RESOURCE UTILIZATION AND MOVEMENTS OF AMERICAN MARTEN (MARTES AMERICANA) IN THE
EASTERN UPPER PENINSULA, MICHIGAN
By
Bradford Richard Silet

A THESIS
Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of
Fisheries and Wildlife – Master of Science
2017

ABSTRACT
RESOURCE UTILIZATION AND MOVEMENTS OF AMERICAN MARTEN (MARTES AMERICANA) IN THE
EASTERN UPPER PENINSULA, MICHIGAN
By
Bradford Richard Silet
The goal of my thesis was to quantify fine scale American marten (Martes americana) space use,
resource selection, and movement behaviors in the Eastern Upper Peninsula of Michigan. I had two
main objectives, to: 1) quantify the relationships among stand-level forest inventory data, Euclidean
distances, and marten space use in the East Unit of the Hiawatha National Forest (HNF), and 2) quantify
marten movement behaviors and relate those behaviors to weather factors. In Chapter 1, I describe
marten space use (quantified from 12 GPS-tagged marten) in the form of utilization distributions, and
relate those distributions to typical forest inventory and available GIS data. I found that forest inventory
data, summarized at the stand level, were poor predictors of marten space utilization. I found some
support for site productivity and tree density as positive correlates of resource use. I suggest that at the
stand-level, site productivity is the variable that best integrated multiple forest attributes, thereby
better representing the complex forest structure knowingly used by marten. In Chapter 2, I examine
daily and seasonal movements and correlated these movements to weather variables. I found that
average daily movement rates were 13.28 m/min (SE = 0.48), and that marten were most active during
winter (and least active during fall). My results indicated that marten in the HNF did not follow
predictable daily activity patterns, and that movement rates were best explained by unmeasured
characteristics related to individual marten and season. I found weak support that temperature and
precipitation affected daily movement rates, with movement rates decreasing as temperature
increased, and movement rates increasing on days with more precipitation. My results indicated that
weather is a poor predictor of marten activity and that movements are greatest in winter, likely
increasing their susceptibility to harvest and predation.

ACKNOWLEDGEMENTS

Funding for this research was provided by the Sault Ste. Marie Tribe of Chippewa Indians
through the Great Lakes Restoration Initiative from the Bureau of Indian Affairs. Additional funding
came from the Emma Shore-Thornton Endowment awarded by the Native American Institute at
Michigan State University.
I would like to express my gratitude to my advisor, Dr. Gary Roloff, for the endless guidance and
encouragement he gave to me during this study. Thank you to my committee members Dr. Henry
(Rique) Campa, III and Dr. Robert Montgomery for their support and involvement during the
development of this study and throughout my research. I am grateful for the help and advice from all
my lab mates in the Applied Forest and Wildlife Ecology Lab and to Erin Zimmer, for field work
assistance. The biologists and staff of the Inland Fish and Wildlife Department of the Sault Ste. Marie
Tribe of Chippewa Indians: Eric Clark, John Powell, Joseph Lautenbach, Rusty Aikens, and Aimee Baier
provided their invaluable time and resources to assist with this study. An additional thank you to Dr.
Maria Spriggs, DVM, for her assistance setting up the handling procedures. I would like to thank the
Hiawatha National Forest and Steve Sjogren for their assistance throughout this research. A special
thanks to John Rickley and Jamie Massey for sharing their lifelong knowledge of marten. And finally,
thank you to Carson LaChapelle for trapping assistance and advice throughout this research project.

iii

TABLE OF CONTENTS
LIST OF TABLES ............................................................................................................................................. v
LIST OF FIGURES .......................................................................................................................................... vi
Introduction .................................................................................................................................................. 1
Study Area .................................................................................................................................................... 3
APPENDIX ..................................................................................................................................................... 5
LITERATURE CITED ........................................................................................................................................ 7
Chapter 1 - Resource Utilization of American Marten in the Hiawatha National Forest in the Upper
Peninsula of Michigan ................................................................................................................................ 11
Abstract .................................................................................................................................................. 11
Introduction ............................................................................................................................................ 12
Methods.................................................................................................................................................. 14
Marten Trapping ................................................................................................................................. 14
Utilization Distributions ...................................................................................................................... 16
Modeling ............................................................................................................................................. 16
Results..................................................................................................................................................... 17
Marten Trapping ................................................................................................................................. 17
Collar Performance ............................................................................................................................. 18
Resource Utilization ............................................................................................................................ 18
Discussion ............................................................................................................................................... 20
APPENDIX ............................................................................................................................................... 24
LITERATURE CITED .................................................................................................................................. 36
Chapter 2 – Weather-related Movements of American Marten in the Hiawatha National Forest,
Michigan ..................................................................................................................................................... 43
Abstract .................................................................................................................................................. 43
Introduction ............................................................................................................................................ 44
Methods.................................................................................................................................................. 46
Marten Trapping ................................................................................................................................. 46
Daily Movement .................................................................................................................................. 47
Weather Related Movements ............................................................................................................. 48
Results..................................................................................................................................................... 49
Marten Trapping ................................................................................................................................. 49
Daily Movements ................................................................................................................................ 49
Weather Related Movements ............................................................................................................. 50
Discussion ............................................................................................................................................... 51
APPENDIX ............................................................................................................................................... 54
LITERATURE CITED .................................................................................................................................. 64
Conclusions ................................................................................................................................................. 69

iv

LIST OF TABLES

Table 1.1. Basal area, stand age, diameter at breast height (DBH), site
productivity, and tree density for stands used by American marten on the
Hiawatha National Forest, eastern Upper Peninsula of Michigan, March 2016 –
July 2017.

…………………………….25

Table 1.2. Distances to nearest openings and maintained, high traffic roads
from American marten telemetry locations on the Hiawatha National Forest,
eastern Upper Peninsula of Michigan, March 2016 – July 2017.

………………………….....27

Table 1.3. Generalized linear mixed model averaged coefficients (95%
confidence intervals), by individual marten, for variables used to describe
resource utilization in the Hiawatha National Forest, eastern Upper Peninsula
of Michigan, March 2016 – July 2017. NA denotes a variable that was not
included in the model averaging for individual marten. Bold-face indicates that
the 95% confidence interval did not overlap 0.

……………………………..28

Table 1.4. Candidate resource utilization models for individual marten with
AICc weight >0, Hiawatha National Forest, eastern Upper Peninsula of
Michigan, March 2016 – July 2017.

………………………………30

Table 2.1. Average (SE) and range of daily weather conditions, by season,
experienced by collared American marten on the Hiawatha National Forest,
eastern Upper Peninsula of Michigan, March 2016 – July 2017.

…………………………….55

Table 2.2. Model selection using AICc for weather related movement rates of
American marten in the Hiawatha National Forest, eastern Upper Peninsula of
Michigan, March 2016 – July 2017. All models with AICc weight >0.

………………………………56

Table 2.3. Generalized linear mixed model averaged coefficients (95%
confidence intervals) for daily weather variables used to describe marten
movement rates in the Hiawatha National Forest, eastern Upper Peninsula of
Michigan, March 2016 – July 2017. NA denotes a variable that was not
included in the model averaging.

……………………………...58

v

LIST OF FIGURES

Figure 0.1. Hiawatha National Forest in the Upper Peninsula of Michigan
(shaded green). Inset: East Unit of the Hiawatha National Forest. This unit
includes approximately 1538 km2 and 70 km of shoreline along lakes Michigan,
Huron, and Superior.

………………………………..6

Figure 1.1. East Units of the Hiawatha National Forest, eastern Upper
Peninsula of Michigan. A. 5.2 km2 grid cells within the East Unit boundary that
were available for random selection. B. Grid cells selected using generalized
random tessellation stratified (GRTS) sampling (purple). Red outlined grid cells
are where marten were collared.

……………………………..34

Figure 1.2. Average daily rate of movement of marten after collaring, Hiawatha
National Forest, eastern Upper Peninsula of Michigan, March 2016 – July 2017.

……………………………..35

Figure 2.1. Selection of Randomly Selected Trapping Grids. A. Grids that fall
completely within the East Unit of the Hiawatha National Forest. B. Grids
selected using generalized random tessellation stratified (GRTS) sampling.
Highlighted grids are where marten were collared.

……………………………..59

Figure 2.2. Average daily marten movement rate after collaring in the
Hiawatha National Forest, eastern Upper Peninsula of Michigan, March 2016 –
July 2017

…………………………….60

Figure 2.3. Movement rate per hour for American marten in a 24-hour day in
the Hiawatha National Forest, eastern Upper Peninsula of Michigan, March
2016 – July 2017. Box plots show the mean (dark bar), 75% quartile (box), 95%
confidence interval (whiskers), and outliers.

……………………………..61

Figure 2.4. Movement rate per hour, by season, for American marten in a 24hour day in the Hiawatha National Forest, eastern Upper Peninsula of
Michigan, March 2016 – July 2017. Box plots show the mean (dark bar), 75%
quartile (box), 95% confidence interval (whiskers), and outliers.

……………………………..62

Figure 2.5. Seasonal average movement rates with 95% confidence intervals
for American marten in the Hiawatha National Forest, eastern Upper Peninsula
of Michigan, March 2016 – July 2017.

……………………………..63

vi

Introduction
American marten (Martes americana) are members of the Mustelidae family and are considered
a species of management concern for several national forests in the United States, including the
Hiawatha National Forest (HNF) in the Upper Peninsula of Michigan. National Forests where marten
occur are required to manage for their habitat (Dumyahn et al. 2007). Additionally, marten are an
important recreational species that is harvested for fur in some U.S. States and Canadian Provinces, and
marten are culturally important to some Native American tribes (e.g., as clan animals). Thus, interest in
marten populations is generally high among natural resource management agencies, Native American
tribes, and the trapping community. Marten are sensitive to environmental change (USDA 2006), and
can thus provide an early warning signal about ecosystem health (USDA 2014, USDA 2006).
The types of habitat selected by marten vary by geographic location (Buskirk and Powell 1994,
Bissonette et al. 1997, Potvin et al. 2000). Although marten located in the western portions of the U.S.
and Canada prefer late-successional coniferous forests (Buskirk and Powell 1994), in the Great Lakes
Region marten may occupy areas that are dominated by deciduous forest types, and mixed coniferous
and deciduous forests (Chapin et al. 1997, Potvin et al. 2000, Payer and Harrison 2002). Marten also rely
on complex structural features near the forest floor (USDA 2014, Buskirk and Powell 1994). Ground
cover that consists of coarse woody debris (CWD) such as downed logs, snags, stumps, and exposed root
masses, and low hanging branches and shrubs, provide denning and resting sites (Buskirk and Powell
1994), habitat for prey (Clark and Campbell 1977, Corn and Raphael 1992, Sturtevant and Bissonette
1996), and escape cover from predators such as great horned owls (Bubo virginianus), eagles (Haliaeetus
spp. and Aquila chrysaetos), fishers (Martes pennanti), red fox (Vulpes vulpes), and coyotes (Canis
latrans) (Clark et al. 1987, Payer and Harrison 2002, Drew and Bissonette 1997, Thompson 1994).
Complex structural features near the forest floor also provide thermal protection during the winter

1

(Taylor and Buskirk 1994), and interstitial spaces beneath the snow layer (Clark and Campbell 1977, Corn
and Raphael 1992, Sherburne and Bissonette 1994).
Martens usually occupy areas with >30% tree canopy cover; but for resting and foraging they
prefer 40% to 60% canopy cover (Spencer et al. 1983, Clark et al. 1987, Watkins 2011). Marten typically
avoid open areas and landscapes where overhead canopy cover and mature trees are absent
(Heinemeyer 2002, Cushman et al.2011). However, vertical stems from regenerating trees and coarse
woody debris can provide adequate security cover when overstory canopy cover is absent (Drew and
Bissonette 1997).
Prior to European settlement, marten occupied both peninsulas of Michigan. Landscape
changes resulting from logging, fires, and human settlements, along with overharvesting, caused marten
to be extirpated, with the last confirmed sighting in the Upper Peninsula of Michigan from Marquette
County in 1939 (Manville 1948, Williams et al. 2007, Skalski et al. 2011). Marten were reintroduced in
the western portion of the Upper Peninsula in the 1950s, and the eastern portion of the Upper
Peninsula in the late 1980s and early 1990s (Williams et al. 2007, Earle et al. 2001). Additionally, 85
marten were reintroduced to the northern Lower Peninsula at two sites in the 1980s. As the marten
population grew in the Upper Peninsula, a trapping season was authorized in 2000 (Williams et al.
2007). As part of the trapping season, Michigan Department of Natural Resources – Wildlife Division
annually monitor marten harvest.
Marten populations are challenged by a rapidly changing climate, and have been identified as a
moderately vulnerable species on the Climate Change Vulnerability Index (CCVI) (Hoving et al. 2013).
This ranking indicates that range and population declines due to climate changes are expected. Some
have speculated that high quality habitat can help alleviate the negative impacts of climate change on
wildlife (e.g., Franklin et al. 2000). Habitat management may be particularly important for populations
2

that are isolated, as is the situation for marten in the HNF. Hence, studies on marten spatial ecology and
resource selection in these areas can provide critical information to guide habitat management.
The goal of my study was to quantify fine scale (both spatially and temporally) marten space
use, resource selection, and movement behaviors in the HNF of the Eastern Upper Peninsula of
Michigan. Here, fine spatial scale refers to forest stands (average ~6 ha in the HNF study area) and
marten GPS telemetry locations (generally 6 m average positional accuracy). Fine temporal scale refers
to attempting a marten location every 15 min for an approximately 2-week period (life of the collar). The
spatial domain was the East Unit of the HNF, and the temporal domain was 2016-2017. I relied on
marten telemetry data, spatially explicit forest inventory data (summarized at the stand level) from the
HNF, Euclidean distance measurements, and Daymet daily weather estimates.

