Evaluation of Maternal Penning to Improve Calf Survival in the Chisana Caribou Herd
Évaluation des Enclos de Maternité pour Améliorer la Survie des Faons du Troupeau de Caribous Chisana
Deceased.
Present address: Oakley Veterinary Services, P. O. Box 1157, Haines, AK 99827, USA.
Retired; present address: 114 Wilson Drive, Whitehorse, YT Y0B 1L0, Canada.
Retired; present address: 1504 Driftwood Drive, Bozeman, MT 59715, USA.
Present address: Alaska Department of Fish and Game, Division of Wildlife Conservation, 802 3rd Street, Douglas, AK 98824, USA.
Retired; present address: Yukon College, Box 10038, Whitehorse, YT Y1A 7A1, Canada.
ABSTRACT
enPredation is a major limiting factor for most small sedentary caribou (Rangifer tarandus) populations, particularly those that are threatened or endangered across the southern extent of the species’ range. Thus, reducing predation impacts is often a management goal for improving the status of small caribou populations, and lethal predator removal is the primary approach that has been applied. Given that predator control programs are often contentious, other management options that can garner broader public acceptance need to be considered.
Substantial calf losses to predation in the few weeks following birth are common for these small caribou populations. Therefore, we employed a novel experimental approach of maternal penning with the goal of reducing early calf mortality in the Chisana Caribou Herd, a declining population in southwest Yukon and adjacent Alaska thought to number around 300 individuals. Maternal penning entailed temporarily holding pregnant females on their native range in a large pen secure from predators from late March through the initial weeks of calf rearing to mid-June. During 2003–2006, we conducted 4 annual penning trials with 17–50 pregnant females each year (n = 146 total), assessed survival of calves born in the pens, and evaluated survival and nutritional effects of penning for females that were held. We also investigated the herd's population dynamics during 2003–2008 to determine effects of maternal penning on calf recruitment and population growth. In addition to information gained during maternal penning, we determined natality and survival patterns via radiotelemetry, conducted autumn age-sex composition surveys each year, and censused the population in mid-October 2003, 2005, and 2007. Based on our penning trials and demographic investigations, we used simulation models to evaluate the effects of maternal penning relative to a population's inherent growth rate (finite rate of increase [λ] without maternal penning) and penning effort (proportion of calves born in penning) to provide perspective on utility of this approach for improving the status of small imperiled caribou populations.
Pregnant females held in maternal penning tolerated captivity well in that they exhibited positive nutritional responses to ad libitum feed we provided and higher survival than free-ranging females (0.993 and 0.951 for penned and free-ranging females, respectively). Survival of pen calves from birth to mid-June was substantially higher than that of free-ranging calves ( = 0.950 and 0.376, respectively). This initial period accounted for 76% of the annual calf mortality in the free-ranging population. Pen-born calves maintained their survival advantage over wild-born calves to the end of their first year ( = 0.575 and 0.192, respectively) during years penning occurred.
Females in the Chisana Herd were highly productive with 57% producing their first offspring at 2 years of age, and annual natality rates averaging 0.842 calves/female ≥2 years old. Age-specific natality rates exceeded 0.900 for 4–9-year-olds, then exhibited senescent decline to 0.467 by 19 years old. Annual survival of free-ranging adult females and calves averaged 0.892 and 0.184, respectively, over all study years; both were reduced during 2004 because of poor winter survival. We noted reduced nutritional condition of caribou late that winter in that females we captured were lighter than in other years and produced lighter calves. We suspect that the reduced survival during winter 2004 and the observed nutritional characteristics resulted from adverse snow conditions in combination with effects of the extreme drought experienced the previous summer. Age-specific survival of adult females was ≥0.900 through 10 years of age, then declined with age.
The Chisana Herd numbered 720 caribou in mid-October 2003, or more than twice that estimated prior to initiating maternal penning, and increased to 766 caribou by mid-October 2007. We calculated that penning added 54.2 yearling recruits, or 40% of calves released from penning. Based on the maternal penning results and the population's vital rates, we determined that the herd would have been stable during 2003–2007 at about 713 caribou without maternal penning; thus, the increase in herd size we observed resulted from maternal penning and was equivalent to the estimate of additional yearling recruits. The improvement in the population trend invoked by maternal penning was limited by the larger than expected population size and resulting low penning effort ( = 11% of calves born in pen).
Our simulations corroborated that maternal penning increased population size by the number of additional recruits provided, even at low penning effort, for inherently stable populations. As the inherent rate of increase dropped below λ = 1.000, more of the additional recruits from penning were needed to offset the downward population inertia, thus requiring increased penning effort to reach stability. For populations declining at λ < 0.890, stability could not be achieved with 100% penning effort given the vital rates in our models.
Maternal penning in its limited application to date has proven to be broadly popular as a nonlethal management action aimed at reducing initial calf mortality from predation in small caribou populations. However, based on the Chisana program and 3 subsequent efforts elsewhere, improvement in population trends have been modest at best and come at a high financial cost. Given the necessity of maximizing penning effort, maternal penning may have a role in addressing conservation challenges for some small caribou populations that are stable or slowly declining, but its application should be primarily driven by objective assessment of the likelihood of improving population trends rather than popularity relative to other management options.
RÉSUMÉ
esLa prédation est un facteur limitant important pour la plupart des petites populations sédentaires de caribou (Rangifer tarandus), et plus particulièrement pour celles qui sont menacées ou en danger d'extinction dans le sud de l'aire de répartition de l'espèce. Ainsi, la réduction des impacts de la prédation est souvent un objectif de gestion pour améliorer le statut des petites populations de caribous et le contrôle létal des prédateurs est l'approche qui a été principalement appliquée. Étant donné que les programmes de contrôle de prédateurs sont souvent critiqués, d'autres options de gestion pouvant recueillir davantage l'acceptation du public doivent être considérées.
Des mortalités substantielles des faons par la prédation surviennent dans les premières semaines après la naissance et sont communes pour ces petites populations de caribous. Pour cette raison, nous avons employé une nouvelle approche expérimentale d'enclos de maternité avec l'objectif de réduire la mortalité précoce des faons du troupeau de caribous Chisana, une population en déclin du sud-ouest du Yukon et de l'Alaska estimée à environ 300 individus. Les enclos de maternité impliquent la captivité temporaire de femelles gestantes dans des enclos excluant les prédateurs situés dans l'aire de distribution du troupeau de la fin mars jusqu’à la fin des premières semaines d’élevage des faons, à la mi-juin. De 2003 à 2006, nous avons conduit 4 essais annuels de mise en enclos avec 17–50 femelles gestantes chaque année (n = 146 au total) pour évaluer d'une part, la survie des faons nés dans les enclos et d'autre part, la survie et les effets nutritionnels des enclos sur les femelles qui étaient maintenues en captivité. Nous avons aussi étudié la dynamique de population du troupeau durant la période 2003-2008 pour déterminer les effets des enclos de maternité sur les taux de recrutement et de croissance de la population. En plus de l'information acquise pendant la période de captivité, nous avons déterminé les taux de natalité et les patrons de survie par radio-télémétrie, conduit des inventaires annuels de composition âge-sexe à l'automne et recensé la population à la mi-octobre en 2003, 2005 et 2007. En se basant sur nos expériences de mise en enclos et sur les inventaires démographiques, nous avons utilisé des modèles de simulations pour évaluer les effets des enclos de maternité sur le taux de croissance inhérent de la population (taux fini d'accroissement [λ] sans les enclos de maternité) et le succès des enclos de maternité (proportion de faons nés dans les enclos) pour mettre en perspective l'utilité de cette approche pour améliorer le statut des petites populations de caribou en péril.
Les femelles gestantes dans les enclos de maternité ont bien toléré la captivité, démontrant une réponse nutritionnelle positive à l'alimentation ad libitum fournie et un taux de survie plus élevée que celui des femelles en liberté (0.993 et 0.951 pour les femelles en enclos et les femelles en liberté, respectivement). La survie de la naissance à la mi-juin était substantiellement plus élevée pour les faons nés en enclos de maternité que pour celle des faons nés en liberté ( = 0.950 et 0.376, respectivement). Pour les faons nés en liberté, 76% de la mortalité annuelle survenait durant cette période. Les faons nés en enclos de maternité ont maintenu leur avantage de survie sur les faons nés en liberté à la fin de leur première année ( = 0.575 et 0.192, respectivement) durant les années pendant lesquelles l'expérience d'enclos de maternité s'est déroulée.
Les femelles du troupeau Chisana étaient très fécondes : 57 % étaient primipares à deux ans et les taux de natalité annuels étaient de 0.842 faon/femelle ≥2 ans en moyenne. Les taux de natalité spécifiques à l’âge dépassaient 0.900 pour la classe 4-9 ans et montraient un déclin dû à la sénescence jusqu’à atteindre 0.467 à 19 ans. La survie annuelle des femelles adultes et des faons en liberté était respectivement de 0.892 et 0.184 pour l'ensemble de la période d’étude; des taux qui étaient réduits en 2004 en raison de la faible survie hivernale. Nous avons constaté une diminution de la condition nutritive des caribous lors de cet hiver, alors que les femelles capturées avaient une masse inférieure et ont mis bas de plus petits faons comparativement aux autres années. Nous soupçonnons que la réduction de la survie hivernale en 2004 ainsi que les caractéristiques nutritionnelles observées ont été causées par des conditions de neige défavorables combinées à une sécheresse extrême survenue l’été précédent. Le taux de survie spécifique à l’âge des femelles adultes était de ≥0.900 jusqu’à l’âge de 10 ans, puis diminuait avec l’âge.
La taille du troupeau Chisana était estimée à 720 individus en 2003 à la mi-octobre, ce qui équivaut à plus de deux fois la taille estimée avant le début de la mise en place des enclos de maternité et a atteint 766 caribous à la mi-octobre 2007. Nos calculs montrent une contribution des enclos de maternité au recrutement de 54.2 caribous d'un an, ou 40% des faons provenaient des enclos de maternité. En se basant sur les résultats obtenus à la suite de la mise en place des enclos ainsi que sur les taux de croissances inhérents à la population, nous avons déterminé que la taille du troupeau se serait probablement stabilisée à environ 713 caribous sans les enclos de mise-bas ; ainsi, l'augmentation de la taille du troupeau observée est attribuable à l'instauration des enclos de maternité et l'augmentation équivaudrait à l'ajout de l'estimé des jeunes d'un an. L'amélioration de la tendance démographique due aux enclos de maternité était toutefois limitée par le fait que la taille du troupeau était sous-estimée, ce qui a réduit l'effort des enclos de maternité ( = 11% des faons sont nés en enclos).
Nos simulations ont confirmé que les enclos de maternité augmentaient la taille de la population dans une proportion similaire au nombre de juvéniles libérés des enclos, et ce, même avec un faible effort de mise en enclos pour les populations relativement stables. Dès que le taux de croissance était inférieur à λ = 1.000, une augmentation de l'effort des enclos de maternité était nécessaire pour compenser le déclin et stabiliser la taille de population. Pour les populations en déclin à λ < 0.890, les modèles incluant 100% de l'effort des enclos de maternité ne permettaient pas de stabiliser la taille de la population en fonction des paramètres démographiques vitaux fixés dans nos modèles.
Même si leur application a été limitée jusqu’à présent, les enclos de maternité ont été popularisés comme une mesure non létale visant à réduire la mortalité des faons due à la prédation chez les petites populations de caribous. Néanmoins, le programme d'enclos de maternité du troupeau de Chisana ainsi que 3 autres programmes ailleurs ont démontré que l'amélioration des tendances démographiques ont été au mieux modestes et sont financièrement très dispendieux. Considérant la nécessité de maximiser l'effort des enclos de maternité, le rôle de cette mesure pourrait permettre de répondre aux enjeux de conservation pour de petites populations de caribous qui sont stables ou en faible déclin. Il n'en demeure pas moins que la mise en place d'une telle mesure doit être justifiée par la probabilité d'améliorer les tendances démographiques plutôt que par sa popularité par rapport aux autres mesures de gestions applicables.
INTRODUCTION
The recovery of small declining populations is a primary component of wildlife conservation (Caughley and Sinclair 1994) and is becoming more important with increasing human impacts on the natural world (Mills 2013). Effective management to recover an imperiled population is dependent on identifying the causes of decline and taking appropriate actions to increase numbers, thus improving the likelihood of long-term persistence (Caughley 1994). However, wildlife managers commonly must direct their efforts at proximate causes of declines to maintain these populations on the landscape, while attempting to address more intractable ultimate causes such as habitat loss and fragmentation (Groombridge 1992, Tilman et al. 1994, Sih et al. 2000, Crooks et al. 2017), overexploitation (Dublin and Wilson 1998, Milner-Gulland et al. 2003, van Velden et al. 2018), agricultural and industrial effects (Foreyt and Jessup 1982, Wehausen et al. 2011, Hebblewhite 2017, Rominger 2018), and changing environmental conditions (Wakelyn 1987, Thuiller et al. 2006, Urban 2015, Rominger 2018). Further, the urgency for management intervention intensifies as a population dwindles in number and becomes increasingly susceptible to extirpation due to environmental fluctuations, demographic stochasticity, and loss of genetic diversity (Caughley and Sinclair 1994, Sih et al. 2000, Mills 2013).
Ungulates are among the most imperiled wildlife worldwide, largely because of overexploitation and habitat loss (Hoffmann et al. 2015; Ripple et al. 2015, 2017; Lindsey et al. 2017). Beyond eliminating harvest, common management tactics to bolster small declining ungulate populations include habitat improvement (Risenhoover et al. 1988, Dolan 2006, Wilson et al. 2006), translocations to augment low numbers (Compton et al. 1995, Singer et al. 2000, Rominger et al. 2004, DeCesare et al. 2010, Poirier and Festa-Bianchet 2018), and predator reductions (Orians et al. 1997; Hervieux et al. 2014; Serrouya et al. 2017b, 2019; Rominger 2018). In the last 2 decades, caribou (Rangifer tarandus) populations across the southern extent of their range that are exhibiting persistent declines have become the focus of intensive conservation efforts (Alberta Woodland Caribou Recovery Team 2005; British Columbia Ministry of Environment [BCME] 2009; Festa-Bianchet et al. 2011; Environment Canada 2012, 2014; Ray et al. 2015) employing all these approaches with limited success (Hervieux et al. 2014, Pyper et al. 2014, Dickie et al. 2017a, Leech et al. 2017, Serrouya et al. 2019). These caribou, defined as the Boreal, Southern Mountain, Central Mountain and Atlantic-Gaspésie designatable units by the Committee on the Status of Endangered Wildlife in Canada (COSEWIC) are currently deemed threatened or endangered across much of the Canadian mainland provinces and the Northwest Territories (COSEWIC 2014a,b). Further, the last remaining caribou herd ranging into the contiguous United States, the small transboundary South Selkirk Mountains population, has been listed as endangered since 1983 (U.S. Fish and Wildlife Service 1994). Despite management efforts to increase their numbers (Wiles 2017), this population has declined from 46 caribou in 2009 to only 1 individual that was taken into captivity in January 2019 (Francovich 2019).
Most caribou populations that inhabit ranges below northern treeline number <3,000 animals, have limited seasonal migrations, and occur at low densities on the landscape (Bergerud 1980, Bergerud et al. 2008). Thus, they exhibit life-history strategies and population dynamics quite different from the few large herds in the region (>25,000 caribou; e.g., Alaska's Fortymile, Mulchatna, and Nelchina herds) or the large, migratory arctic herds. For populations of these sedentary mountain and boreal ecotypes (Hummel and Ray 2008, Festa-Bianchet et al. 2011), predation dominates as a proximate limiting factor (Seip 1991, Adams et al. 1995b, Bergerud 2000, Wittmer et al. 2005a, Bergerud et al. 2008). These caribou commonly share their ranges with wolves (Canis lupus) that prey on all age classes (Bergerud 1980, Mech et al. 1998, Hayes et al. 2003, Valkenburg et al. 2004, Jenkins and Barten 2005). Brown bears (Ursus arctos) and American black bears (U. americanus) are often important predators of young caribou calves (Adams et al. b; Valkenburg et al. 2004; Jenkins and Barten 2005; Pinard et al. 2012). For southerly caribou populations, mountain lions (Puma concolor), coyotes (Canis latrans), and wolverines (Gulo gulo) can also be key predators (Ouellet et al. 1996, Kinley and Apps 2001, Wittmer et al. 2005a, Gustine et al. 2006). The combined losses of calves to these predators are often high during the first few weeks following birth (Page 1985; Adams et al. b; Jenkins and Barten 2005, Pinard et al. 2012). Given the importance of predation in the dynamics of boreal and mountain caribou across North America, reducing the impacts of carnivores is a common management goal whether to reverse declines in caribou numbers to restore harvests in Alaska and Yukon (Gasaway et al. 1983, Farnell et al. 1998, Hayes et al. 2003, Valkenburg et al. 2004), or to maintain small vulnerable populations on the landscapes of southern Canada (Alberta Woodland Caribou Recovery Team 2005, BCME 2009, Environment Canada 2012).
In Alaska and Yukon, where these caribou are generally secure (Harper 2013, Hegel and Russell 2013, COSEWIC 2014a), reducing predation impacts on caribou has largely meant substantially decreasing wolf numbers over a herd's range for several years, usually by aerial shooting (Gasaway et al. 1983, Farnell and McDonald 1988, Orians et al. 1997, Hayes et al. 2003, Valkenburg et al. 2004). Wolf culling programs commonly generate contentious public debate (Harbo and Dean 1983, Stephenson et al. 1995, Orians et al. 1997, Musiani and Pacquet 2004, Hervieux et al. 2015); thus, in the 1990s, both jurisdictions experimented with fertility control as a nonlethal approach to adequately reduce wolf numbers with broader public acceptance (Boertje and Gardner 2003, Hayes et al. 2003, Farnell 2009, Boertje et al. 2017). Although these nonlethal experiments yielded reasonable success in reducing wolf numbers over large areas (≤18,500 km2; Boertje et al. 2017), Alaska has continued with lethal wolf control programs as the mainstay of their intensive management strategies for increasing ungulate populations (Alaska Department of Fish and Game [ADF&G] 2007, 2011). In Yukon, wildlife managers have concluded that aerial wolf control and surgical sterilization has limited utility to meet their management goals given the strong public opposition and high costs, and therefore other management tools with broader public support are needed (Government of Yukon 2012).
Across southern Canada from British Columbia to Labrador, the future for many mountain and boreal caribou populations is uncertain or bleak (Festa-Bianchet et al. 2011; COSEWIC b; Hebblewhite 2017; Cornwall 2018). Along the southern extent of their distribution, caribou have typically occurred at low densities in broad expanses of mature boreal and montane coniferous forests, and in interspersed muskeg and alpine areas. Much of this expansive region historically supported few other ungulates (e.g., moose [Alces americanus], elk [Cervus canadensis], white-tailed deer [Odocoileus virginianus], and mule deer [O. hemionus]), and thus sustained few wolves (Bergerud 1974, 1988; Seip 1992; Festa-Bianchet et al. 2011; Hervieux et al. 2013). Resource development, primarily logging, hydrocarbon exploration and development, mineral extraction, and associated transportation networks have resulted in less suitable habitat for caribou (James and Stuart-Smith 2000, McLoughlin et al. 2005, Courtois et al. 2007, Vors et al. 2007, Johnson et al. 2015). Resulting landscape changes have been conducive to substantial increases in moose and white-tailed deer populations within caribou ranges, and wolf numbers have increased in concert (McLoughlin et al. 2005; Latham et al. b, 2013; Dawe et al. 2014). Also, linear features associated with resource development (e.g., roads, trails, seismic lines, pipelines) facilitate wolf movements and thereby can increase predation risk for caribou (James and Stuart-Smith 2000, Latham et al. 2011a, Whittington et al. 2011, Ehlers et al. 2014, Dickie et al. 2017b). Further, climate warming has been identified as an important driver of white-tailed deer expansion northward into the boreal forest (Dawe et al. 2014). Thus, through the mechanism of apparent competition (Holt 1977, Holt and Lawton 1994), caribou populations have experienced increased predation by wolves that are largely supported by burgeoning populations of moose and white-tailed deer exploiting habitats modified by resource development (Seip 1992; Wittmer et al. 2005b, 2007; Apps et al. 2013; Peters et al. 2013). As a result, many caribou populations have declined markedly, and caribou range has retracted northward (Schaefer 2005, Vors et al. 2007, Hervieux et al. 2013, Ray et al. 2015).
