<%BANNER%>

Population Ecology of the Endangered Vancouver Island Marmot

Permanent Link: http://ufdc.ufl.edu/UFE0023714/00001

Material Information

Title: Population Ecology of the Endangered Vancouver Island Marmot Decline and Potential Recovery
Physical Description: 1 online resource (70 p.)
Language: english
Creator: Aaltonen, Kristen
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2009

Subjects

Subjects / Keywords: ecology, endangered, marmot, population, reintroduction, survival
Wildlife Ecology and Conservation -- Dissertations, Academic -- UF
Genre: Wildlife Ecology and Conservation thesis, M.S.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation

Notes

Abstract: Recovery of the endangered Vancouver Island marmot is contingent upon releases of captive-born marmots into natural habitats. Success of such reintroduction programs largely depends on the ability of released animals to survive in the wild. We used radio-telemetry (1992-2007) and mark-resight (1987-2007) data to estimate seasonal and annual survival rates of the Vancouver Island marmot, to compare survival and cause-specific mortality rates of captive-born marmots that have been released into the natural habitat with those of wild-born marmots, and to test for the effect of age at release on survival of the released marmots. However, annual survival of captive-born marmots released into the wild was low compared to wild-born marmots. Marmots released as two-year-olds or older survived more successfully than those released as yearlings. Predation by golden eagles was the most important cause of mortality for captive-born marmots, whereas predation by wolves and cougars was more important for wild-born marmots. These results suggest that delaying release of captive-born marmots until two years of age may enhance their probability of survival in the wild, and will likely improve the success of the release program. In order to formally test the effects of survival and reproduction on population growth rate, we constructed population projection models and performed elasticity analysis. We estimated age-specific reproductive parameters and used various age-specific survival rates spanning different periods of time and/or representing various subsets of the population. We found that population growth rate in recent years (2003-2008) was only slightly higher than in the past (1987-2002), and currently the population is still declining by at least 12% annually. The population growth rate estimated for a hypothetical population of captive-born marmots corresponds with a 28% decline annually, indicating released individuals alone will not result in a persistent population. However, population growth rate for a population of wild-born individuals in recent years was greater than one, indicating a 3% annual increase in population size. Elasticity analysis confirmed the relative importance of survival for the population, especially adult survival. Our results provide insight into the potential for recovery of the species through reintroductions.
General Note: In the series University of Florida Digital Collections.
General Note: Includes vita.
Bibliography: Includes bibliographical references.
Source of Description: Description based on online resource; title from PDF title page.
Source of Description: This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Statement of Responsibility: by Kristen Aaltonen.
Thesis: Thesis (M.S.)--University of Florida, 2009.
Local: Adviser: Oli, Madan K.
Electronic Access: RESTRICTED TO UF STUDENTS, STAFF, FACULTY, AND ON-CAMPUS USE UNTIL 2011-05-31

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2009
System ID: UFE0023714:00001

Permanent Link: http://ufdc.ufl.edu/UFE0023714/00001

Material Information

Title: Population Ecology of the Endangered Vancouver Island Marmot Decline and Potential Recovery
Physical Description: 1 online resource (70 p.)
Language: english
Creator: Aaltonen, Kristen
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2009

Subjects

Subjects / Keywords: ecology, endangered, marmot, population, reintroduction, survival
Wildlife Ecology and Conservation -- Dissertations, Academic -- UF
Genre: Wildlife Ecology and Conservation thesis, M.S.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation

Notes

Abstract: Recovery of the endangered Vancouver Island marmot is contingent upon releases of captive-born marmots into natural habitats. Success of such reintroduction programs largely depends on the ability of released animals to survive in the wild. We used radio-telemetry (1992-2007) and mark-resight (1987-2007) data to estimate seasonal and annual survival rates of the Vancouver Island marmot, to compare survival and cause-specific mortality rates of captive-born marmots that have been released into the natural habitat with those of wild-born marmots, and to test for the effect of age at release on survival of the released marmots. However, annual survival of captive-born marmots released into the wild was low compared to wild-born marmots. Marmots released as two-year-olds or older survived more successfully than those released as yearlings. Predation by golden eagles was the most important cause of mortality for captive-born marmots, whereas predation by wolves and cougars was more important for wild-born marmots. These results suggest that delaying release of captive-born marmots until two years of age may enhance their probability of survival in the wild, and will likely improve the success of the release program. In order to formally test the effects of survival and reproduction on population growth rate, we constructed population projection models and performed elasticity analysis. We estimated age-specific reproductive parameters and used various age-specific survival rates spanning different periods of time and/or representing various subsets of the population. We found that population growth rate in recent years (2003-2008) was only slightly higher than in the past (1987-2002), and currently the population is still declining by at least 12% annually. The population growth rate estimated for a hypothetical population of captive-born marmots corresponds with a 28% decline annually, indicating released individuals alone will not result in a persistent population. However, population growth rate for a population of wild-born individuals in recent years was greater than one, indicating a 3% annual increase in population size. Elasticity analysis confirmed the relative importance of survival for the population, especially adult survival. Our results provide insight into the potential for recovery of the species through reintroductions.
General Note: In the series University of Florida Digital Collections.
General Note: Includes vita.
Bibliography: Includes bibliographical references.
Source of Description: Description based on online resource; title from PDF title page.
Source of Description: This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Statement of Responsibility: by Kristen Aaltonen.
Thesis: Thesis (M.S.)--University of Florida, 2009.
Local: Adviser: Oli, Madan K.
Electronic Access: RESTRICTED TO UF STUDENTS, STAFF, FACULTY, AND ON-CAMPUS USE UNTIL 2011-05-31

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2009
System ID: UFE0023714:00001


This item has the following downloads:


Full Text

PAGE 1

1 POPULATION ECOLOGY OF THE ENDANGERED VANCOUVER ISLAND MARMOT: DECLINE AND POTENTIAL FOR RECOVERY By KRISTEN AALTONEN A THESIS PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF MASTER OF SCIENCE UNIVERSITY OF FLORIDA 2009

PAGE 2

2 2009 Kristen Aaltonen

PAGE 3

3 To Jack White (and Meg too)

PAGE 4

4 ACKNOWLEDGMENTS I would like to thank my adviser, Madan Oli, for his expert guidance and patience; Andrew Bryant, who brought me into the Vancouver Island marmot project and taught me much about marmots; and Emilio Bruna for his reviews and sage advice. I give m any thanks to the Oli lab for both the intellectual and emotional support! The department of Wildlife Ecology and Conservation also deserves thanks for providing an excellen t environment to conduct r esearch. Project funding was provided in various years by the BC Habitat Conservation Trust Fund, BC Ministry of Environment, World Wildlife Fund (Canada), Forest Alliance of BC, Forest Renewal BC, Cowichan Valley Field Naturalists Society, Nanaimo Field Naturalist s, BC Wildlife Federation, BC Hydro Bridge Coastal Fish and Wildlife Restoration Program, TimberWest Forests, MacMillan Bloedel Limited, Environment Canada (administered through the University of British Columbia), and the Marmot Recovery Foundation. One of us (AAB) was partially supported in some years by a Province of Alberta Graduate Scholarship, University of Calgary Thesis Research Grant, a Canadian Wildlife Service Research Grant, a King -Platt Memorial Award, Franc Joubin Graduate Bursary in Environm ental Science, and University of Victoria (Biology) graduate award. Other essential support was by provided by the University of Florida, Toronto Zoo, Calgary Zoo, Mountain View Breeding and Conservation Center, West Coast Helicopters, Mt. Washington Alp ine Resort, BC Conservation Foundation, BC Conservation Corps, and the Nature Trust of BC. MacMillan Bloedel Limited, Weyerhaeuser, Island Timberlands, TimberWest Forests and Mt. Washington Alpine Resort allowed unrestricted field access to private forest lands.

PAGE 5

5 K. Langelier DVM pioneered radio implant techniques for Vancouver Island marmots in 1992. M. McAdie DVM subsequently performed the majority of radio implant surgeries. H. Schwantje DVM, S. Saksida DVM and M. Smith DVM also performed implantation surgeries. This work would not have been possible without the dedicated work of many individuals who collected and supplied field data: these include: A. Bryant, M. DeLaronde, D. Doyle, C. Jackson, D. Janz, L. Dyck, J. Lewis, K. McDonald, J. MacDermott, D. Milne, M. McAdie, D. Norton, A. Pendergast, S. Pendergast, C. Reid, W. Swain, J. Voller, L. Wilson, and A. Zeeman. Many others too numerous to list here helped with trapping, re -sighting and telemetry work. We thank them all. K. Armitage V. Goswami, J. McCray, and E. Kneip provided helpful comments throughout, for which I am grateful.

PAGE 6

6 TABLE OF CONTENTS page ACKNOWLEDGMENTS .................................................................................................................... 4 LIST OF TABLES ................................................................................................................................ 8 LIST OF FIGURES .............................................................................................................................. 9 ABSTRACT ........................................................................................................................................ 10 CHAPTER 1 POPULATION ECOLOGY, REINTRODUCTION BIOLOGY, AND THE ENDANGERED VANCOUVER ISLAND MARMOT: AN INTRODUCTION.................. 12 2 REINTRODUCTION, AGE -AT RELEASE, AND SURVIVAL ........................................... 15 Introduction ................................................................................................................................. 15 Methods ....................................................................................................................................... 18 Study Species ....................................................................................................................... 18 Study Area ............................................................................................................................ 18 Field Methods ...................................................................................................................... 19 Data Analysis ....................................................................................................................... 20 Analysis of radio telemetry data ................................................................................. 20 Analysis of combined CMR and radio-telemetry data .............................................. 22 Res ults .......................................................................................................................................... 24 Analysis of Radio Telemetry Data ..................................................................................... 24 Cause -Specific Mortality ..................................................................................................... 26 Analysis of Combined (Radio Telemetry and CMR) Data ............................................... 27 Discussion .................................................................................................................................... 28 Survival Rates and Sex and Age Specific Differences .................................................... 28 Seasonal and Site -Specific Variation ................................................................................. 30 Habitat Changes and Vancouver Island Marmots ............................................................. 31 Captive Breeding, Cause-Specific Mortality, Age at Release and Conservation of M. vancouverensis ............................................................................................................ 32 3 POPULATION ECOLOGY OF THE ENDANGERED VANCOUVER ISLAND MARMOT ................................................................................................................................... 44 Introduction ................................................................................................................................. 44 Methods ....................................................................................................................................... 46 Study Species ....................................................................................................................... 46 Study Area ............................................................................................................................ 46 Field Methods ...................................................................................................................... 47 Estimates of Survival and Fecundity Rates ........................................................................ 48 Construction and Analysis of Population Model ............................................................... 49

PAGE 7

7 Results .......................................................................................................................................... 50 Estimates of Surv ival and Fecundity Rates ........................................................................ 50 Population Growth Rate and Elasticities ............................................................................ 51 Discussion .................................................................................................................................... 5 2 Population Decline ............................................................................................................... 52 Potential for Recovery ......................................................................................................... 54 4 POPULATION ECOLOGY OF THE VANCOUVER ISLAND MARMOT: CONCLUSIONS ......................................................................................................................... 64 LITERATURE CITED ....................................................................................................................... 65 BIOGRAPHICAL SKETCH ............................................................................................................. 69

PAGE 8

8 LIST OF TABLES Table page 2 2 A nalysis of the influence of intrinsic and extrinsic factors on survival of the Vancouver Island marmot using radio -telemetry data (1992 2007) and known-fate models. .................................................................................................................................... 35 2 3 Analysis of the influence of intrinsic and extrinsic factors on survival of the Vancouver Island marmot using a subset of the radio -telemetry data (2003 2007) and known-fate models.. ............................................................................................................... 36 2 4 Analysis of the influence of intrinsic and extrinsic factors on surv ival and recapture rate of the Vancouver Island marmot using a combination of radiotelemetry and capture recaptur e data and multi -state models. ................................................................... 37 3 1 Description of data used for parameter estimation time period, segment of population used, and estimates of age-specific survival rates used for each matrix population model .................................................................................................................... 58 3 2 E stimates population growth rate ( ) and elasticities from each matrix population model ....................................................................................................................................... 59