Study Area
The East Unit of the HNF is located in the eastern Upper Peninsula of Michigan (Figure 0.1) and has been
managed by the United States Forest Service (USFS) since 1931. The southern boundary of the East Unit
borders both Lake Michigan and Lake Huron near the city of St. Ignace. The northern boundary includes
Lake Superior, west of Sault Ste. Marie (USDA Forest Service 2017; Figure 0.1). The East Unit includes
approximately 1,538 km2 and 70 km of shoreline along lakes Michigan, Huron, and Superior (Figure 0.1).
At the time of this study, ecosystems included beach ridges and dunes, outwash plains with lowland
coniferous forests, bedrock controlled ground moraines with northern hardwoods, and pine (Pinus spp.)
and aspen (Populus spp.) forests (USDA 2011). Additionally, northern hardwood and mixed forest types
were widespread; the most common tree species included sugar maple (Acer saccharum), red maple
(Acer rubrum), American beech (Fagus grandifolia), red pine (P. resinosa), northern white cedar (Thuja
occidentalis), eastern larch/tamarack (Larix laricina), jack pine (P. banksiana), spruce (Picea spp.), and
balsam fir (Abies balsamea) (USDA Forest Service 2017). Proximity to the Great Lakes has a profound
3

effect on weather in this region. Summer temperatures average 21° C, but along the Great Lakes
shoreline temperatures are cooler by 5° to 8° C. Average snowfall varies from 137 cm along lakes
Michigan and Huron to 610 cm along Lake Superior (USDA Forest Service 2017).
Current land and marten habitat management in the East Unit, originally established as the
Marquette National Forest, has evolved significantly over the last 80 years. After extensive logging in
the late 1800s and early 1900s major fires swept through this region, damaging soils and preventing
natural reforestation (USDA 2017). To prevent future timber shortages, the federal government
subsequently acquired lands throughout the Great Lakes region. Early (late 19th and early 20th centuries),
and forest management practices were established by Congressional Acts to protect wildlife
populations, conserve soil and water, and provide recreational opportunities to the public (USDA 2017).
During the Great Depression (1930s), the government established the Civilian Conservation Corps (CCC)
to create jobs. In the East Unit of the HNF, the CCC began rehabilitating by planting trees, controlling
white pine (P. strobus) blister rust, suppressing fires, and constructing roads, trails, and campgrounds
(USDA 2017). Today in the East Unit, the Forest Service continues to protect Great Lakes shorelines, and
maintain healthy ecosystems through management practices that not only allow timber harvesting but
also protect northern habitats for wildlife. The HNF has an active timber management program. For
example, in 2009 the East Unit harvested approximately 49,554 m3 of lumber dominated by red pine,
jack pine and aspen (USDA 2011).

4

APPENDIX

5

Figure 0.1. Hiawatha National Forest in the Upper Peninsula of Michigan (shaded green). Inset: East Unit
of the Hiawatha National Forest. This unit includes approximately 1538 km2 and 70 km of shoreline
along lakes Michigan, Huron, and Superior.
6

LITERATURE CITED

7

LITERATURE CITED

Bissonette, J. A., D. J. Harrison. C. D. Hargis, and T. F. Chapin. 1997. The influence of spatial scale and
scale-sensitive properties on habitat selection by American marten. pp. 369-385. In: J. A.
Bissonette (ed.). Wildlife landscape ecology: effects of pattern and scale. Springer-Verlag, New
York.
Buskirk S.W. and R. A. Powell. 1994. Habitat ecology of fishers and American martens. In: Buskirk S.W.,
A. S. Harestad, M. G. Raphael, and R. A. Powell (eds) Martens, sables, and fishers: biology and
conservation. Comstock Publishing Associates, Cornell University Press, Ithaca, pp 283–296
Chapin, T. G., D. J. Harrison, and D. M. Phillips. 1997. Seasonal habitat selection by marten in untrapped
forest preserve. Journal of Wildlife Management 61:707-717.
Clark, T. W., E. Anderson, C. Douglas, and M. Strickland. 1987. Mammalian species Martes Americana.
The American Society of Mammologists 289: 1 – 8.
Clark, T. W. and T. M. Campbell. 1977. Population organization and regulatory mechanisms of pine
martens in Grand Teton National Park, Wyoming. Government Report. UNPUBLISHED
Corn, J. G. and M. G. Raphael. 1992. Habitat characteristics at marten subnivean access sites. Journal
of Wildlife Management 56:442-448.
Cushman, S. A., M. G. Raphael, L. F. Ruggiero, A. S. Shirk, T. M. Wasserman, and E. C. O’Doherty. 2011.
Limiting factors and landscape connectivity: the American marten in the Rocky Mountains.
Landscape Ecology 26:1137-1149.
Drew, G. S. (1995). Winter habitat selection by American marten (Martes americana) in Newfoundland:
why old growth? Ph.D. Thesis. Utah State University. Logan, UT.
Drew, G. S., and J. A. Bissonette. 1997. Winter activity patterns of American martens (Martes
americana): rejection of the hypothesis of thermal-cost minimization. Canadian Journal of
Zoology 75.5: 812-816.
Dumyahn, J. B., J. H. Gilbert, and J. H. Gilbert. 2007. Winter Home-range Characteristics of American
Marten (Martes Americana) in Northern Wisconsin. American Midland Naturalist 158:382-394.
Earle, R. D., L. H. Mastenbrook, and T. F. Reis. 2001. Distribution and Abundance of the American
Marten in Northern Michigan. Michigan Department of Natural Resources. Wildlife Report No.
3321.
Foresman, K. R. and D. E. Pearson. 1998. Comparison of Proposed Survey Procedures for Detection of
Forest Carnivores. Journal of Wildlife Management 62(4):1217-1226.

8

Franklin, A. B., D. R. Anderson, R. J. Gutierrez, and K. P. Burnham. 2000. Climate, Habitat Quality, and
Fitness in Northern Spotted Owl Populations in Northwestern California. Ecological Monographs
70(4):539-590.
Heinemeyer, K. S. 2002. Translating individual movements into population patterns: American marten
in fragmented forested landscapes. Dissertation, University of California, Santa Cruz, California,
USA.
Hoving, C.L., Y.M. Lee, P.J. Badra, B.J. Klatt. 2013. Changing climate, changing wildlife: a vulnerability
assessment of 400 species of greatest conservation need and game species in Michigan.
Michigan Department of Natural Resources Wildlife Division Report. Apr;3564.
Manville, R. H. 1948. The vertebrate fauna of the Huron Mountains, Michigan. The American Midland
Naturalist 39.3: 615-640.
Payer, D. C. and D. J. Harrison. 2002. Influence of forest structure on habitat use by American marten in
an industrial forest. Forest Ecology and Management 179:145-156.
Potvin, F., L. Belanger, and K. Lowell. 2000. Marten habitat selection in a clearcut boreal landscape.
Conservation Biology 14:844-857.
Sherburne, S. S. and J. A. Bissonette. 1994. Marten subnivean access point use: response to subnivean
prey levels. Journal of Wildlife Management 58: 400 – 405.
Skalski, J.R., J.J.Millspaugh, M.V. Clawson,J.L. Belant, D.R. Etter, B.J. Frawley, P.D. Friedrich. 2011.
Abundance trends of American martens in Michigan based on statistical population
reconstruction. The Journal of Wildlife Management Nov 1, 75(8):1767-73.
Spencer, W. D., R. H. Barrett, and W. J. Zielinski. 1983. Marten habitat preferences in the northern
Sierra Nevada. Journal of Wildlife Management 47: 1181-1186.
Sturtevant, B. R. and J. A. Bissonette. 1996. Habitat association of Microtus pennsylvanicus within forest
of western Newfoundland as a factor influencing marten habitat quality. In: G. Proulx, H.
Bryant, and P. Woodard (eds.). Presented at the Second International Martes Symposium, Univ.
Alberta, Edmonton Alberta Canada, 12-16 August.
Taylor, S. L., and S. W. Buskirk. 1994. Forest microenvironments and resting energetics of the American
marten Martes americana. Ecography 17.3: 249-256.
Thompson, I. D. 1994. Marten populations in uncut and logged boreal forests in Ontario. Journal of
Wildlife Management 58:272-280.
United States Department of Agriculture (USDA) Forest Service. 2017. Hiawatha National Forest. Forest
Facts. https://www.fs.usda.gov/detail/hiawatha/about-forest/?cid=FSEM_033513

9

United States Department of Agriculture (USDA) Forest Service and The Nature Conservancy of
Michigan. 2011. Land Type Associations of the Hiawatha National Forest. Retrieved from
https://www.conservationgateway.org/Files/Pages/land-type-associations-hi.aspx
United States Department of Agriculture (USDA) Forest Service. Hiawatha National Forest. 2011.
Monitoring and Evaluation Report FY2010.
United States Department of Agriculture (USDA) Forest Service. 2006. Hiawatha National Forest 2006
Forest Plan. Retrieved from:
http://www.fs.usda.gov/detail/hiawatha/landmanagement/planning/?cid=STELPRDB5106336
United States Department of Agriculture (USDA) Forest Service. 2014. Martens, Sables, and Fishers:
New Synthesis Informs Management and Conservation. Science Findings. Issue 160. April 2014.
Watkins, J. A. 2011. Status of reintroduced American marten in the Manistee National Forest within
Michigan’s Northern Lower Peninsula. McNair Scholars Journal.
Williams, B. W., J. H. Gilbert, and P. A. Zollner. 2007. Historical Perspective on the Reintroduction of the
Fisher and American Marten in Wisconsin and Michigan. United States Department of
Agriculture Forest Service. http://www.nrs.fs. fed.us/

10

Chapter 1 - Resource Utilization of American Marten in the Hiawatha National Forest in the
Upper Peninsula of Michigan
Abstract
I examined resource utilization of American marten (Martes americana) in the Hiawatha National Forest
(HNF) in the eastern Upper Peninsula of Michigan. In cooperation with the Inland Fish and Wildlife
Department of the Sault Ste. Marie Tribe of Chippewa Indians, I GPS collared twelve marten (11 males,
one female) from March 2016 through July 2017 and used Brownian bridge movement models to
portray utilization distributions for each marten. I used the utilization value at each GPS location as the
response variable in generalized linear mixed models with forest stand ID as a random effect. Stands in
my study area averaged 5.97 ha (SE = 0.11). Fixed effects included forest inventory measurements
summarized at the stand level (i.e., basal area, stand age, tree diameter, and tree density class), a site
productivity index, and distances from marten locations to nearest opening and high traffic road. I
acquired 7,930 locations (660.8 locations/marten over approximately 14 days), and found that site
productivity was the most consistent correlate of marten utilization, followed by tree density. Marten
use was generally higher in more productive sites and in stands with higher tree densities (although this
relationship was not consistent across all marten). Basal area, stand age, and tree diameter were of
limited value in describing marten utilization. Some marten (n=3) appeared to negatively respond to
openings, and the effect of high traffic roads was mixed. When summarized at the stand level, I found
that individual metrics of vegetation structure typically collected during standard forest inventory were
of limited value for describing marten resource utilization. Rather, I found that an index to site
productivity (that indirectly related to forest structure complexity) was the metric that most consistently
correlated with marten habitat use. At the stand-level, I recommend that managers consider
multivariate measures of forest structure, combined with site productivity to portray marten habitat. I
11

also recommend that managers consider the potential negative effects of openings on marten core use
areas.

Introduction
Marten (Martes americana) tend to occupy older (≥ 80 years) coniferous forests throughout
much of their range (Buskirk and Powell 1994, Powell et al. 2003, Payer and Harrison 2003), however, in
some areas marten have been found using mid-successional, coniferous or mixed coniferous-deciduous
forests (Payer and Harrison 2003), and even areas dominated by young deciduous forests (Poole et al.
2004). Structural attributes found in older forests, like multi-storied canopies, snags, and abundant
downed wood allow marten to avoid predators while resting or hunting (Drew 1995, Bull et al. 2005,
McElhinny et al. 2005, Cushman 2011), and provide habitat for and access to prey (Corn and Raphael
1992, Buskirk and Powell 1994). Presumably, young- and mid-successional (≤ 80 years) forests that
support marten provide these structural attributes (Bowman and Robitaille 1997, Payer and Harrison
2003, Poole et al. 2004, Porter et al. 2005, Hearn et al. 2010, Powell et al. 2009). Variation in forest ages
and cover types used by marten underscores the importance of identifying site-specific features that
contribute to suitable marten habitat (Payer and Harrison 2003, Shirk et al. 2014).
Given that marten are closely associated with complex forest structure, they generally avoid
recently clearcut areas and openings with sparse canopy cover (Bull et al. 2005, Thompson and Harestad
1994, Pereboom et al. 2008). In Maine, use of clearcuts by martens was seasonal (Steventon and Major
1982). During winter marten avoided clearcuts, whereas in summer they used clearcuts to supplement
their diet with raspberries (Rubus spp.; Soutiere 1979). During summer, regenerating hardwood and
softwood trees in clearcuts presumably provided adequate overhead cover (Hawley and Newby 1957,
Herman and Fuller 1974).

