WOOD TURTLE (GLYPTEMYS INSCULPTA) OCCUPANCY AND SPATIAL ECOLOGY IN WORKING FOREST
LANDSCAPES
By
Tricia G. Brockman
A THESIS
Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of
Fisheries and Wildlife – Master of Science
2024
ABSTRACT
The wood turtle (Glyptemys insculpta) is a species of conservation concern and under
consideration for listing under the Endangered Species Act of 1973, as amended (ESA; 88 Fed.Reg.88338),
yet little information is available on wood turtle ecology in private working forests. However, before
effective conservation measures can be applied, it is essential to determine the spatial distribution,
seasonal movement, and home range of wood turtles on these working landscapes. In this thesis, I set out
to address these topics and provide forest managers with information needed to further wood turtle
conservation efforts in working forests.
In Chapter 1, I investigate detection and occupancy of wood turtles on a working forest landscape
in the western Upper Peninsula of Michigan. I applied a single-season occupancy model to parameterize
detection and estimate occupancy probabilities of wood turtles based on riparian conditions. I also
related occupancy status of watershed basins to recent forest management history (~17 years) using non-
parametric testing methods. I concluded that the amount of nesting substrate along a riverbank is a good
indicator of wood turtle occupancy and that there is little relationship between recent forest
management history and wood turtle occupancy on a primarily forested landscape.
In Chapter 2, I compared the seasonal movement distance from flowing water (m) and home
range size (ha) of wood turtles on privately managed forestland between two watershed basins in the
western Upper Peninsula of Michigan and with other studies across the Upper Great Lakes Region. I
accomplished this using radio telemetry to track 10 adult female wood turtles in two watershed basins
during the active seasons of 2021 – 2022. I concluded that, while my estimates were similar to those
across the region, resource availability influenced seasonal movement distance from flowing water and
stream range at the basin level.
I conclude that sustainable forest management and wood turtles can co-exist on the landscape if
forest managers apply management practices that avoid active management in areas and during times
when wood turtles are terrestrial and manage forests in a way that creates a mosaic of mature forest,
young forest, and forest openings that provide wood turtles with essential life requisites. However, more
research is needed to fully understand the ecology of wood turtles on private, working forests, and how
forest managers can contribute to conserving this species.
For my amazing nieces pieces, Gloria (GG) and Elena.
iii
ACKNOWLEDGEMENTS
First and foremost, I would like to thank my major advisor, Dr. Gary J. Roloff, for giving me his
undying trust and support, and for encouraging me to tell my own story. I am also grateful for the
guidance and support of my committee members, Dr. Scott R. Winterstein (MSU), Dr. Jeanette M.
McGuire (MSU), and Dr. Darren A. Miller (NCASI). A special thanks goes to all of the faculty and staff in the
Department of Fisheries and Wildlife at Michigan State University (MSU) who saw me through my years as
an undergrad and again as a master’s student. Natural Resources will always feel like home to me.
This research was funded by the U.S. Fish and Wildlife Service (USFWS) through the National Air
and Stream Improvement (NCASI) Foundation in partnership with the National Alliance of Forest Owners
(NAFO) Wildlife Conservation Initiative. Thank you for entrusting me with this project and for granting me
an additional year to fulfil my vision. Thank you also to my PIs Bradley A. Potter (USFWS) and Dr. Darren A.
Miller (NCASI) for your support and unique perspectives on forestry, policy, and wildlife conservation. This
project would not have been a success without NAFO member companies Lyme Great Lakes Timberlands
and Molpus Woodlands Group who not only provided me land access but taught me the value working
forests have for wildlife conservation. Thank you to American Forest Management, Manulife Investment
Management, and PotlatchDeltic for your participation in my wood turtle workshops. I know that my
wood turtles are in good hands. Thank you to the Michigan Department of Natural Resources for
providing land access, permits, and for giving me valuable experience at various stages throughout my
career.
A very special and heartfelt thank you to my mentor and friend Jim H. Harding, who not only
paved the way for wood turtle research in the Upper Peninsula of Michigan, but fostered in me a passion
for education and is one of my most enthusiastic supporters. Thank you also to the wonderful community
of wood turtle researchers, especially Yu Man Lee (Michigan Natural Features Inventory), Dr. Donald J.
Brown (U.S. Forest Service), Dr. Jennifer A. Moore (Grand Valley State University), Dr. Matthew W. H.
Chatfield (University of Maine), Dr. Jeffrey W. Tamplin (University of Northern Iowa), and Damien I. Mullin
(University of New Brunswick), from across the species range who answered all of my questions, gave me
excellent advice, and welcomed me as a colleague whose research was interesting and important.
I am so very grateful for my dedicated, enthusiastic, and talented field technicians Amelia Cole,
Mackenzie Milner, Quinn Bozek, and Lauren Otterness. I will never forget the adventures we had in the
“Yoop”, from discovering unexplored natural wonders, to encounters with bears and moose, to getting out
of sticky situations involving vehicles. I am so proud of all that you have and will accomplish. Thank you
also to my lab mates in the Applied Forest and Wildlife Ecology Lab at MSU who provided encouragement
iv
and advice, especially to Jillian Rajewski who was my “covid cohort” and was always there when I needed
a lift. Thank you to our lab postdocs Dr. Steven M. Gray and Dr. Evan C. Wilson for always providing me
with thoughtful reviews and r codes to fit my every need. Thank you to the members of the Fisheries and
Wildlife Grad Student Organization, some of whom have become lifelong friends and adventure buddies.
Thank you from the bottom of my heart to the Yooper community, especially to Sylvia and the Friday
night Skipper’s crew in L’anse, who made it so I never felt alone even so far away from home.
None of this would have been possible without the support of my family. My parents, Eric and
Julie Brockman, who always encouraged me to follow my passion and raised me with the tools to do so,
my sister and brother-in-law, Alexa and Brad Allen, who supplied many welcome distractions which
includes my amazing nieces, Gloria and Elena, and my grandparents, Doug and Trudy Green, who always
showed a keen interest in my work. Last but not least, thank you to my partner and best friend, Randy
Zuccaro, for everything.
v
TABLE OF CONTENTS
INTRODUCTION ............................................................................................................................................. 1
LITERATURE CITED ..................................................................................................................................... 5
CHAPTER 1. WOOD TURTLE (GLYPTEMYS INSCULPTA) DETECTION AND OCCUPANCY IN WORKING FOREST
LANDSCAPES IN THE WESTERN UPPER PENINSULA OF MICHIGAN, USA ...................................................... 9
1.1 Introduction ......................................................................................................................................... 9
1.2 Methods ............................................................................................................................................ 10
1.3 Results ............................................................................................................................................... 14
1.4 Discussion .......................................................................................................................................... 15
TABLES AND FIGURES .............................................................................................................................. 20
LITERATURE CITED ................................................................................................................................... 33
APPENDIX 1A: WOOD TURTLE SHELL DIAGRAM ...................................................................................... 37
APPENDIX 1B: WOOD TURTLE CAPTURE DATA ....................................................................................... 38
APPENDIX 1C: DETECTION MODEL RESULTS ........................................................................................... 40
CHAPTER 2. SEASONAL MOVEMENTS AND HOME RANGES OF ADULT FEMALE WOOD TURTLES
(GLYPTEMYS INSCULPTA) IN WORKING FOREST LANDSCAPES IN THE WESTERN UPPER PENINSULA OF
MICHIGAN, USA ........................................................................................................................................... 41
2.1 Introduction ....................................................................................................................................... 41
2.2 Methods ............................................................................................................................................ 43
2.3 Results ............................................................................................................................................... 45
2.4 Discussion .......................................................................................................................................... 47
TABLES AND FIGURES .............................................................................................................................. 52
LITERATURE CITED ................................................................................................................................... 65
APPENDIX 2A: WOOD TURTLE RELOCATION DATA SHEET ...................................................................... 69
APPENDIX 2B: TRANSMITTER ATTACHMENT........................................................................................... 70
APPENDIX 2C: TELEMETERED WOOD TURTLE BIOMETRICS .................................................................... 71
CONCLUSION ............................................................................................................................................... 72
LITERATURE CITED ................................................................................................................................... 74
vi
INTRODUCTION
Of the 8.1 million hectares of forested land in Michigan, over 5 million hectares (62%) are
privately owned and managed (Poudel 2022). Thus, many opportunities exist for wildlife conservation on
private lands. For some wildlife species, basic information on occupancy, movement, and resource use on
privately managed lands is insufficient. By addressing this knowledge gap, private landowners will be
better equipped to employ science-based approaches to managing at-risk species on Michigan’s working
forests.
One of the most influential federal policies affecting working forests and rare species
conservation in the U.S. is the Endangered Species Act (ESA) of 1973, as amended (ESA; 88
Fed.Reg.88338). Prior to a species being listed as threatened or endangered under the ESA, a Species
Status Assessment (SSA) must be conducted to aid the U.S. Fish and Wildlife Service (USFWS) in making
listing determinations (USFWS 2016). However, basic information on candidate species is sometimes
lacking, often leading to precautionary decisions by the USFWS. In many instances, basic knowledge on
candidate species can greatly influence the listing process and decision making, ultimately determining
permissible land management activities in areas where rare species occur.
In recognition of the value in proactively collecting these data, particularly for species presumably
protected under existing best management or sustainable forestry practices, the National Alliance of
Forest Owners (NAFO) Wildlife Conservation Initiative (WCI) was created (NAFO 2020). This initiative is a
collaboration among NAFO and its member companies, the USFWS, the National Council for Air and
Stream Improvement, Inc. (NCASI), and the Association of Fish and Wildlife Agencies (AFWA). This
initiative is supported by the work of NAFO member companies, NCASI staff, USFW staff, universities, and
other partners, to support Science-based approaches to conserving rare species in working forests while
maintaining the economic viability of owning and managing forest lands (NAFO 2020).
In 2020, the WCI identified the North American wood turtle (Glyptemys insculpta) as a focal
species of study. Wood turtles are a semi-terrestrial freshwater species that occupies a historical range in
the United States that includes eastern Minnesota, northern Iowa, Wisconsin, Michigan’s Upper and
northern Lower Peninsulas, New York, Pennsylvania, northern New Jersey, the northern Virginia and
West Virginia, and throughout New England (Harding and Bloomer 1979, Ernst and Lovich 2009). Wood
turtles also occur in southeast Ontario, southern Quebec, New Brunswick, and Nova Scotia in Canada
(Harding and Bloomer 1979, Ernst and Lovich 2009). Wood turtles are listed as endangered by the
International Union for Conservation (IUCN), and while wood turtles are not currently listed as
threatened or endangered under the Federal Endangered Species Act of 1973, as amended, a petition to
1
list them was submitted to the U.S. Fish and Wildlife Service in 2012 and they are currently under review
as a candidate species (van Dijk and Harding 2011, Giese et al. 2012).
Wood turtles are characterized by a dark brown carapace, yellowish plastron with black markings
along the outer edge of each scute, dark brown or black dorsal skin, and yellow to bright orange ventral
skin. Each scute of the shell features raised concentric growth rings (annuli) that may be counted to
estimate a turtle’s minimum age (Harding and Bloomer 1979). Wood turtles are sexually dimorphic, with
males being slightly larger, having a large block-like head, a thick tail, and concave plastron (Lovich et al.
1990, Ernst and Lovich 2009).
Wood turtles use a variety of riparian and terrestrial cover types. Rivers occupied by wood turtles
are typically 3 – 20 m wide, slow moving, with a substrate of sand, gravel, or rock (Ernst and Lovich 2009,
Jones and Willey 2021). Wood turtles will use instream features such as root balls, undercut banks, log
jams and American beaver (Castor canadensis) lodges as brumation sites during the inactive season from
about November - April (Harding and Bloomer 1979, Arvisais et al. 2004, Greaves and Litzgus 2007, 2008).
During the terrestrial portion of their annual life cycle, wood turtles are considered an “edge” species and
thrive in a forested mosaic of varying vegetation structures and compositions. Studies on terrestrial
habitat components have shown that wood turtles use areas that best fulfil their foraging and
thermoregulatory needs during the active season from about April – November (Kaufman 1992, Compton
et al. 2002, Arvisais et al. 2004, Duboise et al. 2009, Brown et al. 2016, Thompson et al. 2018, Latham et
al. 2022). These areas are typically young mixed forests or forest edges with a medium layer of shrub
cover and low canopy cover (Compton et al. 2002, Arvisais et al. 2004, Sweeten 2007, Brown et al. 2016,
Thompson et al. 2018, Latham et al. 2022). Wood turtles are opportunistic omnivores, and their diet
reflects locally and seasonally available plants and animals. They have been observed feeding on
mushrooms, berries, young ferns, aquatic and terrestrial invertebrates, nestling birds, young mice, fish,
and carrion (Harding and Bloomer 1979, Compton et al. 2002, Walde et al. 2003, Arvisais et al. 2004,
Cochran et al. 2014).
Wood turtles go through an annual cycle of seasonal activity periods; brumation (November –
April), emergence/pre-nesting (April – May), nesting (June), post-nesting (July – September), and pre-
brumation (October – November) (Kaufman 1992, Arvisias et al. 2002, Jones and Willey 2021). Variation
in the timing of these activity periods occurs based on location and weather conditions (Ernst and Lovich
2009). During emergence/pre-nesting, wood turtles use rivers as corridors and thermal refuge while
traveling from brumation sites to nesting areas (Arvisais et al. 2004). Nesting occurs in areas of open
canopy and exposed sand/gravel soil that either occur naturally along riverbanks or are anthropogenically
2
created (i.e., clearcuts, gravel pits, and roadsides) (Harding and Bloomer 1979, Buech et al. 1997, Arvisais
et al. 2004, Hughes et al. 2009). Post-nesting is characterized by movement away from the river and
establishment of terrestrial activity areas by adult females who will travel farther from the river on
average than males (Arvisais et al. 2004, Parren 2013, Brown et al. 2016, Otten 2017, Thompson et al.
2018, Latham et al. 2022). Males tend to have larger in-stream ranges than females, presumably to
increase their chances of mating (Jones and Willey 2020, Otten et al. 2021). Wood turtles in the northern
portion of their range and/or in more contiguous mature forests are thought to use larger terrestrial
areas due to thermal and foraging constraints forcing them travel more broadly to meet their resource
needs (Arvisais et al. 2002, Saumure 2004). Pre-brumation marks the return of wood turtles to rivers
(Arvisais et al. 2004). Wood turtles display high site fidelity and will return to the same nesting areas,
terrestrial activity areas, and brumation sites annually (Harding and Bloomer 1979, Quinn and Tate 1991,
Arvisias et al. 2002, Walde et al. 2007, Parren 2013, Thompson et al. 2018, Otten et al 2021). Courtship
and mating occur throughout the active season from April through November, mainly in aquatic systems
(Harding and Bloomer 1979, Ernst and Lovich 2009).
Wood turtles are long-lived, reaching sexual maturity between 11 – 20 years of age, and have a
documented lifespan of 55 years in the wild, but may live longer (Harding and Bloomer 1979, Lovich et al.
1990, Ross et al. 1991, Brown et al. 2015). Similar to other long-lived and slow maturing species, wood
turtle populations depend on high adult survivorship (>97%) to maintain population stability and
therefore may be susceptible to population collapse if even a small proportion of breeding adults are lost
(Congdon et al. 1994, Compton 1999, Lapin et al. 2019). Throughout their range wood turtle populations
are in decline. Increased predation by human subsidized mesopredators, such as raccoons (Procyon
lotor), on nests and adults, vehicle strikes primarily affecting nesting females, and poaching for the
domestic and overseas pet trade are some of the most significant causes of decline (Harding and Bloomer
1979, Jones et al. 2015, Rutherford et al. 2016, Lapin et al. 2019, Latham et al. 2022, Jones and Willey
2021). Habitat alterations, such as invasive plant incursion on nesting areas, urbanization, agriculture, and
changes in hydrology from stream bank stabilization and damming are also detrimental to wood turtle
populations (Harding and Bloomer 1979, Saumure and Bider 1998, Tingley and Herman 2008, Jones et al.
2015, Jones and Willey 2021). However, effects of forest management on wood turtle ecology, are
relatively unknown (Jones and Willey 2021). Wood turtle mortality from encounters with forest
harvesting machinery is identified as a direct threat, though published data are lacking (Saumure 2004,
Tingley and Herman 2008, Jones and Willey 2021). Some hypothesize that forest management, if
conducted in a way that reduces mortality risk and with of consideration wood turtle habitat
3
requirements, can be beneficial in that forest management can create structural diversity by mimicking
natural disturbances (Kaufman 1992, Tingley and Herman 2008, Hughes and Litzgus 2019, Latham et al.
2022). Much of wood turtle range falls within actively managed working forests, and so it is essential to
understand how forest management affects wood turtles and what can be done to mitigate potential
negative effects.