PAGE 9

9 LIST OF FIGURES Figure page 2 1 Historically occupied and recently occupied sites for M. vancouverensis ......................... 38 2 2 Changes in numbers of ear -tagged and radio-telemetered marmots for which annual survival data were available over the study period ............................................................. 39 2 3 Site -specific e stimates of biweekly survival rates by season ............................................. 40 2 4 Estimates of seasonal survival rates for captive born and wild -born marmots .................. 41 2 5 Estimates of cause -specific mortality rates for captive -born and wild-born marmots ...... 42 2 6 Age, sex, and habitat -specific apparent survival rates from combined data analysis. ....... 43 3 1 Historically occupied and recently occupied sites for M. vancouverensis ......................... 60 3 2 Population projection matrix used for all models. ............................................................... 61 3 3 Average age -specific litter size and age of first reproduction. ............................................ 62 3 4 Elasticities from four models, relative importance of age -specific demographic rates to the population growth rate ................................................................................................. 63

PAGE 10

10 Abstract of Thesis Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Master of Science P OPULATION ECOLOGY OF THE ENDANGERED VANCOUVER ISLAND MA RMOT: DECLINE AND POTENTIAL RECOVERY By Kristen Aaltonen May 2009 Chair: Madan K. Oli Major: Wildlife Ecology and Conservation Recovery of the endangered Vancouver Island marmot is contingent upon releases of captive -born marmots into natural habitats. Success of such reintroduction programs largely depends on the ability of released animals to survive in the wild. We used radio -telemetry (19922007) and mark resight (19872007) data to estimate seasonal and annual survival rates of the Vancouver Island marmot, to compare survival and cause -specific mortality rates of captive born marmots that have been released into the natural habitat with those of wild -born marmots, and to test for the effect of age at release on survival of the released marmots. Ho wever, annual survival of captive -born marmots released into the wild was low compared to wild -born marmots. Marmots released as two -year -olds or older survived more successfully than those released as yearlings. Predation by golden eagles was the most i mportant cause of mortality for captive -born marmots, whereas predation by wolves and cougars was more important for wild born marmots. These results suggest that delaying release of captive -born marmots until two years of age may enhance their probability of survival in the wild, and will likely improve the success of the release program. In order to formally test the effects of survival and reproduction on population growth rate, we constructed population projection models and performed elasticity analys is. We estimated

PAGE 11

11 age -specific reproductive parameters and used various age -specific survival rates spanning different periods of time and/or representing various subsets of the population. We found that population growth rate in recent years (20032008) was only slightly higher than in the past (19872002), and currently the population is still declining by at least 12% annually. The population growth rate estimated for a hypothetical population of captive born marmots corresponds with a 28% decline annually, indicating released individuals alone will not result in a persistent population. However, population growth rate for a population of wildborn individuals in recent years was greater than one, indicating a 3% annual increase in population size. El asticity analysis confirmed the relative importance of survival for the population, especially adult survival. Our results provide insight into the potential for recovery of the species through reintroductions.

PAGE 12

12 CHAPTER 1 POPULATION ECOLOGY, REINTRODUCTION BIOLOGY, AND THE ENDANGERED VANCOUVER ISLAND MARMOT: AN INTRODUCTION Understanding causes of population decline is an important first step towards recovery of endangered species (Caughley 1994; Caughley & Gunn 1996). Change in population size is the consequence of changes in one or more vital demographic rates (Caswell 2001; Mills 2007; Oli & Armitage 2004); effects of the environment and management actions on the dynamics of populations are mediated through their influences on demographic variables. Furthermore, reduction in survival can lead to decline of wildlife populations whose growth rate is very sensitive to changes in survival rates (Heppell et al. 2000; Oli & Dobson 2003; Stahl & Oli 2006). Therefore rigorous estimates of survival and knowledge about how various factors influence this key demog raphic rate are necessary for conservation of species at risk and specifically reintroduction planning ( Le Gouar et al. 2008). Captive breeding and reintroduc tion are frequently used for the conservation of rare or endangered species (Caughley & Gunn 1996; Sarrazin & Legendre 2000). However, few reintroduction programs have achieved success, due in part to poor understanding of the populationlevel effects of alternative release strategies (Armstrong & Seddon 2008; Grenier et al. 2007; Sarrazin & Legendre 2000; Seddon et al. 2007). Furthermore, the reintroduction literature largely consists of descriptive accounts of success or failure of reintroduction progra ms; progress in this branch of conservation biology has been slow due to the lack of quantitatively rigorous analysis of data generated from well -designed comparative or experimental studies (Armstrong & Seddon 2008). Demographic models can provide valuab le insights into likely results of conservation efforts, and allow for evaluation of the populationlevel impacts of release strategies; however, few reintroduction programs have benefitted from such models (Caswell 2001; Meretsky et al. 2000; Sarrazin & L egendre 2000).

PAGE 13

13 The Vancouver Island marmot ( Marmota vancouverensis ) is an endangered mammal inhabiting subalpine meadows on Vancouver Island, British Columbia (Janz et al. 2000; Nagorsen 1987). The M. vancouverensis population went through several range restrictions leaving the population in only two locations on the island by the 1970s The marmot population increased to 300 350 individuals during the mid1980s as marmots colonized the openings created by high altitude logging that resembled marmot habit at (natural meadows, Bryant & Janz 1996). This increase in population size was short lived, as most marmot colonies in clearcut habitats disappeared by 2000 and the population declined to an estimated 35 individuals by 2003 (Bryant 2005). The disappearance of marmots from clearcut habitat is likely due to forest regeneration reducing the suitability of clearcuts as marmot habitat (Bryant 1998). Changes in landscape structure following logging probably had direct and indirect impacts on the entire marmot population through predator -prey interactions (Aaltonen et al. in review; Bryant & Page 2005). With the marmot population on the brink of extinction, a captive breeding program was initiated in 1997 with facilities established both on and off Vancouver Island (Bryant 2005). During 20032008, 155 captive -born marmots were released into various sites at both Mt. Washington and Nanaimo Lakes (Kruckenhauser et al. 2009; unpublished minutes, Vancouver Island Marmot Recovery Team, Dec. 2008). The Vancouver I sland marmot is an excellent case of an endangered species for which drastic managem ent efforts have been made with little quantitative support. However, we have the long-term monitoring data available to perform quantitative analysis in order to guide the recovery strategy which is rare for an endangered species We used 20 years of monitoring data to estimate survival, test hypotheses about factors

PAGE 14

14 affecting survivial, examine population growth rate, and consider the potential for recovery of the species.

PAGE 15

15 CHAPTER 2 REINTRODUCTION, AGE -AT RELEASE, AND SURVIVAL Introduction Understanding causes of population decline is an important first step towards recovery of endangered species (Caughl ey 1994; Caughley & Gunn 1996). Change in population size is the consequence of changes in one or more vital demographic rates (Caswell 2001; Mills 2007; Oli & Armitage 2004); effects of the environment and management actions on the dynamics of populations are mediated through their influences on demographic variables. Furthermore, reduction in survival can lead to decline of wildlife populations whose growth rate is very sensitive to changes in survival rates (Heppell et al. 2000; Oli & Dobson 2003; Stah l & Oli 2006). Therefore, rigorous estimates of survival and knowledge about how various factors influence this key demographic rate are necessary for formulation of recovery plans and conservation of species at risk (Le Gouar et al. 2008). The Vancouver Island marmot ( Marmota vancouverensis ) is endemic to Vancouver Island, British Columbia, Canada. Concern brought on by the restricted geographic distribution and low numbers, led to M. vancouverensis being listed as endangered in 1978 (Bryant & Page 2005 ; Shank 1999). Systematic field surveys suggest that the marmot population increased during the early 1980s to a peak of 300350 individuals during the mid 1980s (Bryant & Janz 1996). However, by 2004 the population had declined to approximately 35 individuals in the wild (Bryant 2005). Landscape changes due to clearcut logging and increases in predation related mortality are thought to have contributed to the population decline (Bryant & Page 2005; Janz et al. 2000). Significant conservation accomplish ments towards recovery of M. vancouverensis have been made through planned captive -breeding and release programs (Janz et al. 2000). Captive

PAGE 16

16 breeding facilities were established both on and off the island beginning in 1997, and captive bred marmots have b een released into their natural habitat since 2003 (Bryant 2005, 2007). There have been both reintroductions of marmots into empty but historically occupied sites and re -enforcement of existing colonies with additional marmots. During 2003-2007, 96 marmots were released and monitored using radio-telemetry. Arguably, the most important conservation measure currently in place is the release of captive bred marmots into their natural habitat. Although reintroduction programs have been successful in some ca ses, limitations of captive breeding and re -introduction programs for species conservation are well documented (e.g., Caughley and Gunn 1996; Snyder et al. 1996; Moorhouse et al. 2009). Success of such programs depends to a large extent on the ability of released animals to survive and reproduce in their natural habitats. Often, released animals have a lower probability of survival in the wild than their wild born counterparts, reflecting some cost associated with being raised or held in captivity (Beck et al. 1994; Mathews et al. 2005; Sarrazin and Legendre 2000; Snyder et al. 1996). Any additional mortality may only apply for a period immediately post -release after which survival of released individuals increases to a rate near that of wild -born indivi duals (Bar David et al. 2005; Sarrazin et al. 1994; Maran et al. 2009). Age at which animals are released can also influence their subsequent survival and the success of a reintroduction program (Green et al. 2005; Le Gouar et al. 2008; Sarrazin and Legen dre 2000). Understanding differences between captive -born and wild -born marmots in survival and causes of death can provide information to potentially improve success of the recovery program. However, whether and to what extent survival and causes of mor tality of captive -born marmots differs from those of wild-born marmots, or if age at release influences survival of captive -born marmots, remained unknown until this study.

PAGE 17

17 To provide demographic information needed for effective recovery of M. vancouverens is Bryant and Page (2005) analyzed radiotelemetry data collected through 2004. These authors provided estimates of survival and also examined causes of mortality and seasonal variations in survival. Field study has continued since this work was complet ed, and many more marmots have been monitored via radio-telemetry. Consequently, sample sizes and our ability to test hypotheses regarding factors influencing survival of marmots have increased substantially in recent years. Our goals were to build upon previous work, provide more rigorous estimates of survival, and test hypotheses about factors influencing survival of marmots using radiotelemetry data. Using a subset of radio-telemetry data collected since the first releases of captive born marmots in 2003, we also tested the hypotheses that wild-born marmots would have a higher probability of survival than their captive -born counterparts, and that captive -born marmots released at an older age would survive better than those released as yearlings. Pred ation is the most important cause of mortality of wild Vancouver Island marmots (Bryant & Page 2005). Given that captive born marmots are not exposed to predators during early, possibly important developmental stages as wild-born marmots are, their behavi oral responses to the risk of predation may differ (Blumstein et al. 2001). Alternatively, their ability to escape various predators may differ due to disparities in experience or body condition. Thus, we also estimated cause -specific mortality rates, an d tested for differences in mortality rates between captive born and wild -born marmots for each cause. Pups were seldom implanted with radiotransmitters. Consequently, radio-telemetry data were sparse and it was not possible to estimate survival rates fo r this age -class. We combined the radio telemetry and longterm capture -mark resighting (CMR) data and analyzed the

PAGE 18

18 combined dataset within a multi-state CMR modeling framework (Williams et al. 2002). Combining the data from two sources allowed us to est imate age -specific survival rates, and to test hypotheses regarding the influence of age, sex, habitat type, and geographic location on survival of marmots. Methods Study Species The distribution of M. vancouverensis is restricted to the interior mountaino us zones of Vancouver Island (Nagorsen 1987). Marmots live in open, subalpine meadows characterized by colluvial soils, diverse vegetation, and lookout spots (Heard 1977; Milko & Bell 1986). These small patches of suitable habitat were historically inter spersed within much larger areas of dense old growth forest, creating a pronounced metapopulation structure (Bryant 1998). Vancouver Island marmots are generalist herbivores (Martell & Milko 1986) living in social colonies consisting of one or more famil y groups (Bryant 1998). They exhibit slow maturation, delayed dispersal, and large body size relative to most marmots, but similar to other members of the M. caligata group (Armitage 1999; Barash 1989; Griffin et al. 2008). The annual cycle of Vancouver Island marmots consists of an active season (approximately, early May through early October) and hibernation during the winter in underground burrows for an average of 210 days (SE 7.6 days`, Bryant & McAdie 2003). Study A rea This study was conducted on Vancouver Island, British Columbia, Canada. The regional vegetation is dominated by western hemlock and mountain hemlock, with interspersed meadows. The landscape of Vancouver Island has undergone extensive change since the la te 1950s when forestry companies began harvesting at progressively higher elevations. Clearcuts created open habitats resembling meadows, and marmots began to colonize them extensively during the