12

Martens generally have a negative relationship with roads. In some locations roads can be the
primary source of mortality (e.g., Cervinka et al. 2015, Ruette et al. 2015). Some roads also can serve as
barriers to marten dispersal and movements (e.g., Ruiz-Gonzalez et al. 2014, Li et al. 2014, Howell et al.
2016). For example, martens will use low-traffic roadways for travel and hunting (e.g., Robitaille and
Aubry 2000), but roads with high traffic volumes can serve as movement barriers (e.g., Howell et al.
2016).
Marten unequivocally rely on complex forest structure, but our ability to adequately quantify
that structure is typically restricted to vegetation inventories summarized at the stand level. Here, a
stand refers to an operationally defined patch of relatively homogenous vegetation. In typical natural
resource information systems, stands tend to be mapped at 1’s to 10’s of hectares resolution, with
boundaries ecologically, operationally (e.g., along roads), or administratively defined. Stand-level
summaries form the basis for timber assessments and, by default, often for wildlife habitat assessments.
However, habitat utilization patterns exhibited by animals occur at multiple scales (Johnson 1980). For
resource managers that are required to consider marten habitat, understanding the utility of stand-level
data for management is important.
The goal of my study was to quantify marten resource utilization and relate utilization to standlevel summaries of typical forest inventory data. I also evaluated marten utilization in regards to hightraffic roads and openings. I hypothesized that forest inventory metrics indicative of older forest
conditions (e.g., high basal area, larger tree diameters) would positively correlate with marten use. I also
predicted that site productivity would positively correlate with marten use, as areas with higher
productivity tend to provide more structural complexity (Hargis et al. 1999, Buskirk and Zielinski 1997,
Larson et al. 2008). Lastly, I predicted that marten space use would negatively correlate with openings
and high-traffic roads. Quantifying how the relationships between marten space use, structural

13

attributes that are measured during standard forest inventory, and proximity metrics that can be
derived from GIS provide insights into marten ecology useful to forest planning and habitat
management.
Methods
Marten Trapping
To encompass a variety of forest types, I overlaid a 5.2 km2 square grid on the East Unit of the HNF in the
Upper Peninsula of Michigan (Figure 1.1). Only grid cells that were 100% inside the HNF boundary (not
borders or shoreline) were available for sampling (Figure 1.1A). This cell size is roughly twice the home
range size of marten in the East Unit of the HNF (Sault Tribe Inland Fish and Wildlife Department,
unpublished data), thereby having the potential to encompass multiple home ranges. I selected 30 grids
cells using generalized random tessellation stratified (GRTS) sampling (Stevens and Olsen 2003). This
procedure randomly selects spatially distributed samples from a discrete population while reducing the
chances of spatial clustering (Stevens and Olsen 2003). These grid cells formed the focal locations for
marten trapping. In collaboration with the Inland Fish and Wildlife Department of the Sault Ste. Marie
Tribe of Chippewa Indians, I trapped within selected grid cells from March 2016 through September
2016. Because of low capture rates within the initially selected grid cells, we expanded the trapping
area from September 2016 through July 2017 to a 36.25 km2 hexagon centered on each cell. We also
opportunistically trapped areas that historically supported marten in and around these larger areas.
Once a marten was collared in a grid cell, we moved to a different area. If no marten were trapped in an
area after approximately 5 days, we moved to the next grid cell.
We captured marten using Tomahawk 201 and 204 live traps (Tomahawk Live Trap Company,
Tomahawk, WI). We placed traps next to coarse woody debris and then covered the traps with leaf
litter, conifer boughs, tree bark, fallen logs, or natural debris found in the area. In winter we cut 19 L

14

buckets in half and covered each trap with straw and a half-bucket. Cover over the traps helped keep
marten dry and hidden. We baited traps with meat that included chicken, venison, pork, beef, or fish
heads. After placing the meat at the back of the trap, we placed a commercial call lure called Gusto
(Caven’s Gusto, Pennock, PA) on a branch above the trap.
We anesthetized marten using a facemask on a portable vaporizer with isoflurane in a 100%
oxygen carrier, with an induction setting of 5% and a maintenance setting of 2-3%. We weighed each
marten using a digital scale and monitored temperature, heart rate, and breathing during the collaring
procedure. For marten weighing >833 g we fit a 30g LiteTrack30 collar (SIRTRACK, Hawkes Bay, NZL). I
programmed the collars to attempt a location every 15 min (Moriarty and Epps 2015, Kie 2013). Based
on this setting, I estimated that the GPS battery would last 12 days, with the VHF beacon lasting an
additional 30 days. After collaring, we placed marten in a completely enclosed holding box and waited
for recovery, typically <10 min. Marten were released at the point of capture when fully recovered. I
used VHF triangulations to monitor the general locations of marten for the duration of VHF collar life.
Because these collars required manual downloading to retrieve the GPS data, we started re-trapping
marten after about 2 weeks, corresponding to the time period when GPS battery life was predicted to
end. We used the same procedure except that traps were purposefully placed where VHF tracking
suggested the marten were concentrating their activity. I plotted average daily movement rates since
capture to evaluate effects of the collaring process on marten space use. Locations from days that
showed a capture effect were removed from the analysis. Marten were trapped and handled using
approved protocols from the Institutional Animal Use and Care Committee at Michigan State University
(12/15-185-00) and the Inland Fish and Wildlife Department of the Sault Ste. Marie Tribe of Chippewa
Indians.

15

Utilization Distributions
I used Brownian Bridge Movement Models (BBMM) to analyze the GPS telemetry data because
the data were temporally autocorrelated (Kranstauber et al. 2012, Byrne et al. 2014). The BBMMs were
specifically designed to describe animal movements and portray utilization distributions from temporally
autocorrelated data (Horne et. al. 2007, Marzluff et al. 2004, Moriarty et al. 2016, Aebischer et al. 1993,
Nielson et al. 2013). Prior to analysis, I removed locations with a Dilution of Precision (DOP) >10; DOP
≤10 have acceptable locational error (Lewis et al. 2007, D’Eon and Delparte 2005). I created a utilization
distribution for each collared marten using the kernelbb function in the adehabitatHR package (Calenge
2006) in the program R Studio (R Core Team 2017). I exported the utilization distributions as rasters
from RStudio for use in ArcGIS 10.2.1. Once in ArcGIS, I extracted raster values for each covariate that
corresponded to every marten location. Thus, each marten location was assigned a measure of forest
structure like tree density, basal area, and tree size and proximity value (i.e., distance to nearest
opening, distance to nearest well-traveled road).
Modeling
For each marten, I used generalized linear mixed modeling (GLMM) with HNF stand as a random
effect. I used stand identifier as a random effect because multiple marten locations often occurred
within the same stand (and forest inventory data were summarized at the stand level). Fixed effects
included basal area, stand age, diameter at breast height (DBH), site productivity, tree density classes,
distance to nearest opening, and distance to nearest well-traveled road (Tables 1.1, 1.2). I selected these
attributes based on findings from published marten literature (Hargis et al. 1999, Payer and Harrison
2002, Sturtevant et al. 1997, Tiedemann et al. 2002). Basal area, stand age, DBH and tree size and
density came directly from the HNF. Basal area, stand age, and DBH were continuous data, whereas tree
size and density were ordinal data that ranged from no trees to densely stocked (4 density categories)
and openings to sawtimber sized (4 size categories), respectively (Table 1.1). I used site productivity
16

rankings from Schaetzl et al. (2012) and coded those 1 – 5, representing low to highly productive sites
(Table 1.1). To evaluate the assumption that site productivity positively correlated with forest
complexity, I calculated an additive structural complexity index from tree basal area, DBH, and density
(Barnett et al. 1978), and looked for a positive relationship between the index and site productivity. I
measured Euclidean distance from marten locations to the nearest edge of an opening (as classified in
the HNF stand layer) in ArcGIS (Table 1.2). Openings in the HNF stand layer ranged in size from <0.1 to
454.3 ha, and were defined as having a tree basal area and DBH measurement of zero. I also measured
Euclidean distance from marten locations to nearest high traffic road in ArcGIS. I identified roads with
the potential for higher traffic volumes based on an HNF classification ranging from 1 – 5, with 4 and 5
indicating that the road was routinely maintained and rated for passenger vehicle comfort (Table 1.2).
I produced a correlation table for the continuous fixed effects and identified those with a
correlation > 0.5 as potentially redundant (Dormann et al. 2012); I did not include redundant variables in
the same candidate model. Using all possible combinations of uncorrelated variables, I identified 40 to
76 candidate models for each marten and ranked those models using AICc (Burnham and Anderson
2002). To run the GLMMs, I used the lmer function in the lme4 package in RStudio (Bates et al. 2015).
Each marten potentially had unique models based on correlation results (Table 1.4). For candidate
models with ∆AIC < 7.0 for individual marten (Burnham and Anderson 2002), I used model averaging in
the MuMIn package in R (Bartoń 2016) to produce a single model. I deemed model parameters
significant if 95% confidence intervals did not overlap 0. I subsequently looked for consistency of
parameters (i.e., sign, significance) among marten to identify important predictors.
Results
Marten Trapping
We trapped marten for 5,922 trap nights. We caught 12 marten that were collared (11 males
and 1 female) and 10 marten (1 male and 9 females) that were too small to collar (i.e., <833 g). These
17

trapping results suggested that smaller marten (generally females) were using the same areas as larger
marten (generally males). On average, we caught a marten for collaring every 269 trap nights; trap
success was higher in March, April and May compared to July, August, October, and November (75.5 and
608.1 trap nights/marten, respectively). Martens that were not collared weighed an average of 608.3 g
(505g-706g). The average weight of martens that were collared was 1064.64 g (891g-1335g).
Collar Performance
Prior to deployment, I tested every collar by attempting to acquire at least one location; 100% of
the collars were functional. The GPS batteries lasted on average 14.7 days (SE = 1.09; range = 13-18),
collecting an average of 660.8 locations (SE = 167.6; range 374 – 1061) per marten, with 620.1 (SE =
160.5) of the fixes (94%) having DOP < 10. The average collar fix rate was 45% (SE = 9), comparable to
other GPS studies on marten (Moriarty et al. 2015). These collars were found to have an average
positional accuracy of 6 meters (DOP ≤10). I found no difference in average daily movement rate as time
from capture increased (Figure 1.2), hence all GPS collar locations were used in the resource utilization
model. Additionally, average marten home ranges from my study (male = 9.67 km2, female = 4.46 km2)
approximated home range estimates from other estimates in the Great Lakes Region (e.g., 4.3 – 15.7
km2; Davis 1978, Mech and Rogers 1977).
Resource Utilization
On average, I found that marten GPS locations were in forested stands with a relatively wide
range of basal areas (14.0 – 32.7 m2/ha), including forested stands with no merchantable timber (i.e.,
basal area = 0; Table 1.1). Average stand ages used by marten were >33 (range = 0 – 135) years, but
recently cut and permanent openings were also used (stand age = 0; Table 1.1). I found that average
stem diameters of used stands were >15.1 cm (Table 1.1). Areas used by marten in this study generally
contained small diameter trees overall (upper range = 46 cm, with most <35 cm; Table 1.1). Marten

18

consistently used forested stands that were relatively dense (i.e., stocking code >3), with moderate
productivity (productivity code generally > 2; Table 1.1). Most marten had a broad range of productivity
classes that they used, with codes generally ranging from 1 to 5 (Table 1.1).
I found that marten GPS locations were >167 m (range = 0 – 1,117 m) from openings on average,
with relatively low variability among individuals (most SEs < 14 m; Table 1.2). However, all marten were
documented using non-forested areas during all seasons (i.e., opening distance minimum = 0; Table 1.2).
Marten tended to use habitats that were >440 m from high traffic roads, but this pattern was highly
variable (SE ranged from 13.9 – 50.3 m; Table 1.2). Anecdotally, I found that the marten collared in
November, corresponding to the firearm deer season in Michigan, was farthest from maintained roads
(mean = 5,072 m; Table 1.2).
For each marten, I found that model averaging resulted in relatively simple (i.e., ≤3 parameters)
models for the majority (11 of 12) of marten (Table 1.3). On average, each marten had 8.8 (SE = 0.6)
competing models, with the top-ranking models accounting for 0.26 to 0.96 weight of evidence (Table
1.4). The majority (8 of 12) of top-ranking models included only the random effects, indicating that
unmeasured attributes of the stand or individual marten were explaining the most variability in resource
utilization.
The most consistent (10 of 12 marten) fixed effects that appeared in marten resource utilization
models included site productivity and tree density (Table 1.3). Other variables with some support (2-3
marten) among marten included basal area, distance to openings, and distance to roads (Table 1.3).
Stand age and stem diameter received limited support (1 and 0 marten, respectively; Table 1.3). Of the
most consistently appearing variables, site productivity was significant for 6 (of 10) marten, however the
direction of the effect was variable (Table 1.3). For tree density, the relationship tended to be positive
(i.e., as tree density increased probability of marten use increased), but few (8 of 10) of the relationships
19

were significant (Table 1.3). For the limited number of marten (n=3) that appeared to respond to
openings, the relationship was consistently significant and positive, indicating that high use areas
tended to occur farther from openings (Table 1.3).
Discussion
The positive relationships between marten and complex forest structure is undisputed. Marten
have been shown to use areas with higher basal areas or tree densities (>40m2/ha and 1,054 live
trees/ha; Spencer et al. 1983, Payer and Harrison 2003, Bull et al. 2005, Godbout and Ouellet 2010,
Sanders et al. 2017), older forests (≥ 80 years; Slauson and Zielinski 2009, Wiebe et al. 2013, Sanders et
al. 2017), and forests that contain larger diameter trees (>70 cm; Slauson and Zielinski 2009, Erb et al.
2010, Sanders et al. 2017). However, these structures may not consistently be restricted to mature or
old forest types (e.g., Vigeant-Langlois and Desrochers 2011, Hearn et al. 2010). I found that standard
forest inventory measures that included basal area, stand age, and tree diameter, summarized at the
stand level, were poor predictors of marten space use in the eastern Upper Peninsula of Michigan. This
finding highlights some of the challenges faced by natural resource managers that rely on forest
inventory databases to describe marten habitats. Short-term (i.e., daily, weekly) marten resource
utilization decisions likely occur at fine scales (e.g., individual trees, localized patches of downed wood;
Wiebe et al. 2014, Sanders et al. 2017) that are undetectable in stand-level forest inventories.
Additionally, consistent responses by martens to forest inventory metrics across large landscapes may
not be expected, as marten selection of forest structures can vary depending on landscape context of
home ranges (Hearn et al. 2010, Shirk et al. 2014). In these instances, combinations of factors like site
productivity and forest type, along with information on structural complexity, will likely provide more
consistent insight into why marten are using certain areas.