In 2020, Michigan State University was awarded a research grant from the USFWS through NCASI
Foundation to study wood turtle occupancy and spatial ecology in working forests of the western Upper
Peninsula of Michigan. Partnering with two NAFO member companies, Lyme Great Lakes Timberlands
and Molpus Woodlands Group, I sought to fulfil the following objectives:
1) survey watershed basins with historical wood turtle occupancy records to confirm current
occupancy status;
2) survey proximate basins with no historical occupancy records to confirm current occupancy
status;
3) explore spatial ecology of wood turtles in survey basins; and
4) relate recent forest management and drainage conditions to wood turtle occupancy status and
spatial ecology.
In this thesis, I addressed these objectives in two chapters. In Chapter 1, I surveyed historically
occupied and adjacent watershed basins to parameterize detection and occupancy probability with a
focus on working forests. In Chapter 2, I used very high frequency (VHF) radio telemetry to monitor the
seasonal movements and estimate home range sizes of adult female wood turtles on lands managed by
two NAFO member companies.
4
LITERATURE CITED
Arvisais, M., J. C. Bourgeois, E. Lévesque, C. Daigle, D. Masse, and J. Jutras. 2002. Home range and
movements of a wood turtle (Clemmys insculpta) population at the northern limit of its range.
Canadian Journal of Zoology 80:402–408.
Arvisais, M., E. Lévesque, J. C. Bourgeois, C. Daigle, D. Masse, and J. Jutras. 2004. Habitat selection by the
wood turtle (Clemmys insculpta) at the northern limit of its range. Canadian Journal of Zoology
82:391–398.
Brown, D. J., M. Schrage, D. Ryan, R. A. Moen, M. D. Nelson, and R. R. Buech. 2015. Glyptemys insculpta
(Wood Turtle). Longevity in the Wild. Herpetological Review 46:243–244.
Brown, D. J., M. D. Nelson, D. J. Rugg, R. R. Buech, and D. M. Donner. 2016. Spatial and Temporal Habitat-
Use Patterns of Wood Turtles at the Western Edge of Their Distribution. Journal of Herpetology
50:347–356.
Buech, R. R., L. G. Hanson, and M. D. Nelson. 1997. Identification of wood turtle nesting areas for
protection and management. Proceedings of the 1997 Conservation, Restoration, and
Management of Tortoises and Turtles: An International Conference. New York Turtle and Tortoise
Society, New York, New York. 494 pp.
Cochran, P. A., B. D. Jones, R. Otto, and A. P. Kinziger. 2014. Observations on the Diet of the Wood Turtle
(Glyptemys insculpta) in Northeastern Wisconsin and on Its Potential Role as a Seed Disperser.
28–31.
Compton, B. W. 1999. Ecology and conservation of the wood turtle (Clemmys insculpta) in Maine. Thesis,
University of Maine, Orono, Maine, USA.
Compton, B. W., J. M. Rhymer, and M. McCollough. 2002. Habitat selection by wood turtles (Clemmys
insculpta): An application of paired logistic regression. Ecology 83:833–843.
Congdon, J. D., A. E. Dunham, and R. C. V. L. Sels. 1994. Demographics of common snapping turtles
(Chelydra serpentina): Implications for conservation and management of long-lived organisms.
Integrative and Comparative Biology 34:397–408.
Dubois, Y., G. Blouin-Demers, B. Shipley, and D. Thomas. 2009. Thermoregulation and habitat selection in
wood turtles Glyptemys insculpta: Chasing the sun slowly. Journal of Animal Ecology 78:1023–
1032.
Ernst, C. H., and J. E. Lovich. 2009. Turtles of the United States and Canada. Second Edition. Johns Hopkins
University Press, Baltimore, Maryland, US. 434 pp.
Giese, C. L. A., D. N. Greenwald, and T. Curry. 2012. Petition to list 53 amphibians and reptiles in the
United States as threatened or endangered species under the Endangered Species Act. Petition to
the Secretary of the Interior of the United States. Center for Biological Diversity, Tucson, Arizona,
USA. 454 pp.
5
Greaves, W. F., and J. D. Litzgus. 2007. Overwintering ecology of Wood Turtles (Glyptemys insculpta) at
the species’ northern range limit. Journal of Herpetology 41:32–40.
Greaves, W. F., and J. D. Litzgus. 2008. Chemical, thermal, and physical properties of sites selected for
overwintering by northern wood turtles (Glyptemys insculpta). Canadian Journal of Zoology
86:659–667.
Harding, J. H., and T. J. Bloomer. 1979. The wood turtle, Clemmys insculpta... a natural history. Bulletin of
the New York Herpetological Society 15:9–26.
Hughes, G. N., W. F. Greaves, and J. D. Litzgus. 2009. Nest-site selection by wood turtles (Glyptemys
insculpta) in a thermally limited environment. Northeastern Naturalist 16:321–338.
Hughes, G. N., and J. D. Litzgus. 2019. Impact of natural resource extraction on thermal properties of
wood turtle (Glyptemys insculpta) habitat. Journal of Thermal Biology 84:469–478.
Jones, M. T., and L. L. Willey. 2020. Cross-watershed dispersal and annual movement in adult wood
turtles (Glyptemys insculpta). Herpetological Review 51:208–211.
Jones, M. T., and L. L. Willey, editors. 2021. Biology and conservation of the wood turtle. Northeast
Association of Fish and Wildlife Agencies, Inc., Petersburgh, New York, USA. 235 pp.
Jones, M. T., L. L. Willey, T. S. B. Akre, and P. R. Sievert. 2015. Status and conservation of the wood turtle
in the northeastern United States. Northeastern Association of Fish and Wildlife Agencies. 288
pp.
Kaufmann, J. H. 1992. Habitat use by wood turtles in central Pennsylvania. Journal of Herpetology
26:315–321.
Lapin, C. N., J. W. Tamplin, M. M. Cochrane, J. E. Woodford, D. J. Brown, and R. A. Moen. 2019. A Regional
Analysis of Glyptemys insculpta (Wood Turtle) survival in The Upper Midwest of the USA.
Herpetological Conservation and Biology 14:668–679.
Latham, S. R., A. P. K. Sirén, and L. R. Reitsma. 2022. Space use and resource selection of Wood Turtles
(Glyptemys insculpta) in the northeastern part of its range. Canadian Journal of Zoology 101:20–
31.
Lovich, J. E., C. H. Ernst, and J. F. McBreen. 1990. Growth, maturity, and sexual dimorphism in the wood
turtle, Clemmys insculpta. Canadian Journal of Zoology 68:672–677.
National Alliance of Forest Owners [NAFO]. 2020. Accessed 24 April 2020.
Otten, J. G. 2017. Factors influencing wood turtle (Glyptemys insculpta) home range size in Iowa: A
comparison between suburban and rural populations. Thesis, University of Northern Iowa, Cedar
Falls, Iowa, USA. 163 pp.
6
Otten, J. G., A. C. Hulbert, S. W. Berg, and J. W. Tamplin. 2021. Home Range, Site Fidelity, and Movement
Patterns of the Wood Turtle (Glyptemys insculpta) at the Southwestern Edge of Its Range.
Chelonian Conservation and Biology 20:231–241.
Parren, S. G. 2013. A twenty-five year study of the wood turtle (Glyptemys insculpta) in Vermont:
Movements, behavior, injuries, and death. Herpetological Conservation and Biology 8:176–190.
Poudel, J. 2022. Forest products industries’ economic contributions: Michigan, 2019. Forest Resources
Division, Michigan Department of Natural Resources, Lansing, Michigan, USA. 28 pp.
Quinn, N. W. S., and D. P. Tate. 1991. Seasonal Movements and Habitat of Wood Turtles (Clemmys
insculpta) in Algonquin Park. Journal of Herpetology 25:217–220.
Ross, D. A., K. N. Brewster, R. K. Anderson, N. Ratner, and C. M. Brewster. 1991. Aspects of the Ecology of
Wood Turtles, Clemmys insculpta, in WI. The Canadian Field Naturalist 105:363–367.
Rutherford, J. L., G. S. Casper, and B. Graves. 2016. Factors Affecting Predation on Wood Turtle
(Glyptemys insculpta) Nests in the Upper Peninsula of Michigan. Chelonian Conservation and
Biology 15:181–186.
Saumure, R. A. 2004. Spatial ecology and conservation of the North American wood turtle (Glyptemys
insculpta) in a fragmented agri-forest landscape. Dissertation, McGill University, Montreal,
Quebec, CA. 148 pp.
Saumure, R. A. and J. R. Bider. 1998. Impact of agricultural development on a population of wood turtles
(Clemmys insculpta) in southern Quebec, Canada. Chelonian Conservation and Biology 3:37-45.
Sweeten, S. E. 2007. Home Range, Hibernacula Fidelity, and Best Management Practices for Wood Turtles
(Glyptemys insculpta) in Virgina. Thesis, James Madison University, Harrisonburg, Virginia, USA.
47 pp.
Thompson, D. G., T. Swystun, J. Cross, R. Cross, D. Chartrand, and C. B. Edge. 2018. Fine-and coarse-scale
movements and habitat use by Wood Turtles (Glyptemys insculpta) based on probabilistic
modeling of radiotelemetry and GPS-telemetry data. Canadian Journal of Zoology 96:1153–1164.
Tingley, R., and T. B. Herman. 2008. The effects of agriculture and forestry on the distribution,
movements and survival of wood turtles in an intensively managed landscape. A report for the
Nova Scotia Habitat Conservation Fund. 57 pp.
United States Fish and Wildlife Service [USFWS]. 2016. USFWS Species Status Assessment Framework: an
integrated analytical framework for conservation. Version 3.4. 21 pp.
van Dijk, P.P. and J. Harding. 2011. Glyptemys insculpta. The IUCN Red List of threatened species 2011.
. Accessed 3 May 2023.
7
Walde, A. D., J. R. Bider, C. Daigle, D. Masse, J. C. Bourgeois, J. Jutras, and R. D. Titman. 2003. Ecological
aspects of a Wood Turtle, Glyptemys insculpta, population at the northern limit of its range in
Québec. Canadian Field-Naturalist 117:377–388.
Walde, A. D., J. R. Bider, D. Masse, R. A. Saumure, and R. D. Titman. 2007. Nesting ecology and hatching
success of the wood turtle, Glyptemys insculpta, in Quebec. Herpetological Conservation and
Biology 2:49-60.
8
CHAPTER 1. WOOD TURTLE (GLYPTEMYS INSCULPTA) DETECTION AND OCCUPANCY IN WORKING FOREST
LANDSCAPES IN THE WESTERN UPPER PENINSULA OF MICHIGAN, USA
1.1 Introduction
Wood turtles (Glyptemys insculpta) are a semi-terrestrial riparian species that spend much of
their active season, late April – mid-October in the Upper Peninsula of Michigan, in a mosaic of forest
types throughout their native range, which covers portions of the Midwest and Northeastern United
States, and southeastern Canada (Harding and Bloomer 1979, Compton et al. 2002, Ernst and Lovich
2009). Wood turtles are listed as endangered by the International Union for Conservation (IUCN) and
receive some level of protection in every state where they occur (van Dijk and Harding 2011, Jones and
Willey 2021). Additionally, wood turtles have been petitioned for listing under the Federal Endangered
Species Act of 1973, as amended (ESA; 88 Fed.Reg.88338), in the United States (Giese et al. 2012). Threats
to wood turtle populations include increased nest and adult predation by mesopredators, illegal harvest
for the pet trade, and habitat fragmentation and loss due to urbanization and agriculture (Jones and
Willey 2021).
Forest management has also been suggested as a threat to wood turtles, but little information
exists on effects of forest management on wood turtle populations. While tree harvesting may pose risks
to wood turtle populations through habitat alterations and direct mortality from harvesting equipment,
over the long-term, forest management may be beneficial as it can mimic natural disturbances that
provide a mosaic of vegetation structures and compositions, which provide wood turtles with life
requisites, such as open basking areas and berry-producing vegetation (Congdon et al. 1994, Saumure
2004, Kaufmann 1992, Tingley and Herman 2008, Hughes and Litzgus 2019, Latham et al. 2022). Across
much of wood turtle range, there is significant overlap with private, working forests. Therefore, it is
important to understand how forest management may affect wood turtle occupancy and its implications
for wood turtle conservation.
Robust occupancy estimates require information on survey effectiveness (i.e., the observation
processes; Mackenzie et al. 2017). Wood turtles are a cryptic and secretive species that disperse into
riparian and upland forests during the active season to forage and thermoregulate, making individuals
challenging to detect (Flanagan et al. 2013, Brown et al. 2017). Wood turtle detection has been well-
studied and reliable detection protocols have been implemented by researchers throughout wood turtle
range (Flanagan et al. 2013, Brown et al. 2017, Jones et al. 2015, Northeast Wood Turtle Working Group
2019). Occupancy estimates for wood turtles can apply to aquatic (i.e., river segments, drainage basins)
and terrestrial (upland forest) life stages, with terrestrial conditions varying regionally and locally (Harding
9
and Bloomer 1979, Ernst and Lovich 2009). Buech et al. (1997) identified the availability and quality of
nesting habitat (sand/gravel substrate exposed to direct sunlight) as a limiting factor for wood turtle
occurrence in the Upper Great Lakes Region and McCoard et al. (2016) identified overstory canopy cover
as influencing wood turtle occupancy in West Virgina. Tree density, or basal area, has not yet been used to
predict wood turtle occupancy, but it can serve as a proxy for canopy cover (Buckley et al. 1999). Both
canopy cover and basal area are standard forest inventory measurements that may be used by forest
managers to predict potential wood turtle occupancy.
In the western Upper Peninsula of Michigan, wood turtle populations, including basic
demographic information like occupancy, have not been well documented on private, working forests,
making science-based conservation delivery for landowners difficult. To help fill this information gap, I
identified forest features easily measured using standard forest inventory techniques to predict wood
turtle occurrence in riparian areas and explore the potential effects of forest management on wood
turtles in an occupancy framework (MacKenzie et al. 2018). The ability to estimate wood turtle occupancy
based on standard forest inventory variables would assist forest managers in implementing science-based
management practices to conserve wood turtles and their habitat on private working forests. My
objectives were to 1) estimate detection and occupancy probability of wood turtles in river segments and,
2) relate riparian conditions and recent forest management history to wood turtle occupancy at the basin
level. I hypothesized that wood turtle occupancy was influenced more by localized riparian conditions,
(e.g., the amount of available nesting substrate) than forest management in adjacent uplands. Therefore, I
predicted little to no correlation between occupancy status of watershed basins and recent forest
management history.
1.2 Methods
1.2.1 Study Area
The northern Lower Peninsula and Upper Peninsula of Michigan are dominated by forests and the
majority of wood turtle range overlaps this area (Figure 1.1). Of Michigan’s forested land, 62% is privately
owned and managed (Poudel 2022). My study took place in the western Upper Peninsula of Michigan,
USA from 2020 – 2022 (Figure 1.2). I selected 22 watershed basins at the hydrological unit code (HUC)
level 12 (4,485 - 15,408 ha) as study sites (hereafter “basins”) (Seaber et al. 1987). I selected six basins
based on historical wood turtle occupancy records reported by Michigan Natural Features Inventory
(MNFI) and I selected 16 additional basins given their proximity to historically occupied basins and
National Alliance of Forest Owners (NAFO) ownership adjacent to main river channels (15% - 89% NAFO
ownership within basins). Topography, hydrology, and vegetation composition varied latitudinally. In
10
northern basins, terrain was hilly with glacially exposed bedrock and well-drained sandy soils in the
uplands and saturated organic soils in lowlands. River gradients were higher, compared to southern
basins, and there were few rivers and many lakes (Schaetzl et al. 2013). Upland vegetation was dominated
by hardwoods such as yellow birch (Betula alleghaniensis) and sugar maple (Acer saccharum), whereas
saturated and lowland areas were dominated by black ash (Fraxinus nigra), American elm (Ulmus
americana), red maple (Acer rubrum), black spruce (Picea mariana) and tamarack (Larix laricina)
(LANDFIRE 2022). In southern basins, terrain originated from low drumlins formed by glacial runoff, with
soil composition similar to northern basins. Slow flowing rivers and wetlands dominated the hydrological
landscape. The vegetation in uplands was dominated yellow birch, aspen (Populus spp.), red pine (Pinus
resinosa), and hemlock (Tsuga canadensis) whereas lower areas were comprised primarily of riparian
forest that included balsam fir (Abies balsamea), black ash, American elm, and red maple (LANDFIRE
2022). Average temperatures ranged from -2.6 °C to 10.6 °C in the north, and -0.1 °C to 11.9 °C in the
south of the study area (GLISA 2023a, b). Average annual precipitation ranged from 76.3 cm in the north
to 80.9 cm in the south (GLISA 2023a, b).
1.2.2 Survey Design
For detecting wood turtles, I used a visual encounter survey protocol developed by Brown et al.
