FACTORS INFLUENCING BREEDING SEASON SURVIVAL OF FEMALE
MALLARDS IN THE GREAT LAKES REGION
By
Ryan Adam Boyer

A THESIS
Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of
Fisheries and Wildlife– Master of Science
2015

ABSTRACT
FACTORS INFLUENCING BREEDING SEASON SURVIVAL OF FEMALE MALLARDS
IN THE GREAT LAKES REGION
By
Ryan Adam Boyer
Mallards (Anas platyrhynchos) are one of the most widely studied waterfowl species in North
America, yet there are unknown demographic vital rates that are critical to enhancing the
management of this waterfowl species. Specifically, information on breeding season survival
rates of female mallards in the Great Lakes region is lacking. We estimated breeding season
survival for 484 individually radio-marked female mallards across 9 study sites in the Great
Lakes region from 2001–2003. We modeled the effects of study site, year, state, female age
(after second year [ASY] vs second year [SY]), body condition, and 3 time periods within the
breeding season. Additionally, we created utilization distributions (UDs) (i.e., home ranges),
estimated core areas for 282 individuals, and quantified land cover types in those core areas to
model the effects of those cover types on breeding season survival. Survival ranged from 0.62–
0.85 and the mean across all sites was 0.76. Survival was lowest during the peak nesting period,
but study site, year, state, age, and body condition had no significant effect on survival. Our top
model suggested that breeding season survival decreased as the percent composition of forested
cover within the core areas increased (β= –1.740, 95% CI Lower = –3.282, Upper = –0.197).
Breeding season survival estimates were similar to those estimated elsewhere and we failed to
detect a strong relationship between most land cover types within core areas and survival,
suggesting that female mallard survival in the Great Lakes region may be affected by land cover
factors not assessed as a part of this study or at alternative spatial scales.

Copyright by
RYAN ADAM BOYER
2015

I would like to dedicate the completion of my thesis to everyone in my life who has helped instill
in me the passion that I have for all things wild and to those who taught me to strive to surpass
mediocrity and to persist at all costs.

iv

ACKNOWLEDGMENTS

I would like to thank the Upper Mississippi River Great Lakes Region Joint Venture, Ducks
Unlimited Inc., and numerous private donors for providing funding for this project.
Additionally, I would like to thank my advisor Dr. Scott Winterstein, and my committee
members Dr. Gary Roloff, Dr. Dave Luukkonen, and Dr. Charles Nelson for providing support
and guidance throughout my progression as a graduate research assistant. I would also like to
sincerely thank Dr. John Coluccy of Ducks Unlimited and Ms. Rebecca Humphries who were
instrumental in the initiation of my involvement with this project.
I would like to thank the original investigators of the study from 2001–2003 including all of the
technicians and researchers. Specifically, I would like to thank John Simpson, one of the
original graduate students who willingly provided background information and valuable insight
on his experiences regarding the study. I would like to thank Rob Paige with Ducks Unlimited
for providing valuable insight as I worked through the spatial analysis component of my project.
Additionally, I would like to thank Dr. Robert Montgomery and Kyle Redilla who were an
instrumental part in assisting me with the methodology during the latter stages of the project.
I would be remiss if I didn’t thank my family for instilling in me the values that helped me
achieve this monumental goal. Lastly, but most importantly, I would like to thank my loving
wife Jaime, who was the support network and the inspiration that I needed to achieve the goal of
acquiring my Master’s degree at Michigan State University.

v

TABLE OF CONTENTS
LIST OF TABLES ............................................................................................................ vii
LIST OF FIGURES ........................................................................................................... ix
INTRODUCTION ...............................................................................................................1
CHAPTER 1: Effects of Female-Specific Characteristics on Female Mallard Survival during the
Breeding Season in the Great Lakes Region........................................................................4
Abstract ..............................................................................................................4
Introduction ........................................................................................................5
Study Area .........................................................................................................9
Methods............................................................................................................12
Trapping and Radio-marking ......................................................................12
Statistical Analyses ....................................................................................15
Results ..............................................................................................................18
Discussion ........................................................................................................27
Management Implications ................................................................................31
LITERATURE CITED ....................................................................................33
CHAPTER 2: Effects of Cover Types on Female Mallard Survival during the Breeding Season in
the Great Lakes Region ......................................................................................................39
Abstract ............................................................................................................39
Introduction ......................................................................................................40
Study Area .......................................................................................................41
Methods............................................................................................................45
Capture and Monitoring Techniques ...........................................................45
Utilization Distribution Estimation ..............................................................46
Core Area Estimation ...................................................................................47
Land Cover Types ........................................................................................47
Known-fate Models .....................................................................................49
Results ..............................................................................................................50
Discussion ........................................................................................................56
Management Implications ................................................................................60
LITERATURE CITED ....................................................................................61