PAGE 19

19 1980s. By the mid1980s, more than half of the marmot population was living in clearcut habitats; however, few marmots have inhabited clearcut habitats since 2000. The present distribution of marmots on Vancouver Island consists of two main populations (Bryant 1998). The larger metapopulation, Nanaimo Lakes, is on the southern part of the island, and is characterized by mountains of lower elevation and less rugged terrain than the central and northern regions. All colonies within the Nanaimo Lakes metapopulation are concentrated in an area encompassing 840 km2 w ithin 5 adjacent watersheds (Bryant 1998). The smaller population on Mt. Washington is in central Vancouver Island, 80 95 km north of Nanaimo (Figure 2 1). Based on DNA evidence, dispersal between the two metapopulations is unlikely (Kruckenhauser et al. 2009). Releases of captive born individuals have taken place at both Nanaimo Lakes and Mt. Washington. Within the past two years, there have also been re introductions to Strathcona Provincial Park and Mt. Cain, to the west and north of Mt. Washington respectively. Field M ethods Marmots were captured using single door Havahart traps baited with peanut butter; transferred to a tapered handling bag to restrict marmot movement, and sedated using procedures described in Bryant (1996). Each marmot received a pair of numbered metal ear tags. Individuals were classified into one of four age -classes: pup (0 1 years); yearling (1 2 years); two -year old (2 3 years old); or adult ( captured for the first time a s pups or yearlings; it was estimated for marmots captured for the first time as two year -olds or adults (Bryant 1998). The majority of trapping occurred during the months of July and August. Marmot resightings were made throughout the active season using 60x spotting scopes, usually before 1100 hours when marmots were most active. Resightings were considered visual recaptures, and were included in CMR analysis.

PAGE 20

20 Radio telemetry of marmots began in 1992, with an increasing proportion of the marmot populati on radio-tagged since (Figure 2 2 ). Types of transmitters and methods used for surgical implant of transmitters are described in detail by Bryant and Page (2005). Radio-tagged marmots were tracked from the ground or from a helicopter. Non-dispersing mar mots typically remained close (100 1000 m) to home burrows in most cases, and thus could be tracked from the ground. However, helicopter flights were necessary to reach remote colonies and to track dispersing marmots. The frequency of tracking varied dep ending on funding, weather, road access problems, and research priorities in various years of the study (Bryant and Page 2005). Data A nalysis Analysis of radio -telemetry data We used known fate models implemented in Program MARK (White & Burnham 1999; Williams et al. 2002) to estimate and model survival of marmots. Based on radio-telemetry data, we constructed encounter histories for each marmot for each year with 13 time intervals: 12 active season intervals, each 2 weeks long; and one winter season (hibe rnation) interval, approximately 28 weeks long (Bryant & Page 2005). The differences in time interval length between the active and the winter seasons were accounted for during data analysis. A marmot can survive and be radiotracked for more than one ye ar, each of which was considered a separate encounter history; therefore, sample sizes are reported as marmot -years. We developed and tested an a priori set of candidate models investigating the effects of several factors on survival. First, we tested f or the additive and interactive effects of sex and age on survival. For age effect, we considered a 2 age -cl ass model (yearlings versus older animals) and a 3 age class model (yearlings, two-year -olds, and adults). Limited sample sizes did not permit con sideration of pup survival or additional adult classes. Using the most parsimonious model from the preceding analyses, we tested for seasonal variation in survival using 2 -season

PAGE 21

21 (active season, early May early Oct.; and winter, hibernation), 3 -season (sp ring/early summer, emergence 31 Jul; August/fall, 1 Augimmergence, and winter) and 4-season models (spring/early summer, emergence 31 Jul; August, 1 Aug28 Aug; fall, 29 Augimmergence; and winter). Finally, we tested for the additive and interactive effects of habitat type (natural versus clearcut) and location (Nanaimo versus Washington) using the most parsimonious seasonal model as the base model. A subset of radio telemetry data collected from 20 032007 contained both wild-born and captive -born individuals that were released into natural marmot habitat. Analyses proceeded as described previously, except that we also tested for the effect of origin (captive -born versus wild born) on survival. Additionally, we tested for the effect of release cost, where survival of released individuals was allowed to differ from that of wild -born individuals for one year post release but not thereafter (represented by the covariate, first). Small sample sizes precluded testing for the effect of year or influence of temporal covariates. To test if survival of rele ased marmots was influenced by the age at which a marmot was released, we used a subset of data includin g captive -born individuals only. When a radio tagged marmot died, cause of mortality was determined based on clues left by predators on and around the recovered transmitter (Bryant and Page 2005). Winter mortalities were occasionally confirmed by excavation of hibernacula, although the specific caus e of winter mortality could not be determined. We attributed each mortality event to one of five causes: 1) eagle, 2) cougar, 3) wolf, 4) winter, and 5) unknown (includes unknown predator). We estimated cause -specific cumulative mortality rates from radi o telemetry data using the nonparametric cumulative incidence function estimator (NPCIFE ; Heisey & Patterson 2006). The NPCIFE is a generalization of the staggered-entry Kaplan -Meier method of survival

PAGE 22

22 estimation (Pollock et al. 1989), and uses informatio n on the number and timing of deaths from each cause and the number of radio -collared animals at risk at the time to estimate cause -specific mortality rate. We tested for differences in cause-specific mortality rates between captive -born and wi ld marmots using the Cox proportional hazard model stratified by cause of mortality (Heisey & Patterson 2006). Analysis of combined CMR and radio -telemetry data From 19871992 the M. vancouverensis population was monitored by capture -mark recapture (CMR) only. From 19921999 telemetered; from 20002006, > 60% of the known population was radio-telemetered, with an estimated maximum of 95% radio-tagged in 2006 (Figure 2 2 ). Analysis of radio telemetry data is the method of choice for the estimation and modeling of survival because they provide estimates of true survival rates and also allow determination of cause of death (Williams et al. 2002). However, radio-telemetry data were sparse or non -existent until about a decade ago. Also, only a few pups were radio -tagged, so survival could not be estimated using the same methods. Thus, we merged radio telemetry and CMR data becaus e the combined dataset (1 ) spanned a longe r period of time (19872007), (2) allowed estimation and model ing of survival of pup survival, and (3 ) had larger sample sizes which would increase the precision of estimates of survival. To merge the two data sets, the radio -telemetry data were converted into capture histories with annual intervals between capture occasions. A marmot was considered detected if it was radio -tracked any time during July/August of a given year. A marmot was excluded from the dataset if it was only tracked for a portion of a year which did not overlap the trapping season (July/August) ; however, this was uncommon as most marmots were monitored during some portion of that two month interval. Merging the two datasets inevitably leads to a loss of

PAGE 23

23 temporal resolution and other information contained in the radiotelemetry data, but we gain a cohesive data set over the entire duration of the study (19872007). We analyzed the combined dataset using multi -state Cormack Jolly Seber models implemented in Program MARK (White & Burnham 1999; Williams et al. 2002). The states used were four age c lasses chosen based on the biology of the species, as described previously. Vancouver Island marmots disperse as two -year -olds, and females reproduce for the first time as three or four year -olds (average age of first reproduction = 3.6 yrs ; Bryant 2005). Probability of transition from younger to subsequent older age classes was fixed to 1.0, and probability of transition from older to younger age classes was fixed to zero because those transitions are impossible. The dataset was divided into two groups, based on monitoring method (radio telemetry and mark resight). Recapture rate for the radio telemetry group was fixed to one; this group, therefore, contributed to the estimate of survival rate, but not to recapture rate. Fixing the recapture rate to one in CMR analysis effectively results in a known -fate model. Marmots for which radio transmitters failed were censored. Marmots that were initially monitored by CMR and subsequently radio-tagged were censored from the CMR group and then entered in the rad io tagged group for the appropriate time intervals. Analysis of combined data proceeded as described for the radio telemetry data; one main difference was that pups were also included as an age class using the combined data. We used Akaikes Information Criterion, corrected for small sample size (AICc), for model comparison and for statistical inferences (Burnham & Anderson 2002). Model comparison was based on the differences in AICc values ( AICc), and relative support for each model in a candidate mode l set was based on AICc weight.

PAGE 24

24 Results Analysis of Radio -Telemetry D ata One hundred and thirty two marmots were radio tracked during 19922007 for a total of 367 marmot -years. The number of marmots radio -tracked varied over the study period at each site (Figure 2 2 ). There were 92 marmot -years of radio -tracking recorded for yearlings, 89 marmot years for two year -olds, and 186 marmot -years for adults (Table 2 -1). Although annual survival of females ( S= 0.750; 95% CI = 0.6540.826) was slightly higher t han that of males ( S= 0.675; 95% CI = 0.5840.754), there was no evidence of a sex effect (Table 2 2A ). There was also no evidence that survival was age-specific using two age -classes (yearlings and older marmots classes (yearlings, two -year -olds, and adults), nor was there any evidence for additive or interactive effects of sex and age. The overall annual survival for the entire population usin g the most parsimonious model (m odel 1; Table 2 2 A ) was 0.709 (95% CI = 0.6440.766). Theref ore, for further analysis we used the model with no sex or age effect as the base model. Using the most parsimonious model in the preceding analyses (model 1, Table 2 2A ), as a base model we tested for the effects of seasons (models 4 6, Table 2 2 B). Of th e seasonal models, the four season model was the highest ranked; however, the three season model that combined the August and fall seasons, was equally well supported (<1 c: compare model 4 with model 5 in Table 2 2 B). Using the top ranked seasonal mo del (Model 4, Table 2 2 B), two week estimates of survival were intermediate ( S= 0.982; 95% CI = 0.9730.989) during summer, low during August ( S= 0.963; 95% CI = 0.9380.978) and fall ( S= 0.979; 95% CI = 0.9630.988), and high during the winter ( S= 0.996; 95% CI = 0.9920.998). Using both the most parsimonious seasonal model (model 4, Table 2 2B ), we tested for the effects of habitat type (natural or clearcut) and site (Nanaimo or Washington) on survival

PAGE 25

25 (models 1 3, Table 2 2B ). There was no evidence for the effect of habitat type on survival. The most parsimonious model (model 1 Table 2 2B ) indicated that survival differed between sites during the latter half of the active season (Figure 2 3). Estimated annual survival was higher at the Washington site ( S= 0.788; 95% CI = 0.7100.849) than at Nanaimo ( S= 0.678; 95% CI = 0.6060.742). We used a subset of the radio-telemetry data collected during 2003 2007 to test for the effect of origin (captive born or wild -born) on marmot survival. As in the previous analysis, first we tested for age and sex effect, and found no evidence for either (Table 2 3 A ). We used model 1 in Table 2 3 B as a base model for further analysis because this model differed from the most c < 2, and also had fewer parameters. Next, we tested for seasonal variation in survival using the two, th ree, and four season models as described previously; all of these models were similarly supported (models 5 7, Table 2 3 B). Because the two season model (model 6, Table 2 3 B) was well supported, and also had fewer parameters, we chose this model for furth er analysis. Using this model, active season survival ( S= 0.770; 95% CI = 0.7010.827; over 24 weeks) was lower than winter survival ( S= 0.932; 95% CI = 0.8840.961; over 28 weeks). The two season model (model 6, Table 2 3 B) was used as a base model to tes t for the effect of origin (captive born versus wild-born) and site (Washington versus Nanaimo) on survival. The most parsimonious model (model 1, Table 2 3 B) included additive effect of season and origin, providing substantial evidence for the effect of origin on survival; overall, captive born marmots had a much lower survival ( S= 0.605; 95% CI = 0.5070.696) rate than wild-born marmots ( S= 0.854; 95% CI = 0.7600.915). Survival of captive -born marmots was lower than that of wild born marmots during both seasons, with a more pronounced difference during the