(2017) to detect wood turtles in the Midwest region of the United States. Each survey consisted of two
parallel transects at 0 m and 15 m from the edge of the water on either side of a 1 km segment of river
(Figure 1.3). I placed survey segments along the main river channel in each basin on NAFO ownership and,
when not available, on state and federal ownership. To assess forest and river conditions along the survey
route, I included a data collection location every 50 m along each 0 m transect line and at every 100 m
along the 15 m transect line (Figure 1.3). I drew transect lines and data collection locations using ArcMap
(version 10.8.2, Environmental Systems Research Institute, Redlands, CA, US) and uploaded these as geo-
referenced PDFs onto Samsung tablets (Galaxy Tab A). I conducted detection surveys from early May
through mid-late July. During this time wood turtles remain close to the river and are more likely to be
within the survey area and are more easily detected (Flanagan et al. 2013, Brown et al 2017, Northeast
Wood Turtle Working Group 2019). Using Avenza Maps Pro (Avenza Systems, Inc., Toronto, Ontario, CA),
surveyors followed transect lines and located data collection locations. During a segment survey,
surveyors on each side of the river walked the 0 m transect line first and came back along the 15 m
transect line, searching for wood turtles and recording forest and river conditions. Surveyors were
constrained to searching within 10 m of transect lines but could deviate briefly to search the area more
thoroughly or avoid obstructions.
11
1.2.3 Data Collection
Before beginning a survey, surveyors recorded date, ambient air temperature (°C), water
temperature (°C), and start time. At the end of each survey, surveyors again recorded ambient air
temperature (°C), near-bank water temperature (°C), and end time. On the 0 m transect along the
riverbank, every 50 m, surveyors recorded bank substrate composition and canopy cover (Figure 1.3).
Percentage of bank substrates (silt, sand, gravel, cobble, rock and, organic) were estimated visually within
a 1 m square area at each plot location and overstory canopy cover was recorded by using a fisheye lens
attached to a cell phone at 1.5 m above the ground. As surveyors walked back along the 15 m transect,
they used a factor 10 angle gauge to record tree basal area every 100 m (Figure 1.3).
Upon detecting a wood turtle, surveyors captured it by hand and recorded location and transect
where it was found. For all wood turtles, surveyors recorded number of annuli and percent wear on the
left abdominal scute to determine relative age, and sex (Harding and Bloomer 1979, Parren 2013). Wood
turtles with fewer than 15 annuli were classified as juveniles unless secondary sexual characteristics were
present, or the percent wear was 50% or more (Harding and Bloomer 1979). Surveyors recorded
individual weight (g), shell dimensions (mm), and assigned each wood turtle an individual ID number,
which was notched into the marginal scutes using a triangular file (Appendices 1A; 1B). All wood turtle
handling, processing, and equipment sanitizing protocols were approved by the Institutional Animal Care
and Use Committee at MSU (PROTO202000121) and conducted under a scientific collector permits issued
by the Michigan Department of Natural Resources (Permit numbers: NA (2020), FSCP02222021123151
(2021), FSCP01312022115615 (2022)).
I summarized recent forest management history for each surveyed watershed basin using
National Agriculture Imagery Program (NAIP) aerial imagery in Google Earth Engine (Gorelick et al. 2017).
In each basin, I sampled within 400m of the surveyed river to reflect the maximum distance traveled by
female wood turtles in a concurrent telemetry study (Brockman 2024: Chapter 2). I generated Generalized
Random Tessellation Stratified (GRTS) locations within each 400 m buffer to ensure a spatially balanced
sample (Olsen et.al. 2012). I weighted number of GRTS locations in each basin by area to produce ~1
location per 30 ha. I added an extra 20 locations per basin to offset locations that required censoring (e.g.,
falling within a stream, road, development). I separated by 100 m, which was the estimated length of the
activity area of wood turtles (Compton et al. 2002). For each location, I recorded apparent forest
management treatment (i.e., select cut, shelterwood, clearcut, none), ownership (i.e., NAFO, public,
other), and year of last forest management activity by analyzing a time series of NAIP imagery in Google
Earth Engine starting in 2005 and ending in 2022, with a 2 – 4-year interval between each photo set
12
(Figure 1.4). I used forest harvest data from NAFO member companies and the State of Michigan paired
with NAIP aerial imagery to train my search image and validate designations.
1.2.4 Statistical Analysis
I fit a single-season occupancy model to predict detection probability (p) and occupancy (Ψ) of
wood turtles at the 1 km river segment-level, with paired surveys on either side of the river serving as
spatial replicates. Model assumptions were 1) occupancy status of sites (1 km river segments) does not
change during the sampling period; 2) detections between sites are independent and; 3) detected species
are not misidentified (MacKenzie et al. 2018). These assumptions were not violated because wood turtles
display high site fidelity and regularly occupied the same stretch of river during the sampling period,
wood turtles were marked with unique numbers when detected to avoid double counting, and on-site
training of surveyors and the wood turtles unique appearance made confusion with other species difficult
(Arvisais et al. 2002, Thompson et al. 2018, Brockman 2024: Chapter 2). I selected a single-season model
given that surveys occurred on each side of the river nearly simultaneously and hence segment-level
occupancy status was unlikely to change during the survey period. I fit occupancy models using the
unmarked package (Fiske and Chandler 2011) in program R (R Core Team 2023).
I first parameterized detection using intercept-only occupancy (i.e., Ψ =1). I specified Julian date
(Date), search time (min; Search), air temperature (°C; AirStart) and water temperature (°C; WaterStart) at
the start of each survey, mean air temperature (°C; AirAvg) and water temperature (°C; WaterAvg) during
each survey, and survey effort (m/min; Effort), as each has been documented to influence wood turtle
detection probability (i.e., p; Flanagan et al. 2013, Brown et al. 2017, Akre et al. 2019, Northeast Wood
Turtle Working Group 2019). Before modeling, I standardized all continuous variables to a mean of 0 and
standard deviation of 1.
For modeling Ψ at the segment level, I calculated the mean proportion of sand and gravel along
the riverbank (NestSub), mean proportion canopy cover above 1.5m (CanCov), and mean basal area
(m2/ha, BasalArea) from data collected in segments from both sides of the river. I then fit the occupancy
model using my top ranked detection model. I compared both detection and occupancy models using
AICc (Burnham and Anderson 1998) via the dredge function from the MuMIn package (Bartoń 2023) in
program R. Models within 2 ΔAICc were considered to have equivalent support. I also assessed multi-
collinearity in all models using variance inflation factors (VIF), where I removed variables >3 VIF (Zuur et
al. 2010). I tested for significance in detection and occupancy models using 85% confidence intervals
(Arnold 2010). I used the top-ranked model to predict wood turtle 1 km river segment-level occupancy
probability for the range of covariates measured in this study.
13
I summarized forest management history by calculating percentage of sample locations falling
within each management treatment. I estimated time since last forest management by subtracting the
year of last observed management activity from the last year that wood turtle detection surveys were
conducted (2022). I split basins into two groups, occupied and unoccupied by wood turtles, and I used a
Mann-Whitney U test to investigate the relationship between occupancy status and time since last
management (Mann and Whitney 1947). I also used a Kruskal-Wallis rank-sum test to investigate the
relationship between basin occupancy status and time since last management for each treatment type
(Wilcoxon 1945, Kruskall and Wallis 1952). I used these non-parametric tests to avoid violating
assumptions of normality of my data.
1.3 Results
1.3.1 Detection
I completed 78 replicate surveys (sides) at 39 1 km river segments in 15 of 22 basins from May 6
until July 30 in 2020, June 16 in 2021, and June 26 in 2022. I surveyed 37 segments on land owned by
NAFO member companies and two on land owned by the State of Michigan. Eleven basins had at least
three surveyed segments and four basins had at least one surveyed segment. I detected 24 wood turtles
in six segments (x̅ = 4 turtles, range = 1 – 8 per segment) and four basins. Of the 15 surveyed basins, five
had historical occupancy records, but wood turtles were only detected in two. Two basins without
historical occupancy records were confirmed occupied (Table 1.1). I recorded a range of conditions in
which surveys were conducted to inform my detection and occupancy models (Table 1.2).
For the detection model, I used 39 river segments (6 were occupied by wood turtles) with two
spatial replicates (sides) each. I ranked 35 unique detection models and found 9 competing models (Table
1.3; Appendix 1C). My null detection model (p0) produced an estimated detection probability (p) of 0.67
(SE = 0.18, 85% CI = 0.38 – 0.87), indicating a 67% chance of detecting a wood turtle during a survey if it
was available for detection (i.e., visible in the water, or on land within the survey area) regardless of
survey conditions. My top ranked detection model was p(AirStart + Date); neither AirStart (ß = -2.2, SE =
2.5, 85% CI = -5.3 – 0.9) or Date (ß = -2.9, SE = 2.5, 85% CI = -6.4 – 0.7) were significant. Given low
explanatory power of top ranked detection covariates, multiple competing models, and to reduce
complexity, I used my null detection model (i.e., p0) when fitting occupancy models.
1.3.2 Occupancy
I built occupancy models using p0 and all combinations of occupancy covariates, resulting in 8
models (Table 1.4). My top ranked model with the most weight was Ψ(NestSub), though there were two
competing models, the null model (Ψ0) and the other included NestSub and CanCov (Table 1.4). I found
14
NestSub (ß = 1, SE = 0.6, 85% CI = 0.2 – 0.9) had a positive influence on wood turtle occupancy, with
segments containing a higher proportion of nest substrate having a higher Ψ. My top model predicted a
4% - 42% increase in Ψ within a 1 km segment of river as nest substrate increased from 0% to 81% (Figure
1.5). The percent nesting substrate of occupied 1 km segments ranged from 32% - 79% (Figure 1.5).
1.3.4 Recent Management History
I assessed recent management history in 15 basins at an average of 76 locations (SD = 18.0, range
= 49 – 122) per basin (Table 1.1). I observed a latitudinal shift in forest management treatment from
select cuts dominating in the north to a more clearcut-dominated landscape in the south (Figure 1.6). In
occupied basins, 54% of sampled locations were managed and in unoccupied basins, 43% of sampled
locations were managed sometime within the last 17 years. On average, the proportion of select cuts was
higher in occupied basins (Figure 1.7), though not significantly so. Mean time since last management
trended higher in unoccupied basins (Figure 1.8), but not significantly (Mann-Whitney U, U = 17, P =
0.56).
1.4 Discussion
In my study, I did not find strong evidence that detection variables I measured influenced wood
turtle detection probability, suggesting that visual encounter surveys along river segments can be
effective without additional survey or environmental parameterization (within the domain of
environmental conditions used in this study from May 1 – July 1). I documented nest substrate as an
important predictor of wood turtle occupancy along 1 km river segments, where increased sand and
gravel along the riverbank positively influenced occupancy. This finding indicates that nesting substrate
may be a limiting factor for wood turtle occupancy in primarily forested landscapes, particularly for those
river systems that lack reliable transport of sands and gravels. My results indicate that in-stream and near-
stream (e.g., nesting substrate) habitat elements are more important than variables measured through
standard forest inventory techniques for estimating occupied river segments. I also failed to find a strong
relationship between wood turtle occupancy and recent forest management history at the basin level,
suggesting that wood turtle occurrence and levels of forest management I observed in this study are
compatible. During my study, I documented indicators of robust wood turtle populations; mature adults
(male and female) and juveniles (in 3 of 4 occupied basins). I caution that a longer-term study is needed to
document demographic structure and population trajectories of wood turtle populations in this landscape
as the potential for long-term viability remains unknown.
In my study, I found that the Midwest survey protocol developed by Brown et al. (2017)
performed well (p = 0.67) for detecting wood turtles regardless of survey conditions from May – July. This
15
survey protocol is simple to design and follow, even without using specialized technology like GIS
mapping. I found that those with little prior experience could detect wood turtles at the same rate as
those with years of experience. If implemented during the springtime (May – July) as recommended, this
protocol is an effective way for forest managers and researchers to confirm wood turtle occupancy along
river segments in primarily forested areas of the Midwest (Brown et al. 2017, Northeast Wood Turtle
Working Group 2019). Multiple surveys within a basin will increase reliability of basin-level occupancy
status.
Though there was not strong support to use Julian date and starting air temperature as detection
covariates in my study, field observations and published literature indicate these variables are important
to wood turtle detection. I found that surveys in May tended to be more productive (79% of all
detections, but only 41% of surveys) than June surveys. Additionally, a small subset (n = 4) of segments
were surveyed in multiple years and some surveys on those segments found wood turtles in May but not
June. This peak in detection corresponded with the emergence/pre-nesting activity period in wood
turtles, which takes place between late April and early June in the Upper Peninsula of Michigan (Harding
and Bloomer 1979). Other studies of wood turtle detection have identified Julian date or season, as
important for increasing detection probability. Flanagan et al. (2013) found that detection probability was
69% within 10m of the river until July 1 in New Brunswick, Canada, and Jones et al. (2015) found that
surveys were significantly more successful within 10 m of the river from March 20 - May 27 than during
any other time of year across the Northeast region of the United States. During this period, wood turtles
generally remain within ~12 m to 30 m of the river as it provides thermal refuge during periods of low
temperatures (Kaufman 1992, Dubois et al. 2009). Anecdotal field observations suggested that cooler
starting temperatures resulted in more wood turtle detections. Cool temperatures may increase
detectability by encouraging basking since wood turtles are ectotherms and they need to seek
environmental warmth to maintain their energetic needs (Kaufman 1992, Dubois et al. 2009, Flanagan et
al. 2013, Brown et al. 2017). Jones et al (2015) also noted that, while air temperature did not have a large
influence on detection rates, there was a positive relationship between detection and the change in air
temperature during surveys, which supports my hypothesis that wood turtles will respond to increases in
air temperature during a survey by basking therefore becoming more available for detection.
The proportion of nesting substrate present along riverbanks in my study was a strong predictor
of wood turtle occupancy for 1 km river segments. I did not detect wood turtles in segments where there
was <32% cover of nesting substrate and most (5 of 6) occupied segments contained ~50% or more
nesting substrate cover. A stronger association with nesting substrate rather than canopy cover and basal
16
area as determinants of river segment occupancy aligns with other studies indicating that wood turtles
can tolerate a range of vegetation conditions but require nesting habitat to maintain healthy populations
(Buech et al. 1997, Hughes et al. 2009). Buech et al. (1997) found wood turtle occurrence was strongly
correlated with presence of sand and sandy gravel soils in Minnesota, and that due to the patchy nature
of glacially deposited soils wood turtle populations in the Great Lakes Region could be centered on these
deposits. However, Buech et al. (1997) noted that high quality nesting habitat is not based on availability
of nesting substrate alone. Open canopy allowing direct sunlight to reach the soil, slope, and elevation
above the waterline and distance from the river also play important roles in wood turtle nest site
selection (Buech et al. 1997). Naturally occurring nesting habitat can come in the form of sandy or gravely
areas on the inside of river bends, cutbanks on the outside bends of rivers, and deposits in the stream
channel or on the floodplain (Buech et al. 1997, Jones 2009, Parren 2013). Wood turtles will also nest in
anthropogenically-created openings (Buech et al. 1997, Walde et al. 2007, Jones 2009, Parren 2013). I
observed adult female wood turtles in my study congregating in an anthropogenically-created opening
used for parking in one of my northern basins (where nesting substrate along the riverbank was
uncommon). I also observed wood turtles nesting in openings on dirt roads, and gravel pits used by forest
products companies to source road building material. These areas were located between 5 m and 135 m
from the edges of rivers.
I failed to find an effect of forest harvest (select cut, shelterwood, or clearcut) on wood turtle
occupancy status for forests within 400 m of rivers. Collectively, my results indicate that recent forest
management activities have little to no effect on wood turtle occupancy at the basin level. Basins in my
study area were predominantly forested, typically with >40% of the area not harvested within the last 17
years. Hence, my study indicates that effects of forest harvest on wood turtle occupancy at the basin-level
may be benign if basins remain forest-dominated. I observed radio-tagged adult female wood turtles
choosing active season terrestrial home ranges in areas that had undergone recent forest management.
These females spent much of the active season in naturally regenerating 7 - 23 year-old aspen (Populus
spp.) clearcuts, in an 11 year-old white pine (Pinus strobus) shelterwood, a 26 year-old larch (Larix sp.)
plantation, a 1-6 year-old shelterwood cut dominated by red and sugar maple respectively, and a 17 year-
old sugar maple select cut. These females would also use old haul roads in more mature even-age forests
where canopies were closed, presumably for thermoregulation and foraging. Forest management
activities do not deter wood turtles from utilizing early successional areas, even if little time has passed
since management last occurred.
17
This study provides insights into factors influencing detection and occupancy of wood turtles in
working forest landscapes in the western Upper Peninsula of Michigan. I found that the Midwest wood
turtle survey protocol (Brown et al. 2017) is an effective method for detecting wood turtles to confirm
river segment occupancy. I also found that the amount of nesting substrate had a positive influence on
occupancy probability. I did not find compelling relationships between wood turtle occupancy probability
and recent forest management history, which may indicate little to no negative impacts at the basin level.
However, I caution that my results apply to a primarily forested landscape, and I posit that wood turtles
would be negatively impacted by large scale forest loss (e.g., urban development and conversion to
agriculture). This supposition supports the benefits of sustainably managed working forests as a means
for species conservation.