20

Complex forest structures are more likely to develop on productive sites (Hargis et al. 1999,
Buskirk and Zielinski 1997, Larson et al. 2008). Indeed, in my study a coarse ranking of site productivity
was the most consistent correlate to marten resource use patterns. I found that marten use increased
as site productivity increased. Stands with higher productivity develop complex forest structure more
rapidly than low productivity stands (Larson et al. 2008). Additionally, highly productive forests provide
enough nutrients for quick growth after a disturbance like timber harvest (Page-Dumroese 2010).
Following disturbances on the HNF, early successional trees (including balsam fir, spruce, and aspens)
rapidly colonize highly productive sites (USDA 2011, USDA Forest Service 2017), providing complex
structure suitable for marten foraging (Larsen et al. 2008, Payer and Harrison 2003, Baker 1992). These
densely treed areas provide cover for marten and key habitat for prey like snowshoe hare (Lepus
americanus) and ruffed grouse (Bonasa umbellus) (Vigeant-Langlois and Desrochers 2011, Fuller and
Harrison 2005, Potvin et al. 2000).
Openings with sparse overstory vegetation (<30% cover) generally do not provide enough cover
for marten or their prey (Vigeant-Langlois and Desrochers 2011, Fuller and Harrison 2005, Potvin et al.
2000, Baker 1992). In these low cover areas mortality risks for marten are higher and these places are
generally avoided (Bull et al. 2005, Thompson and Harestad 1994, Pereboom et al. 2008). However, I
found that marten temporarily use and cross openings in HNF. For example, in November a marten
crossed a large (67.9 ha) emergent wetland opening and stayed in a relatively small island (0.18 ha) of
sparse trees. I suspect that the emergent vegetation in this wetland provided enough cover to facilitate
movement to the island of trees (Buskirk and Zielinski 2003, Buskirk and Powell 1994).
Contrary to other studies (Cervinka et al. 2015, Ruette et al. 2015), I failed to find a strong road
effect on marten space use. In my study area, high traffic roads were relatively scarce, but low traffic
roads were common. Roads may correspond to areas where less prey occurs (Vos and Chardon 1998,

21

Rondinini and Doncaster 2002), however I frequently observed marten crossing and traveling down lowtraffic roads. My results highlight the importance of road type when describing marten habitat use. I
documented marten mortality from road kill on high traffic roads during my study, but also observed
marten using low-traffic roads for travel and foraging,
I acknowledge several limitations to my study. My forest inventory data were spatially limited to
the stand scale, which represented average forest structure condition across multiple hectares, a spatial
scale that may not align with short-term marten decisions on resource use. As such, stand-level averages
of forest structure likely represent a scale mismatch and hence it is not surprising that no single variable
was a good predictor of marten space use. Additionally, marten affinity for complex forest structure
suggests that no single variable should be a strong predictor. Rather, marten are likely responding to the
combination of forest structural attributes that includes interactions among the variables I used. These
factors likely contributed to the random effects models outperforming other models for explaining
marten resource utilization. I further acknowledge that my results are based on a limited number of
marten that were mainly males, monitored intensively over a short temporal window. This sampling
design has been shown to reliably portray marten home ranges (Moriarty 2014), but only represents a
sex-biased and relatively short assessment of resource utilization. Over a short temporal window,
proximate factors (e.g., interaction with a predator or conspecific, disturbance by humans or pets) other
than habitat could significantly affect resource utilization.
Given the ecological, recreational, and cultural importance of marten in Michigan,
understanding how marten populations interact with vegetation structure, land use patterns, and site
productivity are important contributions to marten conservation. I encourage natural resource
managers to cautiously use stand-level forest inventory data to quantify marten habitat, as singularly
these measures inadequately portrayed high use areas. My results suggest that site productivity is an

22

important consideration and I encourage managers to combine forest inventory data with site
productivity. I also encourage managers to recognize that some martens positioned high use areas away
from openings, thereby suggesting some sensitivity to non-forested areas. Lastly, my results indicated
that high-traffic roads in the East Unit were not consistent determinants of marten resource utilization.

23

APPENDIX

24

Table 1.1. Basal area, stand age, diameter at breast height (DBH), site productivity, and tree density for stands used by American marten on the
Hiawatha National Forest, eastern Upper Peninsula of Michigan, March 2016 – July 2017.
Basal Area (m2/ha)

Stand Age (yrs)

DBH (cm)

Productivityb

Tree Density
(stems/ha)a

Marten

Month

Mean (SE)

Range

Mean (SE)

Range

Mean (SE)

Range

Mode

Range

Mode

Range

A

March

23.6 (0.7)

0, 46

33 (2)

0, 100

16.5 (0.5)

0, 29.7

4

1, 4

2

1, 5

B

March

24.2 (0.5)

0, 41

65 (3)

0, 97

21.6 (0.3)

0, 26.7

4

1, 4

2

1, 5

C

March

23.2 (1.1)

0, 41

81 (2)

0, 104

17.7 (0.5)

0, 24.9

4

1, 4

2

1, 4

D

April

16.8 (0.4)

0, 32

50 (1)

0, 92

20.9 (0.3)

0, 31.2

3

1, 4

1

1

E

May

28.9 (0.5)

0, 53

78 (2)

0, 125

19.9 (0.3)

0, 29.5

4

1, 4

2

2, 5

F

May

19.3 (0.3)

0, 51

65 (1)

0, 128

18.7 (0.3)

0, 40.6

4

1, 4

1

1, 2

G

July

32.7 (0.5)

0, 49

52 (3)

0, 135

25.6 (0.3)

0, 30.5

4

1, 4

3

3, 5

H

August

28.1 (0.3)

0, 49

50 (2)

0, 112

22.2 (0.3)

0, 27.2

4

1, 4

3

2, 5

I

August

14.0 (0.5)

0, 40

47 (1)

0, 131

15.1 (0.3)

0, 46.0

4

1, 4

2

1, 5

J

August

28.4 (0.3)

0, 48

59 (1)

0, 122

18.5 (0.3)

0, 30.5

4

1, 4

3

2, 5

K

October

22.8 (0.5)

0, 32

65 (4)

0, 96

22.4 (1.0)

0, 40.6

4

1, 4

3

2, 5

25

Table 1.1. (cont’d).
Basal Area (m2/ha)

Stand Age (yrs)

DBH (cm)

Productivityb

Density
(stems/ha)a

a

Marten

Month

Mean (SE)

Range

Mean (SE)

Range

Mean (SE)

Range

Mode

Range

Mode

Range

L

November

15.7 (0.3)

0, 36

75 (1)

0, 130

15.8 (3.0)

0, 29.0

4

1, 4

4

2, 5

Ordinal data, where 1 = non-stocked, 2 = low stocking, 3 = medium stocking, and 4 = high stocking, based on Huron National Forest stocking

codes.
b

Ordinal data, where 1 = Productivity Index 5, 2 = Productivity Index 8, 3 = Productivity Index 9, 4 = Productivity Index 11, and 5 = Productivity

Index 14, based on Schaetzl et al. 2012 productivity indices. Schaetzl et al. (2012) rank productivities from 0 as the least productive to 19 as the
most productive.

26

Table 1.2. Distances to nearest openings and maintained, high traffic roads from American marten
telemetry locations on the Hiawatha National Forest, eastern Upper Peninsula of Michigan, March 2016
– July 2017.
Opening (m)
Marten

Month

Mean (SE)

A

March

306.4 (12.9)

B

March

C

Roads (m)

Range

Mean (SE)

Range

0, 969

494.1 (18.5)

14, 1580

310.0 (13.1)

0, 757

442.3 (20.7)

8, 1586

March

418.8 (32.9)

0, 927

660.4 (37.4)

1, 1393

D

April

527.1 (9.6)

0, 995

1260.3 (22.0)

2, 2317

E

May

246.8 (7.1)

0, 579

612.6 (18.6)

3, 2083

F

May

149.7 (5.0)

0, 591

894.1 (24.0)

3, 2006

G

July

167.0 (4.6)

0, 382

2808.0 (50.3)

729, 3750

H

August

395.7 (7.6)

0, 893

1303.5 (23.6)

72, 2229

I

August

323.6 (7.5)

0, 1117

760.2 (24.9)

18, 2828

J

August

458.9 (12.3)

0, 1058

1155.1 (69.79)

10, 4784

K

October

235.3 (10.7)

0, 675

2204.6 (52.8)

959, 2949

L

November

293.1 (7.9)

0, 831

5072.0 (13.9)

4031, 6149

27

Table 1.3. Generalized linear mixed model averaged coefficients (95% confidence intervals), by individual marten, for variables used to describe
resource utilization in the Hiawatha National Forest, eastern Upper Peninsula of Michigan, March 2016 – July 2017. NA denotes a variable that
was not included in the model averaging for individual marten. Bold-face indicates that the 95% confidence interval did not overlap 0.
Marten

Basal Area

Stand Age

DBH

Density

Productivity

Opening

Roads

A

NA

NA

NA

0.008

0.0196

NA

NA

(-0.08, 0.06)

(0.002, 0.037)

-0.077

0.0191

NA

NA

(-0.16, 0.003)

(0.004, 0.033)

-0.027

NA

NA

NA

NA

NA

0.0006

B

C

NA

NA

NA

NA

NA

NA

(-0.13, 0.106)
D

NA

NA

NA

NA

(0.0005, 0.0007)
E

F

NA

NA

NA

NA

NA

NA

-0.022

-0.021

0.0005

(-0.10-0.056)

(-0.05, 0.014)

(0.0002, 0.0007)

NA

-0.050

NA

(-0.16, 0.06)

28

NA

NA

Table 1.3. (cont’d).
Marten

Basal Area

Stand Age

DBH

Density

Productivity

Opening

Roads

G

NA

NA

NA

0.028

-0.067

NA

-0.0001

(-0.051, 0.107)

(-0.013, -0.011)

0.028

-0.055

0.0004

(-0.051, 0.107)

(-0.09, 0.024)

(0.0003, 0.0006)

0.038

-0.025

NA

NA

(0.019, 0.057)

(-0.088, 0.038)

0.007

0.066

NA

NA

(-0.026, 0.012)

(0.007, 0.126)

0.064

-0.013

NA

NA

(-0.052, 0.180)

(-0.121, 0.095)

0.049

0.121

0.0001

-0.0001

(0.0389, 0.0586)

(0.034, 0.207)

(0.00007, 0.0002)

(-0.0002, -0.0001)

H

I

NA

-0.0009

NA

NA

NA

NA

(-0.0002, -0.0006)
J

K

L

NA

NA

NA

NA

0.0016

0.0014

(0.001, 0.002)

(0.001, 0.002)

NA

NA

NA

29

(-0.0001, -0.00008)
NA

Table 1.4. Candidate resource utilization models for individual marten with AICc weight >0, Hiawatha
National Forest, eastern Upper Peninsula of Michigan, March 2016 – July 2017.
Model
Marten

Number

Model

AICc

Delta

Weight

A

1

UD~ random effect

-493.1

0.00

0.878

2

UD~ productivity

-488.1

4.97

0.073

3

UD~ density

-486.3

6.77

0.030

4

UD~ DBH

-484.0

9.06

0.009

5

UD~ near open

-481.8

11.28

0.003

6

UD~ density + productivity

-481.3

11.80

0.002

7

UD~ basal area

-479.5

13.60

0.001

8

UD~ stand age

-479.4

13.73

0.001

9

UD~DBH + productivity

-479.0

14.05

0.001

10

UD~ near road

-478.5

14.59

0.001

1

UD~ random effect

-333.8

0.00

0.637

2

UD~ density

-331.0

2.75

0.161

3

UD~ productivity

-330.6

3.12

0.134

4

UD~ density +productivity

-328.3

5.49

0.041

5

UD~ DBH

-326.5

7.30

0.017

6

UD~ DBH + productivity

-323.8

10.00

0.004

7

UD~ basal area

-323.3

10.46

0.003

8

UD~ basal area + productivity

-320.8

12.95

0.001

9

UD~ stand age

-320.3

13.46

0.001

1

UD~ random effect

-3.5

0.00

0.903

2

UD~ density

2.5

5.98

0.045

3

UD~ productivity

3.6

7.08

0.026

4

UD~ DBH

4.5

8.02

0.016

5

UD~ near open

7.6

11.11

0.003

6

UD~ stand age

9.2

12.72

0.002

7

UD~ density + productivity

9.6

13.11

0.001

8

UD~ basal area

10.2

13.74

0.001

B

C

30

Table 1.4. (cont’d).
Model
Marten

D

E

F

G

Number

Model

AICc

Delta

Weight

9

UD~ near road

11.4

14.97

0.001

1

UD~ near road

-389.0

0.00

0.958

2

UD~ DBH + near road

-381.5

7.53

0.022

3

UD~ near open + near road

-380.9

8.15

0.016

4

UD~ near road + stand age

-377.1

11.86

0.002

1

UD~ random effect

-352.0

0.00

0.562

2

UD~ near open

-351.1

0.92

0.355

3

UD~ density + near open

-345.4

6.61

0.021

4

UD~ density

-345.1

6.90

0.018

5

UD~ productivity

-345.0

6.93

0.018

6

UD~ near open + productivity

-343.9

8.12

0.010

7

UD~ DBH

-342.8

9.17

0.006

8

UD~ stand age

-342.1

9.88

0.004

9

UD~ DBH + near open

-342.0

9.97

0.004

10

UD~ near open

-338.3

13.72

0.001

11

UD~ density + productivity

-338.2

13.81

0.001

12

UD~ density + near open + productivity

-338.1

13.87

0.001

1

UD~ random effect

-725.1

0.00

0.892

2

UD~ productivity

-719.9

5.17

0.067

3

UD~ density

-717.2

7.91

0.017

4

UD~ near road

-716.4

8.66

0.012

5

UD~ DBH

-715.0

10.03

0.006

6

UD~ stand age

-713.1

11.94

0.002

7

UD~ density + productivity

-712.0

13.12

0.001

8

UD~ near road + productivity

-711.1

13.95

0.001

1

UD~ random effect

-378.6

0.00

0.672

2

UD~ productivity

-376.8

1.82

0.270

3

UD~ density

-372.5

6.16

0.031

31

Table 1.4. (cont’d).
Model
Marten

H

I

J

Number

Model

AICc

Delta

Weight

4

UD~ density + productivity

-370.4

8.20

0.011

5

UD~ DBH

-369.9

8.71

0.009

6

UD~ DBH + productivity

-367.9

10.72

0.003

7

UD~ stand age

-367.3

11.35

0.002

8

UD~ basal area

-364.8

13.83

0.001

1

UD~ near open + productivity

-90.5

0.00

0.783

2

UD~ near open

-87.3

3.21

0.158

3

UD~ density + near open + productivity

-84.0

6.45

0.031

4

UD~ DBH + near open + productivity

-82.1

8.35

0.012

5

UD~ density + near open

-81.0

9.46

0.007

6

UD~ productivity

-79.7

10.81

0.004

7

UD~ DBH + near open

-79.1

11.41

0.003

8

UD~ near open + productivity + stand age

-76.9

13.57

0.001

9

UD~ basal area + near open + stand age

-75.8

14.64

0.001

1

UD~ DBH + near road

-173.4

0.00

0.724

2

UD~basal area + DBH + density + near road

-170.9

2.51

0.416

3

UD~ DBH + near road + productivity

-167.0

6.46

0.029

4

UD~ basal area + DBH + near road

-165.0

8.41

0.011

5

UD~basal+DBH+density+road+productivity

-163.9

9.54

0.006

6

UD~ basal area + density + near road

-162.2

11.24

0.003

7

UD~DBH+ density + near road + productivity

-159.7

13.76

0.001

1

UD~ random effect

-260.6

0.00

0.747

2

UD~ productivity

-258.1

2.55

0.209

3

UD~ density

-253.8

6.79

0.025

4

UD~ DBH

-251.3

9.31

0.007

5

UD~ density + productivity

-251.3

9.36

0.007

6

UD~ DBH + productivity

-248.6

12.04

0.002

7

UD~ stand age

-246.6

12.04

0.001

32

Table 1.4. (cont’d).
Model
Marten

Number

Model

AICc

Delta

Weight

K

1

UD~ random effect

-2.4

0.00

0.843

2

UD~ density

2.3

4.77

0.078

3

UD~ productivity

3.6

6.05

0.041

4

UD~ DBH

4.6

7.03

0.025

5

UD~ density + productivity

8.4

10.85

0.004

6

UD~ near open

9.0

11.47

0.003

7

UD~ basal area

9.9

12.33

0.002

8

UD~ DBH + density

10.0

12.48

0.002

9

UD~ DBH + productivity

10.7

13.16

0.001

10

UD~ stand age

11.2

13.63

0.001

1

UD~ basal area +near road

-590.8

0.00

0.257

2

UD~ density + near road

-590.6

0.22

0.230

3

UD~ basal area + near road + productivity

-590.1

0.75

0.177

4

UD~ density + near road + productivity

-589.7

1.17

0.143

5

UD~density + near open + near road

-588.4

2.41

0.077

6

UD~density+near open+near road+prod.