vi

LIST OF TABLES
Table 1.1

Primary land cover types for the Great Lakes region and associated proportion
(%) within those categories created for the initial selection of study sites in the
Great Lakes region, 2001–2003. Study Site corresponds to the state (i.e., MI =
Michigan, OH = Ohio, WI = Wisconsin, IN = Indiana) and year (2001–2003).
....................................................................................................................11

Table 1.2

Number of individual female mallards monitored (Transmitters) by study site, age
(after second year = ASY, second year = SY, or unknown age = UN), number of
individuals with 1 nesting attempt (Nest), individuals with no recorded nesting
attempts (Nonnest), number of marked individuals on study site for 14 days
(Migrants), and number of individual mortalities (Mortalities) across all sites for
the Great Lakes region, 2001–2003. ..........................................................19

Table 1.3

Female mallard breeding season survival estimates (Ŝ) and associated standard
errors (SE) by study site and year. Survival estimates were initiated when 10
individuals were radio-marked at each site and concluded mid–July across the
Great Lake region, 2001–2003. .................................................................21

Table 1.4

Mortality sources of radio-marked female mallards during the breeding season by
study site in the Great Lakes region, 2001–2003 .......................................22

Table 1.5

Known-fate candidate model results examining the effects of female-specific
characteristics on female mallard survival (n = 484) during the breeding season in
the Great Lakes region, 2001–2003. Models are ranked based on Akaike’s
Information Criterion corrected for sample size (AICc), the difference from the
highest ranked model (∆AICc), and the model weight ..............................26

Table 2.1

Primary land cover types for the Great Lakes region and associated proportion
(%) within those categories created for the initial selection of study sites in the
Great Lakes region, 2001–2003. Study Site corresponds to the state (i.e., MI =
Michigan, OH = Ohio, WI = Wisconsin, IN = Indiana) and year (2001–2003).
....................................................................................................................44

Table 2.2

Model results of candidate models testing the effects of the proportion of
individual land cover types within the core area, distance from the core area to
individual land cover types, number of core areas, and home range size on female
mallard (n= 282) survival during the breeding season in the Great Lakes region,
2001–2003.We used known–fate models in Program Mark to evaluate competing
model likelihood using Akaike’s Information Criterion corrected for sample size
(AICc), the difference from the highest ranked model (∆AICc), and the model
weight.. .......................................................................................................51

vii

Table 2.3

Mean proportions of forested cover (P_For), grassland cover (P_Grass),
agriculture cover (P_Ag), shrubland cover (P_Shrub), and wetland cover (P_Wet)
within individual core areas of female mallards (n = 282) across 9 sites during the
breeding season in the Great Lakes region, 2001–2003 ............................53

Table 2.4

Mean home range size (ha) at the 95 percent isopleth (95_Per_HR), mean core
area isopleth (Core_Iso), mean number of core areas (Num_Cores), and the core
area size (ha) (Core_Area) for each individual female (n = 282) across the 9 study
sites in the Great Lakes region, 2001–2003. ..............................................54

Table 2.5

Mean area (ha) of individual land cover types within individual core areas of
female mallards (n = 282) across 9 sites during the breeding season in the Great
Lakes region, 2001–2003. ..........................................................................55

viii

LIST OF FIGURES
Figure 1.1

Breeding population estimates of mallards in Michigan, Minnesota, and,
Wisconsin from 1968–2013 (derived from the Waterfowl Breeding
Population and Habitat Survey conducted by the U.S. Fish and Wildlife
Service). ...........................................................................................6