PAGE 26

26 active season (Figure 2 4). There was no evidence for the cost of release (compare models 9 and 10 in Table 2 3 B). We also used a subset of data containing only captive born marmots to test for the eff ect of age at -release on survival. There were 46 marmots released as yearlings (74 marmot -years) and 50 marmots released as two -year olds or older (80 marmot -years). Using a 2 -season model (the same seasonal model as in previous analysis), we found that the inclusion of age at release improved the model ( c = 1.71). Annual survival for marmots released as yearlings was lower (0.602; 95% CI = 0.4590.729) than for those released as two-year -olds or adults (0.774; 95% CI = 0.6490.864). Cause Specifi c M ortality There was a significant difference in mortality rates between captive and wild born individuals from eagles, cougars, and over winter ( p = 0.018, 0.039, 0.010, respectively). For each of these causes, mortality was higher for captive -born marm ots than wild -born marmots (Figure 2 5). Mortality due to unknown causes was higher for captive born marmots than wild born marmots (p = 0.054). Wolf predation was the only cause for which wild born marmots had higher mortality than captive -born; however there was not a significant difference between these rates (p = 0.150). Predation by eagles was the most important known cause of death for captive -born marmots, whereas predation by wolves was the primary known cause of death of wildborn marmots (Fig ure 2 5). As a percentage of known mortality (i.e. cause -specific mortality rate/total mortality rate), eagle predation accounted for 25% of captive -born deaths and 13% of wild -born deaths. Cougar predation and winter mortality also were relatively more important sources of mortality for captive -born marmots than wild -born marmots. In contrast, wolf predation accounted for 30% of known wild born marmot mortality and only 6.4% of captive -

PAGE 27

27 born marmot mortality. Mortality due to unknown causes accounted fo r about a third of known deaths for marmots of both origins and was mostly due to predation events where the type of predator could not be identified. Analysis of C ombined ( R adio -Telemetry and CMR) D ata Using the combined dataset, we first tested for the e ffect of age, sex, and additive and interactive effects of the two on survival and recapture rate (Table 2 4 A ). There was evidence to support age -specificity; the most parsimonious model included 3 age -classes for survival (pups, yearling=adult, and two -y ear -olds) and 2 age -classes for resighting rate (yearlings=adults and two -year olds). Estimated annual survival rate was lowest for pups ( S= 0.500; 95% CI = 0.375 0.616), but generally high for yearlings and adults ( S= 0.656; 95% CI = 0.6040.705), and twoyear -olds ( S= 0.649; 95% CI = 0.527=0.917; 95% CI = 0.8440.958) and lower for two-year -olds ( =0.677; 95% CI = 0.4640.836), the dispersing age class (Model 1, Table 2 4 A ) Apparent survival diff ered between sexes only in the two year -old age -class, with females surviving better than males. Survival for two -year -old males was much lower ( S= 0.471; 95% CI = 0.3310.615) than survival of two-year old females ( S= 0.792; 95% CI = 0.6290.895; Figure 2 6 A ). Therefore, we continued our analysis with the model including age -specific survival and recapture rates and sex -specific survival in the two -year -old age -class only (Model 1, Table 2 4 A ). Next, we tested for the effects of habitat type (natural or clearcut) on survival. We found evidence for an effect of habitat type on apparent survival in the two -year -old age -class only. The most parsimonious survival model included the additive effects of both sex and habitat type on the apparent survival of t w o year -old age -class (model 1, Table 2 4B ). Two year -olds in clearcut habitat had lower apparent survival ( S= 0.446; 95% CI = 0.2630.646) than those in

PAGE 28

28 natural habitat ( S= 0.719; 95% CI = 0.5690.833). The additive effects of both sex and habitat type on the apparent survival of two -year old age -class are presented in figure 2 6 B. There was evidence for site -specific dif ferences in apparent survival (m odel 1, Table 2 -4 B), with marmots at Washington surviving better than those at Nanaimo. Discussion Low n umbers, restricted and sparse habitat, small geographic distribution, and anthropogenic influences on marmot habitat continue to cause concern regarding the persistence of Vancouver Island marmots. Given the extremely small population size, the marmot population is at substantial risk of extinction due to demographic stochasticity alone (Caswell 2001; Caughley and Gunn 1996; Morris et al. 2002). This risk is exacerbated by the harsh and dynamic environment M. vancouverensis inhabits, and with changing eco system and predator prey dynamics. Thus, it is likely that conservation measures are necessary in the foreseeable future to ensure longterm persistence of marmots. Our goal was to assist Vancouver Island marmot conservation efforts by providing rigorous estimates of survival rates, evaluating factors influencing survival of marmots, and making recommendations for the reintroduction program. Survival Rates and Sex and Age -Specific D ifferences Using radio -telemetry data (19922007), the overall annual survival of marmots (excluding pups) was 70.9%. There was no evidence that survival differed between sexes or among age classes. Our results regarding the effects of sex on survival of radio tagged marmots as well as estimates of annual survival rate are s imilar to those reported by Bryant & Page (2005). Analyses of combined dataset (i.e., radio telemetry and mark resight) revealed strong evidence for age -specific differences in survival, with annual apparent survival of pups (S= 51.6%) being substantially lower than that of other age -classes ( S

PAGE 29

29 remained close to and hibernated at their natal burrow, and resighting rate of pups (as yearlings the next year) was generally high. Sex effect was apparent only in apparent survival of two -yea r -old marmots. Survival of two -year old males (47%) was much lower than survival of two-year -old females (79%). Vancouver Island marmots typically disperse at two years of age, and males in most species of marmots have a greater propensity to disperse th an females (Armitage 1999; Arnold 1990; Barash 1989; Bryant 1998; Van Vuren and Armitage 1994). We believe that low apparent survival for two-year -old males may be a consequence of male-biased dispersal. This speculation was further supported by the obse rvation that capture probability for two -year olds was substantially lower than that for other age -classes. The low apparent survival rate for the dispersing age class has been observed in other species of marmots (Farand et al. 2002; Ozgul et al. 2006). While it is often thought that dispersing marmots face a greater risk of mortality and therefore have lower survival (Ozgul et al. 2006; Van Vuren and Armitage 1994; Van Vuren 2001), this trend was not apparent in the known-fate analysis of radio -telemet ry data. The lower apparent survival of two -year -old males obtained from CMR analysis than survival estimated from radio-telemetry data most likely was due to the fact that CMR -based methods cannot distinguish between death and permanent emigration (Williams et al. 2002). Our findings (based on the analysis of combined data) that survival of pups was lower than that of other age -classes, and that apparent survival of dispersing age class was lower than that of yearling and adult marmots, are similar to tho se reported for other species of ground -dwelling sciurids (Armitage 2003; Bronson 1979; Bryant 1998; Farand et al. 2002; Griffin et al. 2008; Ozgul et al. 2006; Sherman and Runge 2002). However, annual apparent survival for pups

PAGE 30

30 (50%) was lower than those repor ted for Alpine marmot pups (62%, Farand et al. 2002), yellow bellied marmot pups (Ozgul et al. 2006), and most relevantly, Olympic marmot pups (60% from tagging Griffin et al. 2008). Our estimates of age -specific apparent survival for older age cla sses of M. vancouverensis were generally lower than those reported for other species of marmots (Farand et al. 2002; Ozgul et al. 2006). Seasonal and S ite S pecific V ariation Analysis of radio telemetry data revealed substantial seasonal variation in sur vival (Table 2 2 B, Figure 2 3), with higher survival during hibernation than during late summer/fall (also see Bryant & Page 2005). However, support for two, three, and four -season models was similar, suggesting that the primary cause of seasonal variation was the substantially higher survival during hibernation than during the active season. Marmots are not exposed to predators and other mortality factors linked with activity above ground during hibernation, and are thought to be efficient hibernators (B ryant & McAdie 2003); consequently, survival rates are higher during winter than during the active season. Comparison of our results with those of Bryant & Page (2005) suggests that seasonal variation in survival, throughout the year and particularly during active season, may have become less pronounced in recent years (i.e., mortality is more evenly spread throughout the active season, instead of higher mortality during August and fall relative to spring and early summer). Both the known fate radio telem etry data and combined telemetry and CMR data suggest that survival at the Mt. Washington site was higher than in the Nanaimo Lakes region. The two areas are quite different; Mt. Washington is a ski hill, where most marmots live on ski slopes. These ski slopes are technically clearcuts; however, they more closely resemble natural meadows due to stumps being removed and maintenance to prevent regeneration. Apparently, the ski

PAGE 31

31 slopes are suitable habitat, although differences in vegetative composition and how this affects fitness and survival of marmots have not been formally tested. The other major difference is the amount of human activity at each site. Whereas much of the land inhabited by marmots in the Nanaimo Lakes region is private land inaccessible to the public, Washington is a popular ski slope with lots of human traffic. Griffin et al. (2007) found that high levels of tourist traffic did not negatively affect demographic rates of Olympic marmots in Olympic National Park. Although the majority o f human activities at Mt. Washington occur during winter months while marmots are hibernating, summer activities have become increasingly popular in recent years. While human presence may not have a direct effect on demographic rates of marmots, t here cou ld be indirect effects (Griffin et al. 2007). Habitat C hanges and Vancouver Island M armots The landscape of Vancouver Island has undergone substantial change, especially since logging intensity increased in the 1950s. High elevation clearcut logging crea ted patches of open meadows that resembled natural marmot habitat. These newly created habitats were colonized by marmots, and by the 1990s, nearly half of the known marmot population inhabited clearcuts (Bryant 1996; Bryant and Janz 1996). However, most marmot colonies in clearcut habitats have disappeared and very few marmots have been observed in clearcuts in recent years. Why did marmots disappear from the clearcut habitats where they thrived during 1980s? One explanation is that clearcut habitat se rved as an ecological or evolutionary trap (Robertson and Hutto 2006; Sherman and Runge 2002). While marmots may have persisted in clearcuts that initially resembled natural meadows, subsequent forest regeneration rendered clearcut habitat unsuitable for marmots over a few years or decades. In addition, clearcut logging substantially changed the landscape structure and local marmot population density. These changes could have caused an increase in mortality of marmots due to predation through several

PAGE 32

32 mecha nisms: (1 ) an increased proportion of open habitat in the landscape, making the area more attractive to avian predators; ( 2 ) an increase in secondary growth and edge habitat, where predation by wolves and cougars would be higher; (3) logging roads could h ave acted as conduits for the movement of both predators and marmots, possibly increasing the probability of encounters; and/ or 4 ) an increase in secondary growth altering the abundance of other prey species, and consequently the abundance of predators. W e note that these mechanisms are not mutually exclusive. Captive Breeding, Cause -Specific Mortality, Age-at -R elease an d C onservation of M. vancouverensis Our results clearly show that captive born marmots suffer a higher mortality than wild born marmots, and these results are consistent with findings of Jule et al. (2008). In addition, causes of mortality also differed between the two groups; captive -born marmots were much more vulnerable to avian predators than their wild -born counterparts. These result s suggest that, whereas captive born marmots exhibit anti -predatory responses to mammalian predators (Blumstein et al. 2001; Blumstein et al. 2006), their ability to detect and/or avoid avian predators may be compromised. Pre release anti -predator training improves survival in the wild for some species (Shier and Owings 2006; van Heezik et al. 1999). Exposing captive -born marmots to predators (particularly, golden eagles) before they are released may enhance their survival in natural habitats and contribute to the success of the release program; this however, is unproven and would be financially and logistically challenging to implement. In some species, reintroduced individuals suffer a higher mortality during the first year after release; however, the surv ival may be similar to those of wild -born individuals thereafter (BarDavid et al. 2005; Le Gouar et al. 2008; Sarrazin et al. 1994). We tested for this effect, and

PAGE 33

33 found no evidence that survival of captive -born marmots improved after their first active s eason, winter, or full year in the wild. Age at which captive -bred individuals are released can also influence demography of reintroduced populations (Green et al. 2005; Sarrazin et al. 1994; Sarrazin and Legendre 2000). Marmots released as two -year olds had an approximately 17% higher probability of survival than marmots released as yearlings (survival of marmots released as yearlings : 0.602 0.070; s urvival of marmots released at age two and older : 0.774 0.055). Thus, marmots released at >2 years of age are more likely to survive to reproductive age and contribute to the population growth via reproduction. The initial increase in costs associated with rearing marmots for an extra year may be well worth it if, as our results suggest, it can increase th e probability of success of the release program.