-587.7

3.10

0.055

7

UD~ near road + stand age

-585.7

5.10

0.020

8

UD~ basal area + near open + near road

-585.3

5.56

0.016

9

UD~basal area+near open+near road+prod.

-584.8

6.08

0.012

10

UD~ near road + productivity + stand age

-584.7

6.11

0.012

11

UD~ near open + near road + stand age

-579.9

10.97

0.001

12

UD~ near open + near road + productivity +

-579.1

11.74

0.001

L

stand age

33

Figure 1.1. East Unit of the Hiawatha National Forest, eastern Upper Peninsula of Michigan. A) 5.2 km2
grid cells within the East Unit boundary that were available for random selection. B) Grid cells selected
using generalized random tessellation stratified (GRTS) sampling (purple). Red outlined grid cells are
where marten were collared.
34

Figure 1.2. Average daily rate of movement of marten (n=12) after collaring, Hiawatha National Forest, eastern Upper Peninsula of Michigan,
March 2016 – July 2017.

35

LITERATURE CITED

36

LITERATURE CITED

Aebischer, N. J., P. A. Robertson, and R. E. Kenward. 1993. Compositional analysis of habitat use from
animal radio-tracking data. Ecology 74: 1313–1325.
Baker, J. M., 1992. Habitat use and spatial organization of pine marten on southern Vancouver Island,
British Columbia. M.S. Thesis, Simon Fraser University, Burnaby, British Columbia, Canada.
Barnett, J. L., R. A. How, W. F. Humphreys. 1978. The use of habitat components by small mammals in
eastern Australia. Austral Ecology. Sep 1;3(3):277-85.
Bartoń, K. 2016. MuMIn: Multi-model inference. R package version 1.15.6. Retrieved December 21,
2015, from http://cran.r-project.org/package=MuMIn
Bates, D., M. Mächler, B. Bolker, S. and Walker. 2015. Fitting linear mixed-effects models using
lme4. Journal of Statistical Software, page in press.
Bowman, J. C., and J.-F. Robitaille. 1997. Winter habitat use of American martens (Martes Americana)
within second-growth forest in Ontario, Canada. Wildlife Biology 3.2: 97-105.
Bull, E. L., T. W. Heater, and J. F. Shepherd. 2005. Habitat Selection by the American Marten in
Northeastern Oregon. Northwest Science 79(1)37-43.
Buskirk, S. W. and R. A. Powell. 1994. Habitat ecology of fishers and American martens. Pages 283-296
In S. W. Buskirk, A. S. Harestad, M. G. Raphael, and R. A. Powell, (eds). Martens, sables, and
fishers: biology and conservation. Cornell University Press. Ithaca, NY.
Buskirk, S.W., and W. J. Zielinski. 1997. American marten (Martes americana) ecology and conservation.
The Wildlife Society California North Coast Chapter: 17.
Buskirk, S. W., and W.J. Zielinski. 2003. Small and mid-sized carnivores. Mammal community dynamics:
management and conservation in the coniferous forests of western North America. Cambridge
University Press, Cambridge, UK (2003): 207-249.
Burnham K, P., and D. R. Anderson. 2002. Model selection and multi-model inference. 2nd edition.
Springer-Verlag, New York, New York, USA.
Byrne, M.E., J. C. McCoy, J. W. Hinton, M. J. Chamberlain, and B. A. Collier. 2014. Using dynamic
Brownian bridge movement modelling to measure temporal patterns of habitat selection.
Journal of Animal Ecology, 83(5), pp.1234-1243.
Calenge, C. 2006. The package “adehabitat” for the R software: a tool for the analysis of space and
habitat use by animals. Ecological modelling 197.3: 516-519.
Červinka, J., J. Riegert, S. Grill, and M. Šálek. 2015. Large-scale evaluation of carnivore road mortality:
the effect of landscape and local scale characteristics. Mammal Research 60(3), 233-243.
37

Corn, J. G. and M. G. Raphael. 1992. Habitat characteristics at marten subnivean access sites. Journal of
Wildlife Management 56:442-448.
Cushman, S.A., M.G. Raphael, L.F. Ruggiero, A.S. Shirk, T.N. Wasserman, E.C. O'Doherty. 2011. Limiting
factors and landscape connectivity: the American marten in the Rocky Mountains. Landscape
Ecology 26:1137–1149
D’Eon, R. G. and D. Delparte. 2005. Effects of Radio-collar position and orientation on GPS radio-collar
performance, and the implications of PDOP in data screening. Journal of Applied Ecology
42:383-385.
Davis, M. H. 1978. Re-introduction of the pine marten into the Nicolet National Forest, Forest County,
Wisconsin. M.S. Thesis, University of Wisconsin, Stevens Point. 64 pp.
Dormann, C. F., J. Elith, , S. Bacher, C. Buchmann, G. Carl, G. Carré, J. R. G. Marquéz, B. Gruber, B.
Lafourcade, P.J. Leitão, T. Münkemüller, C. McClean, P. E. Osborne, B. Reineking, B. Schröder, A.
K. Skidmore, D. Zurell, and S. Lautenbach. 2012. Collinearity: a review of methods to deal with it
and a simulation study evaluating their performance. Ecography, 36: 27–46. doi:10.1111/j.16000587.2012.07348.x
Drew, G. S. 1995. Winter habitat selection by American marten (Martes americana) in Newfoundland:
why old growth? PhD Dissertation. Utah State University, Logan, UT.
Erb, J., P. Coy, and B. Sampson. 2010. REPRODUCTIVE ECOLOGY OF FISHER AND MARTEN IN
MINNESOTA. Minnesota Department of Natural Resources Summaries of Wildlife Research
Findings, MNDNR technical Report
Fuller, A. K., and D. J. Harrison. 2005. Influence of partial timber harvesting on American martens in
north-central Maine. Journal of Wildlife Management 69.2: 710-722.
Godbout, G., and J.P. Ouellet. 2010. Fine-scale habitat selection of American marten at the southern
fringe on the boreal forest. ECOSCIENCE 17:(2) 175-185.
Hargis, C. D., J. A. Bissonette and D. L. Turner. 1999. The influence of forest fragmentation and
landscape pattern on American martens. Journal of Applied Ecology 36: 157-172.
Hawley, V. D., and F.E. Newby. 1957. Marten home ranges and population fluctuations. Journal of
Mammalogy 38(2), 174-184.
Hearn, B. J., D. J. Harrison, A.K. Fuller, C.G. Lundrigan, and W. J. Curran. 2010. Paradigm shifts in habitat
ecology of threatened Newfoundland martens. Journal of Wildlife Management 74(4), 719-728.
Herman, T. and K. Fuller. 1974. Observations of the marten (Martes americana), in the Mackenzie
District, Northwest Territories. Can. Field-Nat 88:501-503.
Horne, J. S., E. O. Garton, S. M. Krone, and J. S. Lewis. 2007. Analyzing Animal Movements using
Brownian Bridges. Ecology 88(9):2354-2363.
38

Johnson, D.H., 1980. The comparison of usage and availability measurements for evaluation resource
preference. Ecology, 61: 65-71.
Kie, J.G. 2013. A rule-based ad hoc method for selecting a bandwidth in kernel home-range analyses.
Animal Biotelemetry 1:13.
Kranstauber, B., R. Kays, S. D. LaPoint, M. Wikelski, and K. Safi. 2012. A dynamic Brownian bridge
movement model to estimate utilization distributions for heterogeneous animal movement.
Journal of Animal Ecology, 81(4), 738-746.
Larson, A. J., J. A. Lutz, R. F. Gersonde, J. F. Franklin, and F. F. Hietpas. 2008. Potential site productivity
influences the rate of forest structural development. Ecological Applications 18(4):899-910.
Lewis, J. S., J. L. Rachlow, E. O. Garton, and L. A. Vierling. 2007. Effects of habitat on GPS collar
performance: using data screening to reduce location error. Journal of Applied Ecology 44:663671.
Li, Y., W. Wu, Z. Xiong, Y. Hu, Y. Chang, and D. Xiao. 2014. Effects of forest roads on habitat pattern for
sables in Da Hinggan Mountains, northeasten China. Chinese geographical science 24(5), 587.
Marzluff, J. M., J. J. Millspaugh, P. Hurvitz, and M. S. Handcock. 2004. Relating resources to a
probabilistic measure of space use: forest fragments and Steller's jays. Ecology 85, no. 5: 14111427.
McElhinny, C., P. Gibbons, C. Brack, and J. Bauhus. 2005. Forest and woodland stand structural
complexity: its definition and measurement. Forest Ecology and Management 218(1), 1-24.
Mech, L. D. and L. L. Rogers. 1977. Status, distribution, and movements of martens in northeastern
Minnesota. Research Paper NC-143. St. Paul, MN: U.S. Dept. of Agriculture, Forest Service, North
Central Forest Experiment Station
Moriarty, K. M. 2014. Habitat use and movement behavior of Pacific marten (Martes caurina) in
response to forest management practices in Lassen National Forest, California. Ph.D.
dissertation, Oregon State University. Corvallis.
Moriarty, K. M. and, C. W. Epps. 2015. Retained satellite information influences performance of GPS
devices in a forested ecosystem. Wildlife Society Bulletin 39: 349–357. doi:10.1002/wsb.524
Moriarty, K. M., C. W. Epps, M. G. Betts, D. J. Hance, J. D. Bailey, and W. J. Zielinski. 2015. Experimental
evidence that simplified forest structure interacts with snow cover to influence functional
connectivity for Pacific martens. Landscape Ecology 30:1865–1877.
Moriarty, K. M., C. W. Epps, and W. J. Zielinski. 2016. Forest thinning changes movement patterns and
habitat use by Pacific marten. The Journal of Wildlife Management 80.4: 621-633.

39

Page-Dumroese, Deborah S., Martin Jurgensen, and Thomas Terry. Maintaining soil productivity during
forest or biomass-to-energy thinning harvests in the western United States. Western Journal of
Applied Forestry 25.1 (2010): 5-11.
Payer, D. C., and D. J. Harrison. 2003. Influence of forest structure on habitat use by American marten
in an industrial forest. Forest Ecology and Management 179:145-156.
Payer, D., and D. Harrison. 2005. Relationships between forest structure and habitat use by American
martens in Maine, USA. Martens and fishers (Martes) in human-altered environments: 173-186.
Pereboom, V., M. Mergey, N. Villerette, R. Helder, J. Gerard, and T. Lodé. 2008. Movement patterns,
habitat selection, and corridor use of a typical woodland-dweller species, the European pine
marten (Martes martes), in fragmented landscape. Canadian Journal of Zoology 86:983-991.
Poole, K. G., A. D. Porter, A.D. Vries, C. Maundrell, S. D. Grindal, and C. C. St. Clair. 2004. Suitability of a
young deciduous-dominated forest for American marten and the effects of forest removal.
Canadian Journal of Zoology 82(3), 423-435.
Porter, A. D., C. C. St Clair, and A. de Vries. 2005. Fine-scale selection by marten during winter in a young
deciduous forest. Canadian Journal of Forest Research 35.4: 901-909.
Potvin, F., L. Belanger, and K. Lowell. 2000. Marten habitat selection in a clearcut boreal landscape.
Conservation Biology 14:844-857.
Powell, R.A., S.W. Buskirk, and W.J. Zielinski. 2003. Fisher and marten. In: Feldhammer, G.A., Thompson,
B.C., Chapman, J.A. (Eds.), Wild Mammals of North America. Johns Hopkins Press, Baltimore,
MD, pp. 635–649
Powell, D.C. 2009. A stage is a stage is a stage…or is it? Successional stages, structural stages, seral
stages. White Pap. F14-SO-WP-Silv-10. Pendleton, OR: USDA Forest Ser-vice, Pacific Northwest
Region, Umatilla National Forest. 14 p.
http://www.fs.usda.gov/Internet/FSE_DOCUMENTS/stelprdb5413728.pdf
R Core Team. 2017. R: A language and environment for statistical computing. R Foundation for Statistical
Computing, Vienna, Austria. URL https://www.R-project.org/.
Robitaille, J. F. and K. Aubry. 2000. Occurrence and activity of American martens Martes americana in
relation to roads and other routes. Acta Theriologica 45(1): 137-143.
Rondinini, C., and C. P. Doncaster. 2002. Roads as barriers to movement for hedgehogs. Functional
Ecology 16.4: 504-509.
Ruette, S., J. M. Vandel, M. Albaret, and S. Devillard. 2015. Comparative survival pattern of the syntopic
pine and stone martens in a trapped rural area in France. Journal of Zoology 295(3), 214-222.
Ruiz-González, A., M. Gurrutxaga, S. A. Cushman, M. J. Madeira, E. Randi, and B. J. Gómez-Moliner.
2014. Landscape genetics for the empirical assessment of resistance surfaces: the European pine