In addition to limiting development to maintain currently undisturbed habitats, recovering imperiled southern caribou populations requires allowing succession to restore degraded forests, a process that will likely take decades (Schneider et al. 2010, Environment Canada 2012, Hervieux et al. 2013, Johnson et al. 2015). In the interim, wildlife managers are pursuing a limited set of options for maintaining these small caribou populations, commonly numbering <100 adults (individuals ≥1 year old; COSEWIC b). Management options employed to date have included predator control primarily focused on wolves (Alberta Woodland Caribou Recovery Team 2005, British Columbia Ministry of Forests, Lands and Natural Resource Operations [BCMF] 2014, Hervieux et al. 2014, Serrouya et al. 2019) and mountain lions (BCME 2009, Wilson 2009), attempts to reduce moose populations to limit apparent competition (Steenweg 2011; BCMF 2015; Serrouya et al. 2015b, 2017b, 2019), and translocation of caribou from more robust populations (Bergerud and Mercer 1989, Compton et al. 1995, Warren et al. 1996, DeCesare et al. 2010, Leech et al. 2017). Given limited success in reversing population declines with these methods, additional management approaches are sorely needed.
In response to a marked, persistent decline in the Chisana Caribou Herd, a small population thought to number about 300 individuals in 2002 (Farnell and Gardner 2002) that ranges between southwest Yukon and adjacent Alaska (Fig. 1), we implemented an experimental maternal penning program during 2003–2006 with the goal of increasing chronically low calf survival and thus improving the population's trend. Predator reduction was not an option because the Chisana Herd range was largely encompassed by Wrangell-St. Elias National Park and Preserve in Alaska and the Kluane Wildlife Sanctuary in Yukon; thus, any actions taken to improve the status of the herd had to conform to policies of these protected areas. Moreover, following intensive aerial wolf control programs in the 1980s and 1990s (Farnell and McDonald 1988, Hayes et al. 2003, Farnell 2009), societal and political support for agency-based wolf control to enhance ungulate populations was lacking in Yukon (Government of Yukon 2012). Females within the Chisana Herd generally exhibited good body condition in late winter and were highly productive but experienced substantial losses of calves in the initial weeks following birth (Farnell and Gardner 2002). High levels of predation on calves during this early period is common among small caribou populations (Page 1985; Adams et al. b; Valkenburg et al. 2004; Jenkins and Barten 2005). Therefore, we reasoned that capturing pregnant females in late March and holding them in a predator-free pen on their native range through parturition to mid-June, when their calves would be a few weeks old, would eliminate most of this early calf mortality in the treatment group. If the increases in early survival of calves from maternal penning was sufficient, carried over into higher recruitment at 1 year of age, and provided enough additional recruits, then we expected improvement in the population trend as a result. Further, critically evaluating the efficacy of maternal penning to improve population trends would provide assessment of the utility of this management approach to conservation challenges with other small caribou populations.
Our maternal penning approach was unique in that it involved confining pregnant females and their dependent offspring on native range for a brief period during a crucial life stage. Penning is somewhat analogous to the use of predator exclosures to protect nests of egg-laying species (Sargeant et al. 1974, LaGrange et al. 1995, Ratnaswamy et al. 1997, Maslo and Lockwood 2009) or headstarting to improve juvenile survival for species that provide no parental care (Heppell et al. 1996, King and Stanford 2006, Crane and Mathis 2011, Garcia and Gerber 2016, Starking-Szymanski et al. 2018). However, with these methods only the nests or young are protected; productive females are not constrained from their normal movements. Maternal penning is quite different from captive breeding, a common approach for conservation of critically endangered species (Snyder et al. 1996), including ungulates (Stüwe and Nievergelt 1991, van Dierendonck and Wallis de Vries 1996, Ostrowski et al. 1998, Linklater 2003, Witzenberger and Hochkirch 2011). Breeding populations are maintained long-term in facilities that are most often distant from their native habitats to prevent extinction of species or subspecies that cannot be returned to their historical ranges, or to build up adequate numbers in captivity to provide stock for reintroductions when and where feasible (Snyder et al. 1996, Dolman et al. 2015).
The objectives of our study were 1) to evaluate maternal penning as a tool for increasing caribou calf recruitment in the Chisana Herd, 2) to thoroughly assess the population dynamics of the Chisana Herd to determine the contribution of maternal penning to its demography, and 3) to appraise the utility of maternal penning for improving the status of small at-risk caribou populations. Compared to most other small caribou populations of conservation concern, the Chisana Herd provided a unique opportunity to evaluate maternal penning in that the herd was relatively large, allowing for reasonable sample sizes of females in maternal penning; it inhabited pristine range and did not face the challenges associated with habitat alteration experienced by at-risk southern herds; and the open nature of the vegetation and seasonal distribution of caribou allowed for efficient capture efforts and more thorough population monitoring than is generally feasible for smaller populations in more heavily forested environments (DeCesare et al. 2012, 2016; Serrouya et al. 2017a).
STUDY AREA
The range of the Chisana Caribou Herd extended over approximately 13,000 km2 (centered at 62.0°N, 141.3°W) and predominantly within Wrangell-St. Elias National Park and Preserve in Alaska and the Kluane Wildlife Sanctuary in Yukon (Fig. 1). The entire range was a large, undeveloped wilderness inhabited by only a few people living at remote homesteads and seasonal hunting camps; the only road was the Alaska Highway crossing the northeastern edge of the herd's range. The region is composed of the rugged, glaciated mountains and broad river valleys of the northern St. Elias and eastern Wrangell Mountain Ranges drained by the Chisana, White, and Donjek Rivers. Elevations extend from 800 m at the northeastern limits of the herd's range to ice-covered peaks >3,000 m along its southern and western margins.
Low-elevation caribou habitat consisted primarily of spruce (black spruce [Picea mariana] and white spruce [P. glauca]) woodlands and cottongrass (Eriophorum vaginatum) tundra with scattered stands of aspen (Populus tremuloides), balsam poplar (Populus balsamifera), and Alaska birch (Betula neoalaskana). Shrub communities (predominantly willow [Salix spp.], alder [Alnus spp.], and birch [Betula spp.]) were common along drainages and in the vicinity of treeline at about 1,200 m. Above treeline, mountain slopes were a mosaic of dwarf shrub (predominantly willow, dwarf birch [Betula nana], bog blueberry [Vaccinium uliginosum], and black crowberry [Empetrum nigrum]) and sedge (Carex spp.) communities grading into mixed scree and alpine tundra (predominantly sedge and Dryas spp.) at higher elevations. Fruticose lichens (predominantly Cetraria spp., Cladonia spp., Flavocetraria spp., Peltigera spp., and Stereocaulon spp.) occurred throughout most vegetation types.
The region is also inhabited by moose and thinhorn sheep (Ovis dalli), and the full complement of native carnivores including wolves, grizzly bears, black bears, wolverines, coyotes, lynx (Lynx canadensis), and golden eagles (Aquila chrysaetos). In 2001, wolf abundance was estimated at 5.6 wolves/1,000 km2 within the herd's range (Farnell and Gardner 2002) by aerial snowtracking (Stephenson 1978, Hayes and Harestad 2000). Grizzly bears probably numbered around 15/1,000 km2 based on estimates from similar areas elsewhere in Alaska (Miller et al. 1997, Farnell and Gardner 2002).
The Chisana Herd's range fell within the rain shadow of the Wrangell and St. Elias Mountains and therefore exhibited a semiarid, interior subarctic climate of long cold winters and short growing seasons. Based on remotely sensed climate data for the herd's range (Rieneker et al. 2011, Russell et al. 2013), mean monthly temperatures varied from −18.0°C in January to 8.4°C in July during 1979–2008. During those years, annual precipitation averaged 54 cm of water, with 41% falling as snow (216 cm on average) in October to early May. During our March 2003–May 2008 study period, the region experienced the hottest summer on record in 2004 (Fig. 2A; Wendler et al. 2011). Further, the winters beginning in 2003, 2004, and 2007 were relatively severe, ranking in the upper quartile of winters during 1979–2008 for cumulative snow depth (Fig. 2B).
During the 1990s, wildlife managers, First Nations, hunters, and other local residents became increasingly concerned about the status of Chisana caribou. In 1988, the Chisana Herd was meeting management objectives in that it numbered about 2,000 caribou (Fig. 3A), exhibited adequate calf recruitment (28–31 calves:100 cows [female ≥1 year old] during 1986–1988; Fig. 3B) and bull numbers (36–43 bulls [males ≥1 year old]:100 cows during 1986–1988; Fig. 3C), and supported a small harvest of primarily bulls (Kelleyhouse 1990, Gardner 2003). During a period marked by severe winters (Fig. 2B), autumn calf:cow ratios dropped to 10–12:100 in 1989–1990, then dropped even lower to ≤2:100 during 1991–1993 (Fig. 3B), and caribou numbers were thought to have declined to about 870 by 1993 (Fig. 3A; Gardner 2003). Calf recruitment remained at low levels for the next 9 years, averaging only 8 calves:100 cows during 1994–2002 (range = 4–14:100; Fig. 3B). During this period, adult females were generally in good body condition in late winter and exhibited high productivity (86% pregnant on average), but calf losses in the month following birth were very high (Farnell and Gardner 2002, Gardner 2003). Following the abrupt decline in calf recruitment, adult sex ratios dropped from averaging 38 bulls:100 cows in 1986–1991 to 19:100 during 1995–1999 (Fig. 3C). Population estimates, based largely on minimum aerial counts, fell to 315 caribou in 2002 for an average rate of decline of 12%/year from 1989 (Fig. 3A; Gardner 2003). In response to these declines in caribou numbers, calf recruitment, and adult sex ratios, licensed hunting of Chisana caribou was closed in Alaska and Yukon in 1994 and harvests by First Nations people in Yukon were voluntarily reduced (Farnell and Gardner 2002, Gardner 2003). In 2002, following persistent population decline over 14 years, and with support of local First Nations, the Chisana Herd was designated as a specially protected population under the authority of Yukon Wildlife Act, thus eliminating the remaining legal First Nations harvest (Farnell and Gardner 2002).
METHODS
Research Overview
Given the prognosis of continuing decline for the Chisana Caribou Herd driven predominantly by very low calf recruitment, we initiated a pilot trial of maternal penning in 2003 with 20 females to primarily assess the feasibility of holding caribou in a predator-excluding pen on their native range from late March through parturition to mid-June. With all calves surviving to release in that initial effort and the experience we gained, we scaled up the maternal penning to 29 females in 2004, and then to 50 females each in 2005 and 2006. Although the primary hypothesis underlying maternal penning is that survival will be markedly improved for calves born in the pen, adult females could also benefit from protection from predation during late winter and the calving season when mortality is relatively common (this study; Jenkins and Barten 2005; Wittmer et al. 2005a; L. G. Adams, U.S. Geological Survey [USGS], unpublished data).
Maternal penning may also have nutritional implications for the individuals being held. If caribou females acclimate to penning, the ad libitum provision of high-quality commercial feed could result in improved nutritional condition for the females and their offspring. Alternatively, caribou held for maternal penning that are chronically stressed or do not adequately use novel feed could exhibit reduced nutritional condition. Based on cursory observations of consumption of the pelleted ration and changes in their physical appearance during penning in 2003, we suspected adult females were gaining mass while held in captivity. Thus, we conducted several investigations to evaluate nutritional effects of maternal penning in subsequent years. In 2006, we estimated diets of pen females weekly throughout the penning period via microhistological analyses of feces to understand the timing and extent of use of the commercial ration and lichen we provided compared to native forages available in the pen. In 2005 and 2006, we assessed body mass changes for captive females with platform scales placed at feeders in the pens that allowed us to passively weigh individuals throughout the penning period. We hypothesized that maintaining or gaining mass from the time of capture (late Mar–early Apr) to early May, just prior to the onset of parturition, would indicate improved nutritional condition of females held in the pen, given that free-ranging females were likely to lose mass through the last 5 weeks of winter (Mautz 1978; Leader-Williams 1988; Parker et al. 2005, 2009). To quantify net nutritional gain at the end of penning, we also determined the body mass of females immediately prior to their release from penning in mid-June 2005–2006 and weighed free-ranging adult females at the same time for comparison. During 2004–2006, we compared birth mass of calves born in the pens with those of calves born in the wild, expecting that pen calves would be similar to or possibly heavier than wild-born calves. If nutritional advantages gained during maternal penning persisted well beyond release from the pens, we suspected that autumn body mass of pen calves could be greater than those of wild-born individuals. Thus, in mid-October 2004–2006, we captured and weighed 5-month-old female calves from each group.
To better understand the population dynamics of the Chisana Herd and assess the effects of maternal penning on overall calf survival and herd growth, we intensified population-level monitoring in March 2003. Our investigations relied on a sample of radiocollared females, along with calves instrumented as newborns during maternal penning, to provide information necessary for comparing survival of pen calves and females with their free-ranging counterparts, and to meet other objectives. During March 2003–May 2008, we assessed vital rates (natality, calf survival, adult female survival), conducted sex-age composition surveys in mid-October each year, and estimated population size in mid-October 2003, 2005, and 2007. During all 4 years of maternal penning, we fitted females with radiocollars prior to placement in the pen to facilitate monitoring during penning and following release, and to add to the radiocollared sample in the free-ranging population. In 2003, we did not handle neonates in the pen to minimize disturbance during calving. In the other years, we radiocollared calves shortly after birth. Procedures for capturing, handling, and care in captivity of caribou during this study complied with guidelines established at the time of our research by the Canadian Council on Animal Care (2003) and the American Society of Mammalogists (Animal Care and Use Committee 1998). Net-gunning and aerial darting of caribou followed established protocols of the Yukon Department of Environment (YDE) and USGS, respectively. Research plans were approved by the YDE, USGS, and U.S. National Park Service.
To assess the population-level contribution of maternal penning, we modeled the number of additional calves surviving to mid-October (5 months old) and to mid-May (1 year old) as a result of maternal penning beyond the number that would have survived had they been born in the wild. We estimated the effect of maternal penning on population size by extending these simulations to project the survival of the additional individuals, as well as their resulting progeny, to mid-October each year 2003–2007. Therefore, we could estimate population sizes without maternal penning to compare with actual population estimates during our studies.
We were also interested in assessing the broader applicability of maternal penning to other small imperiled caribou populations. The outcome of any such effort is driven by the population trajectory that exists without penning (inherent population trend or inherent rate of increase [λi]), and the proportion of calves that are born in the pen, or penning effort. To evaluate the utility of maternal penning relative to these factors, we first constructed an age-structured simulation model that projected the female segment of a caribou population for 20 years based on demography of the Chisana Herd. An age-structured model was necessary because productivity and survival of ungulates are known to vary with age, with both tending to be reduced in initial age classes and senescent older age classes (Eberhardt 1985, Adams and Dale 1998a, Gaillard et al. 2000); thus, the effects of an influx of young individuals from penning on population-wide vital rates needed to be accounted for. From this base model, we adjusted calf and adult survival rates to derive simulated populations with λi ranging from 0.850 to 1.000, encompassing observed trends for most imperiled southern caribou populations (Wittmer et al. 2005a, Hervieux et al. 2013). Then we determined population responses to 5-year penning treatments with penning effort ranging from 0% to 100% across that range of λi.
Maternal Penning
Pen construction
We constructed pens at Tchawsahmon Lake, Yukon in 2003, and at Big Boundary Lake, Yukon for 2004–2006 (Fig. 1). Both sites were spruce woodlands near tree line within the winter range of the Chisana Herd and directly adjacent to areas regularly used by caribou in summer. We constructed pens on sites that provided spruce trees for in situ fence posts and some cover for captive caribou but were sufficiently open and on gentle slopes to facilitate monitoring of penned animals. Pens encompassed 6.0, 9.5, 12.0, and 12.2 ha in 2003–2006, respectively, or 0.2–0.3 ha/adult female held. In 2005 and 2006, we reconfigured the Big Boundary Lake pen to ensure that approximately a third of the enclosed area provided previously ungrazed natural forage. After snowmelt, free-flowing water was available to caribou in the pen each year.
We built pen fences out of 2.3-m black geotextile fabric suspended between 3-mm steel cables stretched taut between trees or supplementary posts at ground level and a height of 1.8 m, thus providing an opaque, visual barrier (Fig. 4). We routed fencing along ridgelines and avoided small draws whenever possible to limit the ability of caribou, or their predators, to see over the fence. We cleared fence lines of brush and debris to a width of about 4 m and limbed trees to be used as posts to a height of 2.4 m. We added other posts as necessary to maintain spacing of about 6 m between fence posts. We attached the geotextile fabric to supporting cables by folding approximately 20 cm of material over the top and bottom cables, then pinning the material in place about every meter with 9-cm nails. We filled gaps that invariably occurred between the ground surface and the bottom of the fence with debris from fence-line clearing. We incorporated manufactured corral gates (1.2–1.5 m in width) into the fence line to provide access into the pen. To deter wolves and bears, we surrounded the pen with a solar-charged electric fence placed about 3 m outside the geotextile fence and field staff patrolled the perimeter several times each day (Fig. 4). To aid in monitoring penned caribou, we built 1 (2003) or 2 (2004–2006) elevated observation platforms about 3 m above the ground in the pen or at the fence line near caribou feeding stations that provided reasonable visibility of the entire pen (Fig. 5).
Capture and handling of adult females
We captured female caribou for maternal penning trials by net-gunning from a Bell Jet Ranger 206B helicopter (Bell Helicopter, Forth Worth, TX, USA) during late March–early April 2003–2006. We blindfolded and hobbled captured caribou and trussed them in restraint bags for transport inside 2 additional Bell Jet Ranger 206B helicopters to the pen facility. In 2003, we did not sedate caribou, but in 2004 we evaluated sedation during transport and handling by randomly treating half of the animals with medetomidine hydrochloride (10–13 mg/caribou) via intranasal administration (Cattet et al. 2004), antagonized upon completion of handling with atipamezole hydrochloride (30 mg/caribou). Based on results of the 2004 trial, we sedated all caribou via these procedures in 2005–2006.
Upon arrival at the pen facilities, transport helicopters landed near the pen and we unloaded restrained caribou into a cargo sled towed by a snowmachine for delivery into the pen. Once in the pen, we evaluated each caribou for pregnancy via transrectal ultrasound (Ropstad et al. 1999) and weighed individuals to the nearest 0.5 kg. We fitted pregnant females with mortality-sensing radiocollars with individually numbered marking bands attached; we radiocollared nonpregnant females in 2004 but only eartagged them in other years. In 2004–2006, we removed a canine tooth from each caribou during her initial capture, after administration of lidocaine-epinephrine as a local anesthetic, for age estimation via analyses of cementum annuli (Reimers and Nordby 1968, Miller 1974) by Matson's Laboratory (Manhattan, MT, USA). We retained pregnant females in the pen, whereas we released nonpregnant females. In 2006, some caribou received ultrasound exams in the field (n = 28) to minimize transport of nonpregnant females to the pen facility. As a result, we released 5 nonpregnant individuals at their capture locations that year.
Capture and handling of neonates
With the onset of the birthing season in early May, we closely observed female caribou for behaviors indicative of imminent parturition such as continuous pacing or seclusion from other caribou, or signs of a recent birth including presence of a calf, birth membranes, or a blood-stained rump. We determined birth dates of calves produced in the pen by direct observation or estimated them based on observations of the cow and developmental characteristics of the calf (Adams et al. b). In 2003, we did not handle calves, whereas in 2004–2006, we captured calves by hand usually a few hours after birth and fitted them with expandable mortality-sensing radiocollars. We also sexed and weighed calves to the nearest 0.2 kg, and released them back to their mothers within 3 minutes (Adams et al. b; Adams 2005). For calves that were 1 or 2 days old at capture, we estimated birth mass as described by Adams (2005).