Figure 1.2

Locations of the 9 study sites within the Great Lakes region where we
collected data for research on female mallard survival during the breeding
season, 2001–2003. (OH01 = Port Clinton, OH; MI01 =Riverdale, MI;
WI01 = Ripon, WI; OH02 = Stueben County, IN; MI02 = Battle Creek,
MI; WI02 = Shiocton, WI; OH03 = Akron, OH; MI03 = Big Rapids, MI;
WI03 = New Richmond, WI).. ......................................................10

Figure 1.3

Proportion of 481 radio-marked female mallards nesting during the18week breeding season from 15 March to 18 July grouped by increasing
nesting activity, peak nesting activity, and decreasing nesting activity
across 9 sites in the Great Lakes region, 2001–2003. ....................17

Figure 1.4

Breeding season survival estimates of 484 female mallards by year derived
from known-fate models in Program MARK in the Great Lakes region,
2001–2003......................................................................................24

Figure 1.5

Breeding season survival estimates of 484 female mallards by state (e.g.,
MI = Michigan, OH = Ohio, WI = Wisconsin) derived from known–fate
models in Program MARK in the Great Lakes region, 2001–2003. The
IN02 study site survival estimate was included within the OH Sites.
........................................................................................................25

Figure 2.1

Locations of the 9 study sites within the Great Lakes region where we
collected data for research on female mallard survival during the breeding
season, 2001–2003. (OH01 = Port Clinton, OH; MI01 =Riverdale, MI;
WI01 = Ripon, WI; OH02 = Stueben County, IN; MI02 = Battle Creek,
MI; WI02 = Shiocton, WI; OH03 = Akron, OH; MI03 = Big Rapids, MI;
WI03 = New Richmond, WI).. ......................................................43

Figure 2.2

Relationship between changes in breeding season survival of female
mallards across 18 weeks and percent composition of forested cover
within core use areas based on model results derived from Program
MARK in the Great Lakes region, 2001–2003. .............................52

ix

INTRODUCTION
Factors influencing mallard populations have been the focal point of multiple studies conducted
in the Prairie Pothole Region and results from those studies have long been incorporated in the
management of mallard populations throughout North America. Research suggests that mid–
continent populations of mallards are most sensitive to changes in nest success and female
survival during the breeding season (Johnson et al. 1987, Hoekman et al. 2002). However,
Coluccy et al. (2008) completed a sensitivity analysis on vital rates affecting mallard
productivity in the Great Lakes region using information collected from a 2001–2003 study (see
below), and suggested that the variability in mallard population growth was most sensitive to
changes in non–breeding survival (36%), duckling survival (32%), and nest success (16%), while
other vital rates including breeding season survival accounted for the remaining 16%.
Additionally, harvest derivation records suggested that a large proportion (22%–81%) of
mallards harvested in the Great Lakes region (i.e., Indiana, Illinois, Ohio, Michigan, and
Wisconsin) were produced locally (Zuwerink 2001). Therefore, modeling mallard productivity
using estimates of vital rates collected from mid-continent mallard populations in the Prairie
Pothole Region may not be entirely appropriate for managers in the Great Lakes region.
Although breeding season survival of female mallards accounted for 1% of the variation in Great
Lakes mallard populations Coluccy et al. (2008), we acknowledge that other vital rates such as
nest success and brood survival are dependent upon females surviving the breeding season.
Additionally, understanding the relationship between the effects of land cover types and breeding
season survival can provide important information to regional managers looking to increase a
suite of vital rates through management actions for mallards in the Great Lakes region.

1

For this thesis titled “Factors Influencing Breeding Season Survival of Female Mallards in the
Great Lakes Region” I analyzed data that were initially collected in 2001–2003, as part of a
multi-state study funded by Ducks Unlimited, Inc. (DU). DU was charged with assessing
important vital rates and habitat factors influential to mallard productivity within the Great Lakes
region in an effort to identify opportunities to potentially increase vital rates through strategic
habitat conservation and to address an important information need. The initial data collection,
radio-marking of female mallards, and the monitoring of individuals were completed during
2001–2003 prior to my involvement. Originally, three graduate students were tasked with data
collection and analyzing mallard vital rates, such as brood survival (Simpson 2005), nest survival
(Davis 2008), and a home range analysis (Reichus, unpublished data). Female survival rates
during the breeding season and factors affecting those rates were never fully assessed, but were
later identified as an important information need by DU and the Upper Mississippi River Great
Lakes Region Joint Venture for conservation planning purposes.
I compiled breeding season data and attribute information for each of the individually radiomarked females in the fall of 2012. I created study site maps and utilized nest location data for 9
study sites and visited the study sites to gain perspective on the landscape and land cover types
that were utilized by breeding females. Additionally, I toured a study site with one of the
original graduate students of the project and discussed study design and monitoring protocols
utilized by the researchers. This allowed me to gain valuable insight on specifics of the study
design and methods. I amalgamated location estimate data and attribute information including
age, body condition, and information specific to breeding efforts for each individually radiomarked female, and formatted this information to ensure accuracy and completeness of the data.
The MSU Institutional Animal Care and Use Committee determined that this study was exempt
2

