behavior while walking with its head above the shoulder line. © Laura Puikkonen Marked males were added into the dataset to © Laura Puikkonen 2019

21 increase the sample size of animals without calves. Together these focal watches constituted in 14 hours and 40 minutes of focal observation data.

2.5.3 Other behavioral observations

Some behavioral events, such as nursing, are on average less than half a minute long and occur relatively rarely. Because of this, their occurrence is hard to capture with standard instantaneous scan sampling methods or focal watches. To analyze differences in nursing bout lengths of reindeer mothers, all occurrence approach by Altmann (1974) was used to note down all nursing events observed in marked and unmarked animals during the time in the field both within and outside of scans and focal watches. Relaxed all occurrence approach was used for other interesting behavioral events and contact events between two individual reindeer. These events were not used for any statistical analyses, but they provided further insight for the discussion of the results.

Nursing events were considered successful if a calf suckled five seconds or longer. This was chosen as the limit of successful suckling event, as it has been previously used with reindeer (Engelhardt et al., 2014; Engelhardt et al., 2015). A successful suckling event was considered finished when calf did not contact the mother’s udder with its mouth for 20 seconds or longer (Engelhardt et al., 2015). The total length (including calf pushing mother’s udder), initiator and the terminator of the suckling event (calf or mother) were observed and noted down whenever possible.

2.6 Statistical analyses

From the observations I gained three different datasets.

1. The first dataset consists of time stamped instantaneous scan samples of focal animals including nominal data on activity, numeric data on distance to the focal female’s calf (if applicable) and the nearest neighbor in a group, number of other reindeer in a group and binomial data on reproductive status, vigilance, being in a group, and whether the female and its calf are involved in a same activity or not (if applicable). In addition, this dataset includes additional information not used in the statistical analysis of this study, including notes on the group composition, observation distance, location, and vegetation type around the focal animals. 2. The second dataset includes proportional data on dam’s distance from their calves and vigilance of adult reindeer during the focal watches. 3. The third dataset consists of numeric data on nursing bout lengths (in seconds) collected with all-occurrence principle.

22

The data in these datasets served as response variables for hypothesis testing (Supplementary Table 3). Each dataset was complemented by birth year data collected by Norwegian Polar Institute in the long-term monitoring program and the sex of marked focal animals confirmed either by information from the long-term monitoring program (personal communication, Loe 16.11.2019) or field observations (sex of control animals). For the analysis of time budget in instant scans, I grouped the behavioral states and simultaneously occurring events into three groups: feeding, lying down and other behaviors (including standing, walking and running).

Before statistical analyses, the data was explored visually by using Microsoft Excel (2009). All further statistical analyses were done with R version 3.6.2. (R Core Team, 2019). Each response variable collected by instant scan samples was graphically inspected to detect autocorrelation by plotting the residuals of time series of four continuous instant scan sampling observation periods from individuals 129Y (50 scans on 24.7., no calf), 203B (62 scans on 22.8., calf at heel) and 220W (51 scans on 24.7., with calf). Detected autocorrelation was confirmed by running Ljung-Box test from the package stats (v.3.6.2; R Core Team, 2019), p-value <.05 indicating that autocorrelation was present. To account for autocorrelation, I created separate datasets for each dependent variable with their independent variables and random effects by averaging the variable over the maximum number of autocorrelated lags (l) detected in the time series analysis (Supplementary Table 4). The resulting dataset consisted of data points, each representing l number of scans depending on the response variable. Some data points included l-n scans, because of interruptions in scanning. I built all models without weights because of the inability of the chosen residual diagnostic tests to handle models with weights (Hartig, 2020). Therefore, I removed the data points where scanning was interrupted from the final datasets. However, in the group size dataset I also included data points consisting of only six scans to maintain a larger sample size. As number of individuals at each age was low, I considered categorizing the age into age groups. Defining the category cutoffs is often arbitrary and important information might be lost when categorizing continuous variables (Ranganathan, Pramesh, & Aggarwal, 2017), so I chose not to categorize any continuous variables. Another option to simplify the model structure would have been pooling scans of each individual or reindeer of each age. I did not pool the samples as is not generally considered as a good practice due to violation of the assumptions of independence of the samples and resulting pseudoreplication (Jenkins, 2002; Machlis, Dodd, & Fentress, 1985). In addition, modern statistical methods give good tools to deal with multiple observations from same individuals in the form of mixed models (Jenkins, 2002).

I used generalized linear mixed effect regression models for testing the hypotheses and modelled 14 different response variables with glmmTMB function from the package glmmTMB (Brooks M. E. et

23 al., 2017) (Supplementary Table 4). The glmmTMB fits models with maximum likelihood estimation via Template Model Builder (Brooks M. E. et al., 2017). Besides testing the hypotheses, I created separate control models for each response variable, testing the effect of ear tags and collars. Because of I collected data from females without calves only from July; the rapid growth and development of calves during late summer (Petersson & Danell, 1993); and notable difference in weather conditions between the two observation periods (Norwegian Meteorological Institute and the Norwegian Broadcasting Corporation, 2007-2020), I also tested the effect of observation month (July or August) in the control models. As the datasets include multiple repeated observations of same individuals on different days, individual identity of the focal animal and the observation date were included as random effects in each model. Independent variables in the models were focal animal’s age and reproductive status (calf or no calf at heel), but the latter was excluded from maternal behavior models. In cases where month had a significant effect in a control model, I included it in the respective model as an independent variable (Table 5). To extensively account for the effect of age, quadratic effect of age and the interaction effect between age and calf and age and month were included in each model. However, I removed these effects from the final models if their estimates were non-significant in order to interpret the unbiased estimates of main effects. All fitted models converged and all variance estimates were accessible. I analyzed the degree of possible multicollinearity between the independent variables with check_collinearity function from performance package (Lüdecke, Makowski, Waggoner, & Patil, 2020), and the values for independent variables remained generally low with variance inflation factor (VIF) less than two. Interactions and the quadratic effect of age as well as some the related main effects had moderate to high VIF (> 5). It has been argued, that high VIF values related to use of interactions and quadratic effects are negligible (Allison, 2012; Hanushek E. A. & E., 1977), so I ignored the high VIF values in the presence of these higher order terms. None of the variables were centered (Dedrick et al., 2009).

After fitting the models, I conducted the residual diagnostics for dispersion, homoscedasticity, zero- inflation, deviation, and outliers with a simulation based diagnostic test function simulateResiduals from package DHARMa (Hartig, 2020). I detected overdispersion in 11 out of 14 models. To the models M1a (Calf distance in instant scans) and M5 (Nearest neighbor in instant scans) I included an observation-level random effect to account for overdispersion (Harrison, 2014). In the model M6 (Group size in instant scans) correcting overdispersion with and observation-level random effect would have resulted in quantile deviations in DHARMa residual diagnostics, so I used negative binomial distribution (Harrison, 2015; Ver Hoef & Boveng, 2007). For models M1b (Time close to calf in focal watches), M1c (Time far from calf in focal watches), M4b (Vigilant seconds in focal

24 watches), M7 (Proportion of instant scans in group) and M8a, M8b, M8c (Proportion of scans lying down, feeding or in other activity respectively), overdispersion was accounted for by using betabinomial distribution (Harrison, 2015). Outlier test was significant (p<.01) for the model M4a (Proportion of vigilant scans), which was caused by outliers in the dataset. I did not remove these outliers as based on the field notes it is likely that they are a result of genuine variation in behavior rather than measurement errors (Zuur & Ieno, 2016). In addition, I detected two outliers in the nearest neighbor distance dataset and removed these because it was debatable whether these reindeer had been in a group or alone at the time of the scan and the outliers altered the effect estimates by turning the effect of age significant. Despite the effort made in model fitting, deviation in some of the models remained high in relation to the residual degrees of freedom. This resulted likely from the unbalanced datasets with relatively small effective sample sizes and exclusion of some possibly important explanatory variables such as group composition and size. However, after adjustments in model specifications, each model passed the DHARMa residual diagnostic tests with no significant problems.

I used a significance level of p = 0.05 when estimating the influence of fixed effects. In addition, I evaluated each model based on its AICc value compared to the null model with only random effects. AICc is a version of Akaike information criterion (AIC) that is corrected for small sample sizes and used as a measure to assess the relative quality of statistical models to detect problems such as overfitting (Hurvich & Tsai, 1989). A lower AICc value with ΔAICc > 2 is considered significant and the model is accepted (Hurvich & Tsai, 1989). The AICc values were obtained with function aictab from the package AICcmodavg (Mazerolle, 2019). I also report marginal and conditional pseudo-R2 values calculated with r.squaredGLMM function in the package MuMIn (Barton, 2020) for the models with Poisson, negative binomial and binomial distributions. Marginal pseudo-R2 value shows what proportion of variation is explained by fixed effects in the model and conditional pseudo- R2 value gives the same estimation for the full model including random effects (Barton, 2020).

3 Results

3.1 The control group and effect of observation period

Based on the analysis of the control group and marked study animals’ behavior, no major differences seemed to be related to markings. Ear tags and collar had a significant effect of -0.40 (+- 0.16, p<.02) only to the proportion of time a focal animal was standing (Supplementary Table 5), but the effect of these markings was non-significant for all other response variables, including the proportion of scans in other activity than feeding and lying down, which included the scans standing. The observation

25 month had a significant effect on the vigilance, distance to the nearest neighbor and the proportion of scans the reindeer were feeding or in other activity (Supplementary Table 5). To account for the effect, observation month was included in these models as an independent variable even when the interaction with age was not significant.

3.2 The effect of age and calf at heel on the behavior of Svalbard reindeer females

3.2.1 Maternal behavior

During the field work in Semmeldalen, I did not observe any bonding interactions between adult animals. Instead, all bonding interactions occurred either between females and calves or among calves. Mothers and their calves nuzzled with each other and mothers tended to sniff their calves occasionally. Both mothers and calves often made sounds to recognize and find each other from a distance as earlier documented by Espmark (1971). In addition, I saw calves playing with each other and two separate events in which mother (females 123-66G and 162B) and calf played together by running around for no apparent reason. Apart from playing, nearly all bonding interactions occurred between mother and its own calf, although once an adult female (344R) sniffed a foreign calf and was then threatened away by the calf’s mother. More often, a calf approaching a foreign adult received an agonistic response. I did not detect any signs of allosuckling. Based on the abovementioned observations and the distance between animals, the calf and mother pairs were relatively easy to distinct from others. Overall, both bonding interactions between mothers and calves as well as nursing bouts, were limited and rarely captured by instant scan sampling and focal watches or seen otherwise. During both observation periods, I recorded twelve successful nursing bouts and their full lengths. I observed eight additional nursing bouts, but the length of them is unknown as I noticed them after the start of the suckling. I also observed six rejected suckling attempts in July and five in August. There was no significant difference in the nursing bout length between the mothers of different age (Supplementary Table 4).

Calf was the nearest neighbor of its mother in 74 percent of instant scans, but dams and calves drifted occasionally far away from each other. Calves might lie down while females graze several dozens of body lengths away from them. On one occasion, a calf started to follow a group of strange females and calves and ended up over five hundred meters away from its dam until it laid down to wait that the dam finds it. Although marked, the age of this dam was unfortunately unknown and therefore excluded from the analysis. An average distance between females and their calves in the instant scans was nine body lengths, but within individual variation was high, reaching up to 45 body lengths. The female 169Y stayed closest to its calf with an average distance of one and half body lengths and 26 female 123-66G the furthest with an average distance of 19 body lengths. Young mothers’ distance from their calf increased over the summer from July to August and the interaction between the month and age had a significant effect of 0.12 (±0.05, p<.05) (Supplementary Table 4, Figure 7C) to the mother’s distance to its calf. The size and significance of the effects of month and age were not possible to determine in focal watches because of small sample size, which resulted in relatively high AICc values compared to the null model and high confidence intervals (Supplementary Table 4, Figure 7A & D).

Figure 7 A. Estimates of fixed and random effects with their confidence intervals. B. Predicted synchrony of activities between mother and calf in instant scans both in July and August with original observations plotted with crosses. C. Predicted calf distance in instant scans with original observations plotted with crosses. D. Original observations of proportion of time spent close (<5 bl) and far (>20 bl) from calf in focal watches.

27

On average, mother and calf pairs presented synchronized activity in 71 percent of all the instant scans. However, the variation between individuals was high: Eight-year-old 169Y which was closest to its calf had the highest proportion of synchronized scans (83 percent) and the six-year-old 123- 66G, also having the largest mean distance to its calf, had the lowest proportion of synchronized scans (47 percent). The youngest and oldest mothers tended to synchronize the most and both the quadratic effect of age (B = 0.07 ±0.03, p <.04) and effect of age (B = -0.99 ±0.47, p<.04) were significant, although quadratic effect estimate was low and ΔAICc remained low, difference between the fitted model and null model being indecisive (Supplementary Table 4, Figure 7B).