Animal care and monitoring during captivity
We provided caribou ad libitum a commercial pelleted ration formulated specifically for caribou (Barboza and Parker 2006, Thompson and Barboza 2014; = 2.9 kg dry/adult caribou/day) and terrestrial lichens ( = 0.5 kg dry/adult caribou/day) in plywood feeding troughs each morning and evening. Lichens were predominantly Cladonia arbuscula, C. mitis, and C. stellaris collected elsewhere and transported to pen facilities. Caribou also consumed native forages available within the pen.
We monitored penned caribou at least twice each day by observation or radiotelemetry from the elevated platforms, during regular patrols of the pen perimeter, and along with calf capture efforts. At the end of the penning period (12, 11, 14, and 14 June in 2003–2006, respectively), when nearly all calves were >2 weeks old, we removed a segment of the perimeter fence >100 m and slowly herded the caribou cows and their calves out of the pen by foot (2003, 2004, 2006) or by helicopter (2005). After the release, we monitored cows and calves from maternal penning along with radiocollared females in the free-ranging population to determine their survival.
Diets of adult females in captivity
To evaluate adjustment of caribou to the pelleted ration, we determined diet selection via microhistological analyses of feces (Sparks and Malechek 1968, Free et al. 1970) during the 2006 penning operation. We collected fecal samples from the wild population on 2 April (at the beginning of the penning program) and at weekly intervals from within the pen (9 Apr–11 Jun). For each collection, we gathered 5 composite fecal samples consisting of 5 pellets from each of 5 fresh pellet groups. We placed all samples in plastic bags and stored them frozen until analysis. The Wildlife Habitat Nutrition Laboratory (Washington State University, Pullman, WA, USA) performed microhistological analyses at 100 microscope fields/sample to identify forage classes (ration, lichen, shrubs, grasses-sedges, forbs, and moss) and major forage species that constituted >5% of the diet. To account for differential digestibility among forage classes (Leslie et al. 1983), we corrected the relative densities of plant fragments identified with the apparent dry matter digestibility of major forage groups for caribou from the literature (Boertje 1981, Barboza and Parker 2006, Gustine et al. 2011).
Body mass dynamics of adult females
To assess nutritional effects of providing high-quality commercial feed and forage lichens during maternal penning, we measured body mass of captive females throughout the penning periods in 2005 and 2006 with 4 platform scales (similar to those described by Bassano et al. 2003) at the feeding stations (Fig. 5). Each scale consisted of a 1.2-m × 1.5-m platform constructed of 1.9-cm plywood, with 0.6-m-high side rails, attached to shear-beam load cells at each corner that were connected via a 4-cell summing card to a digital indicator in the elevated observation posts overlooking the feeding stations. In addition to the side rails on each platform, we placed brush around the feed troughs to increase the likelihood that caribou would step onto the scales while accessing the pelleted ration. We put the scales into operation in late April each year and monitored them during twice-daily feeding bouts on most days until we released caribou in mid-June. Personnel in the observation platforms recorded body mass to the nearest 0.5 kg by monitoring the scale displays (Fig. 5).
During periods of observation, we recorded body mass each time an individual caribou stepped onto a platform scale; we averaged multiple measurements acquired from an individual over the course a day to generate a daily mass for that caribou. To estimate body mass of cows just prior to the onset of calving, we averaged the daily measurements obtained during 4–8 May for each female that we observed on a platform scale. We also estimated the proportion of mass lost with calving based on the first daily mass obtained within 10 days following calving (most cows did not return to the feeders for several days after giving birth; there was no notable trend in mass for individuals weighed more than once during this 10-day period) and the last daily mass obtained within the 5 days prior to giving birth for each cow weighed during both periods. Finally, we determined the mass of adult females at release from the pens by averaging daily measurements for each individual obtained during 9–13 June.
Body Mass of Free-Ranging Caribou
To evaluate the influences of maternal penning on nutritional status of caribou, we compared 1) birth mass of pen and wild-born calves, 2) body mass of pen and free-ranging cows in mid-June at release from penning, and 3) body mass of pen and wild-born calves at 5 months of age in mid-October. We obtained the birth mass of calves born in the wild by aerially searching for radiocollared females and other cows with young calves (≤2 days old) throughout the herd's range during 4–5 days in mid-May 2004–2006 encompassing the median calving date, or peak of calving, for the population. When we observed a young calf, we dropped off a person from the helicopter within a few meters and they captured the calf by hand; they weighed calves to the nearest 0.2 kg, sexed calves, and estimated ages based on physical and behavioral characteristics (Adams et al. b; Adams 2005), then they released calves back to their mothers. We estimated birth mass for 1- and 2-day-old calves as described by Adams (2005).
To obtain body mass of free-ranging cows for comparison with those released from the maternal pens, we captured and weighed adult females during 11–16 June 2005–2006 via helicopter darting. Because of the chronic low calf survival in the Chisana Herd, we chose to avoid females with calves at heel but selected radiocollared females that were deemed parturient during peak-calving observations or unmarked females with distended udders that had thus produced calves but recently lost them. We immobilized females with carfentanil citrate and xylazine hydrochloride (3.6 mg and 100 mg/dart, respectively) and weighed them to the nearest 0.5 kg. During these and additional captures, we replaced radiocollars, removed some radiocollars to limit excessive growth of our collared sample, and extracted teeth (as described above) from caribou for which we lacked age estimates. Upon completion of handling, we antagonized immobilizing drugs with naltrexone hydrochloride (100 mg/mg carfentanil) and yohimbine hydrochloride (10 mg each or ~0.11 mg/kg).
In mid-October 2004–2006, we employed helicopter darting to capture, weigh, and radiocollar 5-month-old female calves (half born in the pen, half born in the wild) to evaluate persistence of nutritional effects of improved nutrition during penning. We immobilized these calves with carfentanil citrate and xylazine hydrochloride (1.5 mg and 35 mg/dart, respectively), weighed them to the nearest 0.1 kg, and fitted them with radiocollars. We then injected naltrexone hydrochloride (100 mg/mg carfentanil) and yohimbine hydrochloride (5 mg each or ~0.11 mg/kg) to antagonize the immobilants.
Vital Rates of Free-Ranging Caribou
We used aerial observations of radiocollared caribou and population-wide autumn composition surveys to determine natality and survival rates of free-ranging caribou in the Chisana Herd, and survival rates for pen individuals following their mid-June release. We based our monitoring effort on a biological year beginning on 15 May near the annual median calving date. Our radiotelemetry monitoring schedules varied from year to year, ranging from 9 to 17 relocation periods/year, but we consistently located radiocollared caribou 7 times each year (mid-May, mid-Jun, mid-Aug, mid-Oct, early-Dec, mid-Jan, and mid-Mar). We divided the year into 2 seasons: summer (mid-May–mid-Oct; 3 intervals) and winter (mid-Oct–mid-May; 4 intervals); winters are referred to by the year in which each began. In general, our radiotelemetry monitoring was inadequate to identify causes of all mortalities, but we pinpointed mortality signals from the air and noted any evidence of cause of death at the site. Because of the remote nature of our study area, we investigated mortalities on the ground and retrieved radiocollars at the first opportunity when we were in the area with a helicopter.
Natality
We estimated natality rates, or the proportion of females ≥2 years old that produced offspring, from results of ultrasound pregnancy assessments during captures for maternal penning, and observations of free-ranging radiocollared cows during the peak of calving in mid-May 2003–2008. We radiolocated and observed free-ranging cows from aircraft (Piper Supercub airplane [Piper Aircraft, Vero Beach, FL, USA] only in 2003; Robinson R-44 helicopter [Robinson Helicopter, Torrance, CA, USA] or Supercub in 2004–2008) to determine whether each was parturient, based on the presence of a calf at heel, hard antlers, and-or a distended udder (Bergerud 1964, Espmark 1971, Whitten 1995). We used these observations to determine annual population natality rates and overall age-specific natality rates.
Calf survival
We estimated wild-born calf survival to mid-June, when pen calves were released, by locating all the free-ranging radiocollared females that we deemed parturient and determining those that had a surviving calf at heel at that time. Thus, the survival rate for wild-born calves to mid-June each year was simply the proportion of parturient cows with surviving calves.
To determine survival rates of calves born in the maternal pens from their mid-June release to mid-October, we relied on observations of calves at heel with their radiocollared dams in 2003 and radiocollared calves in 2004–2006. As the study progressed, however, we recognized that the calves-at-heel approach could be biased, resulting in underestimates of calf survival after mid-July. From observations of radioed cow-calf pairs released from the pens in 2004–2006, we noted that 5 calves of females that died during mid-July–September and a calf that was permanently separated from its dam in mid-July all survived beyond mid-October. Also, 4 additional calves permanently split from their mothers between late September and mid-October. Thus, 12% (n = 10 of 83) of radiocollared calves surviving to mid-October 2004–2006 were no longer accompanying their dams. We recognize the potential bias in our summer survival rate for 2003 pen calves following release that could underestimate the population effects of maternal penning in that year.
We determined calf survival rates during winters 2004–2006 based on monitoring of radiocollared calves from maternal penning that survived beyond mid-October, along with additional wild-born female calves captured for weighing in mid-October and instrumented with radiocollars at that time. Because of small samples of wild-born calves each year (n ≤ 11), we did not calculate separate winter survival rates for pen and wild-born calves.
Adult female survival
We estimated survival rates of adult females based on observations of radiocollared individuals. We censored radiocollared females held in maternal penning during mid-March–mid-June from calculations of survival rates for free-ranging caribou. To investigate potential differences in survival between females held in the pens and those that were free-ranging during the penning period, we calculated separate survival rates for the 2 groups.
Estimating Herd Composition and Size
During mid-October 2003–2007, we conducted helicopter surveys each year to determine the sex and age composition of the Chisana Herd, either as stand-alone efforts (2004, 2006) or as part of censuses (2003, 2005, 2007). We conducted all surveys with a Robinson R-44 helicopter, with 1 or 2 observers in addition to the pilot. We delineated search areas based on locations of radiocollared caribou 1–2 days prior to and during the survey. The same experienced observer (L. G. Adams) classified individuals as cows, bulls, and calves by sex (Bergerud 1961) during each survey. In years that maternal penning occurred, we subtracted calves and cows that had been in the pen and were observed during the survey from the totals to derive calf:cow ratios for the herd without maternal penning. To determine herd-wide calf:cow ratios including maternal penning contributions, we used population size and composition to estimate the total number of wild-born calves, added in the pen calves known to be alive at the time of each survey, and recalculated the ratios.
We estimated population sizes in mid-October 2003, 2005, and 2007 with a sightability model to account for differences in detection relative to group size (Samuel et al. 1987, Udevitz et al. 2006). During 1–2 days prior to each census, we determined the distribution of radiocollared females by aerial telemetry, and delineated survey areas that encompassed non-forested, upland regions inhabited by nearly all the marked caribou (≥91% each year). We used obvious landscape features a few kilometers beyond located groups to define boundaries of survey areas. During 2 consecutive days, the helicopter crew systematically searched for caribou groups throughout each survey area without the aid of radiotelemetry. When the crew encountered a group, they identified radiocollared individuals by scanning frequencies or visual observations, then counted and classified all caribou. Simultaneously, the pilot of a Supercub aircraft located all radiocollared caribou in and adjacent to the survey area. The helicopter crew periodically conferred with the Supercub pilot to identify marked groups (i.e., those that included radiocollared caribou) that were missed in areas the helicopter crew had already searched; the helicopter crew then returned to locate and classify those missed groups. As the survey progressed, the Supercub pilot adjusted the survey area boundaries as needed to account for any changes in caribou distribution from the previous few days.
Statistical Analysis
Survival analyses
We expected that calves born in maternal penning would exhibit substantially higher survival to the mid-June pen release than their free-ranging counterparts. Further, adult females held in the maternal pens could experience higher survival in captivity than free-ranging cows. Given that each of the survival rates necessary to address these hypotheses was a simple proportion (i.e., no staggered entry or censoring), we compared rates among years for free-ranging caribou via χ2 analysis, and derived 95% confidence intervals for both their annual rates and the overall mean with the Wilson score method with continuity correction (Newcombe 1998). We considered survival rates to be significantly different if those calculated for the treatment groups from the pens fell outside the 95% confidence intervals of the estimates for their free-ranging counterparts. Similarly, we evaluated annual rates and their mean for survival of pen calves from birth to mid-October relative to those of calves born in the wild.
For other survival analyses of calves and adult females, we used Kaplan-Meier (KM) procedures (Kaplan and Meier 1958, Pollock et al. 1989, DeCesare et al. 2016) to assess survival patterns across seasons or years, and across ages of adult females. To investigate differences in survival among years or related to individual factors, we employed logistic regression as implemented in the known-fate analysis of Program MARK (White and Burnham 1999), and multiple model selection and inference procedures based on Akaike's Information Criterion adjusted for small sample size (AICc; Burnham and Anderson 2002). For each analysis, we started with a limited set of plausible models and ranked model performance based on AICc. We considered models with ∆AICc ≤ 2 as well supported but eliminated any model that met that criteria if it consisted of a higher ranking model with an additional uninformative parameter in accordance with Arnold (2010). We then averaged the resulting set of well-supported models to derive model coefficients, survival rates, and associated error measures (Burnham and Anderson 2002).
We evaluated factors that may have affected post-release survival of pen calves to mid-October, including year, age at release, birth mass, and sex. For these analyses, we treated the entire period as a single interval because all calves entered in mid-June as they were released from the pen and no calves were censored from the analysis to mid-October. For all 4 years of maternal penning, it was obvious that survival was lower in 2006 than other years, so we evaluated an indicator variable for year = 2006 and the age of each calf at release (days) as factors that could have explained variation in calf survival through the summer. We considered 8 models, including an intercept only model and 7 models with intercepts and all combinations of age at release, year = 2006, and their interaction. In 2004–2006, we sexed and weighed nearly all the calves that were released from maternal penning (n = 115 of 119 calves), so we also evaluated effects of those individual factors on post-release survival. We started with the well-supported models from the assessment of age at release and year = 2006 across all years of maternal penning, and added birth mass (kg), an indicator variable for sex (male = 1), and both variables resulting in 8 models to evaluate.
We compared seasonal survival of adult females among the 5 years of our study. We evaluated 6 models for each season including similar survival among all years, and each of the 5 years differing from the other years. Each model included an intercept and coefficients accounting for differences among intervals (3 intervals in summer and 4 intervals in winter based on our radiotracking schedule described above). We derived annual survival rates for adult females as the product of the seasonal predicted rates from model averaging. We assessed winter survival of calves during 2004–2006 in the same fashion.
We also evaluated age-specific survival of adult females during winter and summer seasons for use in our population models for determining the contribution of individuals to the population from maternal penning and for assessing the utility of maternal penning. First, we calculated KM survival rates for each age class from 1 to 18 years. We then evaluated a model set for each season that included 6 models, each with an intercept and terms for intervals during the season. The 6 models included no age effect (constant survival across all age classes), log-linear decline in survival across the entire age range, and 4 senescence threshold models (constant survival prior to the threshold age, then declining survival from the threshold age on). We expected that the actual underlying relationship for senescence would begin to decline at or prior to the first age class that exhibited an obvious drop in survival from our data. Therefore, for the threshold variables, we determined the first age class (X) where the KM survival rate differed from 1 (i.e., 95% CI did not include 1). We created a variable representing a senescence threshold at that age (OLDX = 0 for age < X; OLDX = age – X + 1 for age ≥ X) and similar threshold variables for the 3 age classes younger than X. Models with thresholds older than the 4 we evaluated consistently performed poorly. We multiplied the seasonal age-specific survival rates derived from model averaging to calculate annual age-specific survival rates.
Natality rate assessment
We compared natality rates among years and between methods (i.e., ultrasound exams during late Mar–early Apr captures; field observations in mid-May) via χ2 analyses. To determine age-specific natality rates for use in our population models, we first calculated the proportions deemed parturient for each age class and derived 95% confidence intervals (Wilson score method with continuity correction; Newcombe 1998). We then modeled senescence in natality rates for age classes ≥4 years (i.e., the age females initially reached consistently high natality) by employing logistic regression with multimodel selection and inference methods as described above for survival analyses. We evaluated 6 models including no age effect (constant natality across all age classes), log-linear decline with age, and 4 threshold models with constant natality prior to a threshold age, then a log-linear decline in natality from the threshold on. We noted the age-specific natality rates were consistently high for 4–9-year-olds, then declined notably at 10 years old, so we evaluated senescence models with thresholds of 7–10 years of age.
Population estimation
Other analyses
We conducted χ2 analyses to compare birth sex ratios relative to parity and among years. During our mid-October composition surveys, we classified a large proportion of the wild-born calves estimated to exist in the herd ( = 84%, range = 69–92%), and noted that female calves were more prevalent among wild-born calves than those from penning. If the perceived difference was significant, it would be necessary to account for the different calf sex ratios in our modeling. Therefore, to compare mid-October sex ratios of pen and wild-born calves, we first derived the 95% confidence interval for the wild-born sex ratio (Wilson score method with continuity correction; Newcombe 1998) with a finite population correction. We considered sex ratios to be significantly different if the results for pen calves fell outside the 95% confidence interval for wild-born calves.
We employed analysis of covariance to evaluate effects of penning (pen- vs. wild-born), year, and their interaction on birth mass, while controlling for differences related to sex. We used analysis of variance to evaluate factors that may have influenced mass of 5-month-old females (pen- vs. wild-born, year, and interaction), and mass of pregnant adult females in late March–early April (year). To evaluate significant differences identified for factors with >2 levels, we used Fisher's least significant difference (LSD) procedure (Saville 1990). In 2005–2006 when we weighed female caribou repeatedly throughout the penning period, we used t-tests to compare body mass between years at capture for penning (late Mar–early Apr), prior to the onset of calving (early May), and at release from penning (mid-Jun); we also compared their rates of mass gain (kg/day) between capture and early May between years. We assessed differences in body mass of pen females at release in mid-June with free-ranging females via a t-test. We used Pearson's correlation coefficient to evaluate the relationship between body mass of female pen calves at birth and 5 months of age, and to compare the proportion of the commercial ration in the diets of females in the maternal pens in 2006 relative to the amount of ration we provided. For all frequentist statistical analyses, we considered P ≤ 0.05 indicative of a significant result.
Contribution of Maternal Penning to the Chisana Herd
We determined the contribution of additional calves from maternal penning to mid-October each year as the number of pen calves that survived minus those that would have survived outside the penning program (i.e., number of pen calves born × summer free-ranging calf survival rate). We extended these to estimates of additional yearlings recruited in mid-May each year by multiplying estimates of calves contributed in mid-October by winter calf survival rates. We conducted Monte Carlo simulation of the survival rates and their associated errors through 5,000 iterations to derive 95% confidence intervals for these estimates of additional calves contributed from penning.
To assess the contribution of maternal penning to the size of the Chisana Herd each year, we needed to derive population estimates for mid-October 2003–2007 without the effect of maternal penning. To do so, we constructed a model to project the survival of additional calves contributed from maternal penning, as well as the production and survival of their progeny, through each autumn of our study (Fig. 6). We could then subtract the resulting number of additional individuals stemming from penning from our population estimates to determine the herd sizes without penning. For each cohort from penning, the model was based on 1) outcomes of the maternal penning program (additional calves contributed in mid-October each year, autumn pen calf sex ratio), and 2) vital rates of the Chisana Herd from our field studies (seasonal calf survival rates, summer and annual survival rates of young adult females [1–4 years old] pooled across age classes, age-specific natality rates for 2-, 3-, and 4-year-old females, autumn sex ratio of free-ranging calves). We did not evaluate the survival of adult males in our study, so we used summer and annual survival rates of young males (1–4 years old) from radiotelemetry studies of male caribou survival in the Denali Caribou Herd in central Alaska during 2007–2018 (0.961 [SD = 0.030] and 0.882 [SD = 0.031], respectively; n = 304 caribou-years for 1–4-year-old males; age-specific rates did not differ for these age classes [P > 0.204 and 0.061 for pairwise log-rank tests of summer and annual rates, respectively]; L. G. Adams, unpublished data). We applied annual winter calf survival and age-specific natality rates that differed significantly among categories, otherwise we used average or pooled rates. We derived 95% confidence intervals for our estimates of total caribou contributed from maternal penning in the population each autumn via Monte Carlo simulation of additional pen calves, natality rates, survival rates, and their associated errors, through 5,000 iterations.