40

marten (Martes martes) as a target-species of a regional ecological network. PLoS One 9(10),
e110552.
Sanders, R. L., A. Cornman, P. Keenlance, and J. J. Jacquot, D. E. Unger, and M. Spriggs. 2017. Resting Site
Characteristics of American Marten in the Northern Lower Peninsula of Michigan. The American
Midland Naturalist 177(2), 211-225.
Schaetzl, R.J., F. J. Krist Jr., and B.A. Miller. 2012. A Taxonomically Based, Ordinal Estimate of Soil
Productivity for Landscape-Scale Analyses. Soil Science 177:288-299.
Shirk, A. J., M. G. Raphael, and S. A. Cushman. Spatiotemporal variation in resource selection: insights
from the American marten (Martes americana). Ecological Applications 24.6 (2014): 1434-1444.
Slauson, K. M., and W.J. Zielinski. 2009. Characteristics of summer and fall diurnal resting habitat used
by American martens in coastal northwestern California. Northwest Science 83(1), 35-45.
Spencer, W.D., R.H. Barrett, and W.J. Ziellinski. 1983. Marten habitat preferences in the Northern Sierra
Nevada. Journal of Wildlife Management 47: 1181-1186.
Stevens, D. L., and A. R. Olsen. 2003. Variance estimation for spatially balanced samples of
environmental resources. Environmetrics (14)6: 593-610.
Steventon, J. D., and J. T. Major. 1982. Marten use of habitat in a commercially clear-cut forest. The
Journal of Wildlife Management 1982: 175-182.
Soutiere, E. C. 1979. Effects of timber harvesting on marten in Maine. The Journal of Wildlife
Management 850-860.
Sturtevant, B. R., J. A. Bissonette, J. N. Long, J. and D. W. Roberts. 1997. Coarse woody debris as a
function of age, stand structure, and disturbance in boreal Newfoundland. Ecological
Applications, 7(2), 702-712.
Thompson, I. D. and A. S. Harestad. 1994. Effects of logging on American martens, and Models for
habitat management. Pages 355-367 in Buskirk, S. W., A. S. Harestad, M. G. Raphael, and R. A.
Powell (eds) Martens, sables, and fishers: biology and conservation. Comstock Publishing
Associates, Cornell University Press.
Tiedemann, A. R., J. O. Klemmedson, and E. L. Bull. 2000. Solution of forest health problems with
prescribed fire: are forest productivity and wildlife at risk?. Forest Ecology and Management
127.1: 1-18.
Vigeant-Langlois, C., and A. Desrochers. 2011. Movements of wintering American marten (Martes
americana): relative influences of prey activity and forest stand age. Canadian journal of forest
research 41(11), 2202-2208.
United States Department of Agriculture (USDA) Forest Service. Hiawatha National Forest. 2011.
Monitoring and Evaluation Report FY2010.

41

United States Department of Agriculture (USDA) Forest Service. 2017. Hiawatha National Forest. Forest
Facts. https://www.fs.usda.gov/detail/hiawatha/about-forest/?cid=FSEM_033513
Vos, C. C. and J. P. Chardon. 1998. Effects of habitat fragmentation and road density on the distribution
pattern of the moor frog Rana arvalis. Journal of Applied Ecology 35:44-56.
Wiebe, P. A., J. M. Fryxell, I. D. Thompson, L. Börger, and J.A. Baker. 2013. Do trappers understand
marten habitat?. The Journal of Wildlife Management 77(2), 379-391.
Wiebe, P. A., I. D. Thompson, C.I. McKague, J.M. Fryxell, and J. A. Baker. J. A. 2014. Fine-scale winter
resource selection by American martens in boreal forests and the effect of snow depth on
access to coarse woody debris. Écoscience 21(2), 123-132.

42

Chapter 2 – Weather-related Movements of American Marten in the Hiawatha National Forest,
Michigan

Abstract
As American marten (Martes americana) move they respond to biotic and abiotic factors. For species
that are vulnerable to predation and trapping like marten, movement behaviors in response to weather
and season can inform harvest regulations and research activities. I examined the daily, seasonal, and
weather related movements of marten in the Hiawatha National Forest (HNF) in the eastern Upper
Peninsula of Michigan. I GPS collared 12 (11 males, 1 female) marten and attempted to collect a location
every 15 minutes. I downloaded 7,930 locations (660.8 locations/marten), and for each successive GPS
fix I calculated average hourly rate of movement by season. To model abiotic factors affecting
movement rates, I used hourly rate of movement as a dependent variable in generalized linear mixed
models with fixed effects that included interpolated weather estimates from Daymet. Individual marten
and season were used as random effects in the models. On average, I found that marten moved 13.3
m/min, and that hourly movement rates did not differ throughout a 24-hr day. I also found that marten
moved significantly less in fall than winter, spring, and summer. I failed to find a significant weather
effect on marten movement rates, but limited evidence supported the hypotheses that daily movement
rates increased as: 1) maximum daily temperatures decreased, and 2) as precipitation increased. My
results indicate that fall is a difficult time to capture marten as movement rates are significantly lower,
and that weather cannot be used to reliably predict when marten move.

43

Introduction
Movement ecology of harvested wildlife species can provide insights into capture effort and
animal vulnerability (Gilchrist et al. 2005, Cushman et al. 2011, Nams and Bourgeois 2005). More
broadly, information on animal movements has been used to understand population distributions
(Turchin 1991), resource use (Birchfield and Deters 2005), distribution strategies (Small and Rusch 1989),
and patterns of space use (Kenward et al. 2001). Until recently, studies on animal space use tended to
ignore the connection between successive locations, instead analyzing the characteristics of locations
themselves (Kernohan 2001). Recent advances in telemetry technology and analytical techniques have
increased the ability to quantify animal activities between locations and include analyses on
directionality (Pace 2001), movement rates (Forester et al. 2007, Frair et al. 2005), and relating
environmental variables to movement choices (Cushman et al. 2011). Movement characteristics and
behaviors are often linked directly to critical life history processes like foraging (Heinrich 1979),
reproduction (Martin 1994), predator avoidance (Frair et al. 2005), and dispersal (Broquet et al. 2006).
Thus, understanding factors that relate to animal movement behaviors is important to management and
research.
American marten (Martes americana) are a forest habitat specialist with significant ecological,
recreational, and cultural importance. Generally little is known about marten movement ecology,
especially in regards to seasonal and daily movement rates. Studies indicate that marten are more
active during summer (~60% of the day) compared to winter (~16% of the day; Stone 2010; Bull and
Heater 2000; Powell et al. 2003). In a similar study in Ontario, Canada, Thompson (1986) found that
marten were more active in the spring, summer, and fall (42 – 67% of the day) compared to winter (21%
of the day). Moriarty (2014) speculated that higher summer activity was related to availability of food,
presence of predators, and activities associated with reproduction. Decreases in winter activity may be
44

related to fewer active predators, consumption of larger prey that reduces foraging time, lack of
reproductive activities, and thermal stress during cold temperatures (Stone 2010, Moriarty 2014).
Generally, males travel greater distances than females (Thompson 1986).
Prior to development of satellite-based GPS collars that were light enough for marten, our
knowledge of marten movement behaviors was primarily limited to locations collected through VHF
telemetry. Oftentimes, inference from VHF data was restricted by low fix rates (resulting in large time
intervals between successive fixes) and locational accuracy. Commercially available GPS collars for
marten are relatively new (within the last 5 years) and offer an opportunity to improve our
understanding of movement ecology. For other species, satellite-based technology for studying animal
movements has proven effective for quantifying habitat use (Nielson et al. 2009), identifying stopover
areas (Sawyer et al. 2009, Kochert et al. 2011), migration routes (Takekawa et al. 2010, White et al.
2010), and for recreating movement paths (Witt et al. 2010).
The goal of my study was to characterize marten movement behaviors in the eastern Upper
Peninsula of Michigan to inform capture activities and to better understand environmental factors
affecting movements. Using GPS telemetry data, I described seasonal movement rates and activity
patterns, and correlated weather factors to movement rates. I expected movement rates to be higher in
summer, and because marten are generally thought to be nocturnal (Zalewski 2000, Buskirk 1983,
Strickland et al. 1987), I expected movement rates to be higher at night. Marten are also poor
thermoregulators (Buskirk et al. 1988, Buskirk et al. 1989), so I hypothesized that precipitation would
negatively affect marten movements. Also related to thermoregulation, I hypothesized that as
maximum temperature increased, daily movement rates would decrease. Similarly, when minimum
temperature decreased I expected marten movement rates to decrease. I predicted that as the
difference between daily maximum and minimum temperature increased (i.e., large daily changes in

45

temperature), marten movements would increase. Lastly, I hypothesized that as solar radiation
increased (i.e., bright, sunny days), marten movement rates would decrease. The ability to predict
periods of high marten movements based on weather data could result in greater trapping efficiencies
for research purposes, and provide insight into periods of potentially high capture vulnerability during
public trapping seasons.
Methods
Marten Trapping
To capture marten across the East Unit of the Hiawatha National Forest, I overlaid a 5.2 km2
square grid within the Forest Service property boundary and only used grid cells that were 100% inside
this area (not borders or shoreline; Figure 2.1A). I selected this grid size to encompass multiple marten
home ranges; this area is roughly twice the home range size of marten in the East Unit of the HNF (Sault
Tribe Inland Fish and Wildlife Department, unpublished data). I used generalized random tessellation
stratified (GRTS) sampling to select 30 grid cells for trapping (Figure 2.1B). This procedure randomly
selects spatially distributed samples from a discrete population using an algorithm that avoids spatial
clustering (Stevens and Olsen 2003). I trapped marten in collaboration with the Inland Fish and Wildlife
Department of the Sault Ste. Marie Tribe of Chippewa Indians within selected grids from March 2016
through September 2016. From September 2016 through July 2017, we expanded trapping effort to a
36.25 km2 hexagon centered on the GRTS selected grid because of low capture rates. Towards the end
of the study we also opportunistically trapped areas known to historically support marten. For grid- and
hexagon-based trapping, we moved to a different area once a marten was collared. Our overall strategy
was to collar marten across a broad, partially unbiased area (Figure 2.1).
We captured marten using Tomahawk 201 and 204 live traps (Tomahawk Live Trap Company,
Tomahawk, WI). We placed traps next to coarse woody debris and covered them with leaf litter, conifer

46

boughs, tree bark, fallen logs, or natural debris found in the area. In winter we cut 19 L buckets in half
and covered each trap with straw and a half-bucket. Cover over the traps helped keep marten dry and
hidden. We baited traps with meat: chicken, venison, pork, beef, or fish heads. After placing the meat
at the back of the trap, we placed a commercial call lure, called Gusto (Caven’s Gusto, Pennock, PA), on
a branch above the trap.
We anesthetized marten using a facemask on a portable vaporizer with isoflurane in a 100%
oxygen carrier, with an induction setting of 5% and a maintenance setting of 2-3%. We weighed marten
and constantly monitored their temperature, heart rate, and breathing during the collaring procedure.
We only collared marten weighing >833 g. While under anesthesia, we fit marten with a 30g LiteTrack30
collar (SIRTRACK, Hawkes Bay, NZL). I programmed the collars to attempt a location every 15 min (Kie
2013, Moriarty and Epps 2015,). Based on this setting, I estimated that the GPS battery would last 12
days, with the VHF beacon lasting an additional 30 days. After full recovery in an enclosed container,
marten were released at the point of capture. We used VHF triangulations to monitor marten locations
for the duration of collar life. Because these collars required manual downloading to retrieve data, we
started re-trapping marten after about 2 weeks. We used the same trapping techniques except that
traps were purposefully placed where the VHF tracking results suggested the marten were
concentrating their activities. Marten were trapped and handled using approved protocols from the
Institutional Animal Use and Care Committee at Michigan State University (12/15-185-00) and Inland
Fish and Wildlife Department of the Sault Ste. Marie Tribe of Chippewa Indians.
Daily Movement
I calculated average daily and hourly movement distances for marten using the as.ltraj function
in the adehabitatHR package (Calenge 2006) in RStudio (RStudio Team 2015). I first calculated distances
traveled between successful locations, and then divided this value by the time stamp of the second

47

location in the sequence (expressed as m/min). I summarized these data for all marten by hours within a
day and seasonally. I defined seasons as winter (December, January, February, March), spring (April and
May), summer (June, July, August), and fall (September, October, November). I tested for a seasonal
difference in daily movement rates using a Kruskal-Wallis test (R Core Team 2017).
Weather Related Movements
I acquired weather data from Daymet (Thornton et al. 2016). These data were interpolated from
weather station measurements and portray several daily variables as 1 km2 grid cells across North
America (Thornton et al. 2016). To model the relationships between average daily movement of
martens and weather data, I included total precipitation of the previous day (t-1), precipitation on the
same day (t), precipitation for the next day (t+1), solar radiation (t), snow water equivalent (t),
maximum temperature (t), minimum temperature (t), vapor pressure (t), change in temperature during
the day (tmax-tmin), and change in vapor pressure from previous day. I hypothesized that precipitation
and temperature would play a role in marten movements because they struggle to thermoregulate
(Buskirk et al. 1988). Since marten are a cold adapted species, as maximum temperature increases their
movements should decrease (Cushman et al. 2011), and as low temperature decreases past a
thermoregulatory threshold, their movements should decrease (Buskirk et al. 1988, Buskirk et al. 1989).
Solar radiation refers to the intensity of sunlight reaching the ground and can be used to infer cloud
cover.
I used a generalized linear mixed model (GLMM) to explore relationships between average
marten movement rates (AMMRs) per day and the 10 weather variables (fixed effects). Additionally, I
used individual marten and season as random effects in the GLMM. I initially identified correlated (>0.5;
Dormann et al. 2012) fixed effects and then built a candidate model set using different combinations of
uncorrelated variables. I ranked the candidate model set using AICc (Burnham and Anderson 2002), and

48

for models within AICc <7 from the top ranking model I used model averaging (Bartoń 2016). I deemed a
covariate significant if the 95% confidence interval of the parameter estimate did not overlap 0.