3.2.2 Vigilant behavior

All observed individuals showed vigilant behavior. On average, a female was vigilant in 9.7 percent of the instant scans. There was a near significant trend that a calf at heel decreased vigilance of females in instant scans (B = -0.44 ±0.24, p<.07) (Supplementary Table 4, Figure 8B). On average, mothers were vigilant in 8.4 percent of the scans in July and 4.8 percent of the scans in August, while females without calves in were vigilant 10.8 percent of the scans in July. However, the variation between individuals was high and neither the most vigilant female 117-74G, which was vigilant in 21.3 percent of the scans, nor the least vigilant female 178W, which didn’t show any vigilant behavior during the scans, had a calf. Over six-year-old mothers showed a near significant trend tending to be more vigilant in July than in August in instant scans, and the month and age interaction was a near significant trend (B = 0.26 ±0.14, p<.06). No other significant differences in the vigilance of different age mothers were detected. In focal watches conducted only when animals were not lying down, dams were in total significantly more vigilant than other reindeer (B = 0.61, ±0.22, p<.01) but less vigilant in August than in July (B = 1.40 ±0.26, p<.001), while age had a near significant trend to decrease vigilance (B = -0.07 ±0.04 p<.08) (Supplementary Table 4, Figure 8C). All coefficients with their confidence intervals are visible in Figure 8A.

3.2.3 Social behavior

Mothers were further away from their nearest neighbor than females without calves. They had an average distance of 7.3 body lengths in July and 10.4 in August, compared to the average 5.8 body length distance of females without calves in July. Dams were significantly further away from their nearest neighbor than females without calves (B=0.22 ±0.07, p<.01). In addition, dams were significantly further away from their nearest neighbor in August than in July (B= -0.35 ±0.07, p<.001). (Supplementary Table 4, Figure 9B.)

28

Figure 8 A. Estimates of fixed and random effects with their confidence intervals. B. Predicted vigilance in instant scans with original observations plotted with crosses. C. Predicted vigilance in focal watches with original observations plotted with crosses, two observations of males added in order to increase sample size are circled.

29

Figure 9 A. Estimates of fixed and random effects with their confidence intervals. Data in all models is from instant scans. B. Predicted distance to nearest neighbor with original observations plotted with crosses, two outliers removed from the analysis are circled. C. Predicted time in groups vs. alone with original observations plotted with crosses. D. Predicted group size with original observations plotted with crosses.

30

Females were in a group in 82 percent of the scans. In July, both females with and without calves spent nearly all of the scans in groups; 88 percent and 90 percent respectively, and having or not having a calf did not have significant effect on whether a female was in a group or alone (Supplementary Table 4, Figure 9C). Young dams spent more time alone with calves in August than in July and more than older dams in August. The interaction between month and age was significant (B = -0.50, ±0.21, p<.02) (Supplementary Table 4, Figure 9C). However, confidence intervals for the coefficients in the model were high (Figure 9A). The group sizes of females varied from solidarity up to 36 animals in the same group. Groups larger than twelve individuals were relatively rare, loosely formed and spread out easily. Females without calves were in larger groups than mothers, the average group size being 8.6 animals in July. For females with calves the average group size was 5 in July and 3.2 in August. The average group size was larger for young females without calves compared to old females without calves and old females with calves compared to young females with calves. The interaction of the effects of calf and age was highly significant (B = 0.21, ±0.06, p<.001). In addition, there was the most variation in the group sizes of young females without calves. (Supplementary Table 4, Figure 9D).

3.2.4 Time budget

Dams spent on average 33 percent of scans lying down, 63 percent feeding and 4 percent in other behaviors in August. In July, they spent nearly the same proportion of time lying down (36 percent) with no significant change but proportionately less time feeding (B = -0.46, ±0.20, p<0.03) and more time in other activity than in August (B = 1.07, ±0.34, p<.01). Proportions of these activities in July were 52 and 12 percent, respectively. Females without calves spent on average 43 percent of the time lying down, 49 percent feeding and 8 percent of scans in other behaviors in July. Calf at heel had a significant effect on the time budget of the females by decreasing the proportion of lying down (B = -0.37, ±0.18, p<.05) and a near significant effect (B = 0.48, ±0.27, p<.08) for increasing other behavioral activities, but not feeding. (Supplementary Table 4, Figure 10.)

31

Figure 10 A. Estimates of fixed and random effects with their confidence intervals. Data in all models is from instant scans. B. Predicted time budget for feeding. C. Predicted time budget for other behavioral activities. D. Predicted time budget for lying down.

Original observations in B, C and D are plotted with crosses.

32

4 Discussion

Species with long lifespans have been found to be less negatively affected by environmental variability than short-lived species (Morris et al., 2008). One reason for this is behavioral plasticity stemming from biological evolution, individual learning, and even cultural transmission, as plasticity helps animals to adapt to environmental changes and habitats, where one behavioral phenotype is not necessarily consistently optimal (Mery & Burns, 2009). Svalbard reindeer has been isolated in virtually predator free, highly seasonal environment for at least 5000 years (Van der Knaap, 1989). While no evidence of recent genetic bottlenecks has been found, the population seems genetically impoverished and subdivided within the archipelago, even between valleys located just 20-40 - kilometers apart (Côté et al., 2002; Cronin, Patton, Balmysheva, & Macneil, 2003). This could potentially have decreased the behavioral plasticity in the population (Berrigan & Scheiner, 2004). However, relatively long lifespan with high survival rates especially in females (Tyler & Øritsland, 1989), facilitate possibilities for experience-based learning and increasing behavioral plasticity along ageing. Recently, Svalbard reindeer has been documented to change their behavior in relation to extreme weather events, to which reindeer have responded by changing their home range and/or diet (Hansen, Lorentzen, et al., 2019; Stien et al., 2010). In this study, I investigated the effects of age and rearing a calf to Svalbard reindeer females’ behavior to gain further insight on the effect of experience on their life histories.

The effect of age on the physical fitness, reproductive effort and success of reindeer females has been studied widely (f.ex. (Festa-Bianchet, 1988; Reimers et al., 2005; Rönnegård et al., 2002; Weladji et al., 2002), but there is little information on its effect on behavior, especially in wild reindeer populations (Weladji et al., 2006). The effects of age on the survival and reproduction as well as the behavior of animals can be approached through non-exclusive senescence and terminal investment hypotheses. While senescence deteriorates the physical condition and decreases resources of an ageing individual, terminal investment might also increase its reproductive success in long-lived species (Côté & Festa-Bianchet, 2001; Mirza & Provenza, 1992; Weladji et al., 2006). Although evidence of terminal investment in the behavior was not found in this study, its possibility cannot be disregarded. My observations show that individuals of different ages behave differently, which can be related either to experience gained over the years and learning, changing energy demand and allocation or most likely, both.

Rearing a calf changes female’s energy allocation and demand drastically (Crête & Huot, 1993; Reimers et al., 2005; Rönnegård et al., 2002), and its effect in the behavior of reindeer and caribou

33 females has been studied more widely than that of age. Detailed records of the behavior of dams in the wild have been made already since the 1960s (Lent, 1966; Pruitt, 1960; Thomson, 1977). The findings of this study support the idea that rearing a calf has an effect to the reindeer female behavior through the trade-offs that occur between survival and rearing offspring.

4.1 Effect of age

My first research question was whether the age of Svalbard reindeer female effects its maternal, vigilant and social behavior, and time budget. Based on the idea of terminal investment compensating some of the implications of senescence, I made six hypotheses on the effect of age – and through that experience – to the Svalbard reindeer females’ behavior. In this study’s dataset of females from two to fifteen-year-old animals, I consider ‘young’ and ‘younger’ reindeer to be six years or younger while with ‘old’ and ‘older’ I refer to reindeer of eight years or older. However, when answering the research question, we should keep in mind, that Svalbard reindeer generally start reproducing when they are two to three years old (Weladji et al., 2006; 2008). As the youngest dam sampled in this study was already four years old, it is possible that none of the dams was primiparous. It would mean that having a calf was not a ‘new’ situation for the younger study animals, although older dams had still most likely gained more experience in rearing their offspring. On the other hand, the prime age of reindeer is commonly considered to fall between five- and twelve-years of age as their body mass peaks within that age bracket (Rönnegård et al., 2002; Skogland, 1984; Weladji et al., 2010), but no exact estimate for Svalbard reindeer females have been published.

4.1.1 Age and time of the summer affect maternal behavior

During the field work I saw several interactions between dams and calves, but physical contacts apart from suckling were relatively rare. The observed interactions were similar with interactions documented in previous studies in other subspecies with the exclusion of some behaviors such as head bobbing only known to occur in caribou (de Vos, 1960; Espmark, 1971; Lent, 1966). The hypothesis of no differences in suckling bout lengths of different age mothers was accepted, as age did not have significant effect on the length of these bouts. As the sample size was small and unbalanced due to difficulties in capturing multiple bouts and the full length of successful nursing events, more research would be needed to assess the effect of age in rejecting and accepting suckling initiatives from calf and the proportion of suckling bouts initiated by mother versus calf.

The hypothesis of older dams keeping closer to their calves in order to protect them, based on findings in bison (Bison bison) by Green (1990), was rejected. No significant difference between the overall

34 calf distances of dams of different ages was detected. However, younger dams kept significantly closer to their calves in July than in August, while no difference between months was detected in older dams. I did not distinguish a clear pattern in focal watches due to the low sample size and only individual observations per some of the age classes, although these models suggested a similar trend. Increase in the calf distance over the course of summer has also been shown in caribou (de Vos, 1960), and multiple reasons can contribute to the increasing calf distance over the summer in young dams but not in the old. The difference was especially profound in dams between four and six years of age, which is often considered as the (early) prime age when individual’s body and offspring mass are either at their maximum or still increasing (Loison & Strand, 2005; Rönnegård et al., 2002; Skogland, 1984; Weladji et al., 2008). Hence younger dams might have more resources available and no need to wander far away from their offspring, or their calves might be in better condition and have more energy to follow and keep close to their mother. Van der Wal et al. (2000) showed that in July, Svalbard reindeer prefer high quantity patches of vegetation that have melted from snow early. During sampling, I saw several occasions where a calf had winded apart from its dam either by lying down while its dam walked away while feeding and a couple of times calf itself followed other reindeer far from the dam. These incidences happened mainly in July and ended with either calf starting to vocalize and call its dam or dam itself noticing the missing calf and returning to it. In all cases the female interrupted feeding to watch approaching or walk to the calf. It is possible, that old dams wandered further from their calf in the search of preferred feeding patches because of experience in looking for them and/or dire need for maximizing the energy intake.

In bison, staying close to the calf was anti-predatory behavior (Green, 1990) which rarely persists in predator-free environment due to its costs (Reimers et al., 2011). As arctic foxes and glaucous gulls may prey on young calves (Prestrud, 1992), keeping close to a calf in July might help protect it, and increase reproductive success of the mother, whereas older calves in August would not need so close tending to. This could in turn indicate older mothers of ten years of age to be too restricted by their resource allocation to keep close distance to their calf early on. The crucial part of calf development as a potential prey for foxes and gulls might also have passed at over one month after birth and younger dams are ‘overly careful’. In late July, calves tend to be already over twenty kilograms heavy, which is over six times the weight of either one of these small predators (Petersson & Danell, 1993; Prestrud, 1992). The differences in calf distance could also be related to the changes in the energy and time budget of the females and calves over the course of summer (Bårdsen, Tveraa, Fauchald, & Langeland, 2010; Lent, 1966). In the light of these results, it is unclear whether there is an advantage of staying especially close to the calf or not. Studies looking at the correlates and effects of distance

35 between mothers and calves would be needed, in order to distinguish between alternative explanations.

The hypothesis of older dams being more often involved in the same activity as their calf was partly supported, as ten-year-old dams synchronized behavior more than five to eight-year-old dams. However, the four-year-old dam 220W seemed to synchronize its activity with calf nearly as much as the old dams. Although I based on my hypothesis on that mimicking dam’s activity is important to the development of foraging skills in several ungulate species (Kojola, 1990; Mirza & Provenza, 1992; Thórhallsdóttir et al., 1990; Weladji et al., 2010), there are other more likely reasons for synchronization that occur in both young and old mothers and their calves. While some proportion of unsynchronized scans is unavoidable due to differences in time budgets between dams and calves (de Vos, 1960; Pruitt, 1960), the high degree of synchronization and the proximity of dam and its calf are likely linked (Rands et al., 2014). Hence, it also makes sense that the highest and lowest synchronizing mothers were the same ones being on average the closest and furthest from their calf. Similar energy and time budgets and resulting high activity synchrony has found to play a major role in herd formation in caribou (Lent, 1966; Pruitt, 1960). A high synchrony could similarly be a result of energy and time budget of calves resembling more that of young and old dams than dams in their prime age.

4.1.2 Experience from predator-free environment tends to decrease vigilance

The hypothesis of older females being less vigilant was partly supported. In instant scans, dams’ vigilance decreased significantly with age in August. However, no effect of age was detected between dams of different age in instant scans in July or within females without calves. On the contrary, there was a slight near significant trend of age decreasing vigilance in all animals in focal watches, but due to small number of observations in each age class, this result should be interpreted with caution. The difference in results between these two datasets may originate from a few different reasons. First, instant scans were conducted during both active and resting bouts, while focal watches only during active bouts. Second, instant scans every ten minutes are not the best suited in capturing fine scale differences in the presence of short and frequent vigilance bouts, but they can give a rough picture of overall trends in vigilance when observing multiple individuals. Third, the dataset on focal watches is unbalanced and include two males to increase the sample size, as finding enough focal females for watches and obtaining comparable records during active bouts would have both been too time- consuming.