We had no pretreatment population estimate of the Chisana Herd for October 2002 prior to our maternal penning efforts. However, the number of adults in October 2003 was not affected by maternal penning, and by October 2007, the additional calves from the last maternal penning effort in 2006 had been recruited as adults. Therefore, given that adult sex ratios varied over the course of our study, we focused on changes in adult female numbers to evaluate maternal penning effects on population trend during 2003–2007. To estimate female abundance in 2003 and 2007, and the resulting finite rate of increase with and without maternal penning, we multiplied population estimates by the proportion of adult females in the population as observed during composition surveys. We determined 95% confidence intervals for the rates of increase with Monte Carlo simulations of population estimates, female proportions, additional adult caribou from penning in 2007, and their associated errors through 5,000 iterations.
Assessment of Utility of Maternal Penning
To evaluate the broader applicability of maternal penning as a management tool for improving the status of small at-risk caribou populations, we used deterministic simulation models to assess the population effects of a maternal penning program with our observed calf survival results over a range of λi = 0.850–1.000, and penning effort = 0–100% of the population's calves born in the pen. We constructed an initial model of the female segment of a caribou population based on the average vital rates and other demographic characteristics of the Chisana Herd, and the average results of our maternal penning program (Fig. 7). The initial model used age-specific natality and seasonal survival schedules derived from our field research, and was implemented with 19 age classes, ranging from 0 (calves) to 18 years. We used calf survival results during the penning period and the entire summer to divide summer calf survival into 2 periods (birth to pen release in mid-Jun, release to mid-Oct) and derived separate rates for pen and wild-born calves. We used a common winter survival rate for both pen and wild-born calves. We determined the number of female calves surviving to mid-October based on pooled autumn sex ratios of pen and wild-born calves observed in the field. We established the age structure for the initial model by iteratively running the model without maternal penning until we reached a stable age distribution. We censused the model population in early May each year immediately prior to production of calves, and calculated the finite rate of increase (λ) as described by Caughley (1977). We designed the model to run for 20 years allowing for 5 years of maternal penning at the onset of the simulation period by setting penning effort >0.
Once we configured the initial model based on Chisana Herd demography with no maternal penning, we derived models with λi = 0.850, 0.900, 0.950, and 1.000 by changing adult female survival and calf survival outside the pen such that each contributed equally to the adjustment in λ from the base model (DeCesare et al. 2012, Hervieux et al. 2013). With resulting population models that performed at each λi, we then varied penning effort across its full range (0–100% of calves born in penning) and assessed differences in treatment effects. Similarly, we determined λi where the model population was stable during 5 years of 100% penning effort.
RESULTS
Maternal Penning – Parturition, Calf Mass, and Calf Survival
During late March–early April 2003–2006, we captured 175 adult females to stock the annual maternal penning efforts (Table 1). Of those, 5 died as a result of capture and handling. We released 21 females that we deemed not pregnant via ultrasound. In 2003, we retained 3 females in the pen following equivocal ultrasound exams; they did not produce calves. Overall, 146 pregnant females were included in the maternal penning program, ranging from 17 in 2003 to 50 each in 2005 and 2006 (Table 1). In 2004–2006, caribou held in the pens included 1, 8, and 14 individuals, respectively, that had been in the maternal pens in previous years.
2003 | 2004 | 2005 | 2006 | Total | |
---|---|---|---|---|---|
Cows captured (late Mar–early Apr) | 22 | 37 | 58 | 58 | 175 |
Capture-related mortalities | 1 | 3 | 1 | 0 | 5 |
Not pregnant; released | 1 | 5 | 7 | 8 | 21 |
Cows held to mid-Jun | 20 | 29 | 50 | 50 | 149 |
Not pregnant | 3 | 0 | 0 | 0 | 3 |
Calves born (early May–early Jun) | 17 | 29 | 50 | 50 | 146 |
Calves stillborn, died, or abandoned | 0 | 0 | 5 | 5 | 10 |
Calves released (mid-Jun) | 17 | 29 | 45 | 45 | 136 |
Calves alive (mid-Oct) | 13 | 22 | 37 | 24 | 96 |
Calf survival | |||||
Birth–mid-Jun | 1.000 | 1.000 | 0.900 | 0.900 | = 0.950 |
Mid-Jun–mid-Oct | 0.765 | 0.759 | 0.822 | 0.533 | = 0.720 |
Birth–mid-Oct | 0.765 | 0.759 | 0.740 | 0.480 | = 0.686 |
Birth timing, sex ratios, and mass
The annual onset of calving occurred from 7 to 14 May, calving continued for 23–32 days, and the last calf each year was born between 31 May and 8 June (Fig. 8A). Median calving dates ranged from 17 to 20 May. Overall, 53% of the births occurred during a 7-day period centered on the median (range = 41–59% annually). The timing of calving was positively skewed in that the initial 75% of births occurred in 15 days, whereas the remaining 25% stretched over the succeeding 19 days (Fig. 8B).
During 2004–2006, we determined the sex of 128 of 129 calves born in the maternal pens (1 calf was stillborn or died shortly after birth and was scavenged prior to detection), and 100 wild-born calves captured within 2 days of birth for weighing. Both samples were skewed slightly in favor of females (51.5% and 53.0 % female, respectively), but the overall birth sex ratio did not differ significantly from parity (52.2% female; = 0.44, P = 0.508). Annual birth sex ratios did not vary significantly among years (47.5–56.7% female; = 1.43, P = 0.488).
In 2004–2006, we weighed 121 of 125 neonates that survived beyond birth in the maternal pens, as well as 100 calves born in the free-ranging population for comparison. Overall, calves averaged 8.5 kg at birth but varied widely (Table 2). On average, males were 0.9 kg heavier than females ( = 9.0 vs. 8.1 kg; F1,214 = 21.16, P ≤ 0.001). We detected no difference in birth mass between calves from the pen and those born in the wild ( = 8.6 and 8.5 kg, respectively; F1,214 = 0.00, P = 0.999), but mean birth mass differed among the 3 years (F2,214 = 28.69, P ≤ 0.001; LSD test, P ≤ 0.05), averaging 8.8, 7.8, and 9.2 kg in 2004, 2005, and 2006, respectively. The penning×year interaction was not significant (F2,214 = 1.48, P = 0.229).
Male | Female | |||||||
---|---|---|---|---|---|---|---|---|
n | SD | Range | n | SD | Range | |||
Pen | ||||||||
2004 | 11 | 9.4 | 1.3 | 7.0–11.6 | 16 | 8.0 | 1.1 | 6.0–9.9 |
2005 | 24 | 8.5 | 1.4 | 4.8–11.6 | 24 | 7.5 | 1.3 | 4.5–9.6 |
2006 | 24 | 9.6 | 1.2 | 7.4–11.4 | 22 | 8.7 | 1.0 | 6.6–10.2 |
All years | 59 | 9.1 | 1.4 | 4.8–11.6 | 62 | 8.0 | 1.3 | 4.5–10.2 |
Wild | ||||||||
2004 | 17 | 9.3 | 1.0 | 7.2–11.3 | 12 | 8.5 | 1.3 | 6.2–10.8 |
2005 | 14 | 7.8 | 1.2 | 5.4–9.7 | 26 | 7.3 | 1.0 | 5.2–9.5 |
2006 | 16 | 9.2 | 1.4 | 7.4–11.6 | 15 | 9.3 | 0.9 | 8.0–11.2 |
All years | 47 | 8.8 | 1.3 | 5.4–11.6 | 53 | 8.2 | 1.4 | 5.2–11.2 |
Overall | 106 | 9.0 | 1.4 | 4.8–11.6 | 115 | 8.1 | 1.3 | 4.5–11.2 |
Calf survival to release from pens
In 2003 and 2004, all calves produced in the pen survived to release in mid-June, whereas in 2005 and 2006, 5 calves each year were stillborn, died, or were abandoned (Table 1). These included 4 calves that were stillborn or died shortly following birth because of physiologic problems, and 1 perinatal death within 48 hours of birth of bacterial septicemia (G. A. Wobeser, Canadian Cooperative Wildlife Health Centre, University of Saskatchewan, unpublished report). The remaining 5 cases included 1) a calf abandoned by its dam after capture for radiocollaring and taken permanently into captivity 24 hours later; 2) a sickly calf that died at 8 days old of undetermined causes (autolysis precluded determination of a cause of death but no evidence of trauma or destructive infectious disease was noted; G. A. Wobeser, unpublished report); 3) a calf whose mother died about 48 hours following birth from complications of uterine torsion (the calf exhibited neurological problems consistent with a difficult birth, was briefly held in captivity, and subsequently euthanized), 4) a calf exhibiting neurological problems that suffered a compound leg fracture at 25 days old and was euthanized; and 5) an 11-day-old calf killed by a black bear that got into the pen on 10 June 2006 and we killed shortly after we detected it.
Survival of calves in maternal penning to the mid-June release was 1.000 in 2003–2004, and 0.900 in 2005–2006, or 0.950 on average (Table 1; Fig. 9). In comparison, survival of wild-born calves during the same period in 2003–2007 ranged from 0.300 in 2004 to 0.444 in 2003 (Table 3; Fig. 9) but did not differ significantly among years ( = 1.77, P = 0.778). Among the years of maternal penning (2003–2006), survival of wild-born calves to mid-June averaged 0.378, or 40% of the survival of calves reared in the maternal pens (Fig. 9).
2003 | 2004 | 2005 | 2006 | 2007 | Mean survival | |
---|---|---|---|---|---|---|
Calf survival to mid-June (proportion of parturient radiocollared females with calves at heel) | ||||||
Parturient females | 18 | 30 | 40 | 64 | 111 | |
Calves at heel mid-June | 8 | 9 | 17 | 22 | 41 | |
Calf survival | 0.444 | 0.300 | 0.425 | 0.344 | 0.369 | 0.376 |
95% CI | 0.224–0.692 | 0.154–0.496 | 0.274–0.590 | 0.232–0.474 | 0.281–0.467 | 0.318–0.438 |
Calf survival to mid-October (calculated from autumn calf:cow ratios and other vital rates) | ||||||
Calf survival | 0.270 | 0.222 | 0.197 | 0.221 | 0.170 | 0.216 |
95% CI | 0.215–0.329 | 0.166–0.275 | 0.146–0.247 | 0.173–0.275 | 0.133–0.212 | 0.193–0.239 |
Calf survival to mid-October
Survival of pen calves from release to mid-October was similar during 2003–2005 but reduced in 2006 (Tables 1, 4; Appendix A). Reduced survival in 2006 resulted from higher losses of calves prior to 20 July, or 36 days post-release (Fig. 10). Post-release calf survival increased with age at release from the maternal pens (Table 4; Appendix A). Given the temporal distribution of births (Fig. 8), the median age of the pen calves by the mid-June release was 26 days; however, the youngest 25% of calves were 6–23 days of age (Fig. 11) and these younger calves experienced increased mortality following release. Two models of survival of pen calves from release to mid-October 2003–2006 were well-supported (Table 4), and both included age at release as a variable. The resulting model-averaged relationship demonstrated that survival probability was low for the youngest calves released and improved markedly up to about the median age (Fig. 12A; Appendix A), increasing by 21% with each additional day of age (odds ratio = 1.21; 95% CI = 1.10–1.32). In 2006, post-release survival was reduced across all ages (Fig. 12A). Projections based on the model-averaged relationship between age at release and post-release survival indicated that delaying release until all calves were ≥14 days old, delays of 3–8 days depending on the year, could have increased survival of calves to mid-October by 9–30% (Fig. 12B), with the greatest effect in 2006 when survival to mid-October was particularly low.
Model statisticsa | ||||
---|---|---|---|---|
Model variables | K | AICc | ∆AICc | w |
Age, age×YR06 | 3 | 136.22 | 0.00 | 0.50 |
Age, YR06 | 3 | 137.26 | 1.04 | 0.30 |
Age, YR06, age×YR06b | 4 | 138.09 | 1.86 | 0.20 |
Age | 2 | 147.79 | 11.57 | 0.00 |
YR06 | 2 | 153.92 | 17.70 | 0.00 |
YR06, age×YR06 | 3 | 154.47 | 18.25 | 0.00 |
Age×YR06 | 2 | 156.09 | 19.87 | 0.00 |
Intercept only | 1 | 163.16 | 26.94 | 0.00 |
- a Model statistics include the number of parameters (K), Akaike's Information Criterion corrected for small sample sizes (AICc), change in AICc relative to top model (∆AICc), and AICc weight (w).
- b Model excluded from further consideration. Although ∆AICc ≤ 2, model includes a single additional variable added to a higher ranking model (Arnold 2010).
For 2004–2006 when most pen calves were weighed and sexed (n = 115 of 119 calves released), we also assessed birth mass and sex as individual factors that might influence post-release survival. Two well-supported logistic regression models of calf survival to mid-October included birth mass as a variable, in addition to age at release and a 2006 indicator variable (Table 5; Appendix A). Based on the model-averaged regression results, the odds of surviving were 1.47 times greater for each kg increase in birth mass (odds ratio = 1.47, 95% CI = 1.01–2.16). We found no support for calf sex influencing post-release survival (Table 5).
Model statisticsa | ||||
---|---|---|---|---|
Model variables | K | AICc | ∆AICc | w |
Age, age×YR06, birth mass | 4 | 120.52 | 0.00 | 0.35 |
Age, YR06, birth mass | 4 | 121.94 | 1.42 | 0.17 |
Age, age×YR06, birth mass, sexb | 5 | 122.30 | 1.78 | 0.15 |
Age, age×YR06 | 3 | 122.66 | 2.13 | 0.12 |
Age, YR06, birth mass, sex | 5 | 123.73 | 3.21 | 0.07 |
Age, YR06 | 3 | 123.67 | 3.14 | 0.07 |
Age, age×YR06, sex | 4 | 124.79 | 4.27 | 0.04 |
Age, YR06, sex | 4 | 125.81 | 5.29 | 0.03 |
- a Model statistics include the number of parameters (K), Akaike's Information Criterion corrected for small sample sizes (AICc), change in AICc relative to top model (∆AICc), and AICc weight (w).
- b Model excluded from further consideration. Although ∆AICc ≤ 2, model includes a single additional variable added to a higher ranking model (Arnold 2010).
Given the high survival of calves in the pens, survival of those calves from birth to mid-October exhibited the same pattern as their post-release survival in that rates were similar for 2003–2005 but lower in 2006 (Table 1; Fig. 9B). Survival of wild-born calves from birth to mid-October was markedly lower than pen calves in each year of maternal penning, ranging from 0.197 in 2005 to 0.270 in 2003 (Table 3; Fig. 9B), and averaged 33% (range = 27–46%) that of calves born in the maternal pens within years.
Sex ratios of calves surviving to mid-October
During 2004–2006, both sexes were nearly equally represented among the pen calves that survived to mid-October (49.4% female; n = 83). However, mid-October sex ratios of calves born in the wild throughout our study were biased toward females (58.4% females, 95% CI = 56.1–60.6%, n = 298 calves classified during 2003–2007 composition surveys).
Body mass at 5 months of age
We captured and weighed 61 female calves in mid-October 2004–2006, including 30 pen and 31 wild-born individuals (Table 6). The lightest calf (41.9 kg) was an outlier 11.8 kg lighter than the next smallest calf and we excluded it from analyses; this calf from the penning program in 2006 was orphaned in early July. Pen calves averaged 65.7 kg (Table 6), and mass at birth and in mid-October were positively correlated (r = 0.49, n = 28, P = 0.004). On average, these calves grew at a rate of 0.4 kg/day (n = 28, range = 0.3–0.5 kg/day) from birth to mid-October. Overall, we noted no differences in mass at 5 months between pen and wild-born calves (Table 6; F1,54 = 0.02, P = 0.885) or among cohorts (Table 6; F2,54 = 0.54, P = 0.583). However, the interaction of those variables was significant (F1,54 = 5.03, P = 0.010) in that 2004 wild-born calves were lighter than calves born in the pen, differing by 7.0 kg on average, whereas in other years mass was similar between the 2 groups (Table 6).
n | SD | Range | ||
---|---|---|---|---|
Pen | ||||
2004 | 10 | 68.7Aa | 6.5 | 59.0–81.0 |
2005 | 10 | 64.6A,B | 5.8 | 58.1–74.4 |
2006b | 9 | 63.7A,B | 5.4 | 53.8–69.8 |
All yearsb | 29 | 65.7 | 6.1 | 53.8–81.0 |
Wild | ||||
2004 | 10 | 61.7B | 4.6 | 53.7–69.0 |
2005 | 10 | 68.7A | 7.6 | 55.3–78.9 |
2006 | 11 | 65.9A,B | 5.2 | 56.2–75.1 |
All years | 31 | 65.5 | 6.4 | 53.7–78.9 |
Overall | 60 | 65.6 | 6.2 | 53.7–81.0 |
- a Means with the same letters are not different (least significant difference test; P > 0.05).
- b We excluded an outlier weighing 41.9 kg; individual was orphaned in early July.
Maternal Penning – Adult Female Nutritional Performance and Survival
Diets of adult females in captivity
Adult females placed in captivity quickly accepted the pelleted ration. Based on microhistological analyses of feces in 2006, the commercial feed constituted about 27% of the diet of females in the pen following their first week in captivity (Fig. 13A). The addition of the pelleted feed to their diets primarily resulted in a reduction in the proportion of lichen, which constituted 70% of the diet of free-ranging caribou (Fig. 13A), even though we provided 0.5 kg dry lichen/caribou/day on average and additional lichen was naturally available in the pen.
During early April–mid-June 2006, the pelleted ration comprised an increasing proportion of the diet of females in the pen, reaching 54% of their diet in June, and was offset by comparable declines in the proportion of lichen, the other major diet component (Fig. 13B). The increasing reliance on the ration was corroborated by the strong correlation between the proportion of ration in the diet and the average amount provided each week (kg/adult female/day; r = 0.95, n = 10, P ≤ 0.001) to meet their ad libitum demands, as well as some limited consumption of pellets by calves. Native forages averaged about 18% of the diet of females during maternal penning, composed mainly of mosses and shrubs (Empetrum spp., Vaccinium spp., Salix spp., Dryas spp., and Betula spp., in order of importance). Small increases in vascular forages and decreases in moss consumption were noted after mid-May as emergent green tissues became available (Fig. 13B).
Mass dynamics of adult females during maternal penning
Pregnant females captured in late March–early April 2003–2006 for maternal penning averaged 119.2 kg, varying from 89.8 kg to 147.6 kg (Table 7). Mean mass differed among years (F3,142 = 4.44, P = 0.005) in that cows were lighter in 2005 than 2003 and 2006, and intermediate in 2004 (Table 7).
Year | Capture dates | n | SD | Range | |
---|---|---|---|---|---|
2003 | 25–27 Mar | 17 | 121.6Aa | 6.4 | 111.2–129.4 |
2004 | 2–5 Apr | 29 | 119.1A,B | 11.0 | 96.0–139.4 |
2005 | 30 Mar–5 Apr | 50 | 115.0B | 11.5 | 89.8–140.3 |
2006 | 29 Mar–2 Apr | 50 | 122.5A | 11.0 | 97.8–147.6 |
All years | 146 | 119.2 | 11.1 | 89.8–147.6 |
- a Means with the same letters are not different (least significant difference test; P > 0.05).
In 2005 and 2006, when we weighed individuals repeatedly in the pen, females were lighter at capture in 2005 than 2006 by 7.5 kg on average (Table 8). By early May, just before the onset of parturition, females in 2005 still averaged 3.5 kg lighter than those in 2006, but the difference was no longer significant (Table 8). Thus, 2005 females exhibited a rate of gain during the initial 5 weeks in the pen that exceeded that in 2006 by 50% ( = 0.46 [n = 42, SD = 0.19] and 0.31 [n = 40, SD = 0.19] kg/day, respectively; t80 = 3.67, P ≤ 0.001). On average, females lost 14% of their body mass from delivery of their calves in both years (n = 54, range = 5–27%). At release in mid-June, the pen females averaged 111.4 kg in both years, and were 5.4 kg, or 5%, heavier than their free-ranging counterparts (Table 8).