1.4.1 Management Implications
The availability of suitable nesting substrate appears to be a limiting factor for wood turtles to
persist or colonize portions of river basins, especially in areas where it may be scarce (Buech et al. 1997).
Identifying potential nesting areas can help focus visual encounter surveys (Buech et al. 1997, Walde et al.
2007, Hughes et al. 2009). My study suggests that 1 km river segments with ≥50% cover of sand and
gravel along the edge of the river have a higher chance of being occupied by wood turtles in the western
Upper Peninsula of Michigan. These areas can be readily identified through ground observations or aerial
imagery (Beuch et al. 1997). I recommend prioritizing detection surveys in these areas and conducting
surveys during spring (May -June) using the Midwest wood turtle survey protocol developed by Brown et
al. (2017). After confirmation of occupancy status (repeated surveys may be needed), conservation
measures can be implemented to protect wood turtles where they occur.
A preponderance of human influences has been identified as a threat to wood turtle populations
(Harding and Bloomer 1979, Garber and Burger 1995, Saumure and Bider 1998, Jones 2009, Parren 2013,
Jones et al. 2015). However, private working forests in the western Upper Peninsula of Michigan generally
lack human influences (e.g., agriculture, development, active roads) and landowners can regulate access
and activities on their lands through gating and operating policies. Whereas forest management is may be
viewed as a threat to wood turtles through direct mortality, these relatively large ownership landscapes
with low human impacts offer opportunities for conservation. Forest managers can aid in wood turtle
conservation by being thoughtful about the timing of logging operations (i.e., avoiding heavy equipment
use during the wood turtle active season) where possible, maintaining a primarily forested and
structurally diverse landscape of open canopies, and minimizing human influence in high quality wood
turtle areas. Forest management in this study appeared to provide plant communities with ample nesting,
18
foraging, and thermoregulatory opportunities for wood turtles (Kaufman 1992, Compton et al. 2002,
Tingley and Herman 2008, Hughes and Litzgus 2019, Latham et al. 2022).
19
TABLES AND FIGURES
Figure 1.1. Percent area forested (i.e., timberland; left panel) and wood turtle (Glyptemys insculpta) range
(right panel) at the county level in Michigan, USA (MNFI 2023, Schaetzl R. J., unpublished report).
20
Figure 1.2. Wood turtle study locations (orange) in the western Upper Peninsula of Michigan, USA, 2021-
2022. Within the study boundary 22 watershed basins (HUC 12) were surveyed that included portions of
Baraga, Dickinson, Houghton, Iron, Marquette and Menominee counties.
21
Figure 1.3. Design of visual encounter surveys for wood turtles (Glyptemys insculpta) in the western
Upper Peninsula of Michigan, USA, 2020-2022 (developed by Brown et al. (2017)). Two parallel transects
(white lines), 0 m and 15 m from the edge of the river, were mapped along each side of a 1 km segment of
river in a basin. Percent nesting substrate and percent canopy cover were estimated every 50 m along the
0 m transect, and basal area was measured every 100 m along the 15 m transect.
22
Figure 1.4. Example of different forest management treatments as portrayed by aerial imagery (National
Agriculture Imagery Program (NAIP)) in Google Earth Engine (Gorelick et al. 2017). Forest management
treatments were categorized as select cut (A = preharvest 2005, B = harvested 2009), shelterwood (C =
preharvest 2009, D = harvested 2012), clearcut shelterwood (E = preharvest 2014, F = harvested 2016).
23
Figure 1.5. Predicted occupancy probability of wood turtles (Glyptemys insculpta) based on the
proportion of sand and gravel present on the bank along 1 km segments of river surveyed in the western
Upper Peninsula of Michigan, USA, 2020 - 2022. The grey band represents 85% confidence intervals. The
black circles indicate the occupancy status of 1 km river segments, with 0.00 = unoccupied and 1.00 =
occupied by wood turtles.
24
Figure 1.6. Percentage of locations within 400 m of rivers by forest management treatment (Select =
select cut, Shelter = shelterwood, Clear = clearcut, None = no management) within surveyed basin in the
western Upper Peninsula of Michigan, USA, 2005 – 2022. Basins are arranged from north (left) to south
(right). * indicates basins occupied by wood turtles (Glyptemys insculpta).
25
Figure 1.7. Average (and one standard error) percentage of locations within 400 m of rivers by forest
management treatment (Select = select cut, Shelter = shelterwood, Clear = clearcut, None = no
management) by wood turtle (Glyptemys insculpta) occupancy status (2020 – 2022) of surveyed basins in
the western Upper Peninsula of Michigan, USA, 2005 – 2022. A Kruskall-Wallis rank-sum test did not
identify significant relationships between forest management treatment and basin occupancy status.
26
Figure 1.8. Mean (and one standard error) time since last forest management observed at locations
within 400 m of rivers by treatment (Select = select cut, Shelter = shelterwood, Clear = clearcut) between
2005 and 2022 by wood turtle (Glyptemys insculpta) occupancy status (2020 – 2022) of surveyed basins in
the western Upper Peninsula of Michigan, USA. A Kruskall-Wallis rank-sum test did not identify any
significant relationship between time since last forest management by treatment and basin occupancy
status.
27
Table 1.1. Visual encounter survey results by basin for wood turtles (Glyptemys insculpta) along 39 – 1 km segments of river and mean percent of
GRTS locations falling within forest management treatment types (i.e., select cut, shelterwood, clearcut, and no management) within 400 m of
surveyed rivers in 15 watershed basins (HUC 12) in the western Upper Peninsula of Michigan, USA, 2020-2022.
Basin
Basin-A
Basin-B
Basin-C
Basin-D
Basin-E
Basin-F
Basin-G
Basin-H
Basin-I
Basin-J
Basin-K
Basin-L
Basin-M
Basin-N
Basin-O
Historical
Records
Segments
Surveyed
Segments
Occupied
Wood Turtles
Detected
Sex Ratio
(male:female:juvenile)
Mean %
Select Cut
Mean %
Shelterwood
Mean %
Clearcut
Mean % No
Management
-
-
-
-
-
H
-
-
-
-
H
H
H
H
-
3
3
3
3
3
4
3
3
2
1
1
3
1
3
3
0
1
1
0
0
2
0
0
0
0
0
2
0
0
0
0
8
2
0
0
7
0
0
0
0
0
7
0
0
0
-
1:4:3
0:1:1
-
-
5:2:0
-
-
-
-
-
2:2:3
-
-
-
38.0%
60.0%
33.9%
75.0%
55.1%
32.0%
13.6%
12.3%
14.8%
1.5%
5.9%
21.3%
4.4%
11.4%
12.1%
3.4%
2.7%
4.8%
3.8%
2.0%
0.0%
7.6%
1.2%
3.3%
15.4%
1.5%
13.1%
8.8%
8.0%
8.8%
1.1%
4%
16.1%
1.9%
2.0%
5.3%
18.2%
13.6%
16.4%
36.9%
23.5%
22.1%
15.4%
23.9%
23.1%
57.5%
33.3%
45.2%
19.2%
40.8%
62.7%
60.6%
72.8%
65.6%
46.2%
69.1%
43.4%
71.4%
56.8%
56.0%
28
Table 1.2. Covariates used to inform detection (A) and single-season occupancy (B) models using the
unmarked package (Fiske and Chandler 2011). Detection covariates were recorded during surveys on each
side of the river, and occupancy covariates were derived from mean values measured between paired
surveys along a 1 km segment of river.
Covariate
Mean
SE
Range
Description
A. Detection
Date
Search
157
180.8
2.1
4.6
126 - 212
86 - 304
AirStart
18.4
0.5
7.8 - 32.4
The Julian Date when each survey took place.
Time (min) spent searching for wood turtles during
each survey.
Ambient air temperature (C) measured at the
beginning of each survey.
WaterStart
15.3
0.5
6.1 - 27.5 Water temperature (C) measured at the beginning of
AirAvg
20.3
0.5
9.3 - 30.8
WaterAvg
16.3
0.4
8.2 - 27.2
each survey.
Average ambient air temperature (C) measured
during each survey.
Average water temperature (C) measured during each
survey.
Effort
11.7
0.3
5.9 - 18.2 Meters searched per minute during each survey
(m/min).
B. Occupancy
NestSub
0.4
0.05
0.0 - 0.8
CanCov
0.4
0.03
0.1 - 0.8
BasalArea
22.6
1.2
5.0 - 40.2
Proportion of sand and gravel measured along both
sides a 1 km river segment.
Proportion overstory canopy cover measured in a 1
km river segment.
Basal area (m2/ha) of trees along both sides a 1 km
river segment.
29
Table 1.3. Top 15 of 35 model selection results used to determine the top-ranked predictor of wood turtle
(Glyptemys insculpta) detection probability (p) during surveys of 1 km segments of river in 15 basins in the
western Upper Peninsula of Michigan, USA, 2020 - 2022. Variables include, model structure (Structure),
Akaike information criterion (AICc), delta AICc (ΔAICc), degrees of freedom (df), and model weight (wi).
Structure
p(Date + AirStart)
p(Date)
p(AirDiff)
p(AirStart)
p(AirDiff + WaterStart)
p(WaterStart)
p(AirDiff + Date)
p(WaterAvg + WaterStart)
p(AirAvg + AirStart)
p(AirDiff + AirStart)
p(AirDiff + WaterAvg)
p(AirAvg + Date)
p(AirAvg + AirDiff)
p0
p(Date + Effort)
AICc
47.9
48.0
48.5
48.7
49.5
49.5
49.6
49.7
49.8
49.8
49.9
50.0
50.1
50.3
50.3
ΔAICc
0.0
0.1
0.6
0.8
1.6
1.6
1.7
1.8
1.9
1.9
2.0
2.1
2.2
2.4
2.4
df
4
3
3
3
4
3
4
4
4
4
4
4
4
2
4
wi
0.10
0.09
0.07
0.06
0.04
0.04
0.04
0.04
0.04
0.04
0.04
0.03
0.03
0.03
0.03
30
Table 1.4. Model selection results used to determine the top-ranked predictor of wood turtle (Glyptemys
insculpta) occupancy probability (Ψ) using a null detection model for 1 km segments of river in 15 basins
in the western Upper Peninsula of Michigan, USA, 2020 – 2022. Variables include, model structure,
Akaike information criterion (AICc), delta AICc (ΔAICc), degrees of freedom (df), and model weight (wi).
These models were produced using a single-season occupancy model in the unmarked package (Fiske and
Chandler 2011).
Structure
Ψ(NestSub)
Ψ0
Ψ(CanCov + NestSub)
Ψ(BasalArea + NestSub)
Ψ(BasalArea + CanCov + NestSub)
Ψ(CanCov)
Ψ(BasalArea)
Ψ(BasalArea + CanCov)
AICc
48.69
50.30
50.56
51.17
52.15
52.23
52.65
54.10
ΔAICc
0.0
1.6
1.9
2.5
3.5
3.5
4.0
5.4
df
3
2
4
4
5
3
4
4
wi
0.37
0.17
0.15
0.11
0.07
0.06
0.05
0.02
31
Table 1.5. Kruskal-Wallis rank-sum results comparing forest management treatment and mean time in
years since last forest management activity at locations within 400 m of rivers by treatment (Select =
select cut, Shelter = shelterwood, Clear = clearcut, None = no management) between basins occupied and
unoccupied by wood turtles (Glyptemys insculpta) (2020 – 2022) in the western Upper Peninsula of
Michigan, USA.
Mean Percent Occurrence of
Treatments
Mean Time Since Last Management
(years)
Chi-Squared
P-value
Chi-Squared
P-value
2.45
0.15
0.27
1.70
0.12
0.70
0.60
0.19
0.61
0.15
0.00
-
0.43
0.70
1.00
-
Management
Treatment
Select
Shelter
Clear
None
32
LITERATURE CITED
Akre, T. S., L. D. Parker, E. Ruther, J. E. Maldonado, L. Lemmon, and N. R. McInerney. 2019. Concurrent
visual encounter sampling validates eDNA selectivity and sensitivity for the endangered wood
turtle (Glyptemys insculpta). PLoS One 14:1–23.
Arnold, T. W. 2010. Uninformative parameters and model selection using Akaike’s Information Criterion.
Journal of Wildlife Management 74:1175–1178.
Buech, R. R., L. G. Hanson, and M. D. Nelson. 1997. Identification of wood turtle nesting areas for
protection and management. Proceedings of the 1997 Conservation, Restoration, and
Management of Tortoises and Turtles: An International Conference. New York Turtle and Tortoise
Society, New York, New York. 494 pp.
Bartoń K. 2023. MuMIn: Multi-Model Inference. R package version 1.47.1.
Brockman, T. G. 2024. Wood turtle (Glyptemys insculpta) occupancy and spatial ecology in working forest
landscapes. Thesis, Michigan State University, East Lansing, Michigan, USA 75 pp.
Brown, D. J., M. M. Cochrane, and R. A. Moen. 2017. Survey and analysis design for wood turtle
population monitoring. Journal of Wildlife Management 81:868–877.
Buckley, D. S., J. G. Isebrands, and T. L. Sharik. 1999. Practical field methods of estimating canopy cover,
PAR, and LAI in Michigan oak and pine stands. Northern Journal of Applied Forestry 16:25–32.
Burnham, K. P., and D. R. Anderson. 1998. Model selection and inference: A practical information-
theoretical approach. Springer, New York, New York, USA. 488 pp.
Compton, B. W., J. M. Rhymer, and M. McCollough. 2002. Habitat selection by wood turtles (Clemmys
insculpta): An application of paired logistic regression. Ecology 83:833–843.
Congdon, J. D., A. E. Dunham, and R. C. V. L. Sels. 1994. Demographics of common snapping turtles
(Chelydra serpentina): Implications for conservation and management of long-lived organisms.
Integrative and Comparative Biology 34:397–408.
Dubois, Y., G. Blouin-Demers, B. Shipley, and D. Thomas. 2009. Thermoregulation and habitat selection in
wood turtles Glyptemys insculpta: Chasing the sun slowly. Journal of Animal Ecology 78:1023–
1032.
Ernst, C. H., and J. E. Lovich. 2009. Turtles of the United States and Canada. Second Edition. Johns Hopkins
University Press, Baltimore, Maryland, US. 434 pp.
Fiske, I., and R. Chandler. 2011. unmarked: An R Package for fitting hierarchical models of wildlife
occurrence and abundance. Journal of Statistical Software 43: 1-23.
Flanagan, M., V. Roy-Mcdougall, G. Forbes, and G. Forbes. 2013. Survey methodology for the detection of
wood turtles (Glyptemys insculpta). Canadian Field-Naturalist 127:216–223.
33
Garber, S. D., and J. Burger. 1995. A 20-yr study documenting the relationship between turtle decline and
human recreation. Ecological Applications 5(4):1151-1162.
Giese, C. L. A., D. N. Greenwald, and T. Curry. 2012. Petition to list 53 amphibians and reptiles in the
United States as threatened or endangered species under the Endangered Species Act. Petition to
the Secretary of the Interior of the United States. Center for Biological Diversity, Tucson, Arizona,
USA. 454 pp.
Gorelick, N., M. Hancher, M. Dixon, S. Ilyushchenko, D. Thau, and R. Moore. 2017. Google Earth Engine:
Planetary-scale geospatial analysis for everyone. Remote Sensing of Environment 202:18-27.
Great Lakes Integrated Sciences and Assessments [GLISA]. 2023a. Historical climatology: Marquette,
Michigan. . Accessed 4 October 2023.
Great Lakes Integrated Sciences and Assessments [GLISA]. 2023b. Historical climatology: Iron Mountain
Kingsford WWTP, MI. . Accessed 4 October 2023.
Harding, J. H., and T.J. Bloomer. 1979. The wood turtle, Clemmys insculpta... a natural history. Bulletin of
the New York Herpetological Society 15:9–26.
Hughes, G. N., W. F. Greaves, and J. D. Litzgus. 2009. Nest-site selection by wood turtles (Glyptemys
insculpta) in a thermally limited environment. Northeastern Naturalist 16:321–338.
Hughes, G. N., and J. D. Litzgus. 2019. Impact of natural resource extraction on thermal properties of
wood turtle (Glyptemys insculpta) habitat. Journal of Thermal Biology 84:469–478.
Jones, M. T. 2009. Spatial ecology, population structure, and conservation of the wood turtle, Glyptemys
insculpta, in central new England. Dissertation, University of Massachusetts Amherst, Amherst,
Massachusetts, US. 266 pp.
Jones, M. T., L. L. Willey, T. S. B. Akre, and P. R. Sievert. 2015. Status and conservation of the wood turtle
in the northeastern United States. Northeastern Association of Fish and Wildlife Agencies. 288
pp.
Jones, M. T., and L. L. Willey, editors. 2021. Biology and conservation of the wood turtle. Northeast
Association of Fish and Wildlife Agencies, Inc., Petersburgh, New York, USA. 235 pp.