36

The lower vigilance of older dams in August in instant scans could be an example of the effect of experience to the behavior through the habituation to the lack of predators. The finding is in line with possible predation pressure from arctic foxes and glaucous gulls towards calves being much smaller in August, than earlier in the summer due to their rapid growth. For example, semi-domestic reindeer calves can almost double their weight and gain up to 20 kilograms in between summer (July) and autumn (September) (Petersson & Danell, 1993; Weladji & Holand, 2003a), while arctic foxes’ average summer weight is only around three kilograms (Frafjord, 1992). Although rarely scientifically documented, the threat from these small predators is well illustrated with a scene that I saw in July: An arctic fox ran towards reindeer causing a dam to take an alarm pose and its calf to come next to the mother imitating the reaction. No similar event took place in August. A slight trend of older animals in focal watches supports the assumption, that maintaining a higher degree of anti- predatory behavior in predation-free environment would cause extra costs without benefits (Reimers et al., 2011), which then results in decreased vigilance along ageing and gained experience.

Besides my findings, also earlier studies suggest that ungulates adjust their vigilance and responses to disturbance based on their experience on the level of threat down to a certain limit (Byers, 1997; Reimers et al., 2009; Reimers et al., 2011; Reimers et al., 2012). Reimers et al. (2011) argue that the flight responses of Svalbard reindeer reflect their response threshold inherited from ancestors, representing a possible ‘base level’ of ungulate vigilance. A certain level of vigilance has documented to persist also in predator-free caribou population (Bøving & Post, 1997). In order to compare the current vigilance of females in Semmeldalen with earlier results from Reimers et al. (2011), I excluded vigilance bouts longer than 10 seconds from the following numbers delivered from the focal watches: Vigilant time per minute was on average 0.65 s/min, which is longer than the findings from Reimers et al. (2011) for the females with calves in Reindalen (0.26 s/min), but still within the confidence intervals and in line with their overall results. Frequency of bouts was also higher (every 4.3 minutes), than what Reimers et al. (2011) recorded (every 8.3 minutes). Increased vigilance could therefore indicate increased predation pressure since the study of Reimers et al. (2011), as the level of human activity in the valley has remained relatively unchanged.

Polar bears are opportunistic in regard of their prey and rarely feed on reindeer (Derocher et al., 2000). Recently, observations of polar bears hunting reindeer have increased although they have been a highly local phenomena (Reimers et al., 2011; Stempniewicz, Kidawa, Barcikowski, & Iliszko, 2014). Bears hunt reindeer often by stalking up to resting animals or surprising them with a short, rapid chase (Stempniewicz et al., 2014). This summer, people at the Polish Research Station Hornsund, Southern

37 part of Spitsbergen, documented one polar bear hunting two reindeer in two days by chasing them towards the sea (Hornsund research station, 24.8.2020). Increasing variability in polar bear prey items is thought to be result of diminishing sea ice, which brings challenges to catching enough seals for its diet (Descamps et al., 2017; Kay, Holland, & Jahn, 2011; Stempniewicz et al., 2014). Svalbard reindeer do not migrate, and often prefer to forage in coastal plains and valleys in the summer, as they have the most vegetation (Alendal et al., 1979), and this might make Svalbard reindeer highly prone to polar bear predation in the future (Skogland, 1991). Svalbard reindeer also do not live in multispecies communities where they could obtain protection against predators from each other like ungulate species in African savannah (Sinclair, 1985). However, reindeer has been documented to respond rapidly to ground icing events which limit foraging, by increasing movement toward ice-free areas (Loe et al., 2016; Stien et al., 2010). This shows that behavioral plasticity could also increase movement between grazing grounds, polar bear free refuges and even new calving grounds (Skogland, 1991) or increase vigilance, grouping behavior and the proximity to calf in order to counteract the predation pressure. In the case of experience from increasing polar bear predation, older animals might begin to show higher vigilance than younger ones. This in turn could result in costly implications to body condition due to decrease in the foraging efficiency of older individuals (Altendorf, Laundré, López González, & Brown, 2001), and formerly absent trade-offs in Svalbard reindeer behavior balancing between foraging and predator avoidance.

4.1.3 Average group size increases with age in dams but decreases in females without calves

The hypothesis of older females being further away from other reindeer because of their higher social rank or further habituation to lack of predators than younger females was rejected. Instead, age affected in multiple ways to the association of females with other reindeer, depending on their reproductive status and the observation period. Females gathered in groups kept similar distances irrespective of age. This can be evidence supporting an earlier point of view of Svalbard reindeer as solidary animals brought together outside the breeding season only because of the scarce food resources (Alendal et al., 1979). In semi-domestic reindeer, social ranks among females have been observed in relation to either higher body weight and/or age (Kojola, 1989). Dominance hierarchies persist among females outside of the rutting season and are often related to age and experience also in other ungulate species (Rutberg, 1983; Townsend & Bailey, 1981). On the contrary, it seems that Svalbard reindeer does not necessarily have a strong social hierarchical structure, or it could be determined rather by reproductive status than individual identity or age (Holand et al., 2004; Lent, 1966). Nevertheless, groups are loosely formed aggregations of individuals grazing in the same area,

38 and I observed nearly no physical contact among adult animals during the observation period and no age-related differences in defending grazing spots, as such interactions were rare in first place.

In August, older mothers seem to occur more often in groups than younger mothers, which were more often alone in the scans. This finding further supports rejecting the hypothesis of older females being further away from other reindeer than younger females. It could indicate the role of experience in selecting grazing spots, as old mothers would gather to the ‘good’ grazing spots (Van der Wal et al., 2000). This seems unlikely, as females without calves showed no such differences. Instead, older females could be escaping predation risk by joining groups, however other findings strongly suggest a nearly predator-free habitat in August, when calves are no more as prone to predation of foxes and gulls. Yet another explanation for the behavior could be the approaching breeding season in the autumn, when females aggregate in harems (Heatta, 2009): older females could be gathering to groups more than younger females because of experience spanning over multiple years. However, that no females with calves behaved in similar manner and the observations on group size conflict this idea. Hence, further research is needed to understand the social dynamics and aggregation to groups in Svalbard reindeer.

Interestingly, age had a different effect on group size of females with calves and without calves while no effect of month was detected. For dams, group size increased with increasing age, whereas for females without calves it decreased with increasing age. It could be that older dams search safety from larger groups throughout the season, because they have a high energy demand than other females. Reimers et al. (2011) showed, that increasing group size decreases vigilance. Older dams might thus decrease the need for compromising between vigilant and foraging time by aggregating in larger groups. Females without calves do not have to deal with same trade-off as they do not need to worry about predators as much. Hence, they can choose their grazing area based on solely the forage (Van der Wal et al., 2000), but young animals might still seek safety from groups. However, I could not find similar observations in existing literature, and more research is needed on how the relative costs and benefits of grouping behavior may change with age in (Svalbard) reindeer.

4.1.4 Effects of senescence are not compensated by the female time budget

I predicted that older and younger females would spend proportionately more time feeding rather than lying down compared to prime age females to compensate for the cost of reproduction (Crête & Huot, 1993; Weladji et al., 2002). Interestingly, there was no significant effect of age in the proportion of any measured activity. The differences resulting in higher energy demand could be so fine scale between age groups that they are not detectable, or the relatively low effective sample size and low

39 number of really young and old animals in the dataset might limit the power of this analysis. Also, the observed reindeer with calves are likely to represent a fit subgroup that is in very good condition, as the last winter was very snowy, icy and challenging for reindeer in the area (Norwegian Meteorological Institute and the Norwegian Broadcasting Corporation 2007-2020). This could have caused the weakest individuals already have lost their calves no matter the age, and there might be physiological limitations on the differences of behavior in different ages.

4.2 Effect of calf at heel

My second research question was, does a calf at heel affect the social and vigilant behavior, and time budget of Svalbard reindeer females. Rearing a calf is widely shown to affect the reindeer body condition (Reimers et al., 2005; Rönnegård et al., 2002), although differences between barren and lactating females might converge in the winter after weaning (Allaye Chan-Mcleod et al., 1999). Even if the body size of females does not differ in the following winter, rearing a calf can result in smaller offspring in the following summer (Bårdsen et al., 2010). Based on this evidence on high costs of rearing a calf up to weaning age, it was not surprising that the effect of a calf at heel or the interaction of calf and age had a significant effect for several behavioral variables.

Reproductive status of Svalbard reindeer females reflects on their antler growth. Usually females which have lost their calf already during pregnancy have more developed antlers in the summer, than females with calf at heel or females which have recently lost their calf (Loe et al., 2019). There were both females with shorter and taller antlers in sampled females without calves, indicating that some the short-antlered ones (92B, 198B and possibly 216W and 129Y) might have lost their calves after parturition.

4.2.1 Dams decrease their vigilance towards the end of the summer

Based on earlier studies, I predicted that females with calves would be more vigilant than females without calves. This hypothesis received partial support. In contrast to existing research conducted with Svalbard reindeer earlier (Reimers et al., 2011), females with calves had a near significant trend of being less vigilant in instant scans than females without. This could be related to different age distribution of observed females with and without calves, or to the dams’ different time budget and need for more efficient foraging (Altendorf et al., 2001). In addition, increasing group size can decrease vigilance (Reimers et al., 2011), but dams preferred smaller groups in general. On the finer scale analysis of vigilant seconds during focal scans conducted while reindeer was active, I found dams to be more vigilant than other reindeer in July. In addition, their vigilance decreased radically

40 in August, then being lower than that of all reindeer’s vigilance in July. A number of reasons can explain the differences in results between these two datasets, as explained in the chapter 4.2.3.

According to Bøving and Post (1997), higher predation risk increases the level of vigilance displayed by caribou while feeding, standing, and lying, and decreases time in predation-vulnerable postures such as lying flat. In the absence of predators, the lower vigilance level of dams could be a consequence of the trade-off between foraging and vigilance (Altendorf et al., 2001; Fortin, Boyce, Merrill, & Fryxell, 2004). This seems to be especially prevalent for the month of August. However, in the light of the results from focal scans and previous studies, and as the result from the instant scan dataset is only a near-significant trend, this surprising result is possibly related to a technical aspect of sampling. Reindeer were sampled from varying distances up to one kilometer away from focal animals. Defining vigilance in feeding and otherwise active reindeer was straightforward. When reindeer were laying down, I defined all scans where reindeer had its head up as vigilant. However, this might include also scans in which reindeer are ruminating and not necessarily observing their surroundings as vigilant. Because females without calves spend proportionately more time lying down than dams, it is possible that scans lying down biased the estimates.

4.2.2 Other reindeer are closer to females without, than females with calves

Loe et al. (2006) presented that Svalbard reindeer dams segregate from both males and females without calves due to their different nutritional needs and activity budget throughout calving and lactation periods. Differences in segregation between the sexes or animals with different reproductive statuses were not analyzed in this study, as defining the exact group composition over long distances was often uncertain and the sex of some animals left undefined. Rather, segregation from all other reindeer was examined. Based on findings by Loe et al. (2006) on dams’ social segregation from females without calves and males, suggestion of reproductive status as a basis of domination hierarchy from Lent (1966) and findings by Alendal et al. (1979) on smaller average group size, my hypothesis was that females with calves would be socially more segregated from other adults than females without calves. My findings support this hypothesis.

There were no differences in proportion of time females spent alone whether they had a calf or not. Instead, having a calf significantly increased females’ distance to the nearest neighbor while they were in groups. Females were generally agonistic towards any foreign adult reindeer approaching their calf, which naturally increases the distance between animals (Engelhardt et al., 2014). During my time in Semmeldalen, I did not see any socially bonding interactions such as grooming or playing

41 between adult animals. However, I documented ten agonistic interactions between adults and one from a female towards calf. In one of the ten interactions the individual showing agonistic behavior was a male, and in the rest, it was a female. In five of them the behavior was triggered by feeding on the same patch of vegetation, one by approaching a foreign calf, two for being very close and in two the reason was unclear. Agonistic behaviors included front kicks, antler threats and once a kick jump. All agonistic contacts I observed occurred with virtually no physical contact and resulted in immediate avoidance behavior from the behalf of the recipient. These observations support the idea of solitary animals, which are foremost brought together by food sources, with no significant identity- based dominance structures outside of the rutting season in the summer.