Year | n | SD | Range | Test results | ||
---|---|---|---|---|---|---|
Pen | ||||||
Capture (~1 Apr) | 2005 | 50 | 115.0 | 11.5 | 89.8–140.3 | |
2006 | 50 | 122.5 | 11.0 | 97.8–147.6 | t98 = 3.36, P = 0.001 | |
Pre-calving (~6 May) | 2005 | 42 | 128.0 | 12.6 | 101.5–154.5 | |
2006 | 40 | 131.5 | 11.2 | 110.9–155.8 | t80 = 1.33, P = 0.189 | |
Release (~14 Jun) | 2005 | 39 | 111.4 | 9.5 | 90.6–134.1 | |
2006 | 32 | 111.4 | 9.2 | 91.6–131.2 | t69 = 0.02, P = 0.980 | |
Wild | ||||||
~14 Jun | 2005 | 28 | 105.0 | 8.9 | 82.4–120.9 | |
2006 | 30 | 107.0 | 11.6 | 79.7–124.9 | t56 = 0.76, P = 0.449 | |
Pen vs. wild (~14 Jun) | ||||||
Pen | Both | 71 | 111.4 | 9.3 | 90.6–134.1 | |
Wild | Both | 58 | 106.0 | 10.4 | 79.7–124.9 | t127 = 3.10, P = 0.002 |
Adult female survival during maternal penning
Of the 149 females held in maternal penning during late March–mid-June, only 1 died; that cow succumbed because of birth complications within 48 hours of delivering her calf on 20 May 2006. Thus, survival of females in the pen overall was 0.993. In comparison, survival of free-ranging females with years pooled for the same period was significantly lower at 0.951 (95% CI = 0.910–0.975); survival did not differ significantly among years for free-ranging females (n at risk = 205, 23–83/year; range = 0.911–1.000; = 3.45, P = 0.327).
Other Demographic Rates
Natality
We assessed natality of females ≥2 years old in the Chisana Herd during the 2003–2008 calving seasons based on 625 observations of 222 individuals, including examinations of females from 174 of 175 captures in late March–early April 2003–2006 for maternal penning (a 2003 capture mortality was not examined), and 451 observations of free-ranging radiocollared females during mid-May 2003–2008 (Table 9). Annual natality rates did not differ significantly between the 2 samples in any year ( ≤ 1.21, P ≥ 0.259), although natality of the captured sample tended to be higher. Annual natality estimates based on both data sets ranged from 0.778 in 2003 to 0.888 in 2007 (Table 9) but did not differ significantly among years ( = 5.18, P = 0.394), averaging 0.842 over the 6 years of our study.
Mar–Apr ultrasound exams | Mid-May calving observations | Combined | ||||
---|---|---|---|---|---|---|
n | Rate | n | Rate | n | Rate | |
2003 | 21 | 0.810 | 24 | 0.750 | 45 | 0.778 |
2004 | 37 | 0.838 | 36 | 0.833 | 73 | 0.836 |
2005 | 58 | 0.879 | 50 | 0.800 | 108 | 0.843 |
2006 | 58 | 0.862 | 80 | 0.800 | 138 | 0.826 |
2007 | 125 | 0.888 | 125 | 0.888 | ||
2008 | 136 | 0.882 | 136 | 0.882 | ||
0.847 | 0.826 | 0.842 |
We derived age-specific natality rates from 598 observations of 206 known-age females. We did not have ages for 16 individuals that were captured prior to March 2003 (n = 5), were captured for the 2003 maternal penning (n = 8), or were released as not pregnant and not radiocollared, during March–April 2003 and 2006 captures (n = 3). Of the 67 individuals observed as 2-year-olds, 38 (56.7%) were parturient (Fig. 14; Appendix B). We continued to monitor 19 of the 29 individuals that did not produce calves at 2 years of age; the remaining 10 included 8 that turned 2 years old in May 2008 at the end of our study, and 2 released as nonpregnant and without radiocollars in April 2005. Of the 19, 17 produced their first calf at 3 years; the remaining 2 females produced their first calves at 4 and 5 years.
Natality rates increased to 0.873 for 3-year-olds and remained consistently ≥0.933 for 4–9-year-olds (Fig. 14; Appendix B). Four models of declining natality with age for females ≥4 years old were well-supported with the strongest support for reproductive senescence beginning at 8–9 years of age (Table 10). Estimated natality rates derived from model-averaging ranged from 0.947 for 4–6-year-olds declining to 0.467 for 19-year-olds (Fig. 14; Appendix B).
Model statisticsa | ||||
---|---|---|---|---|
Model variables | K | AICc | ∆AICc | w |
Old8 | 2 | 266.09 | 0.00 | 0.29 |
Old9 | 2 | 266.14 | 0.05 | 0.28 |
Old7 | 2 | 266.72 | 0.63 | 0.21 |
Old10 | 2 | 267.54 | 1.45 | 0.14 |
Age | 2 | 268.46 | 2.37 | 0.09 |
Intercept only | 1 | 287.68 | 21.59 | 0.00 |
- a Model statistics include the number of parameters (K), Akaike's Information Criterion corrected for small sample sizes (AICc), change in AICc relative to top model (∆AICc), and AICc weight (w).
Adult female survival
We estimated survival rates of free-ranging adult females based on 208 radiocollared individuals that were monitored for ≤5 years each during mid-May 2003–mid-May 2008, accumulating 471 caribou-years of survival observations. Our sample included 23 females that were instrumented prior to the beginning of our study in March 2003, 123 individuals captured for the maternal penning program during 2003–2006, 6 nonpregnant females radiocollared and released during 2004 penning captures, and 56 females fitted with radiocollars at 5 months of age in mid-October 2004–2006 that survived beyond 1 year of age. During our study, 48 of these females died of natural causes, and 24 were censored from survival analyses (22 collar removals, 2 capture mortalities). Mortality observations were consistent with predation in 45 cases (36 wolf, 3 bear, 6 wolf or bear in summer) and causes other than predation in 3 cases (1 avalanche, 2 undetermined but not predation).
Summer KM survival rates for adult females ranged from 0.928 to 0.979 during the 5 summers of our study (Table 11). Regression models that were well-supported included the base model and the model for a difference between 2007 and the other years (Table 12) indicating that survival may have been higher in 2007 (0.972 and 0.948 for 2007 and all other years, respectively; Table 11). Winter KM survival rates were ≥0.940 in 4 of 5 winters of our study but only 0.827 during the relatively severe winter 2004 (Table 11). The logistic regression model describing this difference was the only one with support in our assessment (Table 12), providing predicted rates of 0.819 and 0.965 for winter 2004 and all other winters, respectively (Table 11). In winter 2004, adult female survival diverged from other years primarily after mid-January (Fig. 15A). Given these differences in model-derived rates of seasonal survival, annual survival estimates ranged from 0.776 in 2004 to ≥0.915 in the other years (Table 11).
Biological year | Mean n at risk | Deaths | s | 95% CI | Model rate |
---|---|---|---|---|---|
Summer (mid-May–mid-Oct; 3 intervalsa) | |||||
2003 | 36.3 | 3 | 0.930 | 0.887–0.973 | 0.948 |
2004 | 66.0 | 3 | 0.961 | 0.936–0.986 | 0.948 |
2005 | 90.3 | 7 | 0.928 | 0.887–0.970 | 0.948 |
2006 | 127.0 | 6 | 0.952 | 0.920–0.985 | 0.948 |
2007 | 141.0 | 3 | 0.979 | 0.955–1.000 | 0.972 |
Winter (mid-Oct–mid-May; 4 intervalsa) | |||||
2003 | 39.3 | 2 | 0.946 | 0.875–1.000 | 0.965 |
2004 | 67.5 | 12 | 0.827 | 0.760–0.893 | 0.819 |
2005 | 94.3 | 2 | 0.978 | 0.958–0.998 | 0.965 |
2006 | 133.3 | 8 | 0.940 | 0.926–0.954 | 0.965 |
2007 | 136.8 | 2 | 0.986 | 0.966–1.000 | 0.965 |
Annual (mid-May–mid-May; 7 intervalsa) | |||||
2003 | 38.0 | 5 | 0.880 | 0.808–0.952 | 0.915 |
2004 | 66.9 | 15 | 0.794 | 0.721–0.868 | 0.776 |
2005 | 92.6 | 9 | 0.908 | 0.853–0.962 | 0.915 |
2006 | 130.6 | 14 | 0.895 | 0.857–0.933 | 0.915 |
2007 | 138.6 | 5 | 0.965 | 0.934–0.995 | 0.938 |
- a Intervals for survival analysis were defined by our radiotracking schedule each year (mid-May, mid-Jun, mid-Aug, mid-Oct, early Dec, mid-Jan, mid-Mar).
Model statisticsa | ||||
---|---|---|---|---|
Model | K | AICc | ∆AICc | w |
Summer (mid-May–mid-Oct; 3 intervals) | ||||
YR07 | 4 | 226.67 | 0.00 | 0.36 |
Base model | 3 | 227.80 | 1.13 | 0.20 |
YR05b | 4 | 228.09 | 1.42 | 0.18 |
YR03 | 4 | 229.05 | 2.38 | 0.11 |
YR04 | 4 | 229.79 | 3.12 | 0.08 |
YR06 | 4 | 229.81 | 3.14 | 0.07 |
Winter (mid-Oct–mid-May; 4 intervals) | ||||
YR04 | 5 | 254.44 | 0.00 | 0.99 |
YR07 | 5 | 263.18 | 8.74 | 0.01 |
YR05 | 5 | 268.07 | 13.64 | 0.00 |
Base model | 4 | 268.97 | 14.53 | 0.00 |
YR06 | 5 | 270.93 | 16.49 | 0.00 |
YR03 | 5 | 270.97 | 16.53 | 0.00 |
- a Model statistics include the number of parameters (K), Akaike's Information Criterion corrected for small sample sizes (AICc), change in AICc relative to top model (∆AICc), and AICc weight (w).
- b Model excluded from further consideration. Although ∆AICc ≤ 2, model includes a single additional variable added to a higher ranking model (Arnold 2010).
Of the 208 free-ranging caribou females monitored for survival estimation, we determined ages for all but 12 individuals. These 12 caribou were radiocollared prior to May 2003 and died by July 2005, before we had recaptured them and extracted teeth for aging, and no incisors were recovered at their death sites. Age-specific survival rates were based on 453 caribou-years of observation. Age-specific survival of adult females exhibited a pattern of senescence for both seasons (Fig. 16). Logistic regression models exhibited strongest support for a summer senescence threshold at 13 years old, whereas the age-related pattern for winter survival was best described by a consistent decay across all age classes (Table 13). Model rates for summer survival remained ≥0.950 through 12 years of age, then declined sharply for the oldest age classes, whereas winter survival stayed above that level to 9 years of age with a shallower subsequent decline (Fig. 16; Appendix C).
Model statisticsa | ||||
---|---|---|---|---|
Model | K | AICc | ∆AICc | w |
Summer (mid-May–mid-October; 3 intervals) | ||||
Old13 | 4 | 155.85 | 0.00 | 0.29 |
Old11 | 4 | 156.32 | 0.47 | 0.23 |
Old12 | 4 | 156.56 | 0.71 | 0.20 |
Old14 | 4 | 156.67 | 0.83 | 0.19 |
Age | 4 | 158.21 | 2.36 | 0.09 |
Base model | 3 | 184.10 | 28.25 | 0.00 |
Winter (mid-October–mid-May; 4 intervals) | ||||
Age | 5 | 185.65 | 0.00 | 0.54 |
Old8 | 5 | 186.51 | 0.86 | 0.35 |
Old9 | 5 | 189.10 | 3.44 | 0.10 |
Old10 | 5 | 192.65 | 7.00 | 0.02 |
Old11 | 5 | 196.50 | 10.85 | 0.00 |
Base model | 4 | 208.53 | 22.88 | 0.00 |
- a Model statistics include the number of parameters (K), Akaike's Information Criterion corrected for small sample sizes (AICc), change in AICc relative to top model (∆AICc), and AICc weight (w).
Calf winter and annual survival
We estimated survival of calves during winters 2004–2006 based on 114 radiocollared calves (Table 14), including 83 pen calves of both sexes that survived to mid-October (Table 1) and 31 wild-born females captured and radiocollared in mid-October. We censored 9 calves from the analyses (6 dropped their elastic calf collars prior to their first birthday in mid-May; 3 radiocollars deployed in October 2006 on 5-month-old females failed prematurely). Mortality observations were consistent with wolf predation for all 17 calves that died during winter.
Winter | n at risk | Censored | Died | Rate | 95% CI | Model rate |
---|---|---|---|---|---|---|
2004 | 32 | 3 | 9 | 0.690 | 0.548–0.831 | 0.688 |
2005 | 48 | 2 | 5 | 0.893 | 0.826–0.959 | 0.897 |
2006 | 34 | 4 | 3 | 0.900 | 0.798–1.000 | 0.897 |
Winter KM survival rates for calves were similar for 2005 and 2006 and markedly reduced in 2004 (Table 14; Fig. 15B). The logistic regression model of this pattern of winter calf survival was the only one that was well supported (Table 15) and resulted in predicted rates of 0.688 and 0.897 for the relatively severe winter 2004 and other winters, respectively (Table 14). As with adult females, calf survival during winter 2004 exhibited a marked decline after mid-January (Fig. 15B).
Model statisticsa | ||||
---|---|---|---|---|
Model variables | K | AICc | ∆AICc | w |
YR04 | 5 | 133.22 | 0.00 | 0.75 |
Base model | 4 | 137.34 | 4.12 | 0.10 |
YR05 | 5 | 137.77 | 4.56 | 0.08 |
YR06 | 5 | 137.86 | 4.64 | 0.07 |
- a Model statistics include the number of parameters (K), Akaike's Information Criterion corrected for small sample sizes (AICc), change in AICc relative to top model (∆AICc), and AICc weight (w).
Although we only had radiocollared calves for estimating winter calf survival during 3 of the 5 winters of our study, in those winters calf survival and adult female survival exhibited the same pattern of consistently high rates in winters 2005 and 2006, and reduced survival in winter 2004 (Tables 12, 15; Fig. 15). Therefore, we assumed that the model calf survival rates for winters 2005 and 2006 (Table 14) could reasonably be applied to winters 2003 and 2007, given that we noted adult female survival for those winters that was similar to 2005 and 2006. Given this assumption, annual survival of wild-born calves averaged 0.184 for the 5 years (Table 16; Fig. 17). Deaths that occurred between birth and mid-June accounted for most of the mortality of calves, averaging 76% of their annual mortality, compared to 20% and 4% for the remainder of summer, and winter, respectively (Fig. 17). Given these winter survival rates, annual survival of pen calves averaged 0.575 (95% CI = 0.552–0.599), compared to 0.192 (95% CI = 0.168–0.217) for wild-born calves during the 4 years of penning (Table 16).
Pen | Wild | |||
---|---|---|---|---|
s | 95% CI | s | 95% CI | |
2003 | 0.686 | 0.656–0.716 | 0.242 | 0.190–0.294 |
2004 | 0.522 | 0.415–0.629 | 0.153 | 0.104–0.202 |
2005 | 0.663 | 0.634–0.693 | 0.177 | 0.133–0.223 |
2006 | 0.430 | 0.411–0.449 | 0.198 | 0.152–0.245 |
2007 | 0.152 | 0.116–0.189 | ||
0.575 | 0.552–0.599 | 0.184 | 0.164–0.205 |
Herd Growth and Composition
During mid-October censuses of the Chisana Herd in 2003, 2005, and 2007, we detected 67–83% of the marked groups (Table 17); we detected all marked groups of ≥17 caribou (n = 33, range = 17–65 caribou). Detection probabilities increased significantly with group size for each census (Table 17). Overall, unmarked groups tended to be small (n = 61, median = 3 caribou, range = 1–20 caribou). During these censuses, we counted and classified 83–93% of the caribou estimated to occur in the population each year (Table 17). In addition, we conducted classification surveys in mid-October 2004 and 2006 that achieved comparably large sample sizes (Table 18).
2003 | 2005 | 2007 | |
---|---|---|---|
Survey dates | 19–20 Oct | 15–16 Oct | 13–14 Oct |
Area surveyed (km2) | 2,230 | 1,470 | 910 |
Radiocollared caribou (% in survey area) | 39 (100%) | 98 (91%) | 138 (96%) |
Caribou observed (n groups) | |||
Marked groups detected in survey area | 389 (20) | 410 (35) | 615 (25) |
Marked groups missed in survey area | 70 (10) | 66 (10) | 30 (5) |
Unmarked groups observed in survey area | 144 (20) | 123 (25) | 55 (16) |
Marked groups outside of survey | 0 | 47 (3) | 19 (4) |
Total caribou | 603 (50) | 646 (73) | 719 (50) |
Logistic regression resultsa | |||
β0 | −1.216 | −0.280 | −0.390 |
β1 | 0.166 | 0.175 | 0.178 |
G12 (P) | 9.84 (0.002) | 5.88 (0.015) | 7.64 (0.006) |
Population estimates (95% CI) | 720 (603–850) | 706 (646–809) | 766 (719–834) |
- a Coefficients and significance test for logistic regression relationship between detection probability (p) and group size (logit (p) = β0 + β1 group size).
Calves:100 cowsa | ||||
---|---|---|---|---|
n | with MP | without MP | Bulls:100 cows | |
19–20 Oct 2003 | 603 | 24.9 | 23.0 | 37.1 |
11–12 Oct 2004 | 538 | 20.3 | 17.0 | 38.3 |
15–16 Oct 2005 | 646 | 22.2 | 15.6 | 45.9 |
12 Oct 2006 | 628 | 20.6 | 17.6 | 48.8 |
13–14 Oct 2007 | 719 | 13.1 | 49.5 |
- a Calf:cow ratios calculated with and without the contribution of calves from the maternal penning (MP) program.
In October 2003, we estimated the Chisana Herd to number 720 caribou (Table 17; Fig. 18), or more than twice the 315 caribou estimated for 2002 (Gardner 2003), indicating that population size had been substantially underestimated in 2002. Subsequent censuses corroborated the higher 2003 population size with the herd reaching 766 caribou by October 2007 (Table 17; Fig. 18).
During 2003–2006, calf:cow ratios in mid-October averaged 22.0:100 including calves contributed from penning, compared to 18.3:100 without those additional pen calves (Table 18), or nearly 3 times the average of 6.7 calves:100 cows during 1989–2002 (Fig. 3B). The adult sex ratio also increased markedly, from <25 bulls:100 cows during 1995–2002 (Fig. 3C) to >45:100 during 2005–2007 (Table 18).
Contribution of Maternal Penning to the Chisana Herd
We estimated that maternal penning contributed 64.1 additional calves surviving to mid-October (95% CI = 60.0–68.1; 8, 16, 27, and 13 each year 2003–2006, respectively) beyond those that would have survived had they been born in the wild. These additional calves from maternal penning equaled 8–29% of the total calves in the herd in mid-October of those years. Accounting for their overwinter survival, 54.2 (95% CI = 49.9–58.5) additional caribou were recruited as yearlings as a result of maternal penning, or 40% of the 136 calves released during the 4 years of the program.
We estimated that the herd would have been essentially stable across the period from October 2003 to October 2007 (712 and 714 caribou, respectively) without maternal penning, whereas we observed an increase to 766 caribou by October 2007 with penning (Fig. 18). By that time, the maternal penning program had provided 52.2 additional caribou (95% CI = 47.9–56.5), including survivors from maternal penning and their subsequent surviving progeny. The growth rate of the adult female segment of herd improved during 2003–2007 from λ = 0.997 (95% CI = 0.953–1.049) without maternal penning to λ = 1.015 (95% CI = 0.971–1.065) with additional pen individuals.