Kaufmann, J. H. 1992. Habitat use by wood turtles in central Pennsylvania. Journal of Herpetology
26:315–321.
Kruskall, W. H. and Wallis, W. A. 1952. Use of ranks in one-criterion variance analysis. Journal of the
American Statistical Association 42:583-621.
LANDFIRE. 2022. Existing vegetation type layer, LANDFIRE 2.3.0, U.S. Department of the Interior,
Geological Survey, and U.S. Department of Agriculture. . Accessed
20 October 2022.
34
Latham, S. R., A. P. K. Sirén, and L. R. Reitsma. 2022. Space use and resource selection of Wood Turtles
(Glyptemys insculpta) in the northeastern part of its range. Canadian Journal of Zoology 101:20–
31.
MacKenzie, D. I., J. D. Nichols, J. A. Royle, K. H. Pollock, L. Bailey, and J. E. Hines. 2018. Occupancy
estimation and modeling: Inferring patterns and dynamics of species occurrence. Elsevier, San
Diego, California, USA. 641 pp.
Mann, H. B. and D. R., Whitney. 1947. On a test of whether one of two random variables is stochastically
larger than the other. The Annals of Mathematical Statistics 50-60.
McCoard, K. R. P., N. S. McCoard, P. J. Turk, and J. T. Anderson. 2016. Habitat characteristics that
influence the occurrence of wood turtles at the southern limits of their range in the central
Appalachians. Journal of Herpetology 50:381–387.
Michigan Natural Features Inventory [MNFI]. 2023. Glyptemys insculpta.
. Accessed 11
November 2023.
Northeast Wood Turtle Working Group. 2019. Conservation plan for the wood turtle in the northeastern
United States: Population assessment protocol. 9 pp.
Olsen, A. R., T. M. Kincaid, and Q. Payton. 2012. Spatially balanced survey designs for natural resources.
Pages 126-150 in R. A. Gitzen, J. J. Millspaugh, A. B. Cooper and D. S. Licht, editors. Design and
Analysis of Long-Term Ecological Monitoring Studies. Cambridge University Press, Cambridge, UK.
Parren, S. G. 2013. A twenty-five year study of the wood turtle (Glyptemys insculpta) in Vermont:
Movements, behavior, injuries, and death. Herpetological Conservation and Biology 8:176–190.
Poudel, J. 2022. Forest products industries’ economic contributions: Michigan, 2019. Forest Resources
Division, Michigan Department of Natural Resources, Lansing, Michigan, USA. 28 pp.
R Core Team. 2023. R: A language and environment for statistical computing. R Foundation for Statistical
Computing, Vienna, Austria.
Saumure, R. A. 2004. Spatial ecology and conservation of the North American wood turtle (Glyptemys
insculpta) in a fragmented agri-forest landscape. Dissertation, McGill University, Montreal,
Quebec, CA. 148 pp.
Saumure, R. A., and J. R. Bider. 1998. Impact of agricultural development on a population of wood turtles
(Clemmys insculpta) in southern Quebec, Canada. Chelonian Conservation and Biology 3:37-45.
Schaetzl, R. J., H. Enander, M. D. Leuhmann, D. P. Lusch, C. Fish, M. Bigsby, M. Steigmeyer, J. Guasco, C.
Forgacs, and A. Pollyea. 2013. Physiographic map of Michigan.
. Accessed 14
September 2020.
35
Seaber, P. R., F. P. Kapinos, and G. L. Knapp. 1987. Hydrologic unit maps. United States Geological Survey
Water-Supply Paper 2294. US Geological Survey: Anchorage, Alaska, USA.
Tingley, R., and T. B. Herman. 2008. The effects of agriculture and forestry on the distribution,
movements and survival of wood turtles in an intensively managed landscape. A report for the
Nova Scotia Habitat Conservation Fund. 57 pp.
van Dijk, P.P. and J. Harding. 2011. Glyptemys insculpta. The IUCN Red List of threatened species 2011.
. Accessed 3 May 2023.
Walde, A. D., J. R. Bider, D. Masse, R. A. Saumure, and R. D. Titman. 2007. Nesting ecology and hatching
success of the wood turtle, Glyptemys insculpta, in Quebec. Herpetological Conservation and
Biology 2:49-60.
Wilcoxon, F. 1945. Some uses of statistics in plant pathology. Biometrics Bulletin 1:41-45.
Zuur, A. F., E. N. Ieno, and C. S. Elphick. 2010. A protocol for data exploration to avoid common statistical
problems. Methods in Ecology and Evolution 1:3–14.
36
APPENDIX 1A: WOOD TURTLE SHELL DIAGRAM
LAB: Left abdominal scute.
SCL: Straight carapace length.
SCW: Straight carapace width.
SPL: Straight plastron length.
SPW: Straight plastron width.
Figure 1A.1. Diagram of annuli counting location (LAB), morphometric measurement locations (SCL, SCW,
SPL, and SPW) and shell notching sequence for captured wood turtles. Shell height was measured from
the middle of the 3rd vertebral scute to the bottom of the plastron.
37
APPENDIX 1B: WOOD TURTLE CAPTURE DATA
Table 1B.1. Data collected for each wood turtle captured during detection surveys in the western Upper Peninsula of Michigan, USA, between
2020 and 2022. Wood turtles were detected in four of 15 basins surveyed. Each wood turtle was sexed and given a unique ID, which was notched
into the marginal scutes of the shell (See Appendix B). Sex was determined by the presence of secondary sexual characteristics or by counting the
number of annuli and assessing the percent of shell wear on the left abdominal scute. Behavior at the time of capture was also recorded.
Date
5/19/2021
5/19/2021
5/19/2021
5/19/2021
5/19/2021
5/19/2021
5/19/2021
5/19/2021
6/26/2022
6/26/2022
6/1/2020
5/11/2021
5/6/2021
5/6/2021
5/17/2021
5/13/2021
5/13/2021
6/11/2020
6/7/2021
5/5/2022
5/5/2022
5/5/2022
5/23/2022
5/23/2022
Wood
Turtle ID
F20
F22
F24
U15
U17
U19
M21
F23
U54
F56
M2
F12
M8
M10
M13
F9
M14
U6
U28
F41
M42
M39
F50
U45
Transect
ID
Basin ID
Basin-B WT1L
Basin-B WT1L
Basin-B WT1L
Basin-B WT1R
Basin-B WT1R
Basin-B WT1R
Basin-B WT1R
Basin-B WT1R
Basin-C WT3L
Basin-C WT3L
Basin-F WT2L
Basin-F WT2L
Basin-F WT2R
Basin-F WT2R
Basin-F WT3L
Basin-F WT3R
Basin-F WT3R
Basin-L WT3R
Basin-L WT5R
Basin-L WT5R
Basin-L WT5R
Basin-L WT5R
Basin-L WT5R
Basin-L WT5R
Annuli
Count
11
0
19
6
2
2
11
15
7
17
17
13
10
0
16
9
10
5
6
19
24
20
19
3
Wear
(%)
0
100
25
0
0
0
80
10
10
20
0
75
50
100
50
90
10
0
0
0
10
40
10
0
Sex
F
F
F
U
U
U
M
F
U
F
M
F
M
M
M
F
M
U
U
F
M
M
F
U
38
Mass
(g)
530
1390
1430
150
23
25
1390
1050
440
1150
NA
1340
1340
1590
1270
1490
590
NA
130
1040
1100
1240
950
40
SCL
(mm)
154.6
206.1
208.7
99.6
51.3
52.0
209.2
185.2
146.0
207.2
201.5
201.7
216.3
220.3
218.2
208.8
167.1
42.8
101.3
185.4
198.2
213.2
182.3
62.4
SCW
(mm)
121.0
150.7
154.7
83.6
46.6
48.8
152.2
145.7
104.8
149.3
150.5
153.7
142.7
178.9
154.8
159.3
116.9
33.4
79.8
141.5
147.7
151.4
134.2
55.4
SPL
(mm)
144.2
191.6
186.3
91.2
45.9
47.1
179.6
172.1
135.7
185.4
187.1
188.4
184.1
200.6
188.7
190.7
155.4
39.3
92.3
164.6
170.0
176.3
161.7
55.5
SPW
(mm)
68.2
91.6
91.0
46.4
22.4
24.4
80.2
84.8
59.6
90.1
86.7
93.8
80.7
84.9
80.2
91.1
70.8
18.3
43.1
80.7
73.7
81.9
77.2
29.1
SH
(mm) Behavior
52.6
71.1
78.3
36.7
18.7
20.9
74.1
69.4
51.9
68.1
76.1
77.2
73.0
68.7
64.7
77.2
57.6
15.7
40.1
75.8
67.6
72.5
68.0
25.5
RG
BK
RG
RG
BK
BK
BK
RG
BK
BK
RG
WK
FE
BK
BK
BK
BK
RG
RG
BK
BK
BK
BK
BK
Table 1B.1 (cont’d)
Annuli count: Number of visible annuli on the left abdominal scute.
Wear (%): Percentage wear present on the left abdominal scute.
SCL: Straight carapace length.
SCW: Straight carapace width.
SPL: Straight plastron length.
SPW: Straight plastron width.
SH: Shell height, BK: Basking, FE: Feeding, RG: Resting on ground, WK: Walking.
39
APPENDIX 1C: DETECTION MODEL RESULTS
Table 1C.1. Model selection results used to determine the top-ranked predictor of wood turtle (Glyptemys
insculpta) detection probability (p) during surveys of 1 km segments of river in 15 basins the western
Upper Peninsula of Michigan, USA, 2020 - 2022. Variables include, model structure (Structure), Akaike
information criterion (AICc), delta AICc (ΔAICc), degrees of freedom (df), and model weight (wi).
Structure
p(Date + AirStart)
p(Date)
p(AirDiff)
p(AirStart)
p(AirDiff + WaterStart)
p(WaterStart)
p(AirDiff + Date)
p(WaterAvg + WaterStart)
p(AirAvg + AirStart)
p(AirDiff + AirStart)
p(AirDiff + WaterAvg)
p(AirAvg + Date)
p(AirAvg + AirDiff)
p0
p(Date + Effort)
p(Date + WaterAvg)
p(AirAvg + WaterStart)
p(AirDiff + Effort)
p(AirStart + Effort)
p(AirDiff + Search)
p(AirStart + WaterAvg)
p(AirStart + Search)
p(AirStart + WaterStart)
p(Search + WaterStart)
p(WaterAvg)
p(Effort + WaterStart)
p(AirAvg)
p(Search)
p(Effort + WaterStart)
p(Search + WaterAvg)
p(AirAvg + WaterAvg)
p(Effort + WaterAvg)
p(AirAvg + Search)
p(AirAvg + Effort)
p(Effort + Search)
ΔAICc
0.0
0.1
0.6
0.8
1.6
1.6
1.7
1.8
1.9
1.9
2.0
2.1
2.2
2.4
2.4
2.5
2.6
2.6
3.0
3.0
3.1
3.1
3.3
3.4
3.7
3.7
4.0
4.6
4.7
5.8
5.9
6.1
6.3
6.4
6.8
df
4
3
3
3
4
3
4
4
4
4
4
4
4
2
4
4
4
4
4
4
4
4
4
4
3
4
3
3
3
4
4
4
4
4
4
Wi
0.10
0.09
0.07
0.06
0.04
0.04
0.04
0.04
0.04
0.04
0.04
0.03
0.03
0.03
0.03
0.03
0.03
0.03
0.02
0.02
0.02
0.02
0.02
0.02
0.02
0.02
0.01
0.01
0.01
0.01
0.01
0.01
<0.01
<0.01
<0.01
AICc
47.9
48.0
48.5
48.7
49.5
49.5
49.6
49.7
49.8
49.8
49.9
50.0
50.1
50.3
50.3
50.4
50.5
50.5
50.9
50.9
51.0
51.0
51.2
51.3
51.6
51.6
52.0
52.5
52.6
53.7
53.8
54.0
54.2
54.3
54.7
40
CHAPTER 2. SEASONAL MOVEMENTS AND HOME RANGES OF ADULT FEMALE WOOD TURTLES
(GLYPTEMYS INSCULPTA) IN WORKING FOREST LANDSCAPES IN THE WESTERN UPPER PENINSULA OF
2.1 Introduction
MICHIGAN, USA
Wood turtles (Glyptemys insculpta) are listed as endangered by the International Union for
Conservation (IUCN) with potential for listing under the Federal Endangered Species Act of 1973, as
amended (ESA; 88 Fed.Reg.88338) in the United States (van Dijk and Harding 2011, Giese et al. 2012).
Wood turtles are a semi-terrestrial riparian species that spend much of their active season in forested
mosaic landscapes surrounding occupied rivers throughout their range, which includes portions of the
Great Lakes Region and Northeast United States, and southeastern Canada (Harding and Bloomer 1979,
Ernst and Lovich 2009). Many forests in these regions are privately owned and managed and serve as an
important source of wood fiber and economic benefits (Poudel 2022). Co-occurrence of wood turtles and
active forest management poses a potential challenge for wood turtle conservation due to risk of direct
mortality from harvesting equipment and habitat alterations (Saumure 2004, Tingley and Herman 2008).
However, forest management activities may create opportunities for with wood turtle conservation, as
wood turtles are considered an “edge” species and tend to select structurally complex mosaics of open
canopies, young forests, and more mature patches (Kaufmann 1992, Tingley and Herman 2008, Latham et
al. 2022). Potentially negative effects of active forest management on wood turtles are often mitigated
through seasonal restrictions of machine tree harvesting within a certain distance of occupied rivers (i.e.,
riparian buffers). This is particularly important in the Upper Great Lakes Region of the United States, as
over 63% of forested lands are privately owned and managed (Hillard 2021, Kurtz et al. 2019, Poudel
2022, USDA Forest Service 2024). Timing and duration of terrestrial movements of wood turtles and their
corresponding home range sizes vary by latitude, local climate patterns, habitat quality, and sex (Harding
and Bloomer 1979, Kaufmann 1992, Arvisais et al. 2002, Saumure 2004, Remsburg et al. 2006, Thompson
et al. 2018). Therefore, it is important to understand location-specific wood turtle seasonal movements
and home ranges to develop effective management recommendations (Remsburg et al. 2006, Sweeten
2007).
Throughout the year, wood turtles go through five distinct activity periods: emergence/pre-
nesting (hereafter: pre-nesting), nesting, post-nesting, pre-brumation, and brumation (Arvisais et al.
2002). In northern Michigan, the active season for wood turtles (i.e., excluding brumation), typically lasts
from late-April through mid-October (Harding and Bloomer 1979, Remsburg et al. 2006). Emergence/pre-
nesting, when wood turtles emerge from brumation and travel along river corridors to nesting areas, lasts
41
from late April until early June (Harding and Bloomer 1979, this study) Nesting occurs roughly between
June 8 and June 29 (Harding and Bloomer 1979, this study). Post-nesting, when wood turtles are most
terrestrial and at greatest exposure to risks from forestry machinery, lasts from the end of June until
September (Harding and Bloomer 1979, this study). Harding and Bloomer (1979) noted that wood turtles
were found no farther than 152 m from their occupied river in the Upper Peninsula of Michigan during
post-nesting. Remsburg et al. (2006) found 92.5% of wood turtle locations were within 200 m of their
occupied river in the northern Lower Peninsula of Michigan, though distances exceeding 500 m were
observed. Pre-brumation, when wood turtles return to the river and prepare for brumation underwater,
occurs from September until mid-October and may end by late September based on local weather
conditions (Harding and Bloomer 1979). Mean active season home range size for wood turtles in the
northern Lower Peninsula of Michigan was estimated as 30.2 ha using 100% minimum convex polygons
(MCP) (Remsburg et al. 2006). High variance of wood turtle movement from flowing water and home
range size in Michigan illustrates the importance of a more detailed analysis of wood turtle spatial ecology
to inform effective conservation measures.
Few studies have assessed wood turtle seasonal movements from flowing water during their
activity periods in Michigan and made comparisons to other studies in the Upper Great Lakes Region. This
information is critical to inform regulations and management practices to protect wood turtles while
allowing for continuation of active forest activities in working landscapes. To better understand wood
turtle spatial ecology in private, working forests, my objectives were to 1) analyze mean weekly distance
(m) traveled from flowing water throughout activity periods, 2) analyze stream range and terrestrial home
range sizes (ha), and 3) compare results among watershed basins and to other wood turtle research from
the Upper Great Lakes Region. I predicted that mean weekly distances from flowing water would be
similar between basins as all wood turtles were female and, while spatially isolated, basins were within
one degree of latitude. Stream range (i.e., the in-river distance traversed during the active season) would
be smaller in the northern basin where suitable nesting areas were limited (thereby concentrating wood
turtle space use around nesting areas). Summer home range, would be larger in the northern basin where
there were larger stands of contiguous mature forest, perhaps limiting thermoregulatory opportunities
and causing females to range farther. Overall, I predict that my estimates of wood turtle space use will be
comparable to other studies in Michigan and throughout the Upper Great Lakes Region.