Similar situation was described in caribou by Lent (1966), who argued that open groups do not allow learning to recognizing individuals and establish a stable domination hierarchy but rather the reproductive status defines dominant roles. It has been shown that lactating females have a higher reproductive success in the following year, which is a possible proof of generally higher quality of lactating females (Bårdsen et al., 2010). Higher social rank of dams could so be explained by the higher individual quality, manifested by for example size or condition. Holand et al. (2004) presented that hierarchy in Svalbard reindeer population has possibly evolved with visual assessment of physical characteristics rather than an individual identity. Other means of assessing the rank of conspecifics from a distance could be odoriferous substances, likely emitted from reindeer’s interdigital glands which they seem to spread to their head and antlers by a gesture known as head- hindleg contact (HHC) (Espmark, 1977) or vocalizations. I saw head-hindleg contact occasionally performed by both adults and calves. Vocalizations I heard rarely due to the long observation distances, but they are also known to be important in reindeer communication (Espmark, 1971).

As discussed in the chapter 4.2.3, the female group size was significantly affected by age, but its effect was different in dams and females without calf. Overall, females with calves spent proportionately more time in significantly smaller groups than females without. The group sizes of females in this study differed slightly from the observations of Alendal et al. (1979). In most of the observations, I saw the dams with two other reindeer or alone (excluding the calf), whereas Alendal et al. (1979) reported the most frequent group sizes of two and four animals for dams. While the average group size in study of Alendal et al. (1979) was 2.2, the average group size in this study was 5.6. The difference is likely related to different sampling techniques and that only females were included in my dataset. As grouping behavior in ungulates is strongly linked to predation-avoidance, it is reasonable to assume that no increase in predation pressure on the reindeer in Semmeldalen has

42 occurred between the studies. However, according to Reimers et al. (2011), polar bears have hunted reindeer in Edgeøya. Therefore, it could be interesting to see whether the group sizes on Edgeøya have changed from 1970s until now, and if not, study what could be a reason of not aggregating in larger groups to counteract the predation.

4.2.3 Foraging time of dams increases towards the autumn

The hypothesis of females with calves spending proportionately more time feeding and less time resting compared to females without calves, to compensate for the cost of reproduction (Allaye Chan- Mcleod et al., 1999; Reimers et al., 2005; Rönnegård et al., 2002), was partially accepted. The dams did indeed spend proportionately less time lying down and slightly less time in other activities such as moving and standing than females without calves. However, there was no significant difference in time spent feeding between dams and females without calves. The time budget of dams changed significantly between the observation months. While proportion of time dams spent lying down remained approximately the same, proportion of time feeding increased and other activities decreased as the season advanced. Individuals allocating more time on feeding from other activities, can be related to the plant phenology. In July, Svalbard reindeer Figure 10 Weather measurements at Adventdalen weather prefer patches of vegetation which melt the station during the field work (Norwegian Meteorological earliest and have the largest quantity of Institute and the Norwegian Broadcasting Corporation 2007- 2020). July and August observation periods differ from each forage, but the preference fades away in other remarkably by their temperatures. This could be one additional reason for the different time budget of dams August (Van der Wal et al., 2000). Picking between the observation periods. the best patches to feed on would require more moving around. On the other hand, the nutritional content of forage decreases while the growing season proceeds, which might require allocating more time for feeding (Van der Wal et al., 2000). Another reason for increased feeding, approximately same time lying and decreased time in other activities in August might the highly different weather conditions between the observation periods (Figure 9). During winter, Svalbard reindeer survive with their fat reserves from the previous summer by saving as much energy as possible (Cuyler & Øritsland, 1993). The animals might have increased their feeding and saved energy by lying down, while decreasing other less important activity because of the cold weather.

43

In the future, Svalbard reindeer time budget is likely to be affected by the climate change. In addition to the possible predation pressure from polar bears, changing temperatures may cause changes in vegetation in the summer and increase the requirement for robust fat reserves for surviving over the winters. Variation in large-scale winter climate and density dependence have been found to play an important part in Arctic ungulate population dynamics in Greenland (Forchhammer, Post, Stenseth, & Boertmann, 2002). Rain on snow events in Svalbard are increasing due to climate change and have negative consequences to reindeer foraging and survival over winter (Hansen et al., 2011; Stien et al., 2010; Wickström, Jonassen, Cassano, & Vihma, 2020). As winter conditions are harsh and forage scarce, obtaining sufficient fat reserves during the short summer is important (Ringberg, 1979). This can however be more difficult in the future: Climatic warming in the Arctic advances plant phenology and can cause a mismatch between the peak resource demand and availability, which can cause calf mortality and decrease offspring production as already seen in migratory caribou (Post & Forchhammer, 2007). While Svalbard reindeer are non-migratory and would not suffer a spatial mismatch between summer and winter ranges, the short arctic summer requires full capability to utilize the forage resources to survive the next winter. This can make Svalbard reindeer dams and calves highly vulnerable to temporal mismatch.

In semi-domestic reindeer, worse winter conditions have found to decrease reproductive output, lower autumn body mass of dams while increasing the body mass of barren females (Bårdsen et al., 2010). Bårdsen et al. (2010) presented these findings as a proof of risk-sensitive approach, where reindeer reduce their reproductive allocation with worse winter conditions and increase it much more slowly when winter conditions improve. By the risk sensitive approach Bårdsen et al. (2010) challenged the idea of non-plastic life history strategy and argued that adaptation to conditions possible. According to the risk sensitive approach, increasing rain-on-snow events could result in more risk averse life histories and phenotypic plasticity could counteract adverse effects of climate change.

Besides vegetation and winter conditions, climate change can alter the distribution of macroparasites with adverse consequences to ungulate time budget. In mainland populations, harassment by oestrid flies and other parasitic flying insects have been shown to reduce the time reindeer spend feeding and through that their body condition and the autumn weight of calves (Weladji & Holand, 2003b; Witter et al., 2012). Oestrid flies, which cause higher degree of harassment on reindeer and caribou than mosquitoes, are absent in Svalbard (Hagemoen & Reimers, 2002; Loe et al., 2007; Mörschel & Klein, 1997). The most common mosquito in the Arctic, Aedes nigripes is the only mosquito species in Svalbard (Coulson & Refseth, 2004). Warming climate can able faster development resulting in an

44 increased probability of mosquito survival to the adult stage (Culler, Ayres, & Virginia, 2015). In addition, climate change can cause alien species to spread and vagrant species to establish in the archipelago (Coulson & Refseth, 2004). In Greenland, warming has already found to advance mosquito phenology into phenological synchrony with caribou, which has adverse effects for caribou (Culler et al., 2015).

4.2 Limitations of the study

Observational studies of mammalian behavior in the wild are necessary to investigate effects of behavior on fitness and ultimately evolution of behaviors, yet they are restricted by available resources and time, limiting their length and extent. Collecting behavioral samples by human observations is preferred for its non-invasiveness, but time consuming and never entirely free from subjective interpretation. In addition, obtaining sufficient amount of data on animals in the wild requires a wide background research on the species’ behavior, knowledge of the local terrain and weather conditions as well as a fair amount of luck. Although I carefully considered many possible pitfalls in the data collection and handling for this study, I would like to specifically address the following two issues.

Firstly, Svalbard reindeer presents a cyclic behavioral pattern, where active bouts including feeding, moving and standing occur in turns with resting bouts including lying down, sleeping and ruminating. These cycles are approximately five hour long during summer (Loe et al., 2007). Therefore, sampling an individual only during one of the bouts causes issues with autocorrelation and may bias the estimates of the time budget and frequency of behaviors. Problems arise especially, when an individual is only sampled during part of the bout, because consecutive scans are not independent from each other. This might lead to crude over- or underestimates in time proportions of certain activities. However, in practice, sampling a focal individual continuously through one full bout cycle proved to be challenging due to reindeer moving to different directions from each other, further from the observer, and behind hills and ridges while they were grazing. Because of this, sampling was a compromise between obtaining a long continuous series of scans on few individuals, sampling as many individuals on as many days as possible, and sampling as many individuals in total to increase the effective sample size. I addressed autocorrelation in this study by averaging the data over the number of autocorrelated lags for each response variable as explained in the Materials and methods chapter. In addition to autocorrelation, instant scan intervals are prone to be too long to detect behaviors that occur rarely or take a very short time (Engel, 1996). To counteract this, I tested the scanning interval and categories in the study area on the day before starting the sampling as well as

45 conducted separate focal watches and noted down certain events with all-occurrence approach in addition to instant scan sampling.

Secondly, I tried to obtain an even distribution of females of different ages by randomly sampling as large number of females as possible, so that there would inevitably be females of different ages in the sample. I succeeded relatively well in this attempt. However, the age distribution differs slightly between females with and without calves. Svalbard reindeer has low fecundity and females might abort a calf during pregnancy if the environmental conditions are bad (Albon et al., 2002; Hansen, Pedersen, et al., 2019). This was likely during the year of the study, as the winter was especially difficult with over 30 days of above zero temperatures combined with precipitation in 13 different periods between November and April measured at Adventdalen weather station (Norwegian Meteorological Institute and the Norwegian Broadcasting Corporation, 2007-2020). These rain on snow events caused severe challenges to the reindeer population and the winter was publicly referred as the deadliest season for the reindeer in the area since 2008 (Palko, 2019). Accordingly, I observed several dozens of reindeer carcasses in the study area. In the case of severe weather conditions, the youngest, first time reproducing females and the oldest females are the most likely to abort pregnancy (Adams, 2005; Festa-Bianchet, 1988). In addition, females may not reproduce at all before three years of age (Weladji et al., 2006) and can live over 16 years (Hansen et al., 2012). Therefore, having no dams younger than four and only one female older than eleven years old in this study, is a reasonable bias in the sample. It is possible that some age-related differences that could be seen along a wider spectrum of age, are not visible in the age distribution of sampled dams.

Thirdly, instantaneous scan sampling and focal watches are potentially highly detailed and non- invasive study methods for behavioral studies in the field. However, they are time-consuming and expensive methods for gathering data of wild animals, as they require either constant presence of human observer or installing camera arrays to record behavior. Because of the limited resources in this study, I was responsible for collecting all the behavioral data by observations on the spot. Because of time-consuming sampling methods and reindeer often moving out of sight, the effective sample size of different individuals remained low and the length of scanning periods varied daily. In many of the models this caused extremely low variance of the random effects of ID and day of observation. In addition, low and unbalanced sample size resulted in high confidence intervals (Figure 7-10A). Therefore, the effects of individual identity and daily changing weather conditions to the analyzed behaviors cannot be ruled out. Further research is needed to distinguish between age-, identity- and environment-related differences in Svalbard reindeer behavior.

46

5 Conclusion

This study illustrates the effects of age and calf at heel to the behavior of Svalbard reindeer, which inhabits isolated and virtually predator and harassment free Arctic habitat in Svalbard archipelago. It gives a proof that life history traits have an effect to individual’s behavior. The results show that decreasing resources available for survival and reproduction because of senescence can be related in trade-offs in female behavior. On the other hand, no evidence of terminal investment in reproduction through behavior was found. Instead of concluding terminal investment to be an entirely physiological phenomenon, the question of its behavioral implications remains open. As Weladji et al. (2006) argues, greater fitness can evolve to already good females when they learn how to become better mothers. However, further research on what it actually means to ‘be a better mother’ in different populations with different environmental variables is needed, in order to point out terminal investment behavior in a population. In addition, larger sample sizes with even age distribution along the model species whole lifespan is needed.

Although Svalbard reindeer is one of the subspecies of reindeer with an increasing population trend, it is still recovering from past overharvesting (Le Moullec et al., 2019) and sensitive to the climate change related disruptions in the Arctic ecosystem (Kay et al., 2011). In the era of climate change and rapid environmental changes, behavioral plasticity and adjusting behavior based on learning through experience might be in a crucial part in survival of long-living animal species with relatively slow reproduction. This plasticity allows individuals to adapt to rapidly changing habitat faster, than genetic changes on an evolutionary timescale would allow. My results provide further evidence on that Svalbard reindeer has already shown behavioral plasticity in response to predation risk and other habitat variables. Although climate change is likely to bring severe challenges in the form of reducing winter forage due to rain-on-snow events, possible mismatch between plant phenology and high energy demand in calving and rearing season, new predation pressure from polar bears and increasing harassment by insects in the future, Svalbard reindeer may be able to respond to these through experience-based change of behavior. However, increasing sources of trade-offs can decrease especially very old and young animals’ survival and reproductive success.

This study shows the demand for a longitudinal research in wild population to be able to distinguish between effects of age and individual personality. It would be important to also follow up on the offspring of the focal animals to inspect the actual effects of behavior in the reproductive success in the long run. However, longitudinal studies with long-lived wild animals in unrestricted areas are especially hard to implement due to time consuming data collection, remote locations, difficulties in

47 capturing and recapturing or resighting of animals and challenges in choosing efficient but the least invasive study methods which don’t affect survival or behavior. New cutting-edge biotelemetric methods for collecting data from activity patterns to physiological measurements and lighter solutions such as PIT-tags in studying social behavior, offer effective and comprehensive alternatives for observational data collection. In addition, modern machine learning methods offer intriguing possibilities for analyzing larger datasets than ever before and getting trustworthy results even with difficult behavioral datasets if applied by skilled researchers.