Assessment of Utility of Maternal Penning
The initial population model for free-ranging females derived from average vital rates and calf sex ratios from our field studies (Table 19) was essentially stable (λ = 0.995). By adjusting calf and adult female survival, we produced model populations with λi = 0.850, 0.900, 0.950, and 1.000; vital rates for this suite of models (Table 20) were comparable to those from field studies of caribou populations within this range of growth rates (Rettie and Messier 1998, Wittmer et al. 2005a, Hervieux et al. 2013).
Parameter | Values; Sources |
---|---|
Age-specific natality | Appendix B |
Age-specific adult survival (winter, summer) | Appendix C |
Calf survival | |
Wild-born | |
Birth–mid-Jun | 0.378 (2003–2006) |
Birth–mid-Oct | 0.228 (2003–2006) |
Mid-Oct–mid-May | 0.845 (2003–2006) |
Pen | |
Birth–mid-Jun | 0.950 |
Birth–mid-Oct | 0.686 |
Mid-Oct–mid-May | 0.845 (2003–2006) |
Autumn calf sex ratios (mid-October) | |
Pen calves | 49.4% female |
Wild-born calves | 58.4% female |
λi | ||||
---|---|---|---|---|
0.850 | 0.900 | 0.950 | 1.000 | |
Natality (calves/F ≥2 years old) | 0.856 | 0.858 | 0.860 | 0.860 |
Calf survival | ||||
Wild-born | ||||
Birth–mid-Jun | 0.305 | 0.331 | 0.356 | 0.380 |
Mid-Jun–mid-Oct | 0.487 | 0.528 | 0.568 | 0.607 |
Mid-Oct–mid-May | 0.682 | 0.739 | 0.795 | 0.850 |
Annual | 0.101 | 0.129 | 0.161 | 0.196 |
Pen-born (no penning occurred for λi; rates used for modeling penning results) | ||||
Birth–mid-Jun | 0.950 | 0.950 | 0.950 | 0.950 |
Mid-Jun–mid-Oct | 0.583 | 0.632 | 0.680 | 0.726 |
Mid-Oct–mid-May | 0.682 | 0.739 | 0.795 | 0.850 |
Annual | 0.377 | 0.444 | 0.514 | 0.587 |
Calf recruitment without maternal penning (calf F/F ≥1 year old) | ||||
Mid-Oct (5 months old) | 0.074 | 0.086 | 0.098 | 0.110 |
Mid-May (1 year old) | 0.060 | 0.072 | 0.085 | 0.099 |
F annual survival (≥1 year old) | 0.802 | 0.840 | 0.876 | 0.910 |
In our simulations, population responses stemming from maternal penning were determined by the interaction of λi and penning effort (Fig. 19). For λi = 1.000, comparable to the inherent trajectory of the Chisana Herd during our study, calves surviving to recruitment each contributed to a persistent increase in caribou numbers (Fig. 20). Thus, population growth resulting from maternal penning could be demonstrated even at low penning effort. As λi dropped below 1.000, calf survival declined and more of the additional recruits from maternal penning were required to offset increased losses of adult females. Therefore, the penning effort necessary for stability increased (Fig. 19). For example, reaching stability with λi = 0.950 and 0.900 required penning efforts of 38% and 87%, respectively; stability could not be attained with 100% penning effort below a threshold of λi = 0.890 (Figs. 19, 20). Further, populations returned to declining trends similar to those without maternal penning once the management intervention ended (Fig. 20).
Although stabilizing or increasing caribou numbers may be the most obvious goal of a maternal penning program, slowing declines could be a useful outcome for conserving small at-risk populations resulting in their greater persistence on the landscape. As with reaching stability, increases in population persistence were greater for slower inherent declines (Fig. 21). For example, with 50% of the calves born in penning for 5 years, a population decline of 25% could be delayed for 9 years for λi = 0.950. With λi = 0.900, the same population threshold would be postponed for 3 years. For λi = 0.850, only 1 year would be gained, and caribou numbers would decline beyond the 25% threshold during the third year of penning.
DISCUSSION
Maternal penning proved to be effective at eliminating most early mortality for calves born in the pens by protecting them from predation. Only 5% of the pen calves died prior to release in mid-June, and these deaths largely resulted from deficiencies or complications at birth. In comparison, 62% of the wild-born calves were dead by mid-June during 2003–2006 when penning was conducted; these deaths in the few weeks following birth accounted for 76% of the annual calf mortality. Calves born in penning maintained their survival advantage over wild-born calves with annual survival rates averaging 0.575 (95% CI = 0.552–0.599) and 0.192 (95% CI = 0.168–0.217), respectively. We estimated that maternal penning provided 54.2 (95% CI = 49.9–58.5) additional yearling recruits to the herd, or 40% of the 136 calves released from maternal penning.
In October 2003, we determined that the Chisana Herd numbered 720 caribou (95% CI = 603–850; including 8 additional calves from maternal penning), or more than twice that previously estimated. Also, following 14 years of chronically low autumn calf:cow ratios averaging 6.7:100 (1989–2002; range = 0.1–13.8:100), autumn recruitment unexpectedly increased in 2003 and averaged 17.3:100 (without calves contributed from penning) throughout our studies. Even with the improvement noted in recruitment, low calf survival was a key factor limiting population growth given the magnitude of calf mortality observed and the relatively high natality and annual survival of adult females we noted. The Chisana Herd grew to 766 caribou (95% CI = 719–834) by mid-October 2007 as a result of individuals contributed by maternal penning; the herd would have been stable at about 713 caribou without the penning treatment. The improvement in population trend from our management action was limited by the larger than expected herd size and the associated reduced penning effort ( = 11% of calves born in pen).
Our simulation modeling corroborated our field results in that for inherently stable populations like the Chisana Herd, population size increased by the number of additional recruits provided, even at low penning effort. As λi dropped below 1.000, the impact of maternal penning on improving herd growth declined as more additional recruits were needed to offset the downward population inertia; thus, increased penning effort was required to reach stability. Given the vital rates in our models, stability could not be attained with 100% penning effort for populations declining at λi < 0.890. Therefore, although we were able to induce an increase in caribou numbers for the Chisana Herd with low penning effort, that improvement was enabled by inherent stability of the population during our management treatment. Maternal penning may have utility for populations that are slowly declining within a narrow range of λi, but improving herd trend will be dependent on high penning effort.
Proximate Effects of Maternal Penning
Maternal penning greatly increased survival of caribou calves from birth to release into the wild in mid-June. On average, 95% of the calves born in maternal penning were released, whereas only 38% of calves born in the wild were still alive by that time. These initial weeks following birth were the critical period for calf mortality, accounting for 76% of the annual calf deaths for the free-ranging population. Although we did not determine the causes of death for wild-born calves, calf mortality studies of other montane caribou populations in nearby Alaska and British Columbia have reported early losses of similar magnitude, with the vast majority of calf deaths attributed to predation, primarily by bears and wolves (Page 1985; Adams et al. b; Jenkins and Barten 2005; Gustine et al. 2006).
Of the 9 calves that died prior to release from maternal penning, 5 were stillborn or died within 48 hours of birth from health deficiencies and 3 others exhibited obvious physiological problems from birth to their deaths. It is difficult to detect and quantify these early nonpredation losses of newborn ungulates in free-ranging populations because such deaths commonly occur at birth or before neonates can be radiocollared for mortality studies, and offspring with health deficits frequently die of predation as the proximate cause (Whitten et al. 1992, Roffe 1993, Linnell et al. 1995). These deaths are likely the greatest source of bias in studies of neonatal mortality in free-ranging ungulate populations and are rarely accounted for (Linnell et al. 1995). We suspect nonpredation perinatal losses regularly occur and the magnitude we noted (5%) was not unusual. Adams et al. (1995a) studied caribou calf mortality by locating parturient radiocollared females daily and collaring their calves; they noted 11 of 147 calves (7%) were lost to stillbirths (n = 3) and unspecified perinatal causes within 24 hours of birth (n = 8), although some of the unspecified deaths may have resulted from predation (L. G. Adams, unpublished data). Ågren (2001) recorded 6% nonpredation deaths for semidomesticated reindeer (Rangifer tarandus) in Sweden held in a large enclosure for a month during calving, similar to 4% perinatal losses observed in semidomestic reindeer in Scotland (Jorgensen et al. 2015). Stillbirths of roe deer (Capreolus capreolus) fawns increased from 5% to 15% of those observed during a 4-fold increase in population density on a predator-free island off the coast of Norway (Andersen and Linnell 1998). Two studies of white-tailed deer fawn mortality that employed vaginal implant transmitters to locate newborn fawns reported 2–3% losses to stillbirths (Kilgo et al. 2014) or underweight fawns dying within 2 days of birth (Carstenson et al. 2009). Nonpredation perinatal losses of 2–7% are commonly reported for domestic livestock (Patterson et al. 1987, Binns et al. 2002, Mee et al. 2008, Bluel 2011).
Female adults tolerated captivity well during the 10 weeks they were held in the pens. By the end of the first week in captivity in 2006, the commercial feed constituted 27% of their diets, and that doubled by the end of the captive period. That year, the acceptance of the pelleted ration may have been facilitated by the presence of 14 females that had experience in the pen in previous years. With the addition of the pelleted ration and resulting decreases in other diet components by the end of the first week in captivity, the overall digestibility of the diet they consumed increased by 10% (60% vs. 66%) and the nitrogen concentration doubled (0.6% vs. 1.2%) over the diets of free-ranging caribou, based on values for each diet component (Boertje 1981, Barboza and Parker 2006, Gustine et al. 2011). With ad libitum consumption of the high-quality pelleted ration and forage lichens we provided, penned females gained 11 kg on average, or 9% of their initial body mass, during the first 5 weeks of captivity, when free-ranging females were likely losing mass through these last weeks of winter (Mautz 1978, Leader-Williams 1988, Parker et al. 2009). Of note, pregnant females captured for penning in 2005 were lighter than other years but by early May had made up most of their mass deficits on the ad libitum feed we provided. By the mid-June release in 2005–2006, cows from the pen were 5% heavier than their free-ranging counterparts.
Pen females produced calves that were of comparable mass to those born in the wild. We hypothesized that pen calves might be heavier than wild-born calves given the improved nutrition of their mothers late in gestation when most fetal growth occurs (Robbins and Robbins 1979, Skogland 1984). However, Chisana newborns were generally large compared to calves from other caribou populations (Whitten et al. 1992, Valkenburg et al. 2016, Adams 2005, Dale et al. 2008, Couturier et al. 2009), and thus may approach the upper limits of birth mass for the species. Further, Adams (2005) provided evidence that caribou females invest conservatively in their offspring, compared to other ungulates of similar size (Oftedal 1985, Robbins and Robbins 1979), in favor of improving their own nutritional status in late winter. Thus, Chisana females may have preferentially directed any nutritional gain from penning to replace their own body reserves rather than investing in producing larger calves.
Adult females held in the maternal pens also benefitted from the protection from predation. Only 1 female died (0.7% mortality) during the 4 years of maternal penning, compared to 4.9% mortality for free-ranging females during the same period. However, pregnant females maintained in the maternal pens likely underrepresented the older age classes of females that exhibit both lower productivity and lower survival, particularly in late winter during the penning period, and thus were not directly comparable to the population at large. Further, the capture and handling of female caribou for penning and to meet other study objectives presented risks of capture-related mortality (Spraker 1982, Valkenburg et al. 1983, Arnemo et al. 2006). In addition to the 5 capture-related deaths that occurred while stocking the maternal pens, we experienced 3 others associated with other aspects of our study (n = 129 captures of caribou that we radiocollared and monitored subsequent to their capture, and 55 captures of caribou that survived the handling process but were not radiocollared; L. G. Adams, unpublished data). Thus, higher survival of females held in the maternity pens likely offset the capture-related losses we experienced. Given these results, we assumed that survival of females in the pen and in the wild did not differ for our population models.
Calves released from maternal penning in mid-June maintained their survival advantage over wild-born calves through to mid-October; survival of pen calves from birth to autumn averaged 3 times that of wild-born calves. Post-release survival of pen calves was noticeably lower in 2006 compared to other years, but their survival to mid-October was still more than twice that of wild-born calves that year. In 2006, caribou released from maternal penning were distributed widely across the herd's range by early July, whereas in other years, pen caribou remained well aggregated in 2–5 nearby groups within 15 km of the pen sites (L. G. Adams, unpublished data). We suspect this difference in post-release distribution resulted from greater predator activity in 2006 in the vicinity of the Big Boundary Lake penning site. Alternatively, females dispersed more widely in 2006 for some other reason and the higher calf losses resulted from exposure to predators over a broader distribution in early summer 2006.
The age of pen calves at the time of release influenced their subsequent survival through summer. Similar to timing of calving documented for the Denali Caribou Herd (Adams and Dale 1998b), the calving season of the Chisana Herd stretched over about a month with 75% of the births occurring in the first half of the period, and the remaining 25% over the second half. Therefore, although ≥75% of calves were over 3 weeks old by the mid-June release, late-born calves were still relatively young, with a few as young as 6 days. These youngest calves exhibited poor survival, and survival increased with age as calves reached the median age at release of 26 days. We timed the pen release during 11–14 June each year with the goal that nearly all calves would be protected during the critical first 2 weeks of their lives when we expected most mortality to occur (Adams et al. b; Jenkins and Barten 2005); however, 5 of 136 calves released were <2 weeks old. In addition, we were concerned that holding caribou in the pens longer could result in negative effects of insect harassment (Helle and Tarvainen 1984, Russell et al. 1993, Witter et al. 2012) as temperatures warmed in lower-elevation treeline areas where the pens were located; by mid-June, free-ranging caribou were selecting habitat well above treeline. However, our results indicated that calves continued to experience elevated mortality through about 28 days of age. Thus, holding calves longer, even for a few additional days to allow the youngest calves to reach the 2-week threshold and to increase the number of calves that were old enough to experience maximal survival following release, would have likely increased the summer survival of pen calves. We estimated that holding calves an additional 3–8 days, depending on the year, so that all calves released were >2 weeks old may have increased survival over the summer by 9–30%. Further, insect conditions at the pen sites would have been tolerable for caribou held for a few additional days.
Summer survival of individual calves following release from the pens was also positively correlated with their birth mass. Although several studies of caribou calf mortality have reported no relationship between birth mass of individuals and their survival during the initial few weeks of life (Mahoney et al. 1990; Whitten et al. 1992; Adams et al. b; Jenkins and Barten 2005), Jenkins and Barten (2005) did note that calf survival from 15 days of age through 30 September was positively related to birth mass. For calves <2 weeks old, high vulnerability of all individuals to predation (Adams et al. b; Jenkins and Barten 2005) may overwhelm any effects of comparatively subtle physical differences. Beyond that age, the vulnerability of calves to predation declines, and wolves become the main predator still able to capture them (Adams et al. b). Individual characteristics may become more important in determining risks of succumbing to predation. We noted that differences observed in body mass of calves at birth persisted in those surviving to mid-October; thus, low birth mass was indicative of low relative body size throughout the summer. Although young caribou may ultimately compensate for low birth mass (Dale et al. 2008), that compensation occurs well after the initial weeks of life when calves experience most of their predation mortality.
Mid-October body mass of female calves were similar for those born in the pen and in the wild in 2005 and 2006, indicating that although pen calves likely benefitted to some degree from the improved nutrition their mothers experienced while in captivity, any nutritional advantage over wild-born calves did not persist to autumn in those years. By autumn, Chisana calves were generally relatively heavy compared to calves from other Alaskan caribou herds (Dale et al. 2008, Valkenburg et al. 2016), indicating that they commonly experienced lactational support and foraging conditions that facilitated rapid growth during their first summer. In 2004, however, calves born in the pens were heavier than their wild-born counterparts at 5 months of age, by an average of 7 kg or 11%. The 2004 summer was the warmest and third driest in the region over a century of weather observations (Wendler et al. 2011), and markedly warmer across the Chisana range compared to other years of our study. Warm summer temperatures increase insect harassment of caribou, which raises their energy expenditures while reducing foraging time (Helle and Tarvainen 1984, Russell et al. 1993, Colman et al. 2003, Witter et al. 2012), Further, warm and dry conditions can reduce forage quality (Lenart 1997, Lenart et al. 2002). We suspect that the growth of wild-born calves may have been reduced in summer 2004, but the nutritional benefits accrued to pen calves were sufficient to buffer the extreme environmental conditions that summer.
Population Dynamics of the Chisana Herd
The Chisana Herd was one of several small caribou populations across interior Alaska and southwest Yukon that exhibited population declines during 1989–1993 characterized by very poor recruitment and associated with abundant winter snowfall (Adams et al. 1995a, 2006; Valkenburg et al. 1996; Adams and Dale 1998a; Hayes et al. 2003). Winters with deep snow increase energetic costs, reduce forage availability for caribou (Thing 1977, Fancy and White 1985, Russell et al. 1993), and increase risks of wolf predation (Mech et al. 1995, 1998; Mech and Peterson 2003; Adams et al. 2006). Condition-related effects of the unusual winter severity that were detected in interior Alaskan caribou included reduced productivity (Valkenburg et al. 1996, Adams and Dale 1998a), delayed parturition (Adams and Dale 1998b), lower mass at birth (Adams et al. 1995a, Adams 2005) and at 10 months of age (Valkenburg et al. 1996; L. G. Adams, unpublished data), and reduced initial growth of calves (Adams 2003). Although these indicators of nutritional effects disappeared with a return to more normal weather conditions, low recruitment lingered for about a decade in these herds (Valkenburg et al. 1996, 2016; Jenkins and Barten 2005; Adams et al. 2006; Roffler et al. 2012). Increased predation on young calves was clearly a major factor driving these population declines during the period of abnormal winter conditions (Valkenburg et al. 1996; Adams et al. b; Jenkins and Barten 2005), but it is unclear why unusually high predation losses of calves persisted after those conditions ameliorated.
Prior to 1989, there were only sporadic, limited attempts to enumerate the Chisana Herd (Skoog 1968, Hemming 1971, Kelleyhouse 1990) largely because of its remote range with limited access for hunting compared to the much larger, migratory populations in the region (i.e., Nelchina and Fortymile herds; Valkenburg et al. 1996). The first thorough census of the Chisana Herd was conducted in late June 1989; Kelleyhouse (1990; ADF&G, unpublished report) counted 1,660 caribou that were well aggregated in large post-calving groups. Given that few males older than calves are usually associated with these cow-calf groups, the herd probably numbered around 2,000 caribou given the adult sex ratio at the time. In late June 1992, Gardner (1993; ADF&G, unpublished reports) completed a post-calving photocensus (Valkenburg et al. 1985) of well-aggregated caribou and counted 1,234 caribou; the population likely numbered about 1,500 caribou after accounting for underrepresented males. This decline of 9%/year was consistent with the drastic drop in autumn calf recruitment noted from regular composition surveys, culminating with only a single surviving calf observed with 870 cows in late September 1992 (Gardner 1993).
Several additional post-calving censuses of the Chisana Herd were attempted during 1993–2002 (Gardner 2003), but with their numbers declining, caribou tended to be widely scattered in small groups. Further, only a few radiocollars were deployed in the herd to aid in conducting counts. Thus, those efforts were deemed unsuccessful (C. L. Gardner, ADF&G, personal communication and unpublished reports). Without acceptable herd counts, population estimates were based on a simple population model (Valkenburg et al. 2016) combining autumn sex-age composition and plausible adult survival rates to project the herd's trend beyond the 1992 census to roughly track minimum counts from composition surveys (Gardner 2003). Although this modeling approach was intended for “projecting population estimates for a year or 2 after a good census” (Valkenburg et al. 2016:199) and until another census could be conducted, it served as the sole basis for Chisana herd estimates for 10 years to 2002. Without additional censuses to corroborate or adjust the model-based population trend, the estimate of 315 caribou in 2002 (Gardner 2003) was less than half the actual number in the population that year. Thus, from 1992 to 2002, the Chisana Herd declined by about 7%/year, driven largely by continuing low calf recruitment to autumn, averaging 6 calves:100 cows. Along with the low recruitment, the adult male component of the population declined from 39 bulls:100 cows on average during 1986–1991, to an average of 19:100 during 1996–2000, even though annual harvests, composed primarily of males and averaging about 2% of the herd, were largely eliminated after 1994 (Gardner 2003).