Results
Marten Trapping
We trapped marten from March 2016 through July 2017 for 5,922 trap nights. We caught and
collared 12 marten (11 males and 1 female) around 10 of our focal areas (Figure 2.1B). We also caught
10 marten (1 male and 9 females) that were too small to collar, suggesting that our collared marten
were found in habitats similar to other marten. Annual trapping success was 269 nights/marten; trap
success was higher in late winter and spring (83 and 68 trap nights/marten, respectively) than summer
and fall (391 and 1259 trap nights/marten, respectively). Martens that were not collared weighed an
average of 608.3 g (505 g-706 g). The average weight of martens that were collared was 1064.64 g (891
g-1335 g). Telemetry data were collected in late winter (March; n = 3 marten), spring (n = 3), summer (n
= 4), and fall (n = 2).
Daily Movements
I tested every collar prior to deployment by acquiring at least one location; 100% of the collars were
functional as received from the company. The 12 collared marten produced 7,930 locations. The
batteries in the GPS lasted on average 14.7 days (SE = 1.09; range = 13-18), collecting an average of
660.8 locations (SE = 167.6; range 374 – 1061) per marten, with 620.1 (SE = 160.5; 94%) of the fixes with
DOP < 10. The average collar fix rate was 44.9% (SE = 9), comparable to other short-term GPS studies on
marten (Moriarty et al. 2015). The average time lag between successful fixes was 34.7 minutes (SE =
1.17). An average of 43.1 successful locations (SE = 2.4) per day per marten was obtained. I found that
collars had an average positional error of 6 m (DOP ≤10). I found no difference in daily movement rates
related to the trapping and collaring process (Figure 2.2), so all data were used in my analysis.

49

When pooled across all seasons, I found that marten movements throughout the 24-hour day
were highly variable, with no significant differences among hours within a day (Figure 2.3). On average,
marten moved <5 m/min regardless of hour, although rapid movements (>100m/min) were recorded for
all times of the day (Figure 2.3). Maximum observed movement rate exceeded 350 m/min (Figure 2.3).
My results indicated that marten were active throughout the day from late winter to fall, with
variability in seasonal movements lower in fall and summer (Figure 2.4). I found no significant
differences in hourly movement rates of marten within a season (Figure 2.4), however movement rates
among seasons differed (Kruskal-Wallis test, χ2 = 102.41, p < 0.001; Figure 2.5). In contrast to other
studies, I found that average hourly movement rates were higher in late winter (mean =18.16 m/min, SE
= 1.80), but with considerable variability (Figure 2.4). Movement rates in spring and summer were
similar (mean = 16.55 m/min, SE = 1.86; mean = 13.9 m/min, SE = 0.99, respectively), and lowest during
fall (mean = 7.29 m/min, SE = 0.85).
Weather Related Movements
I analyzed daily movement rates and weather for 9 of the 12 marten (Daymet data were only
available for 2016). Summer and fall received higher average daily precipitation than late winter, solar
radiation was comparable across the 3 seasons, and temperatures were cooler in late winter (summer
and fall were comparable; Table 2.1). I observed larger changes in temperature in summer and fall
compared to winter, and lower vapor pressure in winter (Table 2.1). Changes in vapor pressure were
more variable in summer and fall (Table 2.1).
The top-ranking model accounted for 59% weight of evidence and included only the random
effects (Table 2.3), indicating that unmeasured characteristics of martens and seasons explained the
most variation in daily movement rates. Univariate temperature and precipitation models dominated
the competing models (∆AIC < 7.0; Table 2.3), with these models accounting for an additional 33%
weight. Though not significant, precipitation and temperature coefficients from model averaging

50

indicated that as precipitation increased marten daily movement rates increased, and as temperature
increased movement rates decreased (Table 2.4).

Discussion
My results indicated that American marten (mostly adult males in my sample) in the Hiawatha
National Forest do not exhibit daily movement patterns similar to those found in other studies. I found
that average daily movement rates were relatively consistent throughout a 24-hour period, regardless of
season. Others found that both male and female marten activity was higher during daylight hours during
summer and nocturnal in the winter (Zielinski et al. 1983, Martin 1987). Zielinski et al. (1983) suggested
that the nocturnal pattern in winter was potentially related to capturing prey at night. Animal
movements generally result from foraging (Heinrich 1979), reproduction (Martin 1994) and avoiding
predators (Frair et al. 2005), among other factors. For animals with minimal fat reserves and high
metabolism like marten, foraging is a dominant component of daily activities. The lack of variation in
hourly movement rates within a day from my study was likely related to my sample being mostly adult
male marten. Adult male marten move in the spring and summer for mating purposes and during the
winter for foraging (Buskirk and Zielinski 1997, Drew and Bissonette 1997).
I found that marten moved significantly slower during fall compared to other times of the year
(Figure 2.3). In the eastern Upper Peninsula of Michigan, fall corresponds to an increase in the
availability of prey like voles (Getz et al. 2006), red squirrels (Tamiasciurus hudsonicus), and numerous
soft mast species, likely reducing the search times required for marten to successfully forage. Others
have found that marten were more active during summer than winter (Thompson 1986, Thompson and
Colgan 1994), whereas I found that marten were most active during late winter. Buskirk et al. (1988)
suggested that marten were more active in winter to aid in thermoregulation. Another possible
explanation could be the availability of prey. During winter prey availability decreases because the

51

majority of small mammal activity is subnivean, some prey significantly reduce their activity patterns,
many bird species have migrated (Martin 1994), and soft mast is generally unavailable. Limited
availability and access to food during winter may cause marten to travel greater distances.
Because marten are poor thermoregulators due to their small size and moderately effective fur
(Buskirk et al. 1988, Buskirk et al. 1989), I expected weather conditions like precipitation and
temperature to have negative relationships with movement rates. I failed to find a significant weather
factor that influenced adult marten movements on the HNF. Zalewski et al. (2004) suggested that
marten slowed their rate of movements as temperatures decreased, with speeds highest in summer and
lowest in fall and winter. My temperature model-averaged coefficients were all negative, lending
support to the observation by Zalewski et al. (2004). My model-averaged coefficients for precipitation
were positive, indicating that marten movement rates increased as precipitation increased. This was a
pattern I also observed with trapping success, and may be related to thermoregulation or prey activity
but the causal mechanism(s) remains unclear.
The solar radiation variable represented a measure of hotness as influenced by temperature and
sun exposure (Thornton et al. 2016); I failed to find a relationship between marten daily movement
rates and solar radiation. Vapor pressure increases the ability for some animals (rodents) to smell
(Vander Wall 2003), because higher humidity causes higher air conductance (Zalewski 1997). Some
marten prey (e.g., shrews; Otto and Roloff 2012) increase activity with precipitation, which occurs when
water vapor pressure is saturated (Vander Wall 2003). Although I predicted that marten movement
would positively relate to water vapor pressure, I failed to find a significant relationship.
I caution that my results are based on a limited number of marten (mostly males) that were
monitored intensively for relatively short time spans. The combination of frequent fixes over short time
frames (i.e., ~ 2 weeks) has been shown to reasonably represent marten home ranges (Moriarty 2014,
52

Thomasma 1996, Strickland and Douglas 1987), and although I collected hundreds of fixes for each
marten, the ~2-week time frame provided limited opportunities for individual marten to experience a
wide range of weather conditions. Additionally, marten may make movement decisions based on
weather at temporal scales finer than a day. A limitation of the Daymet data was that weather variables
were summarized over a 24 hr period.
This study provides insights into movement rates of American marten in the East Unit of the
HNF, Upper Peninsula of Michigan. The results of my study increased our understanding of marten
movement behavior (quantified rates, by day and season), and demonstrated that readily available
weather data were not useful predictors of marten movement behaviors. Hence, relying on daily
weather forecasting to improve marten capture success (presumably because of increased movements)
will probably not be effective.

53

APPENDIX

54

Table 2.1. Average (SE) and range of daily weather conditions, by season, experienced by collared American marten on the Hiawatha National
Forest, eastern Upper Peninsula of Michigan, March 2016 – July 2017.
Winter
Weather

Mean (SE)

Precipitation (t-1) (mm)

2.1 (0.6)

Precipitation (mm)

Spring

Range

Summer

Mean (SE)

Range

0, 13

NA

2.2 (0.6)

0, 13

Precipitation (t+1) (mm)

2.1 (0.6)

Solar Radiation (W/m2)

Fall

Mean (SE)

Range

Mean (SE)

NA

5.2 (1.4)

0, 35

6.0 (2.6)

0, 38

NA

NA

5.2 (1.4)

0, 35

5.4 (2.6)

0, 38

0, 13

NA

NA

5.8 (1.4)

0, 35

5.3 (2.6)

0, 38

311.1 (16.2)

144, 455

NA

NA

328.1 (13.3)

157, 467

309 (20.8)

154, 435

Snow Weight Equiv. (kg/m2)

129.5 (1.2)

120, 140

NA

NA

0 (0)

0

0 (0)

0

Maximum Temperature (°C)

6.4 (0.5)

2, 10

NA

NA

22.7 (0.6)

10, 30

20.8 (1.0)

11, 26

Minimum Temperature (°C)

-2.1 (0.5)

-11, -1

NA

NA

10.7 (0.6)

-2, 18

9.3 (1.0)

-1, 16

Delta Temperature (°C)

8.5 (0.4)

4, 12

NA

NA

12.0 (0.4)

7, 18

11.5 (0.6)

600, 1800

Vapor Pressure (Pa)

530 (17)

280, 640

NA

NA

1314 (50)

520, 2080

1206 (37)

7, 17

Delta Vapor Pressure (Pa)

12 (17)

-200, 160

NA

NA

-9 (39)

-600, 600

10 (48)

-320, 400

55

Range

Table 2.2. Model selection using AICc for weather related movement rates of American marten in the
Hiawatha National Forest, eastern Upper Peninsula of Michigan, March 2016 – July 2017. All models
with AICc weight >0.
Modela

k

AICc

Delta AIC

Weight

Cum.
Weight

Rate~null

3

670.5

0

0.592

0.592

Rate~tmax

4

674

3.48

0.087

0.679

Rate~prcp

4

674.3

3.78

0.075

0.753

Rate~tmin

4

674.4

3.94

0.069

0.822

Rate~prcpps

4

675.5

5.01

0.041

0.863

Rate~prcppr

4

675.9

5.37

0.034

0.897

Rate~swe

4

676.5

5.97

0.025

0.922

Rate~prcp+tmax

5

677.8

7.29

0.013

0.935

Rate~prcp+tmin

5

678.1

7.64

0.011

0.946

Rate~prcpps+tmax

5

678.8

8.28

0.008

0.954

Rate~prcpps+tmin

5

679.1

8.64

0.007

0.961

Rate~prcppr+tmax

5

679.5

8.95

0.005

0.966

Rate~prcp+prcpps

5

679.5

9.02

0.005

0.971

Rate~prcppr+tmin

5

679.9

9.36

0.004

0.976

Rate~srad

4

680.4

9.91

0.003

0.979

Rate~prcp+swe

5

680.5

10.02

0.003

0.982

Rate~prcpps+prcppr

5

680.7

10.18

0.003

0.986

56

Table 2.2. (cont’d).
Model

k

AIC

Delta AIC

Weight

Cum.
Weight

a

Rate~deltavp

4

680.9

10.43

0.002

0.988

Rate~prcpps+swe

5

681.4

10.85

0.002

0.990

Rate~prcppr+swe

5

681.8

11.28

0.002

0.992

Rate~prcp+prcpps+tmax

6

682.9

12.35

0.001

0.993

Rate~vp

4

683

12.46

0.001

0.995

Rate~prcp+prcpps+tmin

6

683.4

12.85

0.001

0.996

Rate~prcp+prcppr+tmax

6

683.5

12.96

0.001

0.997

Rate~prcp+prcppr+tmin

6

683.7

13.25

0.001

0.998

Rate~srad+tmax

5

683.8

13.25

0.001

0.999

Rate~prcpps+prcppr

5

684

13.55

0.001

1.000

tmax = maximum temperature, prcp = precipitation, tmin = minimum temperature, prcpps =

precipitation (t+1), prcppr = precipitation (t-1), swe = snow weight equivalent, srad = solar radiation, vp
= vapor pressure

57

Table 2.3. Generalized linear mixed model averaged coefficients (95% confidence intervals) for daily
weather variables used to describe marten movement rates in the Hiawatha National Forest, eastern
Upper Peninsula of Michigan, March 2016 – July 2017. NA denotes a variable that was not included in
the model averaging.
Model Variable

Estimate

Std. Error

95% Confidence Interval

(Intercept)

12.47315

2.53361

7.41, 17.54

Precipitation (t-1) (mm)

0.04229

0.06772

-0.09, 0.18

Precipitation (mm)

0.054087

0.07528

-0.10, 0.20

Precipitation (t+1) (mm)

0.05897

0.06660

-0.07, 0.19

Solar Radiation (W/m2)

NA

NA

NA

Snow Weight Equiv. (kg/m2)

0.03661

0.02595

-0.02, 0.09

Maximum Temperature (°C)

-0.10969

0.17367

-0.46, 0.24

Minimum Temperature (°C)

-0.05521

0.16795

-0.39, 0.28

Delta Temperature (°C)

-0.16553

0.26099

-0.69, 0.36

Vapor Pressure (Pa)

NA

NA

NA

Delta Vapor Pressure (Pa)

NA

NA

NA

58

Figure 2.1. Selection of Randomly Selected Trapping Grids. A. Grids that fall completely within the East
Unit of the Hiawatha National Forest. B. Grids selected using generalized random tessellation stratified
(GRTS) sampling. Highlighted grids are where marten were collared.

59

Figure 2.2. Average daily marten movement rate after collaring in the Hiawatha National Forest, eastern Upper Peninsula of Michigan, March
2016 – July 2017.

60

Figure 2.3. Movement rate per hour for American marten in a 24-hour day in the Hiawatha National Forest, eastern Upper Peninsula of
Michigan, March 2016 – July 2017. Box plots show the mean (dark bar), 75% quartile (box), 95% confidence interval (whiskers), and outliers.

61

Figure 2.4. Movement rate per hour, by season, for American marten in a 24-hour day in the Hiawatha National Forest, eastern Upper Peninsula
of Michigan, March 2016 – July 2017. Box plots show the mean (dark bar), 75% quartile (box), 95% confidence interval (whiskers), and outliers.