6 Acknowledgements

I want to wholeheartedly thank my supervisor Emma Vitikainen (The University of Helsinki) for all the invaluable encouragement, support, and advice. In addition, I want to give warm thanks to Mads Forchhammer (The University Centre of Svalbard) for bringing up the idea of this project and great help and assistance during the planning and field work phases. I also want to thank Leif Egil Loe for providing the data of reindeer ages collected by the Norwegian Polar Institute’s long-term monitoring program, and Hannu Pietiäinen (The University of Helsinki) for supporting the writing process during the master’s thesis course and seminar.

I want to thank Sanne Moedt for her astonishing patience and resilience as a polar bear guard during my field work. I also want to thank Audun Tholfsen, Dag Furberg Fjeld & other technicians at UNIS logistics and Stein Tore Pedersen from NPolar logistic department for encouraging and supporting me in taking the challenge to conduct field work in the Arctic. I want to thank you UNIS and NPolar for allowing me and Sanne to use the Tarandus cabin as our base during the field work, the crew from Henningsen Transport & Guiding for the help with organizing the first field trip and Elke from Red Cross Svalbard for the use of their cabin in Reindalen.

I want to thank Jakob Mathias Lothe for listening me talking about reindeer and statistics for hours and hours, and my close friends who helped me to take well-needed breaks from writing. In addition, I want to thank all of you who kept asking me how I was doing, when life threw extra challenges at me and on top of the thesis. Finally, I want to thank my family for being as lovely and understanding as they always are, and give special thanks to Saga and my grandparents for giving me a possibility to establish a strong interest towards nature and ecology in the first place.

48

7 References

Aanes, R., Sther, B.-E., Solberg, E. J., Aanes, S., Strand, O., & ritsland, N. A. (2003). Synchrony in Svalbard reindeer population dynamics. Canadian Journal of Zoology, 81(1), 103-110. doi:10.1139/z02-227 Adamczewski, J. Z. (1987). Body composition in relation to seasonal forage quality in caribou (Rangifer tarandus groenlandicus) on Coats Island, Northwest Territories. In: University of Alberta Libraries. Adams, L., & Dale, B. (1998). Reproductive performance of female Alaskan caribou. Journal of Wildlife Management, 62(4), 1184-1195. doi:10.2307/3801982 Adams, L. G. (2005). Effects of maternal characteristics and climatic variation on birth masses of Alaskan caribou. Journal of Mammalogy, 86(3), 506-513. Albon, S. D., Stien, A., Irvine, R. J., Langvatn, R., Ropstad, E., & Halvorsen, O. (2002). The role of parasites in the dynamics of a reindeer population. Proceedings of the Royal Society B: Biological Sciences, 269(1500), 1625-1632. doi:10.1098/rspb.2002.2064 Alendal, E., de Bie, S., & van Wieren, S. E. (1979). Size and composition of the wild reindeer Rangifer tarandus platyrhynchus population in the Southeast Svalbard Nature Reserve. Ecography, 2(2), 101-107. doi:10.1111/j.1600-0587.1979.tb00687.x Allaye Chan-Mcleod, A., White, R., & Russell, D. (1999). Comparative body composition strategies of breeding and nonbreeding female caribou. Canadian Journal of Zoology, 77(12), 1901-1907. Allison, P. (2012). When Can You Safely Ignore Multicollinearity? Retrieved from https://statisticalhorizons.com/multicollinearity Altendorf, K. B., Laundré, J. W., López González, C. A., & Brown, J. S. (2001). Assessing Effects of Predation Risk on Foraging Behavior of Mule Deer. Journal of Mammalogy, 82(2), 430-439. doi:10.1644/1545-1542(2001)082<0430:Aeopro>2.0.Co;2 Altmann, J. (1974). Observational Study of Behavior: Sampling Methods. Behaviour, 49(3/4), 227-267. Appleby, M. C. (1982). The Consequences and Causes of High Social Rank in Red Deer Stags. Behaviour, 80(3-4), 259-273. doi:10.1163/156853982X00382 Barton, K. (2020). MuMIn: Multi-Model Inference (Version R package version 1.43.17). Retrieved from https://CRAN.R- project.org/package=MuMIn Baskin, M. L., & Skogland, T. (1997). Direction of escape in reindeer. Rangifer, 17(1). doi:10.7557/2.17.1.384 Berrigan, D., & Scheiner, S. (2004). Modeling the evolution of phenotypic plasticity. In (pp. 82-97). Berube, C. H., Festa-Bianchet, M., & Jorgenson, J. T. (1999). Individual Differences, Longevity, and Reproductive Senescence in Bighorn Ewes. Ecology, 80(8), 2555-2565. doi:10.2307/177240 Beumer, L. T., Varpe, Ø., & Hansen, B. B. (2017). Cratering behaviour and faecal C: N ratio in relation to seasonal snowpack characteristics in a High-Arctic ungulate. Polar Research, 36(1), 1286121. Bjørkvoll, E., Lee, A. M., Grøtan, V., Sæther, B.-E., Stien, A., Engen, S., . . . Hansen, B. B. (2016). Demographic buffering of life histories? Implications of the choice of measurement scale. Ecology, 97(1), 40-47. doi:10.1890/15-0317.1 Brooks M. E., Kristensen K., van Benthem K. J., Magnusson A., Berg C. W., Nielsen, A., . . . M., B. B. (2017). glmmTMB Balances Speed and Flexibility Among Packages for Zero-inflated Generalized Linear Mixed Modeling The R Journal, 9(2), 378-400. Byers, J. A. (1997). American Pronghorn: Social Adaptations and the Ghosts of Predators Past: University of Chicago Press. Bårdsen, B.-J., Tveraa, T., Fauchald, P., & Langeland, K. (2010). Observational evidence of risk-sensitive reproductive allocation in a long-lived mammal. Oecologia, 162(3), 627-639. doi:10.1007/s00442-009-1537-0 Bøving, P. S., & Post, E. (1997). Vigilance and foraging behaviour of female caribou in relation to predation risk. Rangifer, 17(2). doi:10.7557/2.17.2.1302 Clutton-Brock, T., Price, O., Albon, S., & Jewell, P. (1992). Early development and population fluctuations in Soay sheep. Journal of Animal Ecology, 381-396. Clutton-Brock, T. H. (1984). Reproductive effort and terminal investment in iteroparous animals. The American Naturalist, 123(2), 212-229. Clutton-Brock, T. H., Guinness, F. E., & Albon, S. D. (1982). Red deer: behavior and ecology of two sexes: University of Chicago press. Collins, W., & Smith, T. (1989). Twenty-four hour behaviour patterns and budgets of free-ranging reindeer in winter. Rangifer, 9, 2-8. Colman, J., Eidesen, R., Gaup, M., Holand, O., Moe, S., & Eigil, R. (2010). Reindeer 24-hr within and between group synchronicity in summer versus environmental variables. Rangifer, 24. doi:10.7557/2.24.1.298 Colman, J., Pedersen, C., Hjermann, D., & Holand, O. (2003). Do wild reindeer exhibit grazing compensation during insect harassment? Journal of Wildlife Management, 67(1), 11-19. Colman, J. E., Pedersen, C., Hjermann, D., Holand, y., Moe, S. R., & Reimers, E. (2001). Twenty-four-hour feeding and lying patterns of wild reindeer Rangifer tarandus tarandus in summer. Canadian Journal of Zoology, 79(12), 2168-2175. doi:10.1139/z01-186 Conradt, L. (1998). Could asynchrony in activity between the sexes cause intersexual social segregation in ruminants? Proceedings of the Royal Society B: Biological Sciences, 265(1403), 1359-1368. doi:10.1098/rspb.1998.0442 Côté, S. D., Dallas, J. F., Marshall, F., Irvine, R. J., Langvatn, R., & Albon, S. D. (2002). Microsatellite DNA evidence for genetic drift and philopatry in Svalbard reindeer. Molecular ecology, 11(10), 1923-1930. doi:10.1046/j.1365-294X.2002.01582.x Côté, S. D., & Festa-Bianchet, M. (2001). Birthdate, mass and survival in mountain goat kids: effects of maternal characteristics and forage quality. Oecologia, 127(2), 230-238. doi:10.1007/s004420000584 Coulson, & Refseth. (2004). The terrestrial and freshwater invertebrate fauna of Svalbard (and Jan Mayen) In Prestrud P., Strøm H., & G. H. (Eds.), A catalogue of the terrestrial and marine animals of Svalbard. Tromsø, Norway Crête, M., & Huot, J. (1993). Regulation of a large herd of migratory caribou: summer nutrition affects calf growth and body reserves of dams. Canadian Journal of Zoology, 71(11), 2291-2296. Cronin, M., Patton, J., Balmysheva, N., & Macneil, M. (2003). Genetic variation in caribou and reindeer (Rangifer tarandus). Animal genetics, 34, 33- 41. doi:10.1046/j.1365-2052.2003.00927.x Culler, L. E., Ayres, M. P., & Virginia, R. A. (2015). In a warmer Arctic, mosquitoes avoid increased mortality from predators by growing faster. Proceedings of the Royal Society B: Biological Sciences, 282(1815), 20151549. doi:doi:10.1098/rspb.2015.1549 Curtsinger, J. W. (2001). Senescence: Genetic Theories. In N. J. Smelser & P. B. Baltes (Eds.), International Encyclopedia of the Social & Behavioral Sciences (pp. 13897-13902). Oxford: Pergamon. Cuyler, L., & Øritsland, N. (1993). Metabolic strategies for winter survival by Svalbard reindeer. Canadian Journal of Zoology, 71(9), 1787-1792. de Vos, A. (1960). Behavior of Barren Ground Caribou on Their Calving Grounds. The Journal of Wildlife Management, 24(3), 250-258. doi:10.2307/3797511 Dedrick, R. F., Ferron, J. M., Hess, M. R., Hogarty, K. Y., Kromrey, J. D., Lang, T. R., . . . Lee, R. S. (2009). Multilevel Modeling: A Review of Methodological Issues and Applications. Review of Educational Research, 79(1), 69-102. doi:10.3102/0034654308325581 Derocher, A. E., Wiig, Ø., & Bangjord, G. (2000). Predation of Svalbard reindeer by polar bears. Polar Biology, 23(10), 675-678. doi:10.1007/s003000000138