from filing an Animal Use form because it only dealt with the retrieval and analysis of data from
paper or electronic records.
My objectives were to 1) estimate breeding season survival rates for female mallards, 2) examine
the effects of site, year, and states in addition to female-specific characteristics such as age (SY
vs. ASY) and body condition, and 3 periods within the breeding season (i.e., increasing nesting
activity, peak nesting activity, and decreasing nesting activity) on survival, and 3) determine
effects of land cover types within core areas, distance to land cover types such as wetland, forest,
and agriculture from core areas, home range size, and number of core areas on survival.

3

CHAPTER 1
Effects of Female-Specific Characteristics on Female Mallard Survival during the Breeding
Season in the Great Lakes Region
Abstract
Survival of breeding mallards (Anas platyrhynchos) has been studied extensively throughout
most of the United States and Canada, however, survival estimates for breeding mallards in the
Great Lakes region are lacking. We estimated the breeding season survival probability for 484
female mallards radio-marked at 9 different sites throughout Indiana, Michigan, Ohio, and
Wisconsin from 2001–2003. We investigated the effects of female- specific characteristics
including age, body condition, and 3 periods within the breeding season (e.g., increasing nesting
activity, peak nesting activity, and decreasing nesting activity) on survival. Our survival
estimates for each of the 9 sites throughout the 18-week breeding season ranged from 0.62 (SE =
0.092) to 0.85 (SE = 0.052) and averaged 0.76 (SE = 0.023). Our best approximating model
suggested that weekly survival of female mallards is lowest (0.97 ± SE = 0.004) during peak
nesting activity when the greatest proportion of females are actively nesting. Our results suggest
that female age and body condition upon arriving on the breeding grounds had no significant
effect on survival. Our survival estimates suggest that breeding season survival in the Great
Lakes region is similar to that from the Prairie Pothole Region, southern Ontario and elsewhere,
suggesting that landscape level variation in land cover types may have little effect on female
mallard breeding season survival across North America. Additional research is needed to
determine the effects of cover types on female survival, however, management efforts aimed at
restoring upland nesting cover and emergent wetlands are likely the greatest ways to increase

4

important breeding season vital rates influential to population dynamics of Great Lakes region
mallards.
Introduction
Mallard (Anas platyrhynchos) populations in the Great Lakes region have increased dramatically
leading up to the mid-1990’s, presumably as a result of eastern expansion associated with
widespread landscape alteration (Heusmann 1991), however, current information on mallard
population size estimates for Michigan, Minnesota, and Wisconsin from the Waterfowl Breeding
Population and Habitat Survey suggests that Great Lakes mallard populations have since
declined (Fig 1.1). Previous studies suggested that the Great Lakes mallard population is
relatively distinct from other populations and has become increasingly important to regional
waterfowl harvests (Anderson and Henny 1972, Munro and Kimball 1982, Zuwerink 2001,
Hoekman et al. 2006a, Coluccy et al. 2008). Mallard harvest in the Great Lakes region exceeds
that of any other duck species (Raftovich et al. 2011) and harvest in the region are primarily
derived from locally produced birds (Munro and Kimball 1982, Zuwerink 2001,).

5

Breeding Population Estimates

600000

500000

400000

300000

Michigan
Wisconsin
Minnesota

200000

100000

0

Year

Figure 1.1 Breeding population estimates of mallards in Michigan, Minnesota, and, Wisconsin from 1968–2013 (derived from the
Waterfowl Breeding Population and Habitat Survey conducted by the U.S. Fish and Wildlife Service).