During our intensive studies, autumn calf:cow ratios without penning improved to 17.3:100 on average, and the prolonged population decline came to an end. Although population size varied across the 5 years of our study, the herd would have been generally stable at about 713 caribou without maternal penning, whereas it increased by 7% to 766 caribou with penning. With higher calf recruitment including calves added from penning, the adult sex ratio increased to ≥46 bulls:100 cows by October 2005.
Similar to other small low-density caribou populations (Seip and Cichowski 1996, Adams et al. 1995b, Adams and Dale 1998a, Jenkins and Barten 2005, Wittmer et al 2005a), Chisana caribou exhibited little evidence of chronic nutritional limitation during our study in that females were large bodied and highly productive, with the majority of females delivering their first offspring at 2 years of age, and 93% of those ≥3 years old producing calves. Females 4–9 years old produced calves at rates exceeding 92%, and even though age-specific rates displayed a senescent decrease beginning at about 8 years of age, the oldest females were still producing calves about 50% of the time. In addition, Chisana calves were large at birth relative to birth mass reported for other caribou populations (Whitten et al. 1992, Adams 2005, Dale et al. 2008, Couturier et al. 2009, Valkenburg et al. 2016). Calves grew rapidly over their first summer and maintained their large relative body size compared to other caribou populations (Dale et al. 2008, Couturier et al. 2009, Valkenburg et al. 2016). We suspect mature Chisana females were commonly at a plane of nutrition by the autumn breeding season that ensured high productivity resulting from their low density on the landscape and the relative abundance of nutritious summer forage (Lenart et al. 2002). Further, high losses of calves to predation within a few weeks of birth reduced the nutritional demands of lactation for the majority of productive females, allowing greater recovery of nutritional reserves before winter and their next reproductive effort (Adams and Dale 1998a, Allaye Chan-McLeod et al. 1999).
Even though calf:cow ratios to mid-October improved compared to those during the population decline that preceded our study, calf losses were still high in the Chisana Herd. Free-ranging calves experienced an average mortality rate of 0.624 between birth and mid-June, accounting for 76% of the calf mortality that occurred throughout the entire year. We are confident that this early mortality of free-ranging calves largely resulted from predation given that calves held in the maternal pens experienced very high survival while protected. In addition, other investigators have consistently documented predation, predominantly attributed to wolves and bears, as the prevalent cause of early calf mortality in other small caribou herds in northwestern North America (Adams et al. b; Jenkins and Barten 2005; Gustine et al. 2006). Such early calf losses were an important and pervasive influence on population growth in that more caribou died each year as young calves during this period of about a month than died in all other age classes throughout the entire year. Calf deaths during the rest of the summer and winter accounted for 20% and 4%, respectively, of total annual calf losses. Given this level of calf mortality, about 18% of the calves survived to their first birthday on average. Although calf survival during winter 2004 was significantly lower than the other years, annual survival was similar among years because of the consistency of other seasonal rates and the limited contribution of winter losses to the annual pattern of the calf survival.
We noted a shift in calf sex ratio from 52% females at birth to 58% females by mid-October, indicating that males experienced more mortality than females during their first 5 months. Given average survival for Chisana calves of 0.216 from birth to mid-October, sex-specific survival rates of 0.242 and 0.188 for females and males, respectively, would account for the observed shift in calf sex ratio. Much of the differential mortality must have occurred before 15 June in that most calf mortality took place then and we noted no effect of sex on calf survival following release from the pen. Caribou commonly have female-biased adult sex ratios (Bergerud 1980), but the age at which differential survival between the sexes begins is uncertain. Bergerud (1971) and Mathisen et al. (2003) provided evidence that male caribou and reindeer calves exhibit risky behavior, including straying farther from their dams and higher activity levels, leading to higher predation mortality. Similar results indicating risky behaviors exhibited by young males have been reported for white-tailed deer (Jackson et al. 1972, Schwede et al. 1992) and free-ranging domestic sheep (Ovis aries; May et al. 2008). Mahoney et al. (1990) reported a male bias in the predation mortality of radiocollared caribou calves over their first year, although other telemetry-based studies of calf mortality have not detected differences in survival related to sex (Adams et al. 1995b, Jenkins and Barten 2005, Mahoney et al. 2016). For other ungulate studies that report on analyses of sex as a factor in predation-caused mortality of young, results either support increased mortality of males or no differences between the sexes (Linnell et al. 1995).
Large herbivores, such as caribou, generally exhibit a pattern of relatively high and stable adult survival compared to that of juveniles (Gaillard et al. 1998). For the Chisana Herd, adult female annual survival averaged 0.892 during the 5 years of our study, higher than most rates reported for other small caribou populations in montane or boreal forest environments (Rettie and Messier 1998, Jenkins and Barten 2005, Wittmer et al. 2005a, Courtois et al. 2007, Hervieux et al. 2013). On average, seasonal survival rates were similar (0.953 vs. 0.936, for summer and winter, respectively) given that the winter rate was calculated over 7 months of the year. Females ≤10 years old experienced annual survival >0.900, with a senescent decline in survival beyond those age classes. Although annual survival exceeded 0.900 in 4 of 5 years, it was lower in 2004, as a result of higher winter losses that year. Calves also experienced poor survival that winter. In addition to reduced survival of caribou, we also observed indicators of reduced nutritional condition from that winter in that females captured in late March–early April 2005 and the offspring they subsequently produced were lighter than in other years. The 2004 winter was characterized by a relatively deep, persistent snowpack; weather conditions prior to mid-February produced ice crusts in the snowpack that may have hampered foraging and mobility of caribou throughout the remainder of the winter (L. L. Larocque, YDE, personal observations). The distribution of caribou was largely limited to the Canadian portion of their range that winter, whereas in other winters they were well distributed throughout. During January–March 2005, much of the herd was further restricted to the Alaska Highway corridor and the mountains within 25 km of the highway to the southwest (83% of 132 radiolocations of adult females; L. G. Adams, unpublished data), areas used little or not at all during other winters. In addition, the extreme temperatures during summer 2004 (Wendler et al. 2011) may have resulted in reduced nutritional condition of caribou entering winter 2004, given the lower mass of free-ranging 5-month-old females in October.
Although adult sex ratios in the Chisana Herd had declined by half in association with the very low recruitment during 1991–2001, the relative abundance of males responded with improved recruitment after 2001 by increasing from an average of 19 bulls:100 cows during 1996–1999 to 48 bulls:100 cows during 2005–2007. Thus, without maternal penning during 2003–2007, adult male numbers would have increased by about 45 individuals, or 27%, based on population size and composition estimates, while female numbers were stable. Other authors have noted similar shifts in adult sex ratios for caribou related to large changes in calf recruitment (Schaefer et al. 1999, Hayes et al. 2003). In any population, adult sex ratios are determined by the relative recruitment and adult survival rates of the 2 sexes. With stable recruitment and sex-specific survival, a population converges to a constant rate of increase, and the proportion in any defined population segment (i.e., sex or age class) will become fixed, as will the ratio of any 2 segments such as the adult sex ratio (Caughley 1977). Thus, sizable shifts in calf recruitment can be expected to evoke corresponding responses in adult sex ratio like we noted in the Chisana Herd both as the population experienced very low calf recruitment during 1989–2001 and as recruitment improved after 2001, even in the absence of harvests directed at males. To date, discussion of adult sex ratios of ungulates has focused largely on selective harvest of males and the resulting effects of reduced male numbers on other population attributes (Ginsberg and Milner-Gulland 1994, Mysterud et al. 2002, Milner et al. 2007). However, there is little recognition that adult sex ratios can fluctuate substantially because of natural shifts in recruitment, a highly variable vital rate in ungulate populations (Gaillard et al. 1998).
Contribution of Maternal Penning to the Chisana Herd
We demonstrated that maternal penning essentially eliminated mortality due to predation while caribou calves were held; we experienced only 1 predation death in 4 years of penning when a black bear breached the pen and was quickly dispatched. However, once released from the pen, calves were subject to mortality that occurred in the population at large throughout the remainder of their first year. Further, some of the calves from penning would have survived to recruitment if born in the wild. Of the 136 calves we released from maternal penning, we estimated that 54.2 (40%) counted as additional yearling recruits. Given that the herd would have been stable without maternal penning, the increase in herd size we estimated between mid-October 2003 and mid-October 2007 that resulted from penning (52.2 caribou) was comparable to that number.
Although we established that maternal penning was a viable approach for increasing caribou calf survival that led to increased herd size, our impact on the Chisana Herd was limited in part because the herd was substantially larger than estimated prior to this management action. As we began maternal penning in 2003, we thought the herd included about 300 caribou, or about 40% of the actual number. Given that smaller herd size with the same maternal penning, results we would have realized an 18% increase in caribou numbers by 2007, or 2.4 times the magnitude of our observed effect. At over 700 caribou, the Chisana Herd was clearly too large for maternal penning at any reasonable scale to result in more than limited population increase.
Utility of Maternal Penning
To evaluate the usefulness of maternal penning for improving the status of small imperiled caribou herds, we focused on assessing the interaction of λi and penning effort relative to attaining population stability or growth. Given that we defined penning effort as the proportion of the calves produced in maternal penning, population size was not a necessary attribute in our modeling exercise. However, in practice, population size is important as a determinant of the number of pregnant females necessary to reach a given penning effort, and thus the cost and feasibility of employing this method.
For inherently stable populations, like the Chisana Herd during 2003–2007, any additional caribou recruited via maternal penning led to population growth and the magnitude of that growth increased with increasing penning effort. Even though our penning effort was low, about 11% on average, we were able to induce a modest increase in the Chisana Herd. Further, increases in modeled populations persisted after penning ended given the inherent stability. For models of declining populations, as λi declined fewer individuals produced from penning were recruited because calf survival following release also declined, and more of the resulting recruits were needed to offset increasing adult female losses. Thus, penning effort necessary to reach stability increased as λi declined. Once maternal penning ended, these model populations resumed their negative trends. Persistence of any gains resulting from penning also dropped along with λi. Given the demographic rates in our models, stability could not be attained with all calves born in penning for λi < 0.890. Therefore, maternal penning has the greatest potential to provide for growth of populations that are inherently stable or experiencing limited declines. As population declines steepened, the efficacy of maternal penning for even slowing declines was limited regardless of penning effort.
Our models for assessing the utility of maternal penning were relatively simple but driven by vital rates derived from our studies and comparable to those reported for at-risk caribou populations (Seip and Cichowski 1996, Rettie and Messier 1998, Wittmer et al. 2005a, Hervieux et al. 2013). Such simple models are important heuristic tools for improving our understanding of interacting processes (Starfield and Bleloch 1986). However, although our models were useful for intellectual exploration of the efficacy of maternal penning, the actual responses of caribou populations to maternal penning can be expected to differ from our model predictions for several reasons. First, each model run was initiated with a stable age structure derived from vital rates that resulted in a particular population trajectory. Vital rates are generally not stable for the life span of individuals within a population and often vary from year to year. Thus, actual age structures undoubtedly deviate from stable age structures and invoke momentum that drive population trends different from those predicted from simple models (Koons et al. 2006). Further, our models did not include any density-dependent influences on vital rates. Inverse density dependence, or Allee effects (Allee et al. 1949), have been reported to influence adult female survival and rates of increase for small caribou populations in British Columbia (Wittmer et al. 2005b, 2007, 2010; McLellan et al. 2010). Where Allee effects are operating, population trajectories would diverge from those derived from our simple models if the magnitude of Allee effects was sufficient to affect population growth over the ranges of abundance modeled. Finally, for small populations, maternal penning is likely to involve few individuals each year. In these cases, stochastic effects can result in actual outcomes for pen calf survival, as well as vital rates of the small populations being treated, that are highly variable and differ substantially from the mean rates driving our models (Mills 2013).
In response to the Chisana program, 3 other maternal penning efforts have been conducted in Alberta and British Columbia in attempts to improve the status of small at-risk caribou populations; these efforts provide additional insights on the utility of maternal penning. In 2006, a pilot trial with 10 pregnant females (~25% penning effort) was performed in the Little Smoky Herd, a population of about 80 caribou in west-central Alberta that was declining at about 6%/year (Smith and Pittaway 2011, Hervieux et al. 2013). Nine calves were released in mid-June (1 calf died in the pen), but survival of pen calves following release was nominally lower than that of wild-born calves by March 2007 (Smith and Pittaway 2011). Further, wolf control was implemented throughout the herd's range beginning in December 2005 prior to maternal penning, with wolf numbers reduced by about 50% that winter (Hervieux et al. 2013). Given poor survival of pen calves following release, the high cost of penning, and the plan for continuation of intensive wolf control, maternal penning ended after 1 year (Smith and Pittaway 2011).
In 2014, a 5-year pilot program was initiated to determine if penning could improve calf survival in the Columbia North caribou population of southeastern British Columbia (Serrouya et al. 2015a). At the time, this population was stable at about 150 caribou because of a management experiment initiated in 2003 to benefit caribou that reduced wolf numbers by 50% in the region via a substantial harvest-driven reduction in abundance of moose, the primary prey for wolves (Serrouya et al. 2015b, 2017b). During 2014–2017, 47 pregnant females were placed in maternal penning (~20% average annual penning effort), and after several calf deaths in the pen, 36 calves were released into the wild in midsummer. Of the calves released, 19 were known to survive to 10 months of age, for about 6 additional recruits resulting from maternal penning (Furk and Serrouya 2017; Revelstoke Caribou Captive Rearing in the Wild [RCRW] 2018; R. Serrouya, University of Alberta, personal communication). The program continued with its fifth year in 2018. Also, the British Columbia government commenced aerial wolf control in February 2017, further reducing wolf abundance in the Columbia North range to assist recovery of the caribou population and increase calf survival from the maternal penning project (BCMF 2017, Cornwall 2018). Recent population estimates indicated the population has been relatively stable from 2004 to late winter 2018 (Furk and Serrouya 2017; R. Serrouya, personal communication); thus, 4 years of maternal penning has not resulted in any detectable improvement in population trend.
Another maternal penning effort was initiated in 2014 in northeast British Columbia as a 5-year emergency measure to stem the precipitous decline of the Klinse-Za (also referred to as Moberly-Scott) caribou population (McNay and Sittler 2013, Seip and Jones 2016a). The targeted population was estimated at about 36 individuals in March 2013 (Seip and Jones 2016a), and on the verge of extirpation after a prolonged decline averaging 13%/year over nearly 2 decades (Seip and Jones 2013, McNay et al. 2017). During 2014–2017, 42 pregnant females were placed in maternal penning (~40% average penning effort); 1 died prior to giving birth and 9 calves were aborted, stillborn, or died in the pen, resulting in 32 calves released in mid-July (McNay et al. 2016, 2017, 2018). Twenty-four of those calves survived to 10 months of age. The fifth year of penning was initiated in March 2018. Limited ground-based wolf removal occurred within and adjacent to the herd's range during 2013–2015 (McNay et al. 2017), but intensive aerial wolf reduction efforts by the British Columbia government commenced in January 2015 with near complete removal of wolves in the treatment area by March 2016 and maintenance of very low wolf numbers in late winter 2017 and 2018 (Seip and Jones 2016b, 2017, 2018). Klinse-Za caribou numbered 42 in March 2015 and increased to 66 individuals by March 2018. Maternal penning has contributed to the observed population growth, but the relative effects of maternal penning and wolf control are unclear (McNay et al. 2017, 2018; Seip and Jones 2016a, 2017, 2018). Based on survival of pen calves to 10 months and reported calf recruitment in the wild (McNay et al. 2016, 2017, 2018; Seip and Jones b), we estimated that about 12 additional caribou were recruited by the Klinse-Za population as a result of maternal penning. Assuming that Klinse-Za caribou would have continued to decline at 13%/year without management intervention, then the combined effect of maternal penning and wolf control has accounted for 48 additional caribou in the population, with 25% attributed to penning.
Compared to the Chisana effort, other penning programs have included small numbers of pregnant females, averaging just 11/year, in part because of the small populations involved. These efforts have also been restrained in response to the risks associated with capturing, transporting, and temporarily holding productive females from these imperiled populations. Such caution presents a dilemma however, in that limited penning effort is less likely to provide the contribution of recruits necessary to improve population trends as illustrated by our modeling. These efforts have also experienced more failed pregnancies and calf losses prior to release (20% overall) than during the Chisana penning program (7%), further limiting the number of calves available for release into the wild. Moreover, the proportion of calves released from these latter efforts that survived to become additional recruits as yearlings was about half that of the Chisana program (23% and 40%, respectively). With these last 2 factors combined, the performance of maternal penning, measured as the number of additional recruits/pregnant female held, was twice as high for the Chisana program compared to these other penning projects (0.37 and 0.18, respectively).
The 3 maternal penning programs in Alberta and British Columbia also differed from the Chisana effort in that all were conducted in concert with management efforts to reduce wolf abundance. Given the critical status of these caribou populations, it is not surprising that multiple management approaches would be applied simultaneously to reduce predation losses and improve population trends in the near term (McNay et al. 2013, Boutin and Merrill 2016, Serrouya and McLellan 2016, Serrouya et al. 2019). However, the population effects of wolf control and maternal penning are not likely to be additive. The efficacy of maternal penning is primarily dependent on the difference between early calf survival in the protection of the pen and that for the free-ranging population. Given that wolves are commonly important predators of caribou calves during their first few weeks (Page 1985, Adams et al. 1995a, Valkenburg et al. 2004, Jenkins and Barten 2005, Gustine et al. 2006) and substantial wolf reductions have resulted in improved calf recruitment in several small caribou populations (Gasaway et al. 1983, Farnell and McDonald 1988, Hayes et al 2003, Hervieux et al. 2013), the gains in calf survival derived from maternal penning will usually be reduced where effective wolf control occurs. The effect of wolf culling on penning results will be dependent on the relative importance of wolves as a mortality agent for young calves compared to other predators in the area and the magnitude of the wolf population reduction. Improved calf recruitment was noted in both the Little Smoky and Klinse-Za populations with the onset of wolf control (Smith and Pittaway 2011; Hervieux et al. 2014; Seip and Jones 2016b, 2017; McNay et al. 2017). Paradoxically, Serrouya et al. (2017b) reported no effect on calf recruitment as wolf numbers declined within the Columbia North caribou range. We suspect the limited contribution of recruits from at least 2 of these maternal penning programs was due in part to improved calf survival overall resulting from reductions in wolf abundance.
We have shown that maternal penning is most likely to be effective at increasing caribou numbers when populations are at or very near stability, and that its efficacy drops quickly over a narrow range of declining λi. Further, maternal penning is applicable only to small populations where a high proportion of pregnant females can reasonably be included in the treatment. Although these constraints define the potential for effective application of maternal penning, other management tactics employed to bolster small declining populations (i.e., translocation and predator reduction) each have their own limitations.
Although wildlife translocations are very common, several reviews have concluded that many such efforts were unsuccessful (Griffith et al. 1989, Wolf et al. 1996, Fischer and Lindenmeyer 2000, Singer et al. 2000). Further, these studies demonstrated that success of translocations was dependent on several factors including the number of individuals released, habitat quality at the release site, and whether the causes driving previous declines still existed. Attempts to augment 2 small imperiled caribou herds via translocation have been stymied by high predation losses of transplanted caribou to mountain lions and lack of appropriate donor stock (Compton et al. 1995, Warren et al. 1996, Leech et al. 2017, Wiles 2017).