49

Descamps, S., Aars, J., Fuglei, E., Kovacs, K. M., Lydersen, C., Pavlova, O., . . . Strøm, H. (2017). Climate change impacts on wildlife in a High Arctic archipelago – Svalbard, Norway. Global Change Biology, 23(2), 490-502. doi:10.1111/gcb.13381 Djakovic, N., Holand, O., Hovland, A., Røed, K., Weladji, R., Fjeldstad, E., & Nieminen, M. (2012). Association patterns and kinship in female reindeer (Rangifer tarandus) during rut. Acta Ethologica. doi:10.1007/s10211-011-0121-x Dolhinow, P., McKenna, J. J., & Laws, J. V. H. (1979). Rank and reproduction among female langur monkeys: Aging and improvement (They're not just getting older, they're getting better. Aggressive Behavior, 5(1), 19-30. doi:10.1002/1098-2337(1979)5:1<19::AID- AB2480050104>3.0.CO 2-7 Douhard, M., Festa-Bianchet, M., Hamel, S., Nussey, D. H., Côté, S. D., Pemberton, J. M., & Pelletier, F. (2019). Maternal longevity and offspring sex in wild ungulates. Proceedings of the Royal Society B, 286(1896), 20181968. Eide, N. E., Eid, P. M., Prestrud, P., & Swenson, J. E. (2005). Dietary responses of arctic foxes Alopex lagopus to changing prey availability across an Arctic landscape. Wildlife Biology, 11(2), 109-122. Engel, J. (1996). Choosing an appropriate sample interval for instantaneous sampling. Behavioural Processes, 38(1), 11-17. doi:https://doi.org/10.1016/0376-6357(96)00005-8 Engelhardt, S. C., Weladji, R. B., Holand, Ø., de Rioja, C. M., Ehmann, R. K., & Nieminen, M. (2014). Allosuckling in reindeer (Rangifer tarandus): Milk-theft, mismothering or kin selection? Behavioural Processes, 107, 133-141. doi:10.1016/j.beproc.2014.08.013 Engelhardt, S. C., Weladji, R. B., Holand, Ø., Røed, K. H., & Nieminen, M. (2015). Evidence of Reciprocal Allonursing in Reindeer, Rangifer tarandus. Ethology, 121(3), 245-259. doi:10.1111/eth.12334 Espmark, Y. (1964). Studies in dominance-subordination relationship in a group of semi-domestic reindeer (Rangifer tarandus L.). Animal Behaviour, 12(4), 420-426. Espmark, Y. (1971). Individual recognition by voice in reindeer mother-young relationship. Field observations and playback experiments. Behaviour, 40(3-4), 295-301. Espmark, Y. (1977). Hindleg—head-contact behaviour in reindeer. Applied Animal Ethology, 3(4), 351-365. doi:https://doi.org/10.1016/0304- 3762(77)90059-1 Evans, R. M. (1990). The relationship between parental input and investment. Animal Behaviour. Festa-Bianchet, M. (1988). Age-specific reproduction of bighorn ewes in Alberta, Canada. Journal of Mammalogy, 69(1), 157-160. Flatt, T., & Heyland, A. (2011). Mechanisms of Life History Evolution: The Genetics and Physiology of Life History Traits and Trade-Offs. Oxford: Oxford University Press. Flydal, K., Eftestøl, S., Reimers, E., & Colman, J.E. (2004). Effects of wind turbines on area use and behaviour of semi-domestic reindeer in enclosures. Rangifer, 24, 55-66. Flydal, K., Korslund, L., E, R., Frode Berntin, J., & Colman, J. (2009). Effects of Power Lines on Area Use and Behaviour of Semi-Domestic Reindeer in Enclosures. International Journal of Ecology, 2009. doi:10.1155/2009/340953 Forchhammer, M. C., Post, E., Stenseth, N. C., & Boertmann, D. M. (2002). Long-term responses in arctic ungulate dynamics to changes in climatic and trophic processes. Population Ecology, 44(2), 113-120. doi:http://dx.doi.org/10.1007/s101440200013 Fortin, D., Boyce, M. S., Merrill, E. H., & Fryxell, J. M. (2004). Foraging costs of vigilance in large mammalian herbivores. Oikos, 107(1), 172-180. doi:10.1111/j.0030-1299.2004.12976.x Frafjord, K. (1992). Size and weight of Arctic foxes Alopex lagopus from the west Svalbard coast. Polar Record, 28(164), 68-70. doi:10.1017/S0032247400020350 Frid, A. (1997). Vigilance by female Dall's sheep: interactions between predation risk factors. Animal Behaviour, 53(4), 799-808. doi:https://doi.org/10.1006/anbe.1996.0345 Gadgil, M., & Bossert, W. H. (1970). Life Historical Consequences of Natural Selection. The American Naturalist, 104(935), 1-24. Gaillard, J.-M., Festa-Bianchet, M., Yoccoz, N., Loison, A., & Toigo, C. (2000). Temporal variation in fitness components and population dynamics of large herbivores. Annual Review of ecology and Systematics, 31(1), 367-393. Girard, I., Swallow, J., Carter, P., Koteja, P., Rhodes, J. S., & Garland Jr, T. (2002). Maternal-care behavior and life-history traits in house mice (Mus domesticus) artificially selected for high voluntary wheel-running activity. Behavioural Processes, 57(1), 37-50. Green, W. C. H. (1990). Reproductive effort and associated costs in bison (Bison bison): do older mothers try harder? Behavioral Ecology, 1(2), 148- 160. Hagemoen, R. I. M., & Reimers, E. (2002). Reindeer Summer Activity Pattern in Relation to Weather and Insect Harassment. Journal of Animal Ecology, 71(5), 883-892. Hansen, B., Aanes, R., & Sæther, B.-E. (2010). Partial seasonal migration in high-arctic Svalbard reindeer (Rangifer tarandus platyrhynchus). Canadian Journal of Zoology, 88(12), 1202-1209. Hansen, B., Veiberg, V., & Aanes, R. (2012). Material from harvested Svalbard reindeer - Evaluation of the material, the data and their areas of application for research and management. Hansen, B. B., & Aanes, R. (2015). Habituation to humans in a predator-free wild ungulate. Polar Biology, 38(2), 145-151. Hansen, B. B., Aanes, R., Herfindal, I., Kohler, J., & Sæther, B.-E. (2011). Climate, icing, and wild arctic reindeer: past relationships and future prospects. Ecology, 92(10), 1917-1923. Hansen, B. B., Lorentzen, J. R., Welker, J. M., Varpe, Ø., Aanes, R., Beumer, L. T., & Pedersen, Å. Ø. (2019). Reindeer turning maritime: Ice‐locked tundra triggers changes in dietary niche utilization. Ecosphere, 10(4), e02672. Hansen, B. B., Pedersen, Å. Ø., Peeters, B., Le Moullec, M., Albon, S. D., Herfindal, I., . . . Aanes, R. (2019). Spatial heterogeneity in climate change effects decouples the long-term dynamics of wild reindeer populations in the high Arctic. Global Change Biology, 25(11), 3656-3668. doi:10.1111/gcb.14761 Hanushek E. A., & E., J. J. (1977). Statistical Methods for Social Scientists. In (pp. 374). New York: Academic Press. Harrison, X. (2014). Using observation-level random effects to model overdispersion in count data in ecology and evolution. PeerJ, 2. Harrison, X. (2015). A comparison of observation-level random effect and Beta-Binomial models for modeling overdispersion in Binomial data in ecology & evolution. PeerJ, 3, e1114. doi:10.7717/peerj.1114 Hartig, F. (2020). DHARMa: Residual Diagnostics for Hierarchical (Multi-Level / Mixed) Regression Models. R package version 0.3.1. . Retrieved from https://CRAN.R-project.org/package=DHARMa Heatta, M. J. (2009). The mating strategy of female Svalbard reindeer. Universitetet i Tromsø, Heitor, F., & Vicente, L. (2008). Maternal care and foal social relationships in a herd of Sorraia horses: Influence of maternal rank and experience. Applied Animal Behaviour Science, 113(1), 189-205. doi:https://doi.org/10.1016/j.applanim.2007.11.005 Hejcmanová, P., Jůnková Vymyslická, P., Koláčková, K., Antonínová, M., Havlíková, B., Stejskalová, M., . . . Hejcman, M. (2010). Suckling behavior of eland antelopes (Taurotragus spp.) under semi-captive and farm conditions. Journal of Ethology, 29, 161-168. doi:10.1007/s10164-010-0241-1 Henriksen, S., Aanes, R., Sæther, B.-E., Ringsby, T. H., & Tufto, J. (2003). Does availability of resources influence grazing strategies in female Svalbard reindeer? Rangifer, 23(1), 25-37.

50

Hoffman, C. L., Higham, J. P., Mas-Rivera, A., Ayala, J. E., & Maestripieri, D. (2010). Terminal investment and senescence in rhesus macaques (Macaca mulatta) on Cayo Santiago. Behavioral Ecology, 21(5), 972-978. Holand, Ø., Gjøstein, H., Losvar, A., Kumpula, J., Smith, M., Røed, K., . . . Weladji, R. (2004). Social rank in female reindeer (Rangifer tarandus): effects of body mass, antler size and age. Journal of Zoology, 263(4), 365-372. Hornsund research station (24.8.2020). [Facebook]. https://www.facebook.com/hornsund/posts/4249755668429548 Accessed on 1.10.2020 Howard, R. D. (1979). Estimating Reproductive Success in Natural Populations. The American Naturalist, 114(2), 221-231. Hurvich, C., & Tsai, C. (1989). Regression and time series model selection in small samples. Biometrika, 76(2), 297–307. Jenkins, S. H. (2002). Data pooling and type I errors: a comment on Leger & Didrichsons. Animal Behaviour, 63(2), F9-F11. doi:https://doi.org/10.1006/anbe.2001.1913 Jepsen, J. U., Eide, N. E., Prestrud, P., & Jacobsen, L. B. (2002). The importance of prey distribution in habitat use by arctic foxes (Alopex lagopus). Canadian Journal of Zoology, 80(3), 418-429. Jingfors, K., Miller, F., Glaholt, R., & Gunn, A. (1983). Caribou Behaviour, Range Use Patterns and Short Term Responses to Helicopter Landings on the Beverly Calving Ground, N.W.T., 1982. Karbø, A. K. (2019). Linking behavior to diet in Svalbard reindeer (Rangifer tarandus platyrhynchus) by use of DNA metabarcoding and GPS- telemetry. Norwegian University of Life Sciences, Ås, Kay, J. E., Holland, M. M., & Jahn, A. (2011). Inter-annual to multi-decadal Arctic sea ice extent trends in a warming world. Geophysical research letters, 38(15), n/a. doi:10.1029/2011gl048008 Kirkwood, T. B. L., & Austad, S. N. (2000). Why do we age? Nature, 408(6809), 233-238. doi:10.1038/35041682 Kjellander, P., Gaillard, J., Hewison, M., & Liberg, O. (2004). Predation risk and longevity influence variation in fitness of female roe deer (Capreolus capreolus L.). Proceedings of the Royal Society of London. Series B: Biological Sciences, 271(suppl_5), S338-S340. Knott, K., Barboza, P., & Bowyer, R. (2005). Growth in arctic ungulates: Postnatal development and organ maturation in Rangifer tarandus and Ovibos moschatus. Journal of Mammalogy - J MAMMAL, 86, 121-130. doi:10.1644/1545-1542(2005)086<0121:GIAUPD>2.0.CO;2 Kojola, I. (1989). Mother's dominance status and differential investment in reindeer calves. Animal Behaviour, 38(2), 177-185. Kojola, I. (1990). Intraherd spacing behaviour of female reindeer: Effects of kinship, age and habituation. Applied Animal Behaviour Science, 26(1-2), 41-47. Kojola, I. (1993). Early maternal investment and growth in reindeer. Canadian Journal of Zoology, 71(4), 753-758. Kojola, I., & Eloranta, E. (1989). Influences of maternal body weight, age, and parity on sex ratio in semidomesticated reindeer (Rangifer t. tarandus). Evolution, 43(6), 1331-1336. Kojola, I., & Helle, T. (1996). Size-related changes in winter condition of female and male reindeer calves. Canadian Journal of Zoology, 74(6), 1174-1177. Kojola, I., Helle, T., Huhta, E., & Niva, A. (1998). Foraging conditions, tooth wear and herbivore body reserves: a study of female reindeer. Oecologia, 117(1-2), 26-30. Kojola, I., & Nieminen, M. (1988). Aggression and nearest neighbour distances in female reindeer during the rut. Ethology, 77(3), 217-224. Kvie, K. S., Heggenes, J., Anderson, D. G., Kholodova, M. V., Sipko, T., Mizin, I., & Røed, K. H. (2016). Colonizing the High Arctic: Mitochondrial DNA Reveals Common Origin of Eurasian Archipelagic Reindeer (Rangifer tarandus). PLOS ONE, 11(11), e0165237. doi:10.1371/journal.pone.0165237 Le Moullec, M., Pedersen, Å. Ø., Stien, A., Rosvold, J., & Hansen, B. B. (2019). A century of conservation: The ongoing recovery of Svalbard reindeer. The Journal of Wildlife Management, 83(8), 1676-1686. doi:10.1002/jwmg.21761 Lee, A. M., Bjørkvoll, E. M., Hansen, B. B., Albon, S. D., Stien, A., Sæther, B. E., . . . Grøtan, V. (2015). An integrated population model for a long‐ lived ungulate: more efficient data use with Bayesian methods. Oikos, 124(6), 806-816. Lent, P. C. (1966). Calving and Related Social Behavior in the Barren-Ground Caribou. Zeitschrift für Tierpsychologie, 23(6), 701-756. doi:10.1111/j.1439-0310.1966.tb01707.x Lenvik, D., Bø, E., & Fjellheim, A. (1988). Relationship between the weight of reindeer calves in autumn and their mother's age and weight in the previous spring. Rangifer, 8(1), 20-24. Loe, L. E. (2019, 16.11.). [ID numbers and birth dates of reindeer ]. Loe, L. E., Bonenfant, C., Mysterud, A., Severinsen, T., Øritsland, N. A., Langvatn, R., . . . Stenseth, N. C. (2007). Activity pattern of arctic reindeer in a predator-free environment: no need to keep a daily rhythm. Oecologia, 152(4), 617-624. Loe, L. E., Hansen, B. B., Stien, A., D. Albon, S., Bischof, R., Carlsson, A., . . . Ropstad, E. (2016). Behavioral buffering of extreme weather events in a high‐Arctic herbivore. Ecosphere, 7(6), e01374. Loe, L. E., Irvine, R. J., Bonenfant, C., Stien, A., Langvatn, R., Albon, S. D., . . . Stenseth, N. C. (2006). Testing five hypotheses of sexual segregation in an arctic ungulate. Journal of Animal Ecology, 75(2), 485-496. Loe, L. E., Pigeon, G., Albon, S. D., Giske, P. E., Irvine, R. J., Ropstad, E., . . . Mysterud, A. (2019). Antler growth as a cost of reproduction in female reindeer. Oecologia, 189(3), 601-609. doi:10.1007/s00442-019-04347-7 Loison, A., & Strand, O. (2005). Allometry and variability of resource allocation to reproduction in a wild reindeer population. Behavioral Ecology, 16(3), 624-633. Lüdecke, Makowski, Waggoner, & Patil. (2020). Assessment of Regression Models Performance. : CRAN. Retrieved from https://easystats.github.io/performance/ Machlis, L., Dodd, P. W. D., & Fentress, J. C. (1985). The Pooling Fallacy: Problems Arising when Individuals Contribute More than One Observation to the Data Set. Zeitschrift für Tierpsychologie, 68(3), 201-214. doi:10.1111/j.1439-0310.1985.tb00124.x Mazerolle, M. J. (2019). AICcmodavg: Model selection and multimodel inference based on (Q)AIC(c). (Version R package version 2.2-2 ). Retrieved from https://cran.r-project.org/package=AICcmodavg McCullough, D. R. (1999). Density dependence and life-history strategies of ungulates. Journal of Mammalogy, 80(4), 1130-1146. doi:10.2307/1383164 McGraw, J. B., & Caswell, H. (1996). Estimation of Individual Fitness from Life-History Data. The American Naturalist, 147(1), 47-64. Meland, M. (2014). Partial migration as a response to ground icing events in a high arctic ungulate. Norwegian University of Life Sciences, Ås, Mery, F., & Burns, J. G. (2009). Behavioural plasticity: an interaction between evolution and experience. Evolutionary ecology, 24(3), 571-583. doi:10.1007/s10682-009-9336-y Meyer‐Holzapfel, M., & Frisch, K. (1956). Über die Bereitschaft zu Spiel‐ und Instinkthandlungen. Zeitschrift für Tierpsychologie, 13(3), 442-462. doi:10.1111/j.1439-0310.1956.tb01678.x Microsoft. (2009). Microsoft Excel. Mirza, S. N., & Provenza, F. D. (1992). Effects of age and conditions of exposure on maternally mediated food selection by lambs. Applied Animal Behaviour Science, 33(1), 35-42. Morris, W. F., Pfister, C. A., Tuljapurkar, S., Haridas, C. V., Boggs, C. L., Boyce, M. S., . . . Menges, E. S. (2008). LONGEVITY CAN BUFFER PLANT AND ANIMAL POPULATIONS AGAINST CHANGING CLIMATIC VARIABILITY. Ecology (Durham), 89(1), 19-25. doi:10.1890/07-0774.1