6

Waterfowl incur exhaustive energetic demands throughout the course of the annual cycle, more
intensely during the breeding season with the inherent risks that are associated with reproduction
(Baldassarre and Bolen 1994). Although the breeding season accounts for a short time period
within the annual cycle for waterfowl, a large percentage of the mortality experienced by
waterfowl populations occurs during the breeding season (Johnson et al. 1992; 446). Effectively
managing waterfowl populations requires that managers understand vital rates that influence
population dynamics. Previous studies suggested that mallard population dynamics are largely
influenced by processes that occur during the breeding season, including female survival
(Hoekman et al. 2002, 2006a, Coluccy et al. 2008). Survival of female mallards during the
breeding season has been investigated extensively in the Prairie Pothole Region (Cowardin et al.
1985, Dufour and Clark 2002, Devries et al. 2003, Brasher et al. 2006) and elsewhere (Losito et
al. 1995, Hoekman et al. 2006b, Kaminski et al. 2013). However, information on this important
vital rate for mallards in the Great Lakes region is lacking.
Survival of female mallards during the breeding season may be affected by intrinsic or “femalespecific” factors in addition to extrinsic factors (e.g., habitat, predators, and weather). Previous
studies have tested the effects of age on survival and found that after-second year (ASY) females
exert a greater nesting effort than do second year females (SY) (Krapu and Doty 1979, Cowardin
el al. 1985, Hoekman et al. 2002) and therefore are believed to be at greater risk of mortality
during the breeding season (Reynolds et al. 1995). Additionally, previous studies have indicated
that female mallards suffer increased risk of mortality while actively nesting compared to other
periods within the breeding season (Johnson and Sargeant 1977, Sargeant et al. 1984, Devries et
al. 2003, Brasher et al. 2006, Hoekman et al. 2006a, Arnold et al. 2012).

7

Female mallard body condition and its effects on survival during hunting and over-wintering
periods have been studied extensively (Reinecke et al. 1987, Bergan and Smith 1993, Jeske et al.
1994), however information on the relationship between body condition and survival during the
breeding season is lacking. Devries et al (2008) tested the effects of body condition on
reproductive measures such as nesting propensity, nest initiation, clutch size, nest survival, and
timing of hatch for mallards, but never directly assessed the impact that body condition may have
on survival. Although female survival is not completely independent of these reproductive
measures, information regarding the effects that body condition may have specific to breeding
season survival is lacking. Understanding the effects of body condition as it relates to breeding
season survival would provide important information for waterfowl managers throughout the
Great Lakes region striving to increase mallard productivity.
Female mallards increase the amount of time spent on nest attendance as individual’s progress
from nest initiation to clutch completion and incubation (Loos and Rohwer 2004). Devries et al.
(2003) and Brasher et al. (2006) found an inverse relationship between the proportion of nesting
females and the probability of survival during the breeding season, suggesting that survival was
lowest when the greatest numbers of females were actively nesting. Hoekman et al. (2006a.)
tested a similar effect on female mallards in Ontario, Canada after noting a mid–season decline
in survival; however, an inverse relationship between proportion of nesting females and survival
was not found. Assessing survival probabilities of female mallards during periods of increased
nesting activity with these data could be beneficial for differentiating between the Great Lakes
region mallard population and mallards elsewhere. Additionally, examining the effects of
survival during times of increased nesting activities could direct the need for further

8

investigation on the effects of land cover types at a spatial and temporal scale that encompass
nest sites and areas of repeated use by females during the breeding season.
From 2012–2014, we analyzed daily location estimates and individual female breeding season
attribute data collected from the Great Lakes region in 2001–2003. Our objectives were to: 1)
estimate weekly breeding season survival rates for female mallards across 9 sites and 2) evaluate
the effects of female-specific characteristics including age and body condition, and evaluate 3
periods within the breeding season (e.g., increasing nesting activity, peak nesting activity, and
decreasing nesting activity) on survival.
Study Area
We selected 9 study sites across 4 different states in the Great Lakes region (Fig. 1.2). We
selected study sites based on incorporating varying percentages of agriculture and forest land
cover types found in the Great Lakes region of the United States (Table 1.1).We utilized
National Land Cover Data (NLCD; USGS 2003) with 30–m resolution to quantify percentages
of cash crops, dairy pasture, and deciduous forest. We used ERDAS IMAGINE 8.4 GIS
software to create an adjusted land cover map of the Great Lakes region by attributing 30 x 30–m
grid cells based on the proportion of the aforementioned cover types within a 2 x 2 km area. We
selected areas that had wetland densities high enough to support breeding populations of
mallards over other potential sites of similar land cover types with lower wetland densities. We
took into consideration logistical concerns during the final selection of the study sites so that
housing existed for field staff along with a series of adequate road networks to maximize access
within the study area.