62

Figure 2.5. Seasonal average movement rates with 95% confidence intervals for American marten in the Hiawatha National Forest, eastern
Upper Peninsula of Michigan, March 2016 – July 2017.

63

LITERATURE CITED

64

LITERATURE CITED

Bartoń, K. 2016. MuMIn: Multi-model inference. R package version 1.15.6. Retrieved December 21,
2015, from http://cran.r-project.org/package=MuMIn
Birchfield, G. L. and J. E. Deters. 2005. Movement paths of displaced northern green frogs (Rana
clamitans melanota). Southeastern Naturalist 4:63-76.
Broquet, T., A, Johnson, E. Petit, I. Thompson, F. Burel, and J. M. Fryxell. 2006. Dispersal and genetic
structure in the American marten, Martes americana. Molecular Ecology 15:1689-97
Bull, E. L. and T. W. Heater. 2000. Resting and denning sites of American martens in northeastern
Oregon. Northwest Science 74(3):179-185.
Buskirk, S. W. 1983. The ecology of marten in southcentral Alaska. Ph.D. dissertation, University of
Alaska.
Buskirk, S. W., H. J. Harlow, and S. C. Forrest. 1988. Temperature regulation in American marten (Martes
americana) in winter. National Geographic Research 4:208-18.
Buskirk, S. W., S. C. Forrest, M. G. Raphael, and H. J. Harlow. 1989. Winter resting site ecology of marten
in the central Rocky Mountains. Journal of Wildlife Management 53:191-96
Buskirk, S. W. and W. J. Zielinski. 1997. American marten (Martes americana) ecology and conservation.
The Wildlife Society California North Coast Chapter Aug 12:17.
Burnham K. P. and D.R. Anderson. 2002. Model Selection and Multimodel Inference. 2. New York:
Springer-Verlag.
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
Cushman, S. A., M. G. Raphael, L. F. Ruggiero, A. S. Shirk, T. M. Wasserman, and E. C. O’Doherty. 2011.
Limiting factors and landscape connectivity: the American marten in the Rocky Mountains.
Landscape Ecology 26:1137-1149.
Drew, G. S. and J. A. Bissonette. 1997. Winter activity patterns of American martens (Martes americana):
rejection of the hypothesis of thermal-cost minimization. Canadian Journal of Zoology 75(5):
812-816.
Dormann, C. F., J. Elith, , S. Bacher, C. Buchmann, , G. Carl, G. Carré, J. R. G. Marquéz, B. Gruber, B.
Lafourcade, P.J. Leitão, T. Münkemüller, C. McClean, P. E. Osborne, B. Reineking, B. Schröder, A.
K. Skidmore, D. Zurell, and S. Lautenbach. 2012. Collinearity: a review of methods to deal with it
and a simulation study evaluating their performance. Ecography, 36: 27–46. doi:10.1111/j.16000587.2012.07348.x
65

Forester, J.D., A.R. Ives, M. G. Turner, D. P. Anderson, D. Fortin, H. L. Beyer, D. W. Smith, and M. S.
Boyce. 2007. State-space models link elk movement patterns to landscape characteristics in
Yellowstone National Park. Ecological Monographs 77(2):285-299
Frair, J. L., E. H. Merrill, D. R. Visscher, D. Fortin, H. L. Beyer, and J. M. Morales. 2005. Scales of
movement by elk (Cervus elaphus) in response to heterogeneity in forage resources and
predation risk. Landscape Ecology 20:273-287.
Getz L. L., M. K. Oli, J. E. Hofmann and B. McGuire. 2006. Vole population fluctuations: factors that
initiate and determine intervals between them in Microtus pennsylvanicus. Journal of
Mammalogy 87: 841–847.
Gilchrist, G., M. Mallory, and F. Merkel. 2005. Can local ecological knowledge contribute to wildlife
management? Case studies of migratory birds. Ecology and Society, 10(1).
Henrich, B. 1979. Resource heterogeneity and patterns of movement in foraging bumblebees.
Oecologia 40:235-245.
Kenward, R. E., S. S. Walls, and K. H. Hodder. 2001. Life path analysis: scaling indicates priming effects of
social and habitat factors of dispersal distances. Journal of Animal Ecology 70:1-13.
Kernohan, B. J. 2001. Analysis of animal space use and movements. Radio tracking and animal
populations: 125-166.
Kie, J.G. 2013. A rule-based ad hoc method for selecting a bandwidth in kernel home-range analyses.
Animal Biotelemetry 1:13.
Kochert, M. N., M. R. Fuller, L. S. Schueck, L. Bond, M. J. Bechard, B. Woodbridge, G. L. Holroyd, M. S.
Martell, and U. Banasch. 2011. Migration patterns, use of stopover areas, and austral summer
movements of Swainson’s Hawks. Condor 113:89-106.
Martin, S.K. 1987. Ecology of the pine marten (Martes americana) in Sagehen Creek,California. Ph.D.
Dissertation, University of California, Berkeley.
Martin, S.K. 1994. Feeding ecology of American martens and fishers. Pp. 297-315 in S. W. Buskirk, A.
Harestad, and M. Raphael, eds. Martens, sables and fishers: Biology and conservation. Cornell
University Press, Ithaca, NY.
Moriarity, K. M. 2014. Habitat Use and Movement Behavior of Pacific Marten (Martes caurina) in
Response to Forest Management Practices in Lassen National Forest, California. Dissertation.
Moriarty, K. M. and C. W. Epps. 2015. Retained satellite information influences performance of GPS
devices in a forested ecosystem. Wildlife Society Bulletin 39: 349–357. doi:10.1002/wsb.524
Moriarty, K. M., C. W. Epps, M. G. Betts, D. J. Hance, J. D. Bailey, and W. J. Zielinski. 2015. Experimental
evidence that simplified forest structure interacts with snow cover to influence functional
connectivity for Pacific martens. Landscape Ecology 30:1865–1877.

66

Nams, V. O., and M. Bourgeois. 2004. Fractal analysis measures habitat use at different spatial scales: an
example with American marten. Canadian Journal of Zoology, 82(11), 1738-1747.
Nielson, R. M., B. F. J. Manly, L. L. McDonald, H. Sawyer, and T. L. McDonald. 2009. Estimating habitat
selection when GPS fix success is less than 100%. Ecology 90:2956-2962.
Otto, C. R. and G. J. Roloff. 2011. Using multiple methods to assess detection probabilities of forestfloor wildlife. The Journal of Wildlife Management 75(2), 423-431.
Pace, R. M. 2001. Estimating and visualizing movement paths from radio-tracking data. In Radio tracking
and animal populations. Edited by J.J. Millspaugh and J.M. Marzluff. Academic Press, London. pp.
189-206
Powell, R.A., S.W. Buskirk, and W.J. Zielinski. 2003. Fisher and marten. In: Feldhammer, G.A., Thompson,
B.C., Chapman, J.A. (Eds.), Wild Mammals of North America. Johns Hopkins Press, Baltimore,
MD, pp. 635–649
R Core Team. 2017. R: A language and environment for statistical computing. R Foundation for Statistical
Computing, Vienna, Austria. URL https://www.R-project.org/.
RStudio Team. 2015. RStudio: Integrated Development for R. RStudio, Inc., Boston, MA URL
http://www.rstudio.com/.
Sawyer, H., M. J. Kauffman, R. M. Nielson, and J. S. Horne. 2009. Identifying and prioritizing ungulate
migration routes for landscape-level conservation. Ecological Applications 19:2016-2025.
Small. R. J. and D. H. Rusch. 1989. The natal dispersal of Ruffed Grouse. Auk 106:72-79.
Stevens, D. L., and A. R. Olsen. 2003. Variance estimation for spatially balanced samples of
environmental resources. Environmetrics (14)6: 593-610.
Stone, K. 2010. Martes americana, American marten. In: Fire Effects Information System. [Online]. U. S.
Department of Agriculture, Forest Service, Rocky Mountain Research Station. Fire Sciences
Laboratory (Producer). Available: http://www.fs.fed.us/database/eis/animals/mammal/maam/all.html. [2017, July 27].
Strickland, M. A., and C. W. Douglas. 1987. Marten. In M. Novak, J. A. Baker, M. E. Obbard, and B.
Malloch, editors. Wild furbearer management and conservation in North America. Ontario
Trappers Association, Toronto, Ontario, Canada. pp 530–546
Takekawa, J. Y., S. H. Newman, X. Xiao, D. J. Prosser, K. A. Spragens, E. C. Palm, B. Yan, F.Li, D. Zhao, D. C.
Douglas, S. B. Muzaffar, and W. Ji. 2010. Migration of waterfowl in the East Asian Flyway and
Spatial relationship to HPAI H5N1 outbreaks. Avian Diseases 54:466-476.
Thomasma, L. E. 1996. Winter habitat selection and interspecific interactions of American martens
(Martes americana) and fishers (Martes pennanti) in the McCormick Wilderness and
surrounding area. Houghton, MI: Michigan Technological University. 116 p. Dissertation. [76871]
67

Thompson, I.D. 1986. Diet choice, hunting behavior, activity patterns, and ecological energetics of
marten in natural and logged areas. Ph.D. Dissertation, Queen's University, Kingston, Ontario,
Canada.
Thompson, I. D., and P. W. Colgan. 1994. Marten activity in uncut and logged boreal forests in Ontario.
The Journal of Wildlife Management 58(2): 280-288.
Thornton, P.E., M.M. Thornton, B.W. Mayer, Y. Wei, R. Devarakonda, R.S. Vose, and R.B. Cook. 2016.
Daymet: Daily Surface Weather Data on a 1-km Grid for North America; Version 3. ORNL DAAC;
Oak Ridge; Tennessee; USA. http://dx.doi.org/10.3334/ORNLDAAC/1328
Turchin, P. 1991. Translating foraging movements in heterogeneous environments into the spatial
distribution of foragers. Ecology 72:1253-1266.
Vander Wall, S. B. 2003. How rodents smell buried seeds: a model based on the behavior of pesticides in
soil. Journal of Mammalogy 84(3): 1089-1099.
White, P. J., K. M. Proffitt, L. D. Mech, S. B. Evans, J. A. Cunningham, and K. L. Hamlin. 2010. Migration of
northern Yellowstone elk: implications of spatial structuring. Journal of Mammalogy 91:827-837
Witt, M. J., S. Akesson, A. C. Broderick, M. S. Coyne, J. Ellick, A. Formia, G. C. Hays, P. Luschi, S. Stroud,
and B. J. Godley. 2010. Assessing accuracy and utility of satellite-tracking data using Argos-linked
Fastlock-GPS. Animal Behaviour 80:571-581.
Zalewski, A. 1997. Factors affecting selection of resting site type by pine marten in primeval deciduous
forests (Bialowieza National Park, Poland). Acta Theriologica, 42:271-288.
Zalewski, A. 2000. Factors affecting the duration of activity by pine martens (Martes martes) in the
Białowieża National Park, Poland. Journal of Zoology 251(4), 439-447.
Zalewski, A., W. Jedrzejewski, and B. Jedrzejewska. 2004. Mobility and home range use by pine martens
(Martes martes) in a Polish primeval forest. Ecoscience 11(1):113-122
Zielinski, W.J., W.D. Spencer, and R.D. Barrett. 1983. Relationship between food habits and activity
patterns of pine martens. Journal of Mammalogy 64:387-396.

68

Conclusions
The purpose of my thesis was to improve our understanding of marten resource utilization and
movement behaviors in the East Unit of the Hiawatha National Forest (HNF). I modeled resource
utilization of marten that were captured across a broad site productivity and forest type gradient
throughout HNF. I found that marten resource utilization in this area could not be explained from standlevel, univariate metrics of forest structure collected as part of standard forest inventory. Rather, my
results indicated that site productivity was a more useful metric, with marten use positively related to
site productivity. I also found that activity centers for some marten were negatively related to distance
from opening, even though all marten were documented to use openings (2.4% of locations). The
effects of high-traffic roads on marten resource utilization were minimal, and for those marten that
were influenced by roads the responses were both positive and negative. My study showed that
resource utilization by individual marten in the HNF is highly variable, and given the variables I evaluated
unpredictable.
I also quantified when and how far marten moved daily, seasonally, and in relation to weather
conditions. Marten were active throughout the day, and more active in late winter than summer and
least active in fall. Movement decisions can be highly individualized based on current and past
experiences and seasonal life history demands. My results support this observation; the random effects
that accounted for individual marten and seasons received the most model support. I observed weak
precipitation and temperature relationships with movement rates, the direction of those relationships
were intuitively correct. I found weak evidence that marten movement rates decreased as temperatures
increased, and that movement rates increased on days with precipitation. My interest in the effects of
weather on movements partially stemmed from observations I made while trying to trap marten. I
observed that marten seemed more trappable in association with precipitation events, so I posited that
weather could be used to improve trapping efficiency. The effects of weather on movement can also be
69

helpful in knowing how marten in the HNF may respond to the weather ramifications of climate change.
Marten movement rates slowed when there was a large change in temperature during the day.
Limitations of this work include: 1) the short-temporal aspect of the collar that recorded marten
locations (GPS collar only active on average 14.7 days), and 2) the use of HNF forest inventory data at
the stand level that were likely coarser than the marten habitat selection process. Additional data more
accurately portraying the forest inventory at a finer scale rather than stand level may be more helpful in
identifying if there are pockets of different measurements within a stand that determine high use.
Results from my work can be used to assist the HNF in forest management decisions regarding
marten management. The HNF should manage for forest structural complexity (e.g., leave snags,
downed wood, facilitate multi-storied forest structure), and recognize that site productivity affects the
complexity of forest structure (i.e., more complex forest structure tends to occur on higher quality sites).
Targeted efforts for increasing forest structural complexity to enhance marten habitat may be
particularly effective on low productivity sites. I also recommend that the HNF include information on
snags and downed wood in their standard forest inventory system, as these habitat elements are used
by multiple wildlife species, including marten. The next phase of research should look at point-level
data (as opposed to stand-level summaries) from the forest inventory as a predictor of marten resource
utilization. Additionally, an analysis of marten movement rates through different forest types could
prove insightful, with the hypothesis that movement rates would be highest through less desirable
forest types (like openings).

70