51

Müller-Schwarze, D., Källquist, L., & Mossing, T. (1979). Social behavior and chemical communication in reindeer (Rangifer t. tarandus L.). Journal of Chemical Ecology, 5(4), 483-517. Mysterud, A., Yoccoz, N. G., Stenseth, N. C., & Langvatn, R. (2001). Effects of age, sex and density on body weight of Norwegian red deer: evidence of density-dependent senescence. Proceedings. Biological sciences, 268(1470), 911-919. doi:10.1098/rspb.2001.1585 Mörschel, F. M., & Klein, D. R. (1997). Effects of weather and parasitic insects on behavior and group dynamics of caribou of the Delta Herd, Alaska. Canadian Journal of Zoology, 75(10), 1659-1670. doi:10.1139/z97-793 Nedberg, T. (2012). The effect of ground icing events on Svalbard reindeer (Rangifer tarandus platyrhynchus) space use and habitat selection. Norwegian University of Life Sciences, Ås, Norwegian Meteorological Institute and the Norwegian Broadcasting Corporation (2007-2020). Semmeldalen. Retrieved from https://www.yr.no/en/statistics/graph/2-2728888/Norway/Svalbard/Semmeldalen Norwegian Polar Institute (2019, 20.6.2019). Svalbard reindeer (Rangifer tarandus platyrhynchus). Retrieved from http://www.mosj.no/en/fauna/terrestrial/svalbard-reindeer-population.html Norwegian Polar Institute n.d. TopoSvalbard basemap. Retrieved from https://toposvalbard.npolar.no/ on 15.11.2019 Omsjoe, E., Stien, A., Irvine, J., Albon, S., Dahl, E., Thoresen, S., . . . Ropstad, E. (2009). Evaluating capture stress and its effects on reproductive success in Svalbard reindeer. Canadian Journal of Zoology, 87(1), 73-85. Ortega-Reyes, L., & Provenza, F. D. (1993). Amount of experience and age affect the development of foraging skills of goats browsing blackbrush (Coleogyne ramosissima). Applied Animal Behaviour Science, 36(2-3), 169-183. Paine, R. (1988). Reindeer and caribou Rangifer tarandus in the wild and under pastoralism. Polar Record, 24(148), 31-42. Palko, K. (2019, 31.7.2019 ). 200 Reindeer Starved to Death. Experts Call It a Sign of Climate Change., Online article. New York Times Retrieved from https://www.nytimes.com/2019/07/31/world/europe/norway-reindeer-climate-change.html Paulsen, I. M. G. (2018). Calving site characteristics and habitat use in Svalbard reindeer. UiT Norges arktiske universitet, Petersson, C., & Danell, B. (1993). Causes of variation in growth rate of reindeer calves. Rangifer, 13(2), 105-116. Pianka, E. R., & Parker, W. S. (1975). Age-Specific Reproductive Tactics. The American Naturalist, 109(968), 453-464. Post, E., & Forchhammer, M. C. (2007). Climate change reduces reproductive success of an Arctic herbivore through trophic mismatch. Philosophical transactions. Biological sciences, 363(1501), 2367-2373. doi:10.1098/rstb.2007.2207 Prestrud, P. (1992). Food habits and observations of the hunting behaviour of Arctic foxes, Alopex lagopus, in Svalbard. Canadian field-naturalist. Ottawa ON, 106(2), 225-236. Pruitt, W. J. (1960). Behavior of the Barren-Ground Caribou. Retrieved from http://hdl.handle.net/11122/1425 Ralls, K., Lundrigan, B., & Kranz, K. (1987). Mother‐young relationships in captive ungulates: Behavioral changes over time. Ethology, 75(1), 1-14. Rands, S. A., Muir, H., & Terry, N. L. (2014). Red deer synchronise their activity with close neighbours. PeerJ, 2, e344-e344. doi:10.7717/peerj.344 Ranganathan, P., Pramesh, C. S., & Aggarwal, R. (2017). Common pitfalls in statistical analysis: Measures of agreement. Perspectives in Clinical Research, 8, 187. doi:10.4103/picr.PICR_123_17 R Core Team (2019). R: A language and environment for statistical computing. Vienna, Austria: R Foundation for Statistical Computing. Retrieved from https://www.R-project.org/ Reimers, E. (1977). Population dynamics in two subpopulations of reindeer in Svalbard. Arctic and Alpine Research, 9(4), 369-381. Reimers, E., & Eftestøl, S. (2012). Response behaviors of Svalbard reindeer towards humans and humans disguised as polar bears on Edgeøya. Arctic, antarctic, and alpine research, 44(4), 483-489. Reimers, E., Holmengen, N., & Mysterud, A. (2005). Life-history variation of wild reindeer (Rangifer tarandus) in the highly productive North Ottadalen region, Norway. Journal of Zoology, 265(1), 53-62. Reimers, E., Loe, L. E., Eftestøl, S., Colman, J. E., & Dahle, B. (2009). Effects of hunting on response behaviors of wild reindeer. The Journal of Wildlife Management, 73(6), 844-851. Reimers, E., Lund, S., & Ergon, T. (2011). Vigilance and fright behaviour in the insular Svalbard reindeer (Rangifer tarandus platyrhynchus). Canadian Journal of Zoology, 89(8), 753-764. Reimers, E., Røed, K., & Colman, J. (2012). Persistence of vigilance and flight response behaviour in wild reindeer with varying domestic ancestry. Journal of Evolutionary Biology, 25(8), 1543-1554. Ringberg, T. (1979). The Spitzbergen reindeer‐a winter‐dormant ungulate? Acta Physiologica Scandinavica, 105(3), 268-273. Roff, D. (1993). The Evolution of Life Histories (Vol. xii). New York: Chapman and Hall. Ropstad, E. (2000). Reproduction in female reindeer. Animal Reproduction Science, 60, 561-570. Rutberg, A. T. (1983). Factors Influencing Dominance Status in American Bison Cows (Bison bison). Zeitschrift für Tierpsychologie, 63(2‐3), 206- 212. doi:10.1111/j.1439-0310.1983.tb00087.x Rönnegård, L., Forslund, P., & Danell, Ö. (2002). Lifetime patterns in adult female mass, reproduction, and offspring mass in semidomestic reindeer (Rangifer tarandus tarandus). Canadian Journal of Zoology, 80(12), 2047-2055. Sinclair, A. R. E. (1985). Does Interspecific Competition or Predation Shape the African Ungulate Community? Journal of Animal Ecology, 54(3), 899-918. doi:10.2307/4386 Skogland, T. (1984). The effects of food and maternal conditions in fetal growth and size in wild reindeer. Rangifer, 4(2), 39-46. Skogland, T. (1988). Tooth wear by food limitation and its life history consequences in wild reindeer. Oikos, 238-242. Skogland, T. (1990). Density dependence in a fluctuating wild reindeer herd; maternal vs. offspring effects. Oecologia, 84(4), 442-450. Skogland, T. (1991). What are the effects of predators on large ungulate populations? Oikos, 401-411. Soulliere, C. E. (2008). Anti-predator behaviour of maternal caribou (Rangifer tarandus) in west-central Newfoundland. Memorial University of Newfoundland, Stearns, S. C. (1989). Trade-offs in life-history evolution. Functional ecology, 3(3), 259-268. Stearns, S. C. (1992). The Evolution of Life Histories: OUP Oxford. Stearns, S. C. (2000). Life history evolution: successes, limitations, and prospects. Naturwissenschaften, 87(11), 476-486. Stempniewicz, L., Kidawa, D., Barcikowski, M., & Iliszko, L. (2014). Unusual hunting and feeding behaviour of polar bears on Spitsbergen. Polar Record, 50(2), 216-219. doi:10.1017/S0032247413000053 Stien, A., Irvine, R., Ropstad, E., Halvorsen, O., Langvatn, R., & Albon, S. (2002). The impact of gastrointestinal nematodes on wild reindeer: experimental and cross‐sectional studies. Journal of Animal Ecology, 71(6), 937-945. Stien, A., Loe, L. E., Mysterud, A., Severinsen, T., Kohler, J., & Langvatn, R. (2010). Icing events trigger range displacement in a high‐arctic ungulate. Ecology, 91(3), 915-920. Taillon, J., Brodeur, V., Festa-Bianchet, M., & Côté, S. (2012). Is mother condition related to offspring condition in migratory caribou (Rangifer tarandus) at calving and weaning? Canadian Journal of Zoology, 90(3), 393-402. Tandberg, A. (2016). Svalbardrein (Rangifer tarandus platyrhynchus) og snøskuterløyper: kvantifisering av unnvikelsesadferd i tid og rom. Norwegian University of Life Sciences, Ås, Thomson, B. R. (1977). The Behaviour of Wild Reindeer in Norway. (PhD), University of Edinburgh, Retrieved from https://onlinelibrary.wiley.com/doi/abs/10.1111/j.1468-2230.1978.tb00792.x

52

Thórhallsdóttir, A. G., Provenza, F. D., & Balph, D. (1990). The role of the mother in the intake of harmful foods by lambs. Applied Animal Behaviour Science, 25(1-2), 35-44. Townsend, T. W., & Bailey, E. D. (1981). Effects of Age, Sex and Weight on Social Rank in Penned White-tailed Deer. The American Midland Naturalist, 106(1), 92-101. doi:10.2307/2425138 Trivers, R. L. (1974). Parent-offspring conflict. Integrative and Comparative Biology, 14(1), 249-264. Tyler, N., & Øritsland, N. (1989). Why don't Svalbard reindeer migrate? Ecography, 12(4), 369-376. Van der Knaap, W. (1989). Past vegetation and reindeer on Edgeoya (Spitsbergen) between c. 7900 and c. 3800 BP, studied by means of peat layers and reindeer faecal pellets. Journal of Biogeography, 379-394. Van der Wal, R., Brooker, R., Cooper, E., & Langvatn, R. (2001). Differential effects of reindeer on high Arctic lichens. Journal of Vegetation Science, 12(5), 705-710. Van der Wal, R., Madan, N., Lieshout, S. v., Dormann, C., Langvatn, R., & Albon, S. D. (2000). Trading Forage Quality for Quantity? Plant Phenology and Patch Choice by Svalbard Reindeer. Oecologia, 123(1), 108-115. Veiberg, V., Mysterud, A., Bjørkvoll, E., Langvatn, R., Loe, L. E., Irvine, R. J., . . . Stenseth, N. C. (2007). Evidence for a trade‐off between early growth and tooth wear in Svalbard reindeer. Journal of Animal Ecology, 76(6), 1139-1148. Ver Hoef, J., & Boveng, P. (2007). Quasi-Poisson vs. negative binomial regression: How should we model overdispersed count data? Ecology, 88, 2766-2772. doi:10.1890/07-0043.1 Verme, L. J. (1989). Maternal investment in white-tailed deer. Journal of Mammalogy, 70(2), 438-442. Vors, L. S., & Boyce, M. S. (2009). Global declines of caribou and reindeer. Global Change Biology, 15(11), 2626-2633. Walter, A., Thomas, R., Leif Egil, L., Irvine, R. J., Erik, R., Vebjørn, V., & Steve, D. A. (2018). Circadian rhythmicity persists through the Polar night and midnight sun in Svalbard reindeer. Scientific Reports, 8(1), 1-12. doi:10.1038/s41598-018-32778-4 Weladji, R., & Holand, Ø. (2003a). Global Climate Change and Reindeer: Effects of Winter Weather on the Autumn Weight and Growth of Calves. Oecologia, 136(2), 317-323. Weladji, R. B., Gaillard, J.-M., Yoccoz, N. G., Holand, Ø., Mysterud, A., Loison, A., . . . Stenseth, N. C. (2006). Good reindeer mothers live longer and become better in raising offspring. Proceedings of the Royal Society B: Biological Sciences, 273(1591), 1239-1244. Weladji, R. B., & Holand, Ø. (2003b). Global climate change and reindeer: effects of winter weather on the autumn weight and growth of calves. Oecologia, 136(2), 317-323. Weladji, R. B., Holand, Ø., Gaillard, J.-M., Yoccoz, N. G., Mysterud, A., Nieminen, M., & Stenseth, N. C. (2010). Age-specific changes in different components of reproductive output in female reindeer: terminal allocation or senescence? Oecologia, 162(1), 261-271. Weladji, R. B., Loison, A., Gaillard, J.-M., Holand, Ø., Mysterud, A., Yoccoz, N. G., . . . Stenseth, N. C. (2008). Heterogeneity in individual quality overrides costs of reproduction in female reindeer. Oecologia, 156(1), 237-247. Weladji, R. B., Mysterud, A., Holand, Ø., & Lenvik, D. (2002). Age-related reproductive effort in reindeer (Rangifer tarandus): evidence of senescence. Oecologia, 131(1), 79-82. Whitehead, H., & Dufault, S. (1999). Techniques for analyzing vertebrate social structure using identified individuals. Adv Stud Behav, 28, 3-74. Wickström, S., Jonassen, M., Cassano, J., & Vihma, T. (2020). Present Temperature, Precipitation and Rain‐on‐Snow Climate in Svalbard. Journal of Geophysical Research: Atmospheres, 125. doi:10.1029/2019JD032155 Witter, L. A., Johnson, C. J., Croft, B., Gunn, A., & Gillingham, M. P. (2012). Behavioural trade-offs in response to external stimuli: time allocation of an Arctic ungulate during varying intensities of harassment by parasitic flies. Journal of Animal Ecology, 81(1), 284-295. doi:10.1111/j.1365-2656.2011.01905.x Zuur, A. F., & Ieno, E. N. (2016). A protocol for conducting and presenting results of regression-type analyses. Methods in Ecology and Evolution, 7(6), 636-645. doi:10.1111/2041-210x.12577