9

Figure 1.2 Locations of the 9 study sites within the Great Lakes region where we collected data
for research on female mallard survival during the breeding season, 2001–2003. (OH01 = Port
Clinton, OH; MI01 =Riverdale, MI; WI01 = Ripon, WI; OH02 = Stueben County, IN; MI02 =
Battle Creek, MI; WI02 = Shiocton, WI; OH03 = Akron, OH; MI03 = Big Rapids, MI; WI03 =
New Richmond, WI).

10

Table 1.1 Primary land cover types for the Great Lakes region and associated proportion (%) within those categories created for the
initial selection of study sites in the Great Lakes region, 2001–2003. Study Site corresponds to the state (i.e., MI = Michigan, OH =
Ohio, WI = Wisconsin, IN = Indiana) and year (2001–2003).
Proportion of
Land Use Category
Region
Study Site
Low Forest-High Agriculture
28.4%
MI01,OH01, WI01, WI02
Intermediate Forest- High Agriculture
14.7%
IN02, WI03
Intermediate Forest- Intermediate Agriculture
12.0%
MI02, OH03
High Forest- Intermediate Agriculture
8.3%
MI03
High Forest- Low Agriculture
7.0%
Intermediate Forest Low Agriculture
6.0%
Low Forest- Intermediate Agriculture
2.0%
Low Forest- Low Agriculture
urban
High Forest- High Agriculture
n/a

11

In 2001, study sites were located near Port Clinton, Ohio (41º30'N, 84º59'W); Riverdale,
Michigan (43º23'N, 84º50'W); and Ripon, Wisconsin (43º50'N, 88º50'W). In 2002, study sites
were located by Stueben County, Indiana (41º38'N, 84º59'W); Battle Creek, Michigan (42º27'N,
85º13'W); and Shiocton, Wisconsin (44º26'N, 88º34'W). In 2003, study sites were located by
Akron, Ohio (41º04'N, 81º13'W); Big Rapids, Michigan (43º37'N, 85º13'W); and New
Richmond, Wisconsin (45º11'N, 92º 23'W).
Methods
Trapping and Radio-marking
We trapped and radio-marked approximately 60 female mallards prior to nesting using
conventional decoy traps at each study site, during late March and early April (Sharp and
Lokemoen 1987). We selected decoy trap site locations by scouting and selecting wetland basins
sufficient in size to hold breeding mallard pairs and selected wetland basins that pairs were
actively using. In certain cases, we trapped several paired hens on the same wetland basin
throughout the spring. Land ownership often limited the ability to randomly distribute our trap
site locations across the study sites. We banded females with standard United States Geological
Survey leg bands and implanted 22–g abdominal Very High Frequency (VHF) transmitters
(Advanced Telemetry Systems, Isanti, MN) using surgical procedures outlined by Korschgen et
al. (1984). We held radio-marked females for one hour following surgery and released
individuals at the trap site location. Additionally, we released males that were captured and
banded with the females to avoid disrupting pair bonds.
We monitored and estimated locations of radio-marked females 1–6 times daily using truckmounted null-array systems (Kenward 1987). We monitored females between the hours of 0600
and 1300 when laying females are likely to be found on the nest (Gloutney et al. 1993).
12