42
2.2 Methods
2.2.1 Study Area
I conducted my study in two watershed basins (hereafter “basins”; hydrological unit code (HUC)
12 level) in the western Upper Peninsula of Michigan (Seaber et al. 1987). These basins were in Baraga
(northern basin) and southern Menominee (southern basin) counties. Average monthly temperatures
ranged from -0.1 °C to 23.9 °C in the northern basin, and -1.6 °C to 27.0 °C in the southern basin during
the active season (GLISA 2023a, b). Average monthly precipitation (cm) ranged from 0.2 cm to 3.1 cm in
the northern basin and 0.2 cm to 4.5 cm in the southern basin during active season (GLISA 2023a, b).
In the northern basin dominant vegetation cover within 400 m of the river occupied by wood
turtles was closed canopy deciduous hardwood forest, mainly yellow birch (Betula alleghaniensis) and
sugar maple (Acer saccharum) (LANDFIRE 2022). Topography was hilly, with exposed glacial bedrock and
well-drained sandy soils in the uplands and poorly drained organic soils in lowlands (Schaetzl et al. 2013).
The occupied river averaged ~20 m across and water was tannic. Riverbed substrate was mostly rock, with
steep densely forested banks, and channelized flow (Figure 2.1). NAFO member companies owned and
managed 43% percent of forest in the northern basin and hardwood select cutting was the dominant
forest harvest technique.
In the southern basin dominant vegetation cover within 400 m of the river occupied by wood
turtles was a mix of open and closed canopy riparian forest, mainly black ash (Fraxinus nigra), American
elm (Ulmus americana), and red maple (Acer rubrum) (LANDFIRE 2022). Topography was dominated by
drumlins, with well-drained sandy loam soils in the uplands and poorly drained organic soils in lowlands
(Schaetzl et al. 2013). The river averaged ~20 m across and water was tannic. Riverbed substrate was
sandy, with exposed sandy banks, and meandering flow (Figure 2.1). In the southern basin, NAFO
member companies owned and managed 53% percent of the land and hardwood select cuts and
clearcutting were the co-dominant forest harvest techniques.
2.2.2 Wood Turtle Capture and VHF Radio Telemetry
From May through June of 2021 and in May of 2022 I captured adult female wood turtles in two
watershed basins during visual encounter surveys or via chance encounters. I focused on adult female
wood turtles as their survival is critical to population viability and they exhibit greater terrestrial
movements during the active season compared to males (Congdon et al. 1994, Parren 2013, Thompson et
al. 2018, Latham et al. 2022). Each turtle was captured by hand, aged, weighed (g), measured (mm), and
given a unique number for identification using a shell notching system (Appendix 2A). I attached Holohil
RI-2B (VHF) radio transmitters (Holohil, Carp, Ontario, CA) weighing 10 g - 15 g (<5% total body weight) to
43
the fourth left or right costal scute using a J-B Weld WaterWeld 2-part epoxy (J-B Weld, Sulphur Springs,
TX, US), with the antenna trailing behind (Appendix 2B). After the epoxy cured for 30 minutes, I released
each wood turtle at point of capture. I relocated tagged female wood turtles using an R-1000 telemetry
receiver (Telonics, Mesa, Arizona, US) and three-element yagi antenna (Advanced Telemetry Systems,
Isanti, MN, US) 1 - 2 times per week from May-October of 2021, and May-September of 2022. During
each relocation, I recorded the GPS location of each turtle, date, weather conditions, vegetation (i.e.,
canopy cover, forest type), turtle behavior (i.e., basking, swimming, mating), and evidence of recent forest
management (e.g., cut stumps) (Appendix 2C). I removed transmitters in September of 2022. All capture
and handling procedures were approved by the Institutional Animal Care and Use Committee at Michigan
State University (PROTO202000121) and conducted under scientific collector permits issued by the
Michigan Department of Natural Resources (Permit numbers: FSCP02222021123151 [2021],
FSCP01312022115615 [2022]).
2.2.3 Seasonal Movements
To evaluate movement distances of wood turtles from flowing water during the active season
(May - October), I first measured Euclidean distance (m) from river edge to each wood turtle relocation
using ArcMap (version 10.8.2, Environmental Systems Research Institute, Redlands, CA, US), and assigned
Julian date to each relocation. Using the “dplyr” package (Wickham et al. 2023) in program R (R Core
Team 2023), I calculated mean weekly distance (m) traveled for all wood turtles by basin across years. I
also incorporated still waterbodies (i.e., temporary ponds, oxbow lakes, backwaters) into my movement
distance analysis due to their observed influence on pre-nesting movements in the southern basin. I used
a Mann-Whitney U test to compare mean weekly distances of female wood turtle relocations from
flowing and still waterbodies between basins throughout the active season and during each activity
period (Wilcoxon 1945, Mann and Whitney 1947).
2.2.4 Home Ranges
I used distance traveled in-stream (stream range) and area used during the post-nesting activity
period when turtles were terrestrial (summer home range) to estimate individual home ranges (Remsburg
et al. 2006, Jones and Willey 2020). Wood turtles in my study exhibited different space-use patterns in-
stream (i.e., linear) compared to terrestrial (i.e., clustered, highly autocorrelated) and I deemed it
important to summarize these distinct space use patterns differently to protected against over-estimating
home range size.
Stream range was measured individually for each wood turtle by measuring distance (m) between
the farthest relocation points in ArcMap within 20 m of the river along the river’s path. I chose relocations
44
within 20 m of the river as this was the average distance of wood turtle relocations during
emergence/pre-nesting and pre-brumation activity periods when they are closely associated with rivers
(e.g., see stream range in Figure 2.2).
For terrestrial home range, I used wood turtle relocations from the post-nesting activity period
(July through late-August) >20 m from the river. I estimated individual terrestrial home ranges using
Brownian bridge kernel method to produce utility distributions using the adehabitatLT and adehabitatHR
packages in R (Calenge 2006). I chose this method to represent core use areas, avoid large areas of un-
utilized space (e.g., see 95% kernel terrestrial home range estimate in Figure 2.2), and to account for high
spatial autocorrelation between successive relocations (Calenge 2006, Row and Blouin-Demers 2006, Silva
et al. 2020). To facilitate comparison of my results to other studies, I also estimated 95% MCPs, using the
adehabitatHR package in R (Calenge 2006) for individuals each year using all relocations (e.g., see 95%
MCP estimate in Figure 2.2). Lastly, I investigated site fidelity for each tagged wood turtle by calculating
percent overlap between terrestrial home ranges in 2021 and 2022 using the Minta Index (Minta 1992). I
used a Mann-Whitney U test to compare stream range length and terrestrial home range size between
basins (Wilcoxon 1945, Mann and Whitney 1947).
2.3 Results
2.3.1 Wood Turtle Capture and VHF Radio Telemetry
In 2021, I captured and tagged seven adult female wood turtles between May 11 and June 9 and
relocated them an average of 27 times each (range = 23 - 30) from May 11 through October 23. In the
northern basin, I captured and tagged two adult females encountered during visual encounter surveys
while basking along the riverbank and three near an unpaved parking area next to the river. In the
southern basin, I captured and tagged two individuals near a gravel pit containing a manmade temporary
pond ~150 m from the river. In 2022, I captured and tagged an additional three individuals in the southern
basin between May 5 and May 23. I relocated all 10 individuals an average of 24 times each (range = 22 -
25) from May 3 through September 24 (Figure 2.3).
2.3.2 Seasonal Movements
Mean weekly Euclidean distance from flowing water in both basins revealed that individuals
remained closest to the river during emergence/pre-nesting (x̅ = 30.3 m, SE = 8.5) and pre-brumation (x̅ =
6.0 m, SE = 4.4) periods (Table 2.1). These movements corresponded with travel along river corridors and
use of rivers for thermal refuge during periods of low air temperatures. During nesting, mean weekly
distance from the river averaged 90.7 m (SE = 16.0, Table 2.1). Movements during this period displayed
the most individual variability as females deposited eggs and then rapidly dispersed into riparian and
45
upland areas between mid-June and early-July. Mean weekly distances from the river were largest during
post-nesting (x̅ = 143.2 m, SE = 5.4), when females established summer home ranges in upland and
riparian areas adjacent to the river (Table 2.1). During this time, females were recorded returning to the
river infrequently. I found that 95% of all wood turtle relocations were within 326.3 m of flowing water
and 237.9 m of all waterbodies for both basins (Table 2.2).
Between basins, mean weekly Euclidean distance from flowing water was >128 m during post-
nesting (Table 2.3). However, mean weekly distances from flowing water were lower in the northern basin
during pre-nesting and nesting periods compared to the southern basin (Figure 2.4A). In the southern
basin, two females were relocated near a temporary pond in a gravel pit created by private forest
managers to source road building material ~150 m from the river. This association with the pond resulted
in higher mean weekly distances from flowing water in the southern basin (Figure 2.4A). Therefore, I
recalculated mean weekly distances of wood turtles in the southern basin from all waterbodies, including
still waterbodies (i.e., temporary ponds, oxbow lakes, backwaters), which resulted in mean weekly
distance estimates during pre-nesting and nesting periods that were more similar for both basins (Figure
2.4B).
I found no significant difference in mean weekly distances from flowing water between basins
throughout the active season (Mann-Whitney U Test, U = 130, P = 0.14, n = 19). Mean weekly distances
from flowing water between basins by activity period did not differ during pre-nesting (U = 4, P = 0.10, n =
5) or nesting (U = 2, P = 0.11, n = 4). However, I found a difference between basins during post-nesting (U
= 7, P = <0.01, n = 7), with wood turtles relocated further from flowing water in the southern basin. Mean
weekly distances from flowing water were significantly larger compared to distances from all waterbodies
in the southern basin during nesting (U = 16, P = 0.03, n = 4) and post-nesting (U = 49, P = <0.01, n = 7).
2.3.3 Home Ranges
For stream range, I found that female wood turtles used larger stretches of river in the northern
basin (Mann-Whitney U, U = 57, P = 0.04). For individuals tracked both years (n = 7), stream ranges were
similar in the northern basin but showed more variation in the southern basin (Figure 2.5A). Mean stream
range size of females in the northern basin was 3,786.7 m (SE = 947.3) and 1,127.8 m in the southern
basin (SE = 518.9) (Table 2.4). Inter-year differences in stream ranges tended to occur during the nesting
period and were likely associated with nest-seeking behavior.
For estimating the annual 95% kernel terrestrial home range (i.e., Brownian bridge kernel), I used
an average of 15 relocation points per individual per year (SD = 3.5, range = 8 – 23) (Figure 2.6). I found
that on average, individual home range sizes did not differ by year (Mann-Whitney U, U = 44, P = 0.41, n =
46
7), and there was no difference between basins (Figure 2.5B). Mean 95% terrestrial home range size of
females in the northern basin was 3.6 ha (SE = 0.8) and 2.5 ha in the southern basin (SE = 0.8) (Table 2.5).
Annual overlap between kernel terrestrial home ranges of females tracked both years (n = 7) showed high
site fidelity, with a mean percent overlap of 59.5% (SD = 0.2, range = 30.4% - 83.3%).
For estimating 95% MCP home ranges, I used all relocations resulting in an average of 25 points
per individual per year (SD = 2.25) (Figure 2.6). I found larger 95% MCP home ranges in the northern basin
compared to the southern basin (Mann-Whitney U, U = 68, P = <0.01). 95% MCP home ranges in the
northern basin averaged 31.5 ha (SE = 8.0), compared to 4.1ha (SE = 6.1) in the southern basin (Table 2.6).
2.4 Discussion
Wood turtles are a species of growing conservation concern throughout their range and
understanding their spatiotemporal ecology in working forest landscapes can inform more effective
management recommendations. This is especially true in Michigan and the Upper Great Lakes Region
where large areas of forested lands are privately managed for timber production. For two watershed
basins in the western Upper Peninsula of Michigan that represented different environmental conditions
and forest management histories, I found that 1) female wood turtles traveled similar distances from
flowing water, though still waterbodies close to an occupied river influenced these distances; 2) stream
range and 95% MCP home range estimates were larger in the northern basin, but 95% kernel terrestrial
home ranges were similar; and 3) distance from flowing water and 95% kernel terrestrial home ranges
were comparable to other areas in the Upper Great Lakes Region, but stream range and 95% MCP home
range were larger. Collectively, these results provide insights into how wood turtles and forest
management can co-exist in working forest landscapes.
Previous studies of wood turtle movements in the Upper Great Lakes Region suggest that adult
females typically remain within 120 – 252 m of flowing water during the active season, though distances
over 500 m were observed for some individuals (Harding and Bloomer 1979, Remsburg et al. 2006, Brown
et al. 2016, Thompson et al. 2018). In this study, 95% of adult female wood turtle relocations were within
326 m of flowing water, and the farthest distance observed was 382 m. In my study, most adult female
wood turtles in both basins remained within 58 m of flowing water during pre-nesting and pre-brumation
periods and then moved farther from flowing water during nesting and post-nesting. However, I
documented that two female wood turtles in the southern basin took advantage of a temporary pond
that formed during the spring in a gravel pit ~150 m (Euclidean distance) from the river, immediately
following emergence from brumation, presumably to seek thermal shelter during the pre-nesting and
nesting periods. Subsequently, these two individuals moved further away from flowing water during pre-
47
nesting, nesting and post-nesting, explaining why my distance estimates were larger than other studies.
This study is potentially one of the first to document how still waterbodies near occupied rivers influence
wood turtle space use patterns leading up to the summer active season. To further substantiate this
claim, in the northern basin where few still waterbodies were present, 95% of all female relocations were
within 238 m of flowing water (max = 326 m), which aligned more with estimates from other studies (e.g.,
Remsburg et al. 2006, Brown et al. 2016). This suggests that, although wood turtles are closely tied to
flowing water, still waterbodies could have a large influence on seasonal movements in watershed basins
where they occur.
Use of still waterbodies by wood turtles during the active season is considered rare but may have
a larger influence on their spatial ecology than previously thought (Quinn and Tate 1991, Brown et al.
2016). Based on my findings, I suggest that this topic warrants further investigation because of potential
implications for wood turtle conservation and forest management. If still waterbodies enable wood
turtles to move farther from flowing water in some areas, protective buffers based solely on flowing
water may not adequately protect terrestrial individuals. Conversely, assuming that wood turtles use all
still waterbodies could be erroneous, resulting in unnecessarily large protective buffers. Therefore,
understanding the relationship between wood turtle movements and still waterbodies close to occupied
rivers is an essential research need.
Estimates of stream range were not available for much of the Upper Great Lakes Region, but
mean stream range for adult female wood turtles was 520 – 750 m in northeast Iowa (Otten 2017, Otten
et al. 2021), 805 m in western Massachusetts (Jones and Willey 2020), 659 m in western Vermont (Parren
2013), and 754 m in western Massachusetts and north-central New Hampshire (Jones 2009). The mean
stream range of all female wood turtles across both years of this study was 2,457 m, farther than
documented stream ranges for female wood turtles in other parts of their range. It is unknown why
stream ranges in my study were larger, but comparison of mean stream ranges between basins in my
study shed some light on the subject.
Adult female wood turtles in the northern basin had a larger mean stream range than those in the
southern basin, likely due to differences in availability of high-quality nesting habitat (i.e., areas with well
drained sand and gravel, open canopies and within 200 m of occupied rivers) (Buech et al. 1997). In the
northern basin, natural nesting areas were limited given general prevalence of steep densely forested
banks throughout the basin. Adult female wood turtles in this basin used a cleared parking area adjacent
to the river, which offered open canopy and bare sandy soil appropriate for nesting (Buech et al. 1997). To
reach this area, females traveled 675 m – 5283 m from their brumation sites in the river. This extended
48
use of the river was also documented by Latham et al. (2022) who observed females in an occupied river
in Maine traveling greater distances to reach nesting beaches in areas where naturally occurring nesting
areas were limited than females occupying rivers where nesting areas were more abundant. I observed
several staging and actively nesting wood turtles at this site along with western painted turtles
(Chrysemys picta belli) and common snapping turtles (Chelydra serpentina). Nests were heavily predated,
though there was evidence of successful hatching (tracks) in the fall. In the southern basin, there was an
abundance of natural nesting habitat that consisted of open sandy banks along the edges of the river
throughout the basin. Adult female wood turtles in the southern basin nested closer to their brumation
sites and were not congregated like females were observed doing in the northern basin.
Mean stream range lengths for adult female wood turtles varied by individual but were similar
among years since females generally used the same brumation and nesting sites throughout the course of
study. However, one female in each basin traveled further along the river in one year during the nesting
season. In the northern basin, a female traveled 2,124 m farther downriver to a patch of gravel on the
riverbank in 2021 and a female in the southern basin traveled 5,670 m downriver to a sandy bank in 2022.
This suggests some plasticity in nest site fidelity and that nest site selection can have a significant
influence on adult female wood turtle stream range.