53

8 Supplementary material

Supplementary Table 1 Observed females, their age, reproductive status (calf at heel or no calf), number of different days individual was observed, number of focal watches, and numbers of instant scans per month. In addition, males 303R (7 years) and 375R (2 years) were both sampled for one focal watch for vigilance.

OBSERVATION FOCAL INSTANT SCANS INSTANT SCANS ID AGE CALF DAYS WATCHES JULY AUGUST 344R 2 3 66 162G 3 1 23 178W 4 2 1 37 216W 4 1 36 220W 4 x 3 2 83 38 181B 5 3 1 57 198B 5 1 17 203B 5 x 3 1 23 113 213B 5 x 5 1 45 51 140Y 6 x 1 1 36 162Y 6 x 1 42 163Y 6 x 3 2 67 41 169Y 6 x 1 1 23 205Y 6 x 1 1 13 117-74G 8 2 61 123-66G 8 x 2 33 291R 8 x 1 15 162B 10 x 6 4 100 58 164B 10 x 4 1 23 79 129Y 11 4 1 109 92B 15 2 30

Control 1 NA x 1 1 26 Control 2 NA x 1 1 26 Control 3 NA x 1 1 30 Control 4 NA x 1 1 22 Control 5 NA x 1 1 16 Control 6 NA x 1 9 Control 7 NA x 1 14 Control 8 NA x 1 25 Control 9 NA x 1 16 Control 10 NA x 1 10 Control 11 NA x 1 7 Control 12 NA x 1 1 15 Control 13 NA x 1 1 16 Control 14 NA x 1 1 42

54

Supplementary Table 2 Preliminary behavioral categories and sub-categories on the basis on existing literature. * peer-reviewed article, ** field report, *** MSc thesis

Behavioral Scan categories Feeding Lying Standing Walking Running Other interval

Sub-categories feeding still lying head up, standing walkin g trotting other 10 min feeding walk lying head galloping down, lying flat

Flydal, feeding lying head up, standing walking running NA 10 min Eftestøl, lying head Reimers & down Colman 2004*

Colman et al. feeding lying standing walking running NA 15 min 2003*

Hagemoen & feeding lying standing walking running other 15 min Reimers 2002* / other

Collins & Smith feeding bedding standing walking trotting cratering 15 min 1989*

Bøving & Post feeding lying head up, NA NA NA searching 15 min 1997* lying head down, lying flat

Witter et al. feeding lying standing / walking running insect 30 min 2012* other avoidanc e

Flydal et al. feeding still lying head up, standing walking running NA 10 min 2009* feeding slow lying head walk down

Jingfors et al. foraging bedded standing walking trotting, NA 20 min 1983** galloping

Colman et al. feeding lying standing walking running annoyance 15 min 2001* behaviors

Soulliere foraging lying head up, standing walking NA other, 15 min 2008*** lying head (inter- cow-calf (includes more down, ruptive & -interaction, categories than lying flat non-inter- aggression, listed) ruptive) insect avoidance, etc.

55

Supplementary Table 3 Hypotheses with each response variable used in models for hypothesis testing.

Hypothesis Response variable Model H1a Older mothers keep Distance between the focal mother and its calf in instant scans in body lengths (bl) M1a closer to their calves Time the focal mother spends close (<5 bl) to calf during focal watches (s) M1b Time the focal mother spends far (>20 bl) from calf during focal watches (s) M1c H1b Older mothers are more Proportion of instant scans of the focal mother and calf in same behavioral category M2 often involved in the same activity as their calf H1c No age-related Length of all recorded successful nursing bouts (s) M3 differences in nursing bout lengths H1d Older females are less Proportion of instant scans of the focal female showing vigilant behavior M4a vigilant Time the focal reindeer shows vigilant behavior during focal watches (s) M4b H1e Older females are Distance between the focal female and its nearest neighbor in instant scans (bl) M5 further away from other Proportion of instant scans of the focal female in a group versus alone M6 reindeer than younger Number of reindeer (excluding own calf) in the focal female’s group in instant scans M7 females H1f Older and younger Proportion of instant scans of the focal female feeding M8a females spend Proportion of instant scans of the focal female laying down M8b proportionately more Proportion of instant scans of the focal female standing M8c time feeding and less time standing, moving, or laying down compared to prime age mothers H2a Females with calves are Proportion of instant scans of the focal female showing vigilant behavior M4a more vigilant Time the focal reindeer shows vigilant behavior during focal watches (s) M4b H2b Females with calves are Distance between the focal female and its nearest neighbor in instant scans (bl) M5 socially more segregated Proportion of instant scans of the focal female in a group versus alone M6 from other reindeer Number of reindeer (excluding own calf) in the focal female’s group in instant scans M7 H2c Females with calves Proportion of instant scans of the focal female feeding M8a spend proportionately Proportion of instant scans of the focal female laying down M8b more time feeding and Proportion of instant scans of the focal female in other activity state M8c less time standing, moving, or laying down than females without

56

Supplementary Table 4. Effect of age and calf at heel to the maternal, vigilant and social behavior and time budget of Svalbard reindeer females in summer. Table includes effect estimates with p-values and details about the model specification. Significant and near-significant effects are highlighted. N.S. = non-significant effect, in case of Month it was not included in the model based on preliminary analysis including control group, and in case of Quadratic effect of age or Interaction terms they were dropped out from final model in order to interpret the main effects. In the presence of interaction, the significant main effects are not highlighted. *Explanation of response variables: NB = nursing bout length in seconds, CD = distance of the focal female to its calf in body lengths in instant scans, CD-FC = proportion of time spent closer than 5 body lengths to own calf in focal scans, CD-FF = proportion of time spent further than 20 body lengths in focal scans, SYNC = proportion of instant scans of focal female with same activity as its calf, VIG = proportion of vigilant instant scans, VIG-FM = proportion of vigilant time in focal scans, NN = distance of the focal female to its nearest neighbor in body lengths in instant scans, GS = group size of focal animal’s group in instant scans, IG = proportion of instant scans of focal female in a group, TB-L = proportion of instant scans of focal female laying down, TB-F = proportion of instant scans of focal female feeding, TB-S = proportion of instant scans of focal female standing, TB-M = proportion of instant scans of focal female walking or running. // ** Observation level random effect was used to account for overdispersion in the models with Poisson distribution. Overdispersion parameter was used with Negative binomial and Betabinomial distributions. // *** Lags mean the number of scans that have been averaged into one observation in order to avoid temporal autocorrelation. // **** Distribution of the response variable: P = Poisson, NB = Negative binomial, B = Binomial, BB = Betabinomial.

Maternal behavior Vigilance Social behavior Time budget Model M1a M1b M1c M2 M3 M4a M4b M5 M6 M7 M8a M8b M8c Response variable* CD CD-FC CD-FF NB SYNC VIG VIG-FM NN IG GS TB-L TB-F TB-S Effects of independent variables Effect of Calf at heel -0.44 0.61 0.22 -0.65 -1.96 -0.37 0.13 0.48 <.07 <.01 <.01 <.29 <.001 <.05 <.32 <.08 Effect of Month (July) -1.16 5.96 -7.70 -1.11 -1.40 -0.35 4.38 0.40 -0.46 1.07 n.s. n.s. n.s. <.01 <.10 <.06 <.24 <.001 <.001 <.01 <.14 <.03 <.01 Effect of Age -0.02 0.23 -0.29 <0.01 -0.99 -0.25 -0.07 <0.01 0.43 -0.08 -0.03 0.03 0.01 <.80 <.32 <.40 <.93 <.04 <.07 <.08 <.82 <.03 <.02 <.30 <.50 <.73 Quadratic effect of Age 0.07 n.s. n.s. n.s. n.s. n.s. n.s. n.s. n.s. n.s. n.s. n.s. n.s. <.04 Interaction effect Age:Calf 0.21 n.s. n.s. n.s. n.s. n.s. n.s. n.s. <.001 Interaction effect Age:Month 0.12 -0.85 1.16 0.26 -0.50 n.s. n.s. n.s. n.s. n.s. n.s. n.s. n.s. <.05 <.06 <.03 <.06 <.02 Number of groups in random effects Random effect I: Day 17 - - - 17 17 - 17 17 17 17 17 17 Random effect II: Individual 12 9 9 7 12 21 14 21 21 21 21 21 21 Observation level random effect** 239 ------260 - - - - - Information about the model Number of observations 239 14 14 12 830 1297 17 260 50 168 410 410 410 Lags*** 3 - - - 1 1 - 3 1d 7 3 3 3 Distribution **** P BB BB P B B BB P BB NB BB BB B Model validation Model AICc 1504.20 177.52 136.24 93.69 1001.57 736.23 152.16 1349.19 209.94 882.78 1014.07 1115.07 484.73 Null model AICc 1505.07 165.56 125.93 90.03 1001.35 737.67 159.86 1371.05 209.79 893.28 1014.98 1118.46 488.22 Delta AICc 0.87 11.96 10.31 3.66 0.22 1.44 7.7 21.86 0.15 10.5 0.91 3.39 3.48 AICc Weights 0.61 0 0.01 0.14 0.47 0.67 0.98 1 1 1 0.61 0.84 0.85 Pseudo R2m 0.069 - - <0.01 0.01 0.08 - 0.22 - 0.51 - - 0.14 Pseudo R2c 0.87 - - 0.33 0.02 0.13 - 0.45 - 0.95 - - 0.23

Supplementary Table 5 Effect of marking and the observation month (July), on the response variables which were tested with the control group. All control models passed dHARMA diagnostic tests. N.S. = not significant. * In addition to response variables, the ‘other’ category of time budget was split up and proportion of time standing and moving were tested separately.

Corresponding model & response variable Effect of month Effect of markings M1a Distance between the focal mother and its calf in B=-0.32 ±0.16 N.S. instant scans in body lengths p<.05 M2 Proportion of instant scans of the focal mother and N.S. N.S. calf in same behavioral category M3 Length of all recorded successful nursing bouts N.S. N.S. M4a M4a Proportion of instant scans of the focal female B=0.5 ±0.28 N.S. showing vigilant behavior p<.05 . M4b M4b Time the focal reindeer shows vigilant behavior B=0.81 ±0.17 N.S. during focal watches p<.001 M5 M5 Distance between the focal female and its nearest B=0.25 ±0.11 N.S. neighbor in instant scans (bl) p<.03 M6 Proportion of instant scans of the focal female in a N.S. N.S. group versus alone M7 Number of reindeer (excluding own calf) in the focal N.S. N.S. female’s group in instant scans M8a Proportion of instant scans of the focal female lying down N.S. N.S.

M8b Proportion of instant scans of the focal female feeding B=0.40 ±0.17 N.S. p<.02 M8c Proportion of instant scans of the focal female in other activity B=0.79 ±0.25 N.S. p<.01 Standing* N.S. B=-0.40 ±0.16, p<.02 Moving* B=1.31 ±0.43 N.S. p<.01

58