PAGE 34

34 Table 2 1 Summary of records included in each analysis. For known fate analyses, value represents number of marmot -years (one record per marmot per year); for the analysis of combined radio -telemetry and mark resight data, values represent number of mark recapture histories (roughly equal to number of marmots; however, one marmot can be included in the capture -mark resight group in one year and the radio -telemet ry group in another year depending on implantations and radio failures). Age Class Sex Data Set Pups Yearlings Two Year O lds Adults Male Female Radio T elemetry (1992 2007) 0 92 89 186 206 161 Radio T elemetry (20032007)* 0 66 74 115 148 107 Combination 70 98 102 316 136 137 Habitat Type Site Origin Data Set Clearcut Natural Nanaimo Washington Captive Wild Radio T elemetry (1992 2007) 31 336 262 105 154 213 Radio T elemetry (20032007)* 0 255 178 77 154 101 Combination 93 180 33 240 48 225 *Subs et of radio -telemetry data were used to test for the effect of origin

PAGE 35

35 Table 2 2 Analysis of the influence of (A) intrinsic (sex and age) and (B ) extrinsic factors on survival (S) of the Vancouver Island marmot using radio-telemetry data (19922007) and known -fate models. Akaikes Information Criterion corrected for small sample size (AICc), differences in AICc values ( AICc), AICc weights (wi), and the number of parameters (np) are shown for each model. The most parsimonious model in the first set of analysis testing (for intrinsic differences in survival) was used as the base model for the second set of analysis testing for effects of extrinsic factors. In both analyses, the most parsimonious models are highlighted in bold. No. Model AICc w i np A )* 1 S(.) 593.65 0.00 0.275 1 2 S(sex) 594.21 0.56 0.208 2 3 S(2 age class) 595.17 1.52 0.129 2 4 S(2 age -class+sex) 595.26 1.61 0.123 3 5 S(2 age class+sex) 595.74 2.08 0.097 3 6 S(3 age class) 597.09 3.44 0.049 3 7 S(2 age -class*sex) 597.25 3.59 0.046 4 8 S(3 age class+sex) 597.61 3.96 0.038 4 9 S(3 age class+sex 2yr ) 597.74 4.09 0.036 4 B )** 1 S(4 season+site A,F ) 552.75 0.00 0.645 5 2 S(4 -season+site) 555.59 2.84 0.156 5 3 S(4 -season+site S,A,F ) 556.69 3.94 0.090 5 4 S(4 season) 557.86 5.11 0.050 4 5 S(3 -season) 558.27 5.51 0.041 3 6 S(2 -season) 560.10 7.34 0.016 2 7 S(time) 567.32 14.57 0.000 13 8 S(site) 590.91 38.16 0.000 2 9 S(.) 593.65 40.90 0.000 1 10 S(habitat) 595.22 42.47 0.000 2 *Age -classes are yearlings, two -year olds, and adults for the 3 age -class models. Age classes for 2 age -class models are yearlings and adults. Subscripts indicate age -class(es) for which the model allows the age class to affect survival ( -yrs of age or older; 2yr = 2 -yr olds; ad = adults). **The 4 seasons were spring/early summer ( S emergenceJuly 31), August ( A August 1 28), fall ( F August 29 immergence), and winter ( W hibernation); the 3 season model combined August and fall; and the two season model combined the entire active season. Subscripts indicate allowing survival to differ between sites only during season(s) noted. Pluse s (+) indicate additive effects and asterisks (*) indicate interactive effects. The dot (.) indicates a constant parameter value.

PAGE 36

36 Table 2 3 Analysis of the influe nce of (A ) intrinsic (sex and age) and ( B) extrinsic factors on survival ( S ) of the Vancou ver Island marmot using a subset of the radio-telemetry data (20032007) and known-fate models. Akaikes Information Criterion corrected for small sample size (AICc), differences in AICc values ( AICc), AICc weights (wi), and the number of parameters (np) are shown for each model. The most parsimonious model in the first set of analysis testing (for intrinsic differences in survival) was used as the base model for the second set of analysis testing for effects of extrin sic factors. In both analyses, the most parsimonious models are highlighted in bold. No. Model AICc wi np A )* 1 S(sex) 488.66 0.00 0.322 2 2 S(2 age class+sex) 489.79 1.13 0.183 3 3 S(.) 490.21 1.54 0.149 1 4 S(2 age class) 490.77 2.11 0.112 2 5 S(3 age class+sex) 491.43 2.77 0.081 4 6 S(2 age -class+sex) 491.44 2.77 0.080 3 7 S(3 age class) 492.56 3.89 0.046 3 8 S(3 age class+sex 2yr ) 493.58 4.92 0.028 4 B )** 1 S(2 season+origin) 455.82 0.00 0.368 3 2 S(2 -season*origin) 456.19 0.37 0.305 4 3 S(2 season+origin+first A,F ) 457.25 1.43 0.180 4 4 S(2 season+origin+first W ) 457.69 1.87 0.145 4 5 S(4 -season) 467.92 12.10 0.001 4 6 S(2 season) 468.29 12.46 0.001 2 7 S(3 season) 470.24 14.42 0.000 3 8 S(origin) 479.60 23.78 0.000 2 9 S(.) 490.21 34.38 0.000 1 10 S(first) 491.79 35.97 0.000 2 *Age -classes are yearlings, two -year olds, and adults for the 3 age -class models. Age classes for 2 age -class models are yearlings and adults. Subscripts indicate age -class(es) for which the model allows the age class to affect survival ( -yrs of age or older; 2yr = 2 -yr olds; ad = adults). **The 4 seasons were spring/early summer ( S emergenceJuly 31), August ( A August 1 28), fall ( F August 29 immergence), and winter ( W hibernation); the 3 season model combined August and fall; and the two season model combined the entire active season. Subscripts indicate allowing survival to differ between sites only during season(s) noted. The first covariate refers to the release cost effect, where survival is different the first year for released marmots (or first season only as indicated by subscript) than post -fi rst year. Pluses (+) indicate additive effects and asterisks (*) indicate interactive effects. The dot (.) indicates a constant parameter value.

PAGE 37

37 Table 2 4 Analysis of the influence of (A) intrinsic (sex and age) and (B ) extrinsic factors on survival ( S ) and recapture rate ( ) of the Vancouver Island marmot using a combination of radio telemetry and capture -re capture data and multi -state models. Akaikes Information Criterion corrected for small sample size (AICc), differences in AICc values ( AICc), AICc weights (wi), and the number of parameters (np) are shown for each model. The most parsimonious model in the first set of analysis (testing for intrinsic differences in survival) was used as the base model for the second set of analysis testing for e ffects of extrinsic factors. In both analyses, the most parsimonious m odels are highlighted in bold. No. Model AICc wi np A )* 1 S(3 age class+sex 2yr age) 765.81 0.00 0.644 6 2 S(3 age class+sex 2yr age+sex 2yr ) 767.59 1.77 0.265 7 3 S(2 age) 771.44 5.62 0.039 4 4 S(4 age+sex age) 773.02 7.21 0.018 7 5 S(3 age) 773.15 7.33 0.016 5 6 age) 774.73 8.92 0.007 3 7 S(4 age) 775.18 9.37 0.006 6 8 age) 776.95 11.14 0.002 4 9 778.97 13.16 0.001 3 10 779.54 13.73 0.001 2 11 780.56 14.75 0.000 4 12 780.81 15.00 0.000 3 B ) 1 S(3 age+sex 2yr +site+habitat 2yr age) 755.28 0.00 0.753 8 2 S(3 age+sex 2yr age) 758.04 2.76 0.189 7 3 S(3 age+sex 2yr +habitat 2yr age) 761.49 6.21 0.034 7 4 S(3 age+sex 2yr +site age) 763.72 8.45 0.011 7 5 S(3 age+sex 2yr +habitat age) 765.12 9.85 0.005 7 6 S(3 age+sex 2yr age) 765.81 10.54 0.004 6 7 S(3 age+sex 2yr age) 766.02 10.75 0.003 7 *Age -classes are pups, yearlings, twoyear -olds, and adults. For recapture rate, 3 age -class models consider yearling, two year -old, and adult differently; and, 2 age -class models consider yearling and adult recapture rate to be equal but different from t wo -year old recapture rate. For apparent survival, 2 age -class models are pups and older marmots (>1 yr); 3 age -class models are where pup and two year -old apparent survival differ, and yearling and adult survival are set equal. All subscripts refer to the age -class(es) for which the individual covariate was considered ( -yrs of age or older; 2yr = 2 yr olds; ad = adults). Pluses (+) indicate additive effects and asterisks (*) indicate interactive effects. The dot (.) indicates a constant paramet er value.

PAGE 38

38 Figure 2 1 Historically occupied (squares) and recently occupied (circles) sites for M. vancouverensis The rectangle encompasses recently active colonies within the Nanaimo Lakes region, from which the majority of survival data were obtained. Locations of the isolated Mt. Washington sub-population, and recent reintroductions at Mt. Cain and Greig Ridge are also shown.

PAGE 39

39 Figure 2 2 Changes in numbers of ear tagged (hatched bars) and radio -telemetered (solid bars) marmots for which annual survival data were available. Almost half of the sampled marmots were telemetered, with the Mt. Washington sub -population being particularly well -represented. The overall trend illustrates greatly increased reliance upon radio -telemetry in recent years, largely driven by growing numbers of reintroduced captive -born marmots.

PAGE 40

40 Figure 2 3 Estimates of biweekly survival rates ( 1SE) estimated using the most parsimonious model of the known-fate analysis (Table 1 2b, model no. 1). Survival differ ed between sites during August and fall, but not during spring/early summer or winter. For the winter interval, survival was estimated over a two -week interval as a derived parameter; therefore, winter survival is directly comparable to biweekly survival rates during other seasons.

PAGE 41

41 Figure 2 4 Estimates of seasonal survival rates ( 1SE) for captive -born and wild-born marmots estimated using the most parsimonious model from the subset analysis (Table 1 3b, model no. 1) Active season survival spans 24 weeks, and the winter season spans 28 weeks.