I found that 95% kernel terrestrial home ranges varied by individual and year but differences
between basins were negligible. However, three females with the largest terrestrial home ranges utilized
mature stands of trees with denser canopies (i.e., mature sugar maple (Acer saccharum) select cut in
2004, larch plantation established in 1995). Larger home range sizes may correspond to poorer habitat
quality where resources are limited (Kaufmann 1992, Remsburg et al. 2006, Saumure 2004, Arvisais et al.
2002), suggesting that these closed canopy conditions were lower quality for wood turtles. Forest
management practices that result in a mosaic of forest structures that juxtapose open canopy with
mature forest to create more structurally diverse edges could be beneficial to wood turtles (Kaufmann
1992, Tingley and Herman 2008, Saumure 2004, Compton et al. 2002). Understanding how harvest
methods, stand sizes, and wood turtle home ranges interact is a subject that warrants further study.
My mean 95% MCP estimates for all female wood turtle across basins and years (18.0 ha), was
greater than mean estimates for females in northern Wisconsin (100% MCP = 0..82 ha; Ross et al. 1991)
but smaller than those in the northern Lower Peninsula of Michigan (100% MCP = 30.2 ha; Remsburg et al
2006) and Ontario (100% MCP = 24.3 ha; Quinn and Tate 1991). Thompson et al. (2018) found that mean
female wood turtle 95% MCP home range varied (6.4 ha and 21.6 ha) between two independent
watershed basins in northern Ontario, though there was no explanation provided to account for this. I
49
suspect that these differences partially stem from potential bias in using MCPs (i.e., risk of over-estimating
space use) and in differing environmental conditions at the watershed scale (e.g., availability of nesting
areas). I found that mean 95% MCP estimates in the northern basin were larger than those in the
southern basin but noted that greater use of the river corridor (i.e., stream range in this study) strongly
influenced those results. As a result, 95% MCP estimates from my study included large areas of terrestrial
areas not used by radio-tagged wood turtles. Relying on 95% MCPs alone to estimate home range size
may provide inaccurate space-use estimates.
Female wood turtles in this study displayed high site fidelity, with a mean 95% kernel terrestrial
home range overlap of 59% from year to year with most females returning to the same nesting area,
terrestrial home range, and brumation sites each year. Similar to observations of site fidelity by Harding
and Bloomer (1979), one female in my study was observed basking on the same stump located on the
edge of an old haul road in 2021 and 2022 (Figure 2.7). I also observed similar overland travel routes used
by individuals between the river and their terrestrial home ranges. One female occupied an area managed
using a shelterwood treatment the year before, but it was unknown if she frequented that area
previously. I hypothesize that this kind of disturbance during the winter inactive season (mid-October –
late April) would have little effect on site fidelity, but this needs more study.
In conclusion, this study provided insights into the spatial ecology of adult female wood turtles in
the western Upper Peninsula of Michigan and its implications for forest management in the state and
across the Upper Great Lakes Region. While 95% kernel terrestrial home range was similar across study
basins, seasonal movement from flowing water and stream range differed due to resource availability.
Adult female wood turtles in the southern basin took advantage of still waterbodies to move farther
across the landscape and those in the northern basin traveled farther along the river to reach limited
nesting areas. Compared to other studies from the Upper Great Lakes Region, female wood turtles made
similar seasonal movements from flowing water and 95% MCP estimates fell within the range reported by
others. Overall, while there were similarities between my results and those of others, there was
substantial variation between populations at the watershed scale. Therefore, to effectively conserve wood
turtles in working forests, more work is needed to understand how landscape features and resource
availability at the watershed scale affect space use.
2.4.1 Management Implications
One of the most common recommendations provided to conserve wood turtles in working forests
is implementing riparian buffers that limit machine intensive harvests around occupied rivers. NAFO
member companies in my study area maintain a 30.5 m wide riparian buffer, or riparian management
50
zone, around flowing water to protect water and aquatic habitat quality (MDNR 2023). However, to
conserve wood turtles, riparian buffers around occupied rivers should reflect seasonal movements from
flowing water made by wood turtles, using local or regional data to inform their widths throughout the
active season. Findings from my study indicated that landscape features, such as still waterbodies at the
watershed scale, may influence movement distances from flowing water. Therefore, implementing a
buffer that extends to the maximum observed distance from flowing water throughout the active season
may account for variation between watersheds across a broader area. In my study area, a single buffer
that restricts use of heavy machinery 400 m from flowing water from late April – mid-October would
protect 100% of female wood turtles. A static 400 m buffer is more restrictive to active forest
management, so adjusting buffer widths to reflect seasonal differences in distances from flowing water is
suggested. In my study, buffers that would encompass 95% of female wood turtle relocations during the
pre-nesting (170 m), nesting (330 m), post-nesting (360 m) and pre-brumation (60 m) may be less
restrictive. Riparian buffers need not be utilized during the inactive season from mid-October – late April
or if there is persistent snow cover.
Stream range length can also inform the length of riparian buffers that limit machine-intensive
harvests during the active season. These buffers are often extended a certain distance upstream and
downstream from confirmed wood turtle observations. In the western Upper Peninsula of Michigan, using
the stream range of wood turtles from other parts of their range may not provide adequate protection
given that the mean stream range of female wood turtles in this area are larger. In my study area, buffers
should extend along the river approximately 2.5 km on either side of wood turtle observations. I found
stream range length in this study was strongly influenced by nest-seeking behavior, with larger stream
ranges in the northern basin suggesting that suitable nesting areas were limited. Basins in which stream
range is longer may be candidates for nest site restoration, creation, and management; a proven, effective
way to conserve populations (Buhlmann and Osborn 2011).
Three of my female wood turtles occupied larger 95% kernel terrestrial home ranges in areas of
mature forest with dense canopies. Apparently, female wood turtles occupying these forests ranged
farther to reach areas of open canopy, focusing their use on old haul roads. These roads provided open
canopy, dense ground cover and forage in the form of fruit bearing plants like wild red raspberry (Rubus
idaeus). Using haul roads within the range of female wood turtle movement from flowing water should be
avoided in the active season. Closing haul roads to motorized vehicle traffic after a harvest has been
completed can eliminate the risk of mortality from vehicle strikes near occupied rivers, especially in areas
where female wood turtles may regularly seek them out.
51
TABLES AND FIGURES
Figure 2.1. Rivers occupied by wood turtles in northern (left) and southern (right) basins in the western
Upper Peninsula, Michigan, USA, 2021-2022. These watershed basins were approximately 71 km apart and
differed in adjacent vegetation structure and composition, topography, hydrology and river substrate.
52
Figure 2.2. Home ranges for adult female wood turtles were estimated using three techniques: stream
range, 95% kernel terrestrial home range (Brownian bridge kernel method), and 95% minimum convex
polygons (MCP). Stream range (m) was measured between the two farthest relocations within 20 m of the
river along the rivers path to portray riparian space use during pre-nesting and pre-brumation activity
periods. The 95% kernel terrestrial home range (ha) was estimated from a utilization distribution based on
relocations >20 m from the river during the post-nesting period (July – August) to portray terrestrial space
use. 95% MCP (ha) was calculated using all relocations to facilitate comparisons with other studies.
53
Figure 2.3. Relocations for each adult female wood turtle in northern (top panel) and southern (bottom
panel) watershed basins in the western Upper Peninsula, Michigan, USA, 2021-2022. Individuals were
relocated 1 - 2 times per week through the active season (May - October) using VHF radio telemetry.
54
A
B
Figure 2.4. Mean weekly Euclidean distances (m) and standard errors of adult female wood turtles from
flowing water (A) and from all waterbodies (B) in two watershed basins in the western Upper Peninsula,
Michigan, USA, 2021-2022. Five individuals in each basin were relocated 1 - 2 times per week through the
active season (May - October).
55
A
B
Figure 2.5. Stream range distance (m) (A) and 95% kernel terrestrial home range area (ha) (B) of adult
female wood turtles in the western Upper Peninsula, Michigan, USA, 2021-2022. Stream range was
measured between the two farthest relocations within 20 m of the river along the rivers path to reflect
riparian space use. 95% kernel terrestrial home range was estimated based on 95% Brownian bridge
kernels for relocations >20 m from flowing water during post-nesting (July – August) to reflect terrestrial
space use. Open points indicate mean values, horizontal bars indicate median values, shaded areas
represent the inter-quartile range, and vertical bars indicate the 1.5 inter-quartile range.
56
Figure 2.6. 95% kernel terrestrial home range and 95% minimum convex polygon (MCP) home ranges of
adult female wood turtles in the western Upper Peninsula, Michigan, USA, 2022. 95% kernel terrestrial
home ranges were estimated based on 95% Brownian bridge kernels for relocations >20 m from flowing
water during post-nesting (July – August), and 95% MCPs were estimated using all relocations during the
active season (May – mid-October) in 2022.
57
Figure 2.7. F25, an adult female wood turtle in the western Upper Peninsula of Michigan observed
basking on the same rotten stump on the edge of a closed haul road in 2021 (left panel) and 2022 (right
panel).
58
Table 2.1. Mean Euclidean distance (m) from flowing water for all female wood turtles by activity period
in the western Upper Peninsula, Michigan, USA, 2021-2022.
Activity Period
Pre-nesting
Nesting
Post-nesting
Pre-brumation
Mean Distance (m)
30.3
90.7
143.2
6.0
SE
8.5
16.0
5.4
4.4
Range (m)
0 - 183.9
0 - 381.6
0 - 373.0
0 - 58.9
59
Table 2.2. Euclidean distance (m) within which 95% of relocations of adult female wood turtles occurred
from flowing water and all waterbodies (i.e., temporary ponds, oxbow lakes, backwaters) by activity period
in the western Upper Peninsula, Michigan, USA, 2021-2022.
Activity Period
Pre-nesting
Nesting
Post-nesting
Pre-brumation
Distance (m) from
River
165.5
325.9
360.0
58.4
All Waterbodies
54.9
241.6
237.9
58.4
60
Table 2.3. Mean Euclidean distance (m) from flowing water and all waterbodies of female wood turtles in
two watershed basins by activity period in the western Upper Peninsula, Michigan, USA, 2021-2022.
* indicates distances calculated from all waterbodies (i.e., temporary ponds, oxbow lakes, backwaters).
Activity Period
Pre-nesting
Nesting
Post-nesting
Pre-brumation
North
13.2 (3.8, 0.0 - 114.5)
72.9 (20.8, 0.0 - 325.9)
128.8 (8.5, 0.0 - 248.8)
6.8 (5.6, 0.0 - 58.9)
Mean (SE, range) distance (m)
South
72.3 (21.6, 0.0 - 183.9)
118.0 (9.5, 0.0 - 381.6)
162.7 (7.7, 0.0 - 373.0)
5.1 (2.9, 0.0 - 36.8)
South*
9.9 (3.0, 0.0 - 40.4)
51.6 (12.2, 0.0 - 193.4)
88.6 (5.9, 0.0 - 282.0)
5.3 (2.9, 0.0 - 36.8)
61
Table 2.4. Stream ranges (m) of 10 adult female wood turtles by basin, individual, and year in the western
Upper Peninsula, Michigan, USA, 2021-2022. Stream range was measured between the two farthest
relocations <20 m of the river along the rivers path to reflect riparian space use.
Basin
North
South
Both
Wood Turtle ID
F09
F12
F25
F27
F29
Intra-year Mean
F30
F32
F41
F50
F52
Intra-year Mean
Intra-year Mean
2021
5229.5
674.7
2855.7
7190.5
3960.6
3982.2
17.5
193.5
-
-
-
105.5
2874.6
2022
5283.3
754.2
2890.6
5066.2
3962.3
3591.3
0.0
5863.2
1183.5
420.2
998.3
1693.0
2642.2
Annual Mean
5256.4
714.4
2873.2
6128.3
3961.4
3786.7
8.8
3028.3
1183.5
420.2
998.3
1127.8
2457.3
62
Table 2.5. Terrestrial home ranges (ha) of 10 adult female wood turtles by basin, individual, and year in
the western Upper Peninsula, Michigan, USA, 2021-2022. Estimated based on 95% Brownian bridge
kernels for relocations >20m from flowing water recorded during the Post-nesting (July – August).
Basin
North
South
Both
Wood Turtle ID
F09
F12
F25
F27
F29
Intra-year Mean
F30
F32
F41
F50
F52
Intra-year Mean
Intra-year Mean
2021
2.3
3.1
4.8
0.3
4.0
2.9
4.0
2.5
-
-
-
3.3
3.0
2022
6.7
2.0
3.4
2.2
7.0
4.3
2.9
1.7
1.2
0.9
5.1
2.3
3.3
Annual Mean
4.5
2.5
4.1
1.3
5.5
3.6
3.5
2.1
1.2
0.9
5.1
2.5
3.1
63
Table 2.6. Minimum Convex Polygon (MCP) home ranges (ha) of 10 adult female wood turtles by basin,
individual, and year in the western Upper Peninsula, Michigan, USA, 2021-2022. Estimates represent 95%
MCPs using all relocations to facilitate comparisons with other studies.
Basin
North
South
Both
WOTU ID
F09
F12
F25
F27
F29
Intra-year Mean
F30
F32
F41
F50
F52
Intra-year Mean
Intra-year Mean
2021
50.5
6.0
25.7
55.2
40.2
35.5
4.9
2.9
-
-
-
3.9
26.5
2022
62.7
5.6
12.3
15.7
40.7
27.4
2.0
4.4
1.0
2.3
12.3
4.4
15.9
Annual Mean
56.6
5.8
19.0
35.4
40.5
31.5
3.5
3.6
1.0
2.3
12.3
4.1
18.0
64
LITERATURE CITED
Arvisais, M., J. C. Bourgeois, E. Lévesque, C. Daigle, D. Masse, and J. Jutras. 2002. Home range and
movements of a wood turtle (Clemmys insculpta) population at the northern limit of its range.
Canadian Journal of Zoology 80:402–408.
Brown, D. J., M. D. Nelson, D. J. Rugg, R. R. Buech, and D. M. Donner. 2016. Spatial and Temporal Habitat-
Use Patterns of Wood Turtles at the Western Edge of Their Distribution. Journal of Herpetology
50:347–356.
Buech, R. R., L. G. Hanson, and M. D. Nelson. 1997. Identification of wood turtle nesting areas for
protection and management. Proceedings of the 1997 Conservation, Restoration, and
Management of Tortoises and Turtles: An International Conference. New York Turtle and Tortoise
Society, New York, New York. 494 pp.
Buhlmann, K. A., and C. P. Osborn. 2011. Use of an Artificial Nesting Mound by Wood Turtles (Glyptemys
insculpta): A Tool for Turtle Conservation. Northeastern Naturalist 18(3):315-334.
Calenge, C. 2006. The package adehabitat for the R software: a tool for the analysis of space and habitat
use by animals. Ecological Modelling 197:516-519.
Compton, B. W., J. M. Rhymer, and M. McCollough. 2002. Habitat selection by wood turtles (Clemmys
insculpta): An application of paired logistic regression. Ecology 83:833–843.
Congdon, J. D., A. E. Dunham, and R. C. V. L. Sels. 1994. Demographics of common snapping turtles
(Chelydra serpentina): Implications for conservation and management of long-lived organisms.
Integrative and Comparative Biology 34:397–408.
Ernst, C. H., and J. E. Lovich. 2009. Turtles of the United States and Canada. Second Edition. Johns Hopkins
University Press, Baltimore, Maryland, US. 434 pp.
Giese, C. L. A., D. N. Greenwald, and T. Curry. 2012. Petition to list 53 amphibians and reptiles in the
United States as threatened or endangered species under the Endangered Species Act. Petition to
the Secretary of the Interior of the United States. Center for Biological Diversity, Tucson, Arizona,
USA. 454 pp.
Great Lakes Integrated Sciences and Assessments [GLISA]. 2023a. Historical Climatology: Marquette,
Michigan. . Accessed 04 October 2023.
Great Lakes Integrated Sciences and Assessments [GLISA]. 2023b. Historical Climatology: Iron Mountain
Kingsford WWTP, MI. < https://glisa.umich.edu/station/iron-mt-kingsford-wwtp/ >. Accessed 04
October 2023.
Harding, J. H., and T. J. Bloomer. 1979. The wood turtle, Clemmys insculpta... a natural history. Bulletin of
the New York Herpetological Society 15:9–26.
Hillard, S. 2021. Minnesota’s Forest Resources 2019. Minnesota Department of Natural Resources,
Division of Forestry, Resource Assessment Office, Grand Rapids, Minnesota, USA. 135 pp.
65
Jones, M. T. 2009. Spatial ecology, population structure, and conservation of the wood turtle, Glyptemys
insculpta, in central new England. Dissertation, University of Massachusetts Amherst, Amherst,
Massachusetts, US. 266 pp.
Jones, M. T., and L. L. Willey. 2020. Cross-watershed dispersal and annual movement in adult wood
turtles (Glyptemys insculpta). Herpetological Review 51:208–211.
Kaufmann, J. H. 1992. Habitat use by wood turtles in central Pennsylvania. Journal of Herpetology
26:315–321.