Control of large carnivores (i.e., wolves and mountain lions) can be effective for reversing ungulate population declines (Orians et al. 1997, Rominger 2018). Wolf populations are resilient to moderate annual harvests (Adams et al. 2008), so control efforts must be intensive to reduce wolf numbers and sustain them at low levels for several years to evoke a population response in their ungulate prey (Orians et al. 1997). Only 3 of the 8 aerial wolf control programs reviewed by Orians et al. (1997) could demonstrate an increase in the target ungulate population; these programs reported ≥69% reductions in wolf abundance and lasted for 6–7 years (Gasaway et al. 1983, Farnell and McDonald 1988, Hayes et al. 2003, Farnell 2009). Further, wolves rapidly recover to pre-control levels once harvest pressure is reduced (Ballard et al. 1987, Hayes and Harestad 2000). Recent intensive aerial wolf control programs in central British Columbia and adjacent Alberta have been successful at arresting declines of 5 small caribou herds and increasing their numbers (Serrouya et al. 2019), but despite those successes, wolf culling in the region remains highly controversial (Hervieux et al. 2015, Pagé 2018). In British Columbia, efforts to reduce wolf numbers on the ranges of the 4 imperiled caribou populations through substantial harvest-driven reductions in the moose populations (Steenweg 2011; Serrouya et al. 2015b, 2017b) have shown equivocal results in that the Columbia North population has stabilized, whereas the others have continued to decline (Serrouya et al. 2019). Mountain lions present different challenges than wolves in that they occur at low densities and are cryptic, so it is difficult to determine their population size or the magnitude of harvest-driven reductions (Choate et al. 2006, McKinney 2011, Robinson et al. 2015). Harvest mortality is generally additive for mountain lion populations so their numbers can be reduced under sustained hunting pressure, but they also can recover quickly once harvests are reduced (Lindzey et al. 1992, Stoner et al. 2006, Robinson et al. 2014). For both of these large carnivores, management efforts to reduce their abundance remain highly controversial (Orians et al. 1997, Pierce and Bleich 2003, Hervieux et al. 2015, Mitchell et al. 2018).
MANAGEMENT IMPLICATIONS
The concept of maternal penning as a management tool for reducing predation effects on small caribou populations grew out of the contentious public debate surrounding lethal wolf control programs in Alaska and Yukon during the 1980s and 1990s (Stephenson et al. 1995, Orians et al. 1997, Todd 2002, Farnell 2009, Hayes 2010). To address the concerns at that time, the Government of Yukon (1992) engaged in a public planning effort to establish principles of wolf conservation and management within the territory and articulate specific guidelines for conducting wolf reduction programs (Todd 2002). As part of that process, wildlife managers were challenged to develop nonlethal methods to bolster depressed ungulate populations that could garner broad social acceptance (Government of Yukon 1992, Farnell 2009). Continued acrimony over the intensive wolf reductions in the Aishihik region of southwestern Yukon during 1993–1997 (Hayes et al. 2003, Hayes 2010) further strengthened interest in pursuing approaches other than lethal wolf control to recovering declining ungulate populations in the territory as the plight of the Chisana Herd was being recognized.
Maternal penning has proved to be popular among a broad and diverse array of constituencies. The Chisana program was quick to receive endorsement by government agencies, First Nations, and the public in and around the herd's range, and received positive media attention. That support did not waver as we came to realize that the caribou population was larger than we originally thought and therefore our ability to affect the population trajectory was limited. Rather than curtailing our effort, funding organizations unanimously supported a shift in our primary objective to vetting the maternal penning approach for possible application to imperiled southern caribou populations. Following closely on our program, the Little Smoky penning effort received substantial support from a consortium of resource development companies and the Alberta provincial government (Smith and Pittaway 2011). Ongoing penning efforts for the Columbia North and Klinse-Za populations are community-led programs that have attracted considerable funding from government agencies, First Nations, resource development industries, winter recreation companies, and environmental organizations, and have directly involved local people in project management and day-to-day operations (Furk and Serrouya 2017, McNay et al. 2017). Further, they have employed social media to engage a widespread interested public (Peace Northern Caribou Recovery 2017, RCRW 2017). These latter programs have continued to receive substantial financial and community support despite limited population responses to date.
The popularity of maternal penning is derived from its perceived role as an alternative, or at least a nonlethal adjunct, to wolf culling and its primary characteristic of actively fostering caribou young through their critical first weeks of life in imperiled populations where few calves normally survive. The initial calf survival results of penning are dramatic, and visions of these young calves and their mothers resonate strongly with those directly involved with penning operations and the broader public. However, the realized population effect of maternal penning is much reduced in that some calves released will die over the remainder of their first year, and some proportion of surviving pen calves would have been recruited as yearlings anyway. Further, it is important to acknowledge that, unlike the more contentious option of wolf control, maternal penning presents risks to the caribou being handled and held in penning (Hayek et al. 2016); these risks have the potential of being catastrophic (e.g., disease outbreak or predator incursion in the pen).
Although broadly popular, maternal penning is an expensive undertaking and programs to date have provided modest contributions at best to improving herd trends. With maternal penning as the only management action applied, our results from the Chisana program are the clearest; with an average annual cost of $500,000 CAD, we produced 54 additional recruits at a cost of about $37,000 CAD each. Annual costs of the other 3 penning efforts have ranged from $360,000 to $470,000 CAD (Smith and Pittaway 2011; McNay et al. 2014, 2016; Serrouya and McLellan 2016). The 1-year Little Smoky program provided no additional recruits (Smith and Pittaway 2011), whereas the individuals contributed by maternal penning for the other 2 efforts over 4 years (12 and 6 for Klinse-Za and Columbia North, respectively) came at a price of about $200,000 CAD each.
Recovery of small imperiled caribou populations across their southern distribution will require active population management in the near term to arrest declines and bolster numbers where possible to maintain them on the landscape as habitat conditions improve over the next few decades (U.S. Fish and Wildlife Service 1994; Festa-Bianchet et al. 2011; Environment Canada 2012, 2014). Maternal penning may have a role for populations ≤100 caribou given the high costs and the possibility of maximizing penning effort for such small populations. In the southern and central mountains of British Columbia and Alberta where many caribou conservation challenges occur, 20 of 26 populations fall below that threshold (Ray et al. 2015), so ample opportunity exists. However, we have demonstrated that maternal penning is most likely to be effective over a narrow range of λi near stability, and may be of little use for populations that are declining more rapidly. As additional maternal penning programs are considered, it is critical that implementation is primarily driven by their biological value rather than broad popularity compared to other management options. Expectations for the contribution of maternal penning to improving population trends must be realistic, accounting for the inherent population trajectory, feasible penning effort, and the effects of other concurrent management actions. These population projections need to be clearly articulated to the involved stakeholders and other interested publics. For small declining populations where maternal penning is likely to be applied, a large majority of pregnant females must be included for there to be any likelihood of producing a meaningful return to the population. Further, penning projects must be accompanied by adequate demographic monitoring to clearly evaluate population effects and add to continuing critical evaluation of the utility of maternal penning.
ACKNOWLEDGMENTS
We greatly appreciate the financial support provided by YDE, USGS, U.S. National Park Service, Canadian Wildlife Service, U.S. Fish and Wildlife Service, White River First Nation, and Yukon Fish and Wildlife Enhancement Trust. Of the many people who assisted with pen and camp construction; caribou capture, transport, and processing; and animal care in captivity, we particularly thank D. Broeren, D. J. Demma, D. Drummond, K. Egli, M. Kienzler, M. Larsen, P. Merchant, W. A. C. Nixon, S. Oakley, and K. Russell. We thank W. B. Collins for the innovative fencing design that could be quickly constructed at our remote penning sites. Net-gun capture and transport of caribou to the pens were conducted safely and efficiently with helicopter pilots D. Makkonen, A. Morrison, K. Scholz, D. Washington, and K. Ziehe. Additional captures and population surveys were ably conducted with helicopter pilot R. Swisher. C. Drinnen and E. Onoserychuk safely flew many trips in and out of camp delivering people, equipment, and supplies. We particularly want to thank Supercub pilot H. McMahan for his continued interest in this project and the insights he provided while logging hundreds of flight hours radiotracking and assisting with surveys and captures. Veterinary support provided by G. P. Adams, S. J. Kutz, and D. M. Mulcahy was much appreciated. We are grateful for efforts of students from F. H. Collins Secondary School, Whitehorse, Yukon and St. Elias Community School, Haines Junction, Yukon to collect hundreds of bags of lichen for caribou feed. D. Dickson was instrumental in drawing attention to concerns over the Chisana Herd's status and graciously provided the use of his camp in 2003. We thank S. Swisher for use of her cabin as the Chisana, Alaska base of operations; we also appreciate the assistance provided by A. Jones, E. Jones, and I. Thorall in Chisana. We thank the White River First Nation and Northway Tribal Council for enabling our work within their traditional territories. Personnel of U.S. Customs and Border Protection at Alcan, Alaska and Canada Border Services Agency at Beaver Creek, Yukon graciously facilitated our transboundary fieldwork. The advice and assistance provided by C. L. Gardner and R. G. White are greatly appreciated. R. S. McNay and R. Serrouya kindly provided information on the ongoing Klinse-Za and Columbia North maternal penning programs, respectively. We thank B. C. Adams, C. L. Gardner, T. M. Hegel, G. V. Hilderbrand, H. E. Johnson, C. D. Mitchell, J. M. Pearce, D. R. Seip, and B. D. Uher-Koch for their helpful comments on previous manuscript drafts. We appreciate the constructive criticism provided by M. Ben-David, P. S. Barboza, and S. Boutin that improved the final version of this monograph. The French translations of the title and abstract were graciously completed by S. D. Côté, with the assistance of A. Brodeur, and F. Déry. Any use of trade, firm, or product names is for descriptive purposes only and does not imply endorsement by the U.S. or Canadian Government.
APPENDIX A
85% CI | ||||
---|---|---|---|---|
Variable | Β | SE | Lower | Upper |
Survival of pen calves, mid-Jun release to mid-Oct, all yearsa | ||||
Intercept | −2.939 | 1.107 | −4.533 | −1.344 |
Age | 0.187 | 0.048 | 0.118 | 0.255 |
YR06 | −0.570 | 0.752 | −1.652 | 0.512 |
Age×YR06 | −0.040 | 0.034 | −0.089 | 0.008 |
Survival of pen calves, mid-Jun release to mid-Oct, 2004–2006 onlyb | ||||
Intercept | −5.482 | 1.841 | −8.132 | −2.831 |
Age | 0.159 | 0.052 | 0.084 | 0.234 |
YR06 | −0.562 | 0.790 | −1.699 | 0.576 |
Age×YR06 | −0.049 | 0.038 | −0.104 | 0.005 |
Birth mass | 0.383 | 0.195 | 0.102 | 0.664 |
Age-specific natality of adult females ≥4 years oldc | ||||
Intercept | 2.889 | 0.254 | 2.523 | 3.254 |
Old7 | −0.053 | 0.083 | −0.172 | 0.066 |
Old8 | −0.079 | 0.110 | −0.238 | 0.080 |
Old9 | −0.084 | 0.120 | −0.257 | 0.088 |
Old10 | −0.046 | 0.079 | −0.160 | 0.068 |
Summer survival of adult females by yeard | ||||
Intercept | 3.570 | 0.315 | 3.116 | 4.023 |
t1 | 1.092 | 0.651 | 0.155 | 2.029 |
t2 | 0.566 | 0.647 | −0.365 | 1.498 |
YR07 | 0.633 | 0.689 | −0.360 | 1.626 |
Winter survival of adult females by yeare | ||||
Intercept | 3.867 | 0.321 | 3.404 | 4.329 |
t1 | 1.453 | 0.576 | 0.625 | 2.282 |
t2 | 2.138 | 0.762 | 1.040 | 3.236 |
t3 | 1.014 | 0.498 | 0.298 | 1.731 |
YR04 | −1.756 | 0.404 | −2.337 | −1.174 |
Summer age-specific survival of adult femalesf | ||||
Intercept | 4.351 | 0.382 | 3.800 | 4.901 |
t1 | 1.191 | 0.707 | 0.174 | 2.209 |
t2 | 1.110 | 0.624 | 0.211 | 2.009 |
Old11 | −0.121 | 0.191 | −0.396 | 0.153 |
Old12 | −0.161 | 0.244 | −0.512 | 0.190 |
Old13 | −0.248 | 0.343 | −0.741 | 0.246 |
Old14 | −0.204 | 0.326 | −0.674 | 0.265 |
Winter age-specific survival of adult femalesg | ||||
Intercept | 5.232 | 0.879 | 3.967 | 6.498 |
t1 | 1.484 | 0.665 | 0.526 | 2.442 |
t2 | 1.881 | 0.781 | 0.757 | 3.005 |
t3 | 1.457 | 0.665 | 0.498 | 2.415 |
Age | −0.167 | 0.140 | −0.368 | 0.034 |
Old8 | −0.140 | 0.174 | −0.390 | 0.110 |
Winter calf survival, 2004–2006h | ||||
Intercept | 2.971 | 0.463 | 2.304 | 3.638 |
t1 | 2.249 | 1.084 | 0.688 | 3.811 |
t2 | 2.219 | 1.084 | 0.657 | 3.780 |
t3 | 0.332 | 0.547 | −0.755 | 0.822 |
YR04 | −1.286 | 0.511 | −2.021 | −0.550 |
- a Variables included in model were age at release (age, days) and an indicator variable for year = 2006 (YR06).
- b Years calves were weighed at birth. In addition to age and YR06, birth mass (kg) was included in the final model.
- c Variables included in model defined thresholds for the onset of senescence in natality (oldX where X = 7, 8, 9, or 10; age < X, oldX = 0; age ≥ X, oldX = age – X + 1).
- d Model included an intercept and 2 terms (ti) to account for 3 intervals defined by our radiotracking schedule during summer (mid-May, mid-Jun, mid-Aug, and mid-Oct); the other variable was an indicator variable for year = 2007 (YR07).
- e Model included an intercept and 3 terms (ti) to account for 4 intervals defined by our radiotracking schedule during winter (mid-Oct, early Dec, mid-Jan, mid-Mar, mid-May); the other variable was an indicator variable for year = 2004 (YR04).
- f Model included an intercept and 2 terms (ti) to account for 3 radiotracking intervals during summer; other variables included defined thresholds for the onset of senescence as described above (oldX where X = 11, 12, 13, or 14).
- g Model included an intercept and 3 terms (ti) to account for 4 radiotracking intervals during winter; other variables included were age (years) and old8 that defined a threshold at 8 years for the onset of senescence as described above.
- h Model included an intercept and 3 terms (ti) to account for 4 radiotracking intervals during winter; the other variable was an indicator variable for year = 2004 (YR04).
APPENDIX B
Age | n | Parturient | Calculated rate | Model rate | |||
---|---|---|---|---|---|---|---|
(yrs) | KAa | CEMa | Total | (n) | Rate | 95% CI | (≥4 years old) |
2 | 57 | 10 | 67 | 38 | 0.567 | 0.441–0.686 | |
3 | 44 | 27 | 71 | 62 | 0.873 | 0.768–0.937 | |
4 | 28 | 38 | 66 | 62 | 0.939 | 0.844–0.980 | 0.947 |
5 | 17 | 43 | 60 | 56 | 0.933 | 0.830–0.978 | 0.947 |
6 | 7 | 47 | 54 | 52 | 0.963 | 0.862–0.994 | 0.947 |
7 | 2 | 45 | 47 | 45 | 0.957 | 0.843–0.993 | 0.945 |
8 | 2 | 50 | 52 | 50 | 0.962 | 0.857–0.993 | 0.937 |
9 | 53 | 53 | 50 | 0.943 | 0.834–0.985 | 0.923 | |
10 | 31 | 31 | 25 | 0.806 | 0.619–0.919 | 0.903 | |
11 | 30 | 30 | 26 | 0.867 | 0.684–0.956 | 0.877 | |
12 | 16 | 16 | 14 | 0.875 | 0.604–0.978 | 0.846 | |
13 | 14 | 14 | 11 | 0.786 | 0.488–0.943 | 0.809 | |
14 | 12 | 12 | 10 | 0.833 | 0.509–0.971 | 0.765 | |
15 | 9 | 9 | 6 | 0.667 | 0.309–0.910 | 0.714 | |
16 | 8 | 8 | 6 | 0.750 | 0.356–0.955 | 0.658 | |
17 | 5 | 5 | 2 | 0.400 | 0.073–0.830 | 0.597 | |
18 | 2 | 2 | 1 | 0.500 | 0.027–0.973 | 0.533 | |
19 | 1 | 1 | 1 | 1.000 | 0.055–1.000 | 0.467 | |
Total | 157 | 441 | 598 | 517 |
- a KA = known age; initially captured as calves; CEM = age from cementum analysis.
APPENDIX C
Summer (mid-May–mid-Oct) | Winter (mid-Oct–mid-May) | Annual (mid-May–mid-May) | ||||||||||
---|---|---|---|---|---|---|---|---|---|---|---|---|
KM survival rate | KM survival rate | KM survival rate | ||||||||||
Age (yr) | Mean n at risk | s | 95% CI | Model rate | Mean n at risk | s | 95% CI | Model rate | Mean n at risk | s | 95% CI | Model rate |
1 | 56.0 | 1.000 | 0.979 | 56.0 | 0.982 | 0.948–1.000 | 0.990 | 56.0 | 0.982 | 0.948–1.000 | 0.969 | |
2 | 46.0 | 1.000 | 0.979 | 46.5 | 1.000 | 0.988 | 46.3 | 1.000 | 0.968 | |||
3 | 45.0 | 0.961 | 0.925–0.998 | 0.979 | 48.0 | 0.980 | 0.941–1.000 | 0.986 | 46.7 | 0.942 | 0.879–1.000 | 0.965 |
4 | 44.0 | 1.000 | 0.979 | 48.8 | 0.979 | 0.939–1.000 | 0.983 | 46.7 | 0.979 | 0.939–1.000 | 0.963 | |
5 | 38.0 | 0.975 | 0.927–1.000 | 0.979 | 38.3 | 1.000 | 0.980 | 38.1 | 0.975 | 0.927–1.000 | 0.960 | |
6 | 30.7 | 0.971 | 0.917–1.000 | 0.979 | 32.3 | 1.000 | 0.977 | 31.6 | 0.971 | 0.916–1.000 | 0.957 | |
7 | 36.3 | 0.976 | 0.929–1.000 | 0.979 | 39.8 | 1.000 | 0.973 | 38.3 | 0.976 | 0.928–1.000 | 0.953 | |
8 | 36.7 | 1.000 | 0.979 | 38.0 | 0.974 | 0.924–1.000 | 0.963 | 38.3 | 0.974 | 0.924–1.000 | 0.943 | |
9 | 30.0 | 0.912 | 0.821–1.000 | 0.979 | 30.5 | 0.933 | 0.847–1.000 | 0.951 | 30.4 | 0.851 | 0.762–0.940 | 0.931 |
10 | 21.7 | 1.000 | 0.979 | 22.0 | 0.955 | 0.870–1.000 | 0.934 | 21.9 | 0.955 | 0.870–1.000 | 0.914 | |
11 | 17.0 | 0.933 | 0.807–1.000 | 0.977 | 15.3 | 0.760 | 0.593–0.928 | 0.912 | 16.0 | 0.710 | 0.526–0.893 | 0.890 |
12 | 11.0 | 1.000 | 0.969 | 11.3 | 0.833 | 0.623–1.000 | 0.883 | 11.1 | 0.833 | 0.623–1.000 | 0.855 | |
13 | 6.3 | 1.000 | 0.948 | 6.8 | 1.000 | 0.845 | 6.6 | 1.000 | 0.802 | |||
14 | 6.0 | 0.750 | 0.554–0.946 | 0.897 | 6.0 | 0.714 | 0.380–1.000 | 0.798 | 6.0 | 0.625 | 0.290–0.961 | 0.715 |
15 | 5.0 | 0.833 | 0.561–1.000 | 0.801 | 4.8 | 0.800 | 0.449–1.000 | 0.739 | 4.9 | 0.667 | 0.290–1.000 | 0.592 |
16 | 5.3 | 0.500 | 0.217–0.783 | 0.644 | 4.0 | 0.505 | 0.154–0.847 | 0.669 | 4.6 | 0.250 | 0.050–0.450 | 0.431 |
17 | 3.0 | 0.500 | 0.010–0.990 | 0.432 | 2.0 | 1.000 | 0.587 | 2.4 | 0.500 | 0.010–0.990 | 0.254 | |
18 | 1.3 | 0.500 | 0.000–1.000 | 0.220 | 1.0 | 1.000 | 0.497 | 1.1 | 0.500 | 0.000–1.000 | 0.109 |