PAGE 42

42 Figure 2 5 Estimates of cause -specific mortality rates (mean 1SE) for captive born and wild born marmots estimated using the nonparametric cumulative incidence function estimator. Significant diffe rences are indicated by an asterisk (*). Most of the unknown cases of mortality represented cases in which predation was known to have been the cause of death, but for which forensic clues precluded identification of the predator species involved

PAGE 43

43 Figure 2 6 Appa rent survival rates from combined data analysis. A ) Age -specific annual apparent survival rates (means 1SE)) estimated using model no. 1, Table 1 4a. Apparent survival rates differed between males and feamales only for two -year olds. B) Sex -specific annual apparent survival rates for two -year -old ma rmots by habitat type at Nanaimo Lakes based on model no. 1 in Table 1 4b. B A

PAGE 44

44 CHAPTER 3 POPULATION ECOLOGY OF THE ENDANGERED VANCOUVER ISLAND MARMOT Introduction Captive breeding and reintroduction are frequently used for the conservation of rare or endangered species (Caughley & Gunn 1996; Sarrazin & Legendre 2000) However, few reintroduction programs have achieved success, due in part to poor understanding of the populationlevel effects of alternative release stra tegies (Armstrong & Seddon 2008; Grenier et a l. 2007; Sarrazin & Legendre 2000; Seddon et al. 2007) Furthermore, the reintroduction literature largely consists of descriptive accounts of success or failure of reintroduction programs; progress in this branch of conservation biology has been slow due to the lack of quantitatively rigorous analysis of data generated from well -designed comparative or experimental studies (Armstrong & Seddon 2008) Demographic models can provide valuable insights into likely results of conservation efforts, and allow for evaluation of the populationlevel impacts of release strategies; however, few reintroduction programs have benefitted from such models (Caswell 2001; Meretsky et al. 2000; Sarrazin & Legendre 2000) The Vancouver Island marmot ( Marmota vancouverensis ) is an endangered mammal inhabiting subalpine meadows on Vancouver Island, British Columbia (Janz et al. 2000; Nagorsen 1987) The M. vancouverensis population went through several range restrictions leaving the population in only two locations on the island by the 1970s (Figure 3 1). The marmot population increased to 300350 individuals during the mid1980s as marmots colonized the openings created by high altitude logging that resembled marmot habitat (natural meadows, Bryant & Janz 1996) This increase in population size was short lived as most marmot colonies in clearcut habitats dis appeared by 2000 and the population declined to an estimated 35 individuals by 2003 (Bryant 2005) The disappearance of marmots from clearcut habitat is likely

PAGE 45

45 due to forest regeneration reducing the suitabilit y of clearcuts as marmot habitat (Bryant 1998) Changes in landscape structure following logging probably had direct and indirect impacts on the entire marmot population through predator -prey interactions (Aaltonen et al. in review; Bryant & Page 2005) With the marmot population on the brink of extinction, a captive -bre eding program was initiated in 1997 with facilities established both on and off Vancouver Island (Bryant 2005) During 20032008, 155 captive -born marmots were released into various sites at both Mt. Washington and Nanaimo Lakes (Kruckenhauser et al. 2009; unpubl ished minutes, Vancouver Island Marmot Recovery Team, Dec. 2008) The reintroduction program has contributed to an increase in population size in the wild (COSEWIC 2008). However, recent work has shown that captive -born marmots suffer substantially higher mortality compared to their wild -born counterparts, and that causes of mortality differed substantially between captive born and wild -born marmots (Aaltonen et al. in review). Moreover, survival of released marmots in the wild was found to be influenced by the age at which they were released. Marmots released as two -year olds or older survived better than those released as yea rlings. These findings may have important consequences for the reestablishment of a viable population and for the consideration of alternative management scenarios. Although some estimates of demographic rates exist (Aaltonen et al. in review; Bryant 2005; Bryant & Page 2005; CO SEWIC 2008) only limited information on the population ecology of Vancouver Island marmots is currently available. Our intent was to narrow this gap in knowledge by providing better information on population ecology of Vancouver Island marmots and evaluating potential population level impacts of alternative release strategies.

PAGE 46

46 In this paper we (1) use estimates of demographic variables to parameterize age -structure d matrix population models, (2) analyze the matrices to calculate population growth rat es and (3) calculate the elasticity of the population growth rate to vital demographic parameters. We also estimate target demographic rates for recovery of the species and investigate population level effects of releasing marmots either as yearlings or as two -year -olds, and the effects of a reduction in mortality of captive -born marmots after they have been released into their natural habitats. Our findings are expected to guide future release strategy and other management efforts aimed at the conservat ion of this critically endangered species. Methods Study Species Vancouver Island marmots live in small, subalpine meadows characterized by colluvial soils, diverse vegetation, and rocky outcrops (Heard 1977; Milko & Bell 1986) They are generalist herbivores (Martell & Milko 1986) and are active during the spring and summer (approximately, early May through early October). They hibernate in underground burrows during the winter for an a verage of 210 days (SE 7.6 days, Bryant & Mc Adie 2003) They exhibit slow maturation, delayed dispersal, and large body size relative to most marmots, but similar to other members of the M. caligata group (Armitage 1999; Barash 1989; Griffin et al. 2008) Study Area This study was conducted on V ancouver Island, British Columbia, Canada. The present distribution of marmots consists of two main populations: Nanaimo Lakes and Mt. Washington (COSEWIC 2008). The Mt. Washington population is relatively small, and data were insufficient for estimating age -structured demographic parameters. Thus, we focused on the data collected from the larger population, Nanaimo Lakes. The Nanaimo Lakes population is located

PAGE 47

47 on the southern part of the island, and is characterized by mountains of lower elevation and less rugged terrain than the central and northern regions. All colonies within the Nanaimo Lakes population are concentrated in an area encompassing 840 km2 within 5 adjacent watersheds (Bryant 1998). Marmot habitat in the Nanaimo Lakes region is descri bed in detail elsewhere (Bryant & Janz 1996) Field Methods From 19872008, m armots were captured using single door Havahart traps baited with peanut butter, transferred to a tapered handling bag to restrict marmot movement, and sedated using procedures de scribed in Bryant (1996). Each marmot received a pair of numbered metal ear tags. Individuals were classified into one of four age -classes: pup (0 1 years), yearling (1 2 years), two -year old (23 years old), or adult ( known for marmots captured for the first time as pups or yearlings; it was estimated for marmots captured for the first time as two -year -olds or adults (Bryant 1998) The majority of trapping occurred during the months of July and August. Marmot resightings were made throughout the active season using 60x spotting scopes, usually before 1100 hours when marmots were most active. Radio telemetry of marmots began in 1992, with an increasing proportion of the ma rmot population radio-tagged since 2000. Types of transmitters and methods used for surgical implant of transmitters are described in detail by Bryant and Page (2005). Radio-tagged marmots were tracked from the ground or from a helicopter. Non-dispersin g marmots typically remained close ( flights were necessary to reach remote colonies and to track dispersing marmots. The frequency of tracking varied depending on funding, weather, road access problems and research priorities in various years of the study (Bryant & Page 2005)

PAGE 48

48 Marmota vancouverensis pups are born in early June, emerge from their natal burrows in early July, and typically s tay near natal burrows for the first few weeks post -emergence (Bryant 1996) The number of weaned pups per female was counted when pups first emerged from their natal burrows during the first weeks of July. Pups were not trapped immediately post -emergence due to small size (Bryant 1998) therefore, there is a possibility of pup mortality both pr ior to emerging from the natal burrow and post -emergence but prior to tagging. Two year -old females breed infrequently; most female marm ots breed for the first time as 3 or 4 -year olds (Bryant 2005) Estimates of Survival and Fecundity Rates From 19871992, the M. vancouverensis population was monitored by capture -mark resight (CMR) method only. Resightings were considered visual reca ptures, and were included in CMR analysis. From 19921999, tagged; from 20002006, > 60% of the known population was radio-tagged, with an estimated maximum of 95% radio tagged in 2006. Analysis of radio telemetry d ata is the method of choice for the estimation and modeling of survival because it provides estimates of true survival rates and also allow s determination of cause of death (Williams et al. 2002) However, radio telemetry data were sparse or non -existent until about a decade ago. Also, only a few pups were radio tagged, so pup survival could not be es ti mated using radio -telemetry Thus, we estimated age-specific survival rates using both capture resight and radio-telemetry data (Aaltonen et al. in review) We merged radio -telemetry and CMR data because the combined dataset (a) spanned a longer period of time (19872007), (b) allowed estimation and modeling of survival of pup survival, and (c) had larger sample sizes which would increase the precision of estimates of survival. We used multi -state Cormack Jolly -Seber models implemented in Program MARK (White & Burnham 1999; Williams et al. 2002) to estimate stage -specific survival from the

PAGE 49

49 combined dataset. The states used were four age classes chosen based on the biology of the species: pup (0 1 years), yearling (12 years), twoyear -old (2 3 years), and adult ( estimated survival over the entire study period (19872007) and also over the populati on decline before reintroduction began (19872002). We used known fate models implemented in Program MARK (White & Burnham 1999; Williams et al. 2002) to estimate and model survival of marmots using radiotelemetry data. The subset of radio-telemetry data collected from 2003 2008 contained both wildborn and captive born individuals that were released into natural marmot habitat. This subset was used to estimate age -specific survival rates for the entire population, and the wild born and released individuals separately over the past five years. We used pup survival estimates from CMR methods for this time period as pups were not tracked by radiotelemetry. Using data from females in the wild, we estimated age-specific litter size as the number of weaned pups per litter produced by a female of age x and fe cundity rate as one -half times the average number of pups produced by females (including females that did not reproduce that year) of each age -class. There was no evidence of temporal variation in litter size; the primary sex ratio in the wild did not var y significantly from 1:1 (Bryant 2005) Construction and Analysis of Population Model The parameter estimates described above were used to construct several age -structured matrix population models (Table 3 1). We used four age -based stages: 1) pup, age 0 1; 2) yearling, age 1 2; 3) twoyear -old, age 2 3; and adult, ages 3 and older. Age -specific fertility rate was estimated using the post breeding census formulation (Caswell 2001): Fi = Pimi where mi is age -specific fecundity and Pi is age -specific survival rate.

PAGE 50

50 Using these estimates of survival and fertility rates, we constructed a population projection matrix A (Figur e 3 2 ), where subscripts refer to age -class (p=pups, y=yearlings, t=two -year olds, a=adults). Deterministic populat ion growth rate, (lambda), was calculated as the dominant eigenvalue of the population projection matrix. We calculated sensitivity of to vital demographic rates as partial derivative of with respect to vital rates, and elasticity of to vital demograp hic variables as (Caswell 2001): where eij is the elasticity of to ijth entry of the pop ulation projection matrix, and / aij is the sensitivity of to ijth entry of t he matrix. Result s Estimates of Survival and Fecundity R ates Survival rates varied depending on time period and origin of the population (wild-born or captive -born marmots released into the wild). Pup survival ranged from 0.460 (0.3390.585; 19872007; model 1) to 0.492 (0.3720.613; 19872002; model 4), depending on the time period over which survival was estimated (Table 3 1). Yearling survival ranged from 0.618 (0.5590.674; 19872007; model 1) to 0.893 (0.7560.958; 20032008; model 6). Twoyear -old survival ranged from 0.736 (0.4250.913) to 0.923 (0.7560.979). Adult survival ranged from 0.456 (0.3070.613) to 0.812 (0.6880.894). For yearlings, two-year -olds, and adults, the highest estimate of survival rate was based on the model that used just the post release subset of radio telemetry data for wild -born individuals only ( model 6 Table 3 1). For two -year olds and adults, the lowest survival rate was estimated from the same data and survival model, but for captive -born marmots only ( model 7 Table 3 1). ij ij ija e a

PAGE 51

51 Fecundity did not vary over time or with origin ( wild or captive -born), but did vary with age. Reproductive output increased with age, two -year old fecundity being much lower than that of older females ( mx = 0.129, 0.420, and 0.616, for 2 year -olds, 3 -year olds, and -year -olds respectively). A much smaller proportion of two -year -old females breed and they had a smaller average litter size compared to older females (Figure 3 3A ). Overall, average litter size was 3.3 9 (0.13; N=66 litters; Figure 3 3A ) and average age of first reproduction was 3. 3 3 ( 0.22; N=24; Figure 3 3B ). Population Growth Rate and E lasticities Asymptotic population growth rate, ranged from 0.721 (model 7, Table 3 -1 ) to 1.038 (model 6, Table 3 1), depending primarily on the subset of the data used to estimate survival rates. Bo th of these growth rates represent hypothetical populations in which populations consist of entirely captive -born, released individuals ( =0.721) or entirely of wildborn marmots in recent years (20032008; =1.038). The population growth rate estimated f or the entire population (both wild and captive -born) since releases began in 2003 was 0.876 (model 5, Table 3 1 ). Population growth estimated using parameters estimated from combined (i.e. radio telemetry and CMR) data over the entire study period (model 1, Table 3 1 ) was 0.821. All estimates of population growth rate for the entire population over any duration, using parameters from either combined or radiotelemetry data and from any model, were in the range of 0.821 to 0.878. This indicates populatio n decline of at least 12 % and up to 18%, encompassing various time periods and methods of parameter estimation. In general, population growth rate was more sensitive to proportional changes in survival rates than fertility rates. Specifically, the population growth rate was proportionally most sensitive to changes in survival of adults, followed by pups, yearlings, and twoyear -olds. These trends of relative importance of various demographic rates to population growth rate was

PAGE 52

52 consistent for all the model s we tested ( e ( Pa) > e ( Pp) > e ( Py) > e ( Pt) > e ( Fa) > e (Ft) > e ( Fy); Table 3 2). The population growth rate was proportionally much more sensitive to changes in survival of adults than to changes in any other parameter for all models, with e ( Pa) twice as large as any other elasticity (Table 3 2 Figure 3 4 ), except for the model representing captive -born individuals only (Table 3 2, model 7). For released marmots, survival of younger age -classes was similar in importance to adult survival (Figu re 3 4) Discussion Population modeling is a necessary, but often missed, step in recovery planning for endangered species (Meretsky et al. 2000; Sarrazin & Legendre 2000) We must have an understanding of what the population dynamics were like during the decline of a species, including which demographic rates drove the population decline (Caswell 2001) If population dynamics in recent years are similar to dynamics during population decline, the n the goal of ensuring population persistence in the wild may not be met unless conditions change. Thr ough population modeling, we can determine goals for demographic rates in the wild and evaluate alternative reintroduction scenarios (Bar David et al. 2005; Caswel l 2001; Sarrazin & Legendre 2000) ; this information is needed but was not available for the Vancouver Island marmot population until this study. Population D ecline The Vancouver Island marmot population crashed from an estimated 300 350 marmots in the m id 1980s to less than 50 marmots by 2002, representing a drastic decline in just 15 years. We estimated population growth rate over the entire study period (19872007) to be between 0.82 and 0.85. Our estimates may slightly underestimate the population g rowth rate due to the limitations of CMR -based survival parameters; however, a 15% decline annually appears reasonable when compared to the observed trend.