Kurtz M., Cassandra, T. A. Albright, B. D. Anderson, B. J. Butler, C. Buntrock, D. Buckler, S. J. Crocker, M. D.
Nelson, A. L. Diss-Torrance, T. C. Goff, B. J. Hemmer, R. S. Morin, R. J. Piva, S. A. Pugh, R. I.
Riemann, S. L. Schanning, B. F. Walters, C. W. Woodall. 2022. Wisconsin forests 2019: summary
report. U.S. Department of Agriculture, Forest Service, Northern Research Station, Madison,
Wisconsin, USA. 24 pp.
LANDFIRE. 2022. Existing vegetation type layer, LANDFIRE 2.3.0, U.S. Department of the Interior,
Geological Survey, and U.S. Department of Agriculture. . Accessed
20 October 2022.
Latham, S. R., A. P. K. Sirén, and L. R. Reitsma. 2022. Space use and resource selection of Wood Turtles
(Glyptemys insculpta) in the northeastern part of its range. Canadian Journal of Zoology 101:20–
31.
Mann, H. B. and D. R., Whitney. 1947. On a test of whether one of two random variables is stochastically
larger than the other. The Annals of Mathematical Statistics 50-60.
Michigan Department of Natural Resources [MDNR] 2023. Michigan forestry best management practices
for soil and water quality. Forest Resources Division, Michigan Department of Natural Resources,
Lansing, Michigan, USA. 70 pp.
Minta, S.C., 1992. Tests of spatial and temporal interaction among animals. Ecological Applications 2:178-
188.
Otten, J. G. 2017. Factors influencing wood turtle (Glyptemys insculpta) home range size in Iowa: A
comparison between suburban and rural populations. Thesis, University of Northern Iowa, Cedar
Falls, Iowa, USA. 163 pp.
Otten, J. G., A. C. Hulbert, S. W. Berg, and J. W. Tamplin. 2021. Home Range, Site Fidelity, and Movement
Patterns of the Wood Turtle (Glyptemys insculpta) at the Southwestern Edge of Its Range.
Chelonian Conservation and Biology 20:231–241.
Parren, S. G. 2013. A twenty-five year study of the wood turtle (Glyptemys insculpta) in Vermont:
Movements, behavior, injuries, and death. Herpetological Conservation and Biology 8:176–190.
Poudel, J. 2022. Forest products industries’ economic contributions: Michigan, 2019. Forest Resources
Division, Michigan Department of Natural Resources, Lansing, Michigan, USA. 28 pp.
66
Quinn, N. W. S., and D. P. Tate. 1991. Seasonal Movements and Habitat of Wood Turtles (Clemmys
insculpta) in Algonquin Park. Journal of Herpetology 25:217–220.
R Core Team. 2023. R: A language and environment for statistical computing. R Foundation for Statistical
Computing, Vienna, Austria. Accessed 1 January 2023.
Remsberg, A. J., T. L. Lewis, P. W. Huber, and K. A. Asmus. 2006. Home ranges of wood turtles (Glyptemys
insculpta) in Northern Michigan. Chelonian Conservation and Biology 5:42–47.
Ross, D. A., K. N. Brewster, R. K. Anderson, N. Ratner, and C. M. Brewster. 1991. Aspects of the Ecology of
Wood Turtles, Clemmys insculpta, in WI. The Canadian Field Naturalist 105:363–367.
Row, J. R., and G. Blouin-Demers. 2006. Kernels are not accurate estimators of home-range size for
herpetofauna. Copeia 2006:797–802.
Saumure, R. A. 2004. Spatial ecology and conservation of the North American wood turtle (Glyptemys
insculpta) in a fragmented agri-forest landscape. Dissertation, McGill University, Montreal,
Quebec, CA. 148 pp.
Schaetzl, R. J., H. Enander, M. D. Leuhmann, D. P. Lusch, C. Fish, M. Bigsby, M. Steigmeyer, J. Guasco, C.
Forgacs, and A. Pollyea. 2013. Physiographic map of Michigan.
. Accessed 14
September 2020.
Seaber, P. R., F. P. Kapinos, and G. L. Knapp. 1987. Hydrologic unit maps. United States Geological Survey
Water-Supply Paper 2294. US Geological Survey: Anchorage, Alaska, USA. 66 pp.
Silva, I., M. Crane, B. M. Marshall, and C. T. Strine. 2020. Reptiles on the wrong track? Moving beyond
traditional estimators with dynamic Brownian Bridge Movement Models. Movement Ecology
8:1–14.
Sweeten, S. E. 2007. Home Range, Hibernacula Fidelity, and Best Management Practices for Wood Turtles
(Glyptemys insculpta) in Virgina. Thesis, James Madison University, Harrisonburg, Virginia, USA.
47 pp.
Thompson, D. G., T. Swystun, J. Cross, R. Cross, D. Chartrand, and C. B. Edge. 2018. Fine-and coarse-scale
movements and habitat use by Wood Turtles (Glyptemys insculpta) based on probabilistic
modeling of radiotelemetry and GPS-telemetry data. Canadian Journal of Zoology 96:1153–1164.
Tingley, R., and T. B. Herman. 2008. The effects of agriculture and forestry on the distribution,
movements and survival of wood turtles in an intensively managed landscape. A report for the
Nova Scotia Habitat Conservation Fund. 57 pp.
United States Department of Agriculture [USDA] Forest Service. 2024. Forest inventory and analysis one-
click factsheet. <
https://public.tableau.com/views/FIA_OneClick_V1_2/StateSelection?%3AshowVizHome=no>.
Accessed 4 February 2024.
67
van Dijk, P.P. and J. Harding. 2011. Glyptemys insculpta. The IUCN Red List of threatened species 2011.
. Accessed 3 May 2023.
Wickham H., R. François, L. Henry, K. Müller, D. Vaughan. 2023. dplyr: A Grammar of Data Manipulation.
. Accessed 19 October 2023.
Wilcoxon, F. 1945. Some uses of statistics in plant pathology. Biometrics Bulletin 1:41-45.
68
APPENDIX 2A: WOOD TURTLE RELOCATION DATA SHEET
Figure 2A.1. Sample datasheet used to record data at each turtle relocation point.
69
APPENDIX 2B: TRANSMITTER ATTACHMENT
Figure 2B.1. I attached Holohil RI-2B (VHF) radio transmitters (Holohil, Carp, Ontario, CA) weighing 10 g-15
g (<5% total body weight) to the fourth left or right costal scute using a J-B Weld WaterWeld 2-part epoxy
(J-B Weld, Sulphur Springs, TX, US), with the antenna trailing behind to reduce the risk of entanglement.
Sand was rubbed into the epoxy as it dried to camouflage the bright white epoxy. Epoxy was allowed to
cure for 30 min before release.
70
Table 2C.1. Each turtle was captured by hand, aged, weighed, measured, and given a unique number for identification using a shell notching
system.
APPENDIX 2C: TELEMETERED WOOD TURTLE BIOMETRICS
Basin
North
South
Wood turtle
ID
F09
F12
F25
F27
F29
F30
F32
F41
F50
F52
Capture
Date
5/13/2021
5/11/2021
5/20/2021
5/20/2021
5/21/2021
6/9/2021
6/9/2021
5/5/2022
5/23/2022
5/23/2022
Annuli
Count Wear (%) Weight (g)
9
13
0
0
0
6
19
19
19
21
90
75
100
100
100
95
15
0
10
5
1490.0
1340.0
1270.0
1330.0
1430.0
930.0
1150.0
1040.0
950.0
1200.0
Annuli Count: Number of visible annuli on the left abdominal scute.
Wear (%): Percentage of wear present on the left abdominal scute.
SCL: Straight carapace length.
SCW: Straight carapace width.
SPL: Straight plastron length.
SPW: Straight plastron width.
SH: Shell height.
SCL (mm)
208.8
201.7
205.0
199.4
201.7
161.4
195.3
185.4
182.3
184.6
SCW (mm)
159.3
153.7
142.9
149.6
147.2
128.4
145.1
141.5
134.2
142.8
SPL (mm)
190.7
188.4
179.6
183.8
188.4
159.5
171.6
164.6
161.7
167.6
SPW (mm)
91.1
93.8
91.1
81.5
88.1
74.4
79.3
80.7
77.2
82.1
SH (mm)
77.2
77.2
76.9
77.9
77.6
67.6
75.7
75.8
68.0
71.1
71
CONCLUSION
In my thesis, I explored occupancy, seasonal movements, and home ranges of wood turtles
(Glyptemys insculpta) in working forests of Michigan’s western Upper Peninsula, USA. My primary
contributions included 1) parameterizing detection and occupancy of wood turtles and relating occupancy
to riparian conditions and recent forest management history, and 2) describing seasonal movements and
estimation of home range size for adult female wood turtles. My study occurred at multiple spatial scales,
including river segments (i.e., 1 km reaches of shoreline; detection probability and occupancy), annual
wood turtle home ranges (quantifying seasonal distances from occupied rivers and estimating riparian
and terrestrial space use), and watershed basins (hereafter: basins; confirming occupancy status and
assessing its relationship to forest management history). Collectively, my results inform survey protocols
that NAFO member companies may use to determine basin occupancy status, and provide information on
seasonal movements and use of terrestrial areas to inform best management practices and reduce
negative effects of forest activities on wood turtles.
In my first chapter, I used a protocol for conducting visual encounter surveys in the Midwest
developed by Brown et al. (2017) to detect wood turtles in 15 basins from 2020 – 2022. I used the
unmarked package (Fiske and Chandler 2011) in program R (R Core Team 2023) to model detection and
occupancy using covariates collected in the field. I also built a recent management history (~17 years)
using time series aerial imagery in Google Earth Engine (Gorelick et al. 2017) and used it to compare
management in occupied and unoccupied basins. I found that use of the Midwest protocol was an
effective method for detecting wood turtles in the spring (May – June), resulting in a detection probability
(p) of 0.67. Occupancy probability (Ψ) was influenced primarily by the proportion of sand and gravel
(nesting substrate) making up the riverbank substrate, with most occupied segments containing >49%
sand and gravel. This indicated that availability of nesting habitat within a 1 km segment of river could be
used to help refine locations of visual encounter surveys used to confirm wood turtle occupancy of a river
segment or basin. Analysis of recent forest management history did not provide compelling evidence of a
relationship between wood turtle occupancy and forest management treatment or time since last
management activity at the basin level. However, I caution that these results came from a region that is
primarily forested with few other human influences affecting wood turtle populations.
In my second chapter, I tracked 10 adult female wood turtles from 2021 – 2022 in two disjunct
basins in the north and south of my study area to quantify seasonal movements and measure distances
from flowing water, stream range lengths, and terrestrial home range sizes throughout the active season
(May – October). I compared these estimates between basins and to those in other studies from the
72
Upper Great Lakes Region. Seasonal movement distances from flowing water differed between basins due
to use of a still waterbody in the southern basin, resulting in longer distances from the river during the
pre-nesting and nesting periods. Distances from flowing water were larger than others documented for
wood turtles throughout the region (Harding and Bloomer 1979, Remsburg et al. 2006, Brown et al. 2016,
Thompson et al. 2018), due to the influence of southern basin turtles using still waterbodies away from
the occupied river early in the active season. Mean stream range length (i.e., stretches of river used by
telemetered turtles) also differed between basins, with turtles in the northern basin using greater
stretches of river to access (and potentially historic) nesting sites. While there were no stream range
estimates available for the Great Lakes Region, mean stream range was larger for females in my study
area than those in Iowa, Massachusetts, Vermont, and New Hampshire (Jones 2009, Parren 2013, Otten
2017, Jones and Willey 2020, Otten et al. 2021). Mean 95% kernel terrestrial home ranges did not differ
between basins, though there was some correlation between larger summer home range sizes and forest
structure, with larger ranges in more contiguous mature forest. Larger home ranges are thought to
correspond with areas of limited resources (i.e., forest openings) as wood turtles need to use more area
to fulfill life requisites (Kaufmann 1992, Remsburg et al. 2006, Saumure 2004, Arvisais et al. 2002). Mean
95% minimum convex polygon (MCP) home ranges that used all locations (aquatic and terrestrial) were
strongly influenced by using the river as a travel corridor, grossly inflating home range size estimates and
including areas not utilized by telemetered wood turtles. My mean 95% MCP estimate was within the
range of others reported in other studies in the Upper Great Lakes Region (Quinn and Tate 1991, Ross et
al. 1991, Remsburg et al 2006, Thompson et al. 2018). These results emphasize the importance of
understanding how basin and regional level factors can influence wood turtle movements and space use,
so that effective wood turtle conservation measures can be implemented.
Overall, the results of my research provide preliminary evidence that wood turtles can persist
with forest management and forest managers can make significant contributions to wood turtle
conservation by implementing science-based management practices. However, more studies on the acute
and long-term effects of active forest management on occupancy and space use in the Upper Great Lakes
Region and across the range of wood turtles are still needed, as is research on population dynamics. This
area of study would benefit from a long-term documentation of wood turtle population trajectories and
demographic structure on working forestlands as well as wood turtle response to forest management at
different intensities and/or in areas of mixed-use.
73
LITERATURE CITED
Arvisais, M., J. C. Bourgeois, E. Lévesque, C. Daigle, D. Masse, and J. Jutras. 2002. Home range and
movements of a wood turtle (Clemmys insculpta) population at the northern limit of its range.
Canadian Journal of Zoology 80:402–408.
Brown, D. J., M. D. Nelson, D. J. Rugg, R. R. Buech, and D. M. Donner. 2016. Spatial and Temporal Habitat-
Use Patterns of Wood Turtles at the Western Edge of Their Distribution. Journal of Herpetology
50:347–356.
Brown, D. J., M. M. Cochrane, and R. A. Moen. 2017. Survey and analysis design for wood turtle
population monitoring. Journal of Wildlife Management 81:868–877.
Fiske, I., and R. Chandler. 2011. unmarked: An R Package for fitting hierarchical models of wildlife
occurrence and abundance. Journal of Statistical Software 43: 1-23.
Gorelick, N., M. Hancher, M. Dixon, S. Ilyushchenko, D. Thau, and R. Moore. 2017. Google Earth Engine:
Planetary-scale geospatial analysis for everyone. Remote Sensing of Environment 202:18-27.
Harding, J. H., and T. J. Bloomer. 1979. The wood turtle, Clemmys insculpta... a natural history. Bulletin of
the New York Herpetological Society 15:9–26.
Jones, M. T. 2009. Spatial ecology, population structure, and conservation of the wood turtle, Glyptemys
insculpta, in central new England. Dissertation, University of Massachusetts Amherst, Amherst,
Massachusetts, US. 266 pp.
Jones, M. T., and L. L. Willey. 2020. Cross-watershed dispersal and annual movement in adult wood
turtles (Glyptemys insculpta). Herpetological Review 51:208–211.
Kaufmann, J. H. 1992. Habitat use by wood turtles in central Pennsylvania. Journal of Herpetology
26:315–321.
Otten, J. G. 2017. Factors influencing wood turtle (Glyptemys insculpta) home range size in Iowa: A
comparison between suburban and rural populations. Thesis, University of Northern Iowa, Cedar
Falls, Iowa, USA. 163 pp.
Otten, J. G., A. C. Hulbert, S. W. Berg, and J. W. Tamplin. 2021. Home Range, Site Fidelity, and Movement
Patterns of the Wood Turtle (Glyptemys insculpta) at the Southwestern Edge of Its Range.
Chelonian Conservation and Biology 20:231–241.
Parren, S. G. 2013. A twenty-five year study of the wood turtle (Glyptemys insculpta) in Vermont:
Movements, behavior, injuries, and death. Herpetological Conservation and Biology 8:176–190.
Quinn, N. W. S., and D. P. Tate. 1991. Seasonal Movements and Habitat of Wood Turtles (Clemmys
insculpta) in Algonquin Park. Journal of Herpetology 25:217–220.
R Core Team. 2023. R: A language and environment for statistical computing. R Foundation for Statistical
Computing, Vienna, Austria.
74
Remsberg, A. J., T. L. Lewis, P. W. Huber, and K. A. Asmus. 2006. Home ranges of wood turtles (Glyptemys
insculpta) in Northern Michigan. Chelonian Conservation and Biology 5:42–47.
Ross, D. A., K. N. Brewster, R. K. Anderson, N. Ratner, and C. M. Brewster. 1991. Aspects of the Ecology of
Wood Turtles, Clemmys insculpta, in WI. The Canadian Field Naturalist 105:363–367.
Saumure, R. A. 2004. Spatial ecology and conservation of the North American wood turtle (Glyptemys
insculpta) in a fragmented agri-forest landscape. Dissertation, McGill University, Montreal,
Quebec, CA. 148 pp.
Thompson, D. G., T. Swystun, J. Cross, R. Cross, D. Chartrand, and C. B. Edge. 2018. Fine-and coarse-scale
movements and habitat use by Wood Turtles (Glyptemys insculpta) based on probabilistic
modeling of radiotelemetry and GPS-telemetry data. Canadian Journal of Zoology 96:1153–1164.
75