PAGE 53

53 Several populations of other marmot species are not in decline. Yellow -bellied marmots (M. flaviv entris ) in western North America and various alpine marmot ( M. marmota ) populations throughout Europe appear stable or even slightly growing (Farand et al. 2002; Ozgul et al. 2006; Ozgul et al. 2009; Stephens et al. 2002) Our estimates of age -specific survival were generally lower than thos e estimated for these populations (Oli & Armitage 2004; Ozgul et al. 2006) Reproductive parameters were also slightly lower; however, both the yellow -bellied and alpine marmot exhibit a faster life history so this is to be expected (Armitage 2003) Reproductive rates of yellow -bellied and alpine marmots are comparable to those es timated for the Vancouver Island marmot only in situations of high population density. Reproductive suppression has been a commonly cited explanation for low proportions of females breeding in colonies of high population density (Griffin et al. 2008; Ozgul et al. 2006) Low densities of Vancouver Island marmots make reproductive suppression an unlikely mechanism for any decrease in reproductive output. On the other hand, the population growth rate recently estimated for the Olympic marmot (M. oly mpus ) suggests a decreasing population at a rate of 7% annually from 20022006 (Griffin et al. 2008) The Olympic marmot, l ike the Vancouver Island marmot is a species with limited geographic range and slow life -history compared to other species of marmot. Survival of Vancouver Island marmot pups was slightly lower than survival of Olympic marmot pups ; however, survival rate s for all other age -classes were similar (Griffin et al. 2008; Griffin et al. 2007b) Reproductive rates of Vancou ver Island marmots were also similar to those reported for Olympic marmots in recent years and slightly lower than historic estimates, from a time of presumed stability for the Olympic marmot population (Barash, 1973; Griffin et al. 2008; Griffin

PAGE 54

54 et al. 2007a) Griffin (2008) considers the slight decline i n reproductive output a secondary effect of increased mortality of adult females. This was also the case in the collapse (>95% decrease from 1987 1999) of a Northern Idaho ground squirrel population where decreased reproductive rates was a result of a low proportion of adult females in the population (Sherman & Runge 2002) It is apparent that low survival due to predation is the proximate cause of the Vancouver Island marmot pop ulation crash. As primary evidence, population growth rate was much more sensitive to proportional changes in survival than to proportional changes in reproduction. Furthermore, any decrease in reproductive output of the population is most likely a resul t of high adult mortality through two mechanisms. First, high adult mortality leads to lower adult population density, which may lead to mate -finding Allee effects. Second, fewer adult females in the population causes a shift in age structure towards younger individuals of lower fecundity. These demographic causes of proximate population decline are similar to those cited for both the crash of a Northern Idaho ground squirrel population (Sherman & Runge 2002) and the recent decline of the Olympic marmot population (Griffin et al. 2008) Potential for R ecovery In recent years (20032008), population growth rate has improved slightly to 0.876. While there has been an increase in population size due to release of captive -born marmots, models still predict a 12% decline per year. If population size has increased in the wild solely due to addition of released marmots, this effect will persist only for the duration of releases, unles s the population growth rate improves. To investigate what level of demographic success is necessary to create a persistent population in the wild, we estimated two hypothetical population growth rates representing best case and worst -case scenarios. The first, a population of captive -born, released individuals only;

PAGE 55

55 and the second, using demographic rates of only wild -born marmots in recent years. Estimated growth rate for a population of entirely captive born individuals was 0.721. With a decline of ne arly 30% a year, the current demographic rates of released individuals will not lead to a persistent reintroduced population. However, some hope lies with the marmots born in the wild in recent years. The population growth rate we estimated for a popula tion of these wild born marmots was 1.03; the only model we examined which predicted a growing population. If enough marmots are released into the wild, and if enough of these marmots survive to reproductive age, than their offspring might contribute to a viable population. This strategy may require the release of a large number of individuals over a number of years, if we must wait for the next generation of marmots to survive sufficiently in the wild to produce a persistent population. It also requires that marmots are released and remain in a spatial distribution at high enough densities to minimize mate finding Allee effects. Over the past 3 years, some released female marmots which do survive to reproductive age and are in the presence of a male have reproduced in the wild. As additional evidence, although sample sizes are small, we were unable to detect an effect of origin (captive or wild -born) on reproductive rates. It also seems as though a second generation effect of captivity on survival is un likely because the high survival rate estimated for wild -born marmots in recent years is comprised of both offspring of wild born individuals and of offspring of released marmots,. Survival of captive born marmots released into the wild must be improved to produce a persistent population in the wild. If marmots continue to be released as yearlings, with an extremely low survival rate (lower than those released at age two or older, Aaltonen et al. in review) the probability that they contribute to p opulation growth through reproduction is low.

PAGE 56

56 Releasing marmots at an older age (i.e. shorter time required to survive until age with maximum reproductive value) would likely increase the success of the reintroduction program. We do not have sufficient data on predator populations nor sufficient understanding of the interaction between the altered landscape and predator -prey dynamics on the island to know if the cause of low overall marmot survival (and hence the proximate cause of population decline) ha s been removed. It is possible that the difference in survival rates between captive -born and wild -born marmots is due to selective predation on captive -born marmots because of some decreased ability of released marmots to avoid or escape predators parti cularly avian predators If released individuals can survive long enough to produce an increasingly wild -born population, there are multiple possible outcomes. As captive -born individuals become less abundant, wildborn individuals could become increasingly de predated and survival of wild -born individuals could decrease. Alternatively, wild -born offspring could be more capable of avoiding predation and survival of wild born individuals could remain high. Currently, reproductive rates of the Vancouver Isl and marmot seem adequate, although the only true evidence is by comparison with historic reproductive rates of the Olympic marmot and comparison with other members of the M. caligata group (Bryant 2005; Griffin et al. 2008) If reproduction is insufficient at all, it is most likely due to fewer older aged females (with higher reproductive value) being alive to breed. Matrix population models project a continuing declin e of greater than 10% annually unless vital rates improve. We have found that low survival is an extremely important driver of population decline, and this is likely to be the case if survival of captive -born marmots in the wild is not increased. These f indings suggest that management strategies aimed at increasing survival rates would be required to ensure the longterm persistence of the Vancouver Island

PAGE 57

57 marmot. Further population mode ling and use of alternative reintroduction strategies within an expe rimental design are necessary to improve knowledge about survival in the wild, the overall recovery strategy, and the probability of achieving a persistent wild population of Vancouver Island marmots.

PAGE 58

58 Table 3 1 Description of data used, time period, and estimates of age -specific survival rates used in each matrix population model Parameter Model Dataset* Time Period Population P p P y P t P a 1 Combined 1987 2007 entire 0.460 (0.339 0.585) 0.618 (0.559 0.674) 0.807 (0.638 0.908) 0.618 (0.559 0.674) 2 Combined 1987 2007 wild born 0.490 (0.370 0.611) 0.646 (0.588 0.700) 0.825 (0.662 0.919) 0.646 (0.588 0.700) 3 Combined 19872007 entire 0.490 (0.370 0.611) 0.63 (0.573 0.684) 0.799 (0.631 0.902) 0.63 (0.573 0.684) 4 Combined 1987 2002 wild born 0.492 (0.372 0.613) 0.638 (0.572 0.699) 0.825 (0.613 0.933) 0.638 (0.572 0.699) 5 Radio telemetry 2003 2008 entire 0.490 (0.370 0.611) 0.777 (0.604 0.888) 0.835 (0.588 0.947) 0.628 (0.510 0.733) 6 Radio telemetry 20032008 wild -born 0.490 (0.370 0.611) 0.893 (0.756 0.958) 0.923 (0.756 0.979) 0.812 (0.688 0.894) 7 Radio telemetry 2003 2008 captive born 0.490 (0.370 0.611) 0.651 (0.441 0.815) 0.736 (0.425 0.913) 0.456 (0.307 0.613) 8 Radio telemetry 1992 2007 entire 0.490 (0.370 0.611) 0.712 (0.540 0.838) 0.802 (0.534 0.935) 0.659 (0.566 0.742) *Data set used to estimate survival rates (either radio telemetry data only or a combined data set including radio telemetry and capture -mark resight data) Population or subset of population for which survival rates were estimated (Model 4 is automatically wild -born only, because no captive -born marmots had been released yet) Age -specific survival rates: Pp is pup survival from first tagging through July/August the following year; Py is survival of yearlings; Pt, survival of two -year olds; Pa, survival of adults, age 3 and older

PAGE 59

59 Table 3 2 Estimates population growth rate ( ) and elasticities from each matrix population model *Data set used to estimate survival rates (either radio telemetry data only or a combined data set including radio telemetry and capture -mark resight data) Population or subset of population for which survival rates were estimated Elasticity Model Dataset Time period Population Pp Py Pt Pa Fy Ft Fa 1 Combined 1987 2007 entire 0.165 0.156 0.127 0.388 0.009 0.029 0.127 0.821 2 Combined 1987 2007 wild born 0.165 0.156 0.127 0.388 0.009 0.029 0.127 0.858 3 Combined 1987 2007 entire 0.166 0.157 0.127 0.384 0.009 0.029 0.127 0.839 4 Combined 1987 2002 wild born 0.166 0.157 0.128 0.383 0.009 0.029 0.128 0.851 5 Radio telemetry 20032008 entire 0.180 0.169 0.133 0.337 0.012 0.036 0.133 0.876 6 Radio telemetry 20032008 wild -born 0.152 0.144 0.121 0.433 0.008 0.023 0.121 1.038 7 Radio telemetry 20032008 captive born 0.212 0.196 0.140 0.240 0.017 0.056 0.140 0.721 8 Radio telemetry 19922007 entire 0.166 0.156 0.127 0.385 0.010 0.029 0.127 0.878

PAGE 60

60 Figure 3 1. Historically occupied (squares) and recently occupied (circles) sites for M. vancouverensis The rectangle encompasses recently active colonies within the Nanaimo Lakes region, from which the majority of survival data were obtained. Location s of the isolated Mt. Washington population and recent reintroductions at Mt. Cain and Greig Ridge are also shown, but were not included in analyses

PAGE 61

61 Figure 3 2. Population pro jection matrix used for all models. Pi is the age -specific survival rate and mi is age -specific fecundity. Matrix constructed using post -breeding census technique (Caswell 2001). 0 000 000 00yyttaa p y ta PmPmPm P P PP A

PAGE 62

62 Figure 3 3 Average age -specific litter size and age of first reproduction. A ) Average (SE) litter size for two year -olds, adults (three -year olds), and older (four years and older ) Number of litters displayed. B ) Distribution of age at first reproduction (N=24) A B

PAGE 63

63 Figure 3 4. Elasticities from four models Columns from left to right represent models 1, 5, 6, and 7 (Table 3 2). Model 5 represents the entire population from 20032008; models 6 and 7 represent noted hypothetical sub population over the same time period (20032008).

PAGE 64

64 CHAPTER 4 POPULATION ECOLOGY OF THE VANCOUVER ISLAND MARMOT: CONCLUSIONS Survival of captive born marmots released into the wild must be improved to produce a persistent population in the wild. If marmots continue to be released as yearlings, with an extr emely low survival rate (lower than those released at age two or older, Aaltonen et al. in review), the probability that they will contribute to population growth through reproduction is low. Releasing marmots at an older age (i.e. shorter time required t o survive until age with maximum reproductive contribution ) would likely increase the success of the reintroduction program. Currently, reproductive rates of the Vancouver Island marmot seem adequate, although the only true evidence is by comparison with historic reproductive rates of the Olympic marmot and comparison with other members of the M. caligata group (Bryant 2005; Griffin et al. 2008). If reproduction is insufficient at all, it is most likely due to fewer older aged females (with higher reprod uc tive value) being alive to breed, as suggested to be the case for Olympic marmots and a population of Idaho ground squirrels (Griffin et al. 2008; Sherman & Runge 2002) Matrix population models project a continuing decline of greater than 10% annually unless vital rates improve. We have found that low survival is an extremely important driver of population decline, and this is likely to be the case if survival of captive -born marmots in the wild is not increased. These finding suggest that management strategies aimed at increasing survival rates would be required to ensure the longterm persistence of the Vancouver Island marmot. Further population modeling and experimental design of reintroductions are necessary to improve knowledge about survival i n the wild, to improve the overall recovery strategy, and to increase the probability of achieving a persistent wild population of Vancouver Island marmots.

PAGE 65

65 LITERATURE CITED Aaltonen, K. L., A. A. Bryant, J. A. Hostetler, and M. K. Oli. in review. Reintroductions, age at release, and survival of the endangered Vancouver Island marmot. Biological Conservation. Armitage, K. B. 1999. Evolution of sociality in marmots. Journal of Mammalogy 80:1 10. Armitage, K. B. 2003. Marmots. Pages 188209 in G. Feldhamer, B. Thompson, and J. Chapman, editors. Wild mammals of North America: Biology, management, and conservation. Johns Hopkins University Press, Baltimore. Armstrong, D. P., and P. J. Seddon. 2008. Directions in reintroduction biology. Trends In Ecology & Evolution 23:20 25. BarDavid, S., D. Saltz, T. Dayan, A. Perelberg, and A. Dolev. 2005. Demographic models and reality in reintroductions: Persian fallow deer in Israel. Conservation Biology 19:131138. Barash, D. P. 1989. Marmots: so cial behavior and ecology. Stanford University Press. Bryant, A. A. 1996. Reproduction and persistence of Vancouver Island marmots ( Marmota vancouverensis ) in natural and logged habitats. Canadian Journal of Zoology 74:678687. Bryant, A. A. 1998. Metapopulation ecology of Vancouver Island marmots ( Marmota vancouverensis ). University of Victoria, Victoria, B.C. Bryant, A. A. 2005. Reproductive rates of wild and captive Vancouver Island marmots (Marmota vancouverensis ). Canadian Journal of Zoology 83:664. Bryant, A. A. 2007. Recovery efforts for Vancouver Island marmots, Canada. Pages 3032 in P. S. Soorae, editor. Re introduction News. No. 26. IUCN/SSC Re Introduction Specialist Group, Abu Dhabi. Bryant, A. A., and D. W. Janz. 1996. Distribution and abundanc e of Vancouver Island marmots (Marmota vancouverensis ). Canadian Journal of Zoology 74:667677. Bryant, A. A., and M. McAdie. 2003. Hibernation ecology of wild and captive Vancouver Island marmots ( Marmota vancouverensis ). Pages 159 166 in R. Ramousse, D. Allaine, and M. Le Berre, editors. Adaptive strategies and diversity in marmots. International Marmot Network, Lyon, France. Bryant, A. A., and R. E. Page. 2005. Timing and causes of mortality in the endangered Vancouver Island marmot ( Marmota vancouverens is ). Canadian Journal of Zoology 83:674. Burnham, K. P., and D. R. Anderson 2002. Model Selection and Multimodal Inference: A Practical Information Theoretic Approach. Springer -Verlag, New York.

PAGE 66

66 Caswell, H. 2001. Matrix population models: construction, ana lysis, and interpretation. Sinauer Associates, Sunderland, Mass. Caughley, G. 1994. Directions in conservation biology. Journal of Animal Ecology 63:215244. Caughley, G., and A. Gunn 1996. Conservation biology in theory and practice. Blackwell Science, In c. COSEWIC. 2008. COSEWIC assessment and update status report on the Vancouver Island marmot Marmota vancouverensis in Canada. Page vii + 29. Committee on the Status of Endangered Wildlife in Cananda, Ottawa. Farand, E., D. Allaine, and J. Coulon. 2002. Va riation in survival rates for the alpine marmot (Marmota marmota ): effects of sex, age, year, and climatic factors. Canadian Journal of Zoology 80:342349. Grenier, M. B., D. B. McDonald, and S. W. Buskirk. 2007. Rapid population growth of a critically end angered carnivore. Science 317:779779. Griffin, S. C., M. L. Taper, R. Hoffman, and L. S. Mills. 2008. The case of the missing marmots: Are metapopulation dynamics or range -wide declines responsible? Biological Conservation 141:12931309. Griffin, S. C., M. L. Taper, and L. S. Mills. 2007a. Female Olympic marmots ( Marmota olympus ) reproduce in consecutive years. The American Midland Naturalist 158:221 225. Griffin, S. C., T. Valois, M. L. Taper, and L. S. Mills. 2007b. Effects of tourists on behavior and d emography of Olympic marmots. Conservation Biology 21:10701081. Heard, D. 1977. The behavior of Vancouver Island marmots ( Marmota vancouverensis ). University of British Columbia, Vancouver. Heisey, D. M., and B. R. Patterson. 2006. A review of methods to estimate cause-specific mortality in presence of competing risks. Journal of Wildlife Management 70:15441555. Heppell, S. S., H. Caswell, and L. B. Crowder. 2000. Life histories and elasticity patterns: Perturbation analysis for species with minimal demographic data. Ecology 81 :654665. Janz, D. W., A. A. Bryant, N. Dawe, H. Schwantje, B. Harper, D. Nagorsen, D. Doyle, M. deLaronde, D. Fraser, D. Lindsay, S. Leigh Spencer, R. McLaughlin, and R. Simmons. 2000. National recovery plan for the Vancouver Island marmot (2000 update). Rep. No. 19, Recovery of Nationally Endangered Wildlife, Ottawa, Ont. Jule, K. R., L. A. Leaver, and S. E. G. Lea. 2008. The effects of captive experience on reintroduction survival in carnivores: A review and analysis. Biological Co nservation 141:355 363.

PAGE 67

67 Kruckenhauser, L., A. A. Bryant, S. C. Griffin, S. J. Amish, and W. Pinsker. 2009. Patterns of within and between -colony microsatellite variation in the endangered Vandcouver Island marmot ( Marmota vancouverensis ): implications for conservation. Conservation Genetics. Le Gouar, P., A. Robert, J. P. Choisy, S. Henriquet, P. Lecuyer, C. Tessier, and F. Sarrazin. 2008. Roles of survival and dispersal in reintroduction success of griffon vulture ( Gyps fulvus ). Ecological Applications 18:859872. Martell, A., and R. Milko. 1986. Seasonal diets of Vancouver Island marmots. Canadian Field Naturalist 100:241245. Meretsky, V. J., N. F. R. Snyder, S. R. Beissinger, D. A. Clendenen, and J. W. Wiley. 2000. Demography of the California Condor: I mplications for reestablishment. Conservation Biology 14:957967. Milko, R., and A. Bell. 1986. Subalpine meadow vegetation of south central Vancouver Island. Canadian Journal of Botany 64:815821. Mills, L. S. 2007. Conservation of wildlife populations. B lackwell Publishing. Moorhouse, T. P., M. Gelling, and D. W. Macdonald. 2009. Effects of habitat quality upon reintroduction success in water voles: Evidence from a replicated experiment. Biological Conservation 142:53 60. Nagorsen, D. 1987. Marmota vancouverensis Mammalian Species 270:1 5. Oli, M. K., and K. B. Armitage. 2004. Yellow bellied marmot population dynamics: Demographic mechanisms of growth and decline. Ecology 85:2446. Oli, M. K., and F. S. Dobson. 2003. The relative importance of life -history variables to population growth rate in mammals: Cole's prediction revisited. American Naturalist 161:422 440. Ozgul, A., K. B. Armitage, D. T. Blumstein, and M. K. Oli. 2006. Spatiotemporal variation in survival rates: Implications for population dynamics of yellow -bellied marmots. Ecology 87:1027. Ozgul, A., M. K. Oli, K. B. Armitage, D. T. Blumstein, and D. H. Van Vuren. 2009. Influence of Local Demography on Asymptotic and Transient Dynamics of a Yellow -Bellied Marmot Metapopulation. American Naturalist 173:517530. Pollock, K. H., S. R. Winterstein, C. M. Bunck, and P. D. Curtis. 1989. Survival analysis in telemetry studies: The staggered entry design. Journal of Wildlife Management 53:7 15. Sarrazin, F., and S. Legendre. 2000. Demographic approach to r eleasing adults versus young in reintroductions. Conservation Biology 14:488500.

PAGE 68

68 Seddon, P. J., D. P. Armstrong, and R. F. Maloney. 2007. Developing the science of reintroduction biology. Conservation Biology 21:303312. Shank, C. 1999. The committee on t he status of endangered wildlife in Canada (COSEWIC): a 21year retrospective. Canadian Field Naturalist 113:318341. Sherman, P. W., and M. C. Runge. 2002. Demography of a population collapse: The Northern Idaho ground squirrel ( Spermophilus brunneus brunneus ). Ecology 83:28162831. Stahl, J. T., and M. K. Oli. 2006. Relative importance of avian life -history variables to population growth rate. Ecological Modelling 198:23 39. Stephens, P. A., F. Frey Roos, W. Arnold, and W. J. Sutherland. 2002. Model compl exity and population predictions. The alpine marmot as a case study. Journal of Animal Ecology 71:343361. White, G. C., and K. P. Burnham. 1999. Program MARK: survival estimation from populations of marked animals. Bird Study 46:120138. Williams, B. K., J. D. Nichols, and M. J. Conroy 2002. Analysis and management of animal populations: modeling, estimation, and decision making. Academic Press, San Diego.

PAGE 69

69 BIOGRAPHICAL SKETCH Kristen was born in West Lafayette, Indian a and grew up a Purdue fan there in the academic shadow of her older brother. High sch ool tardies, frequent tears in the hallways, sleeping in class, and running with the infamous KLAS crew didnt keep her out of college. Much to her mothers chagrin, she left home to attend a small private unive rsity in Indianapolis for her undergraduate education. She earned a bachelo rs in science from Butler University, majoring in biology with minors in mathematics a nd Spanish. She also walked on to the Butler Bulldogs soccer team (sans white cleats) and loved pl aying on the team for all four years. It was rumored that she had the best foot skills in th e nation and while she made life-long friends from the team, there is speculation th at they still ridicule her on occasion out of jealousy of her athleticism. After Kristen graduated from Butler, much to her Finnish fathers chagrin, she went to Uppsala, Sweden to do forest genetics resear ch on the lignin biosynthetic pathway. She soon decided that white biology was not for her, and she needed to do something greener in nature. She had a wonderful time in Sweden, in fact th e only failure of her stay in Sweden was not hooking a Swedish husband to dilute her Finnish blood. Kristen started graduate school at the University of Florida in 2006 and quickly grew to love both ecology and conservation. Gainesville has provided her with both a beloved lifestyle of simplicity and an exceptionally diverse group of friends which she has learned much from. While she has yet to decide her future path, sh e is certain it will include conservation of our natural resources and sustainabi lity. She hopes to spread awar eness of the consequences our decisions have beyond our immediate perspectives and a respect for all living creatures, for all things in this world are connecte d. In the spirit of in terconnectedness, she is very grateful for all

PAGE 70

70 the much needed emotional support from friends and family over the years, she surely would have missed an insurmountable proportion of deadlin es without their love and encouragement.