Researchers used hand-held tracking equipment when locations could not be obtained from the
vehicle with null-array systems and to confirm nesting attempts by breeding females. We
utilized aircraft to locate missing females after extensive ground searching throughout the study
area was unsuccessful. We estimated Universal Transverse Mercator (UTM) locations via
triangulation using LOCATE software (Pacer, Truro, NS, Canada).
We measured the body mass (nearest 10 g) for each radio-marked female, with a 1.5-kg Pesola
hand-held spring scale. Additionally, we recorded a series of structural measurements including
skull length (nearest 0.1 mm) from the occipital protuberance to the bill tip, central culmen
(nearest 0.1 mm) from the intersection of skin and premaxilla to the bill tip, tarsus length (nearest
0.1 mm) from the proximal to the lateral condyles of the metatarsus, keel length (nearest 0.1
mm) from the tip of the cranial process to the end of the medial caudal process, and wing chord
(nearest 1 mm) from the wrist on bent wing to the tip of the longest primary. To aid in
accurately classifying age as either second year (SY) or after second year (ASY) based on
feather characteristics, we collected the greater secondary coverts from the right wing (Krapu et
al. 1979, R. Clark, Canadian Wildlife Service, unpublished data). We used body mass as an
index of body condition but body mass can vary in response to structural size (Alisauskas and
Ankeny 1987, Ankeny and Afton 1988, Ankney and Alisauskas 1991). Thus, we conducted a
principle components analysis using four of the structural measurements to correct body mass for
structural size. We failed to record culmen measurements for 31 females and thus we excluded
the culmen measurement from the principal component analysis. We calculated a bodycondition index as the first principle component (PC1) of the four structural variables measured
on each female using PROC PRINCOMP in SAS (SAS Institute 2004). To determine if body
mass was related to body size, we regressed body mass on PC1. Body mass was related to body
13

size (F = 148.02, P < 0.0001, R2 = 0.22) as indicated by the following relationship: Body mass =
1,097.10±3.32 + 33.32±2.74 × (PC1). We used residuals from the regression to derive a size
corrected body mass (Yi) using the following equation (Ankney and Alisauskas 1991):

Where Yobs is unadjusted body mass and

is the mean of unadjusted body mass for all

females.
We collected location data and information regarding breeding season behavior for individuals
up to 24 weeks after radio-marking. Survival across all study sites was constant (1.0) beyond
week 16, therefore we truncated the length of the study period from 24 weeks (mid-Mar – endAug) to 18 weeks (mid-Mar – mid-Jul) for the survival analyses. Similar studies estimating
breeding season survival of female mallards in other regions estimated survival over the course
of a 13–week (Devries et al. 2003) and 16–week period (Hoekman et al. 2006a). However, since
estimating brood survival was one of the original emphases of the 2001–2003 study (Simpson
2005), we had location estimates and breeding season data on individual females up to 24 weeks.
We estimated the 18-week breeding season survival for 484 radio-marked female mallards (n =
53 at MI01, n = 56 at OH01, n = 53 at WI01, n = 52 at MI02, n = 54 at IN02, n = 56 at WI02, n =
56 at MI03, n = 54 at OH03, n = 50 at WI03) at 9 study-areas from 2001–2003.
We initiated survival estimates by site, year, and state when ≥ 10 radio-marked individuals had
entered the study following radio-marking. Following a failed nesting attempt, we monitored
females that were found grouped with other individuals and not paired with a male for 2
consecutive days to ensure no additional nesting attempts occurred. We classified individuals no
longer exhibiting breeding season behavior as molt migrants and concluded the monitoring for
14

those individuals. We right–censored individuals classified as molt migrants and those that had
emigrated from the study sites. Thirty–five individuals were believed to have suffered radio
failure throughout the course of the study. We determined the cause of mortality upon location
of the transmitter and the carcass when remains were found. In many cases the cause of death
could not be determined and we classified the source of mortality as unknown.
Statistical Analyses
We utilized Program MARK (White and Burnham 1999) known–fate models (White and Garrott
1990) to determine the maximum likelihood estimates of female survival at weekly intervals
throughout the 18–week study period. We coded individuals as having survived the weekly
interval in the analysis if they were located during the weekly interval and known to have
survived throughout the entire 7-day period. We temporarily censored individuals whose fate
was not recorded throughout one or multiple weekly intervals until fate could be determined.
We estimated survival by study area, state, and year prior to estimating the entire sample (i.e., all
individuals across all sites) to test for random effects that may have influenced the resulting
estimates. We used generalized linear models and the logit link function for testing the effects of
female-specific characteristics of age and body condition, and the 3 periods within the breeding
season. We modeled survival using a binomial distribution of errors and the logit link function
due to our response variable of weekly survival being formatted in a binomial fashion (e.g., 0 =
alive, 1 = dead) (McCullagh and Nelder 1989). We incorporated an identity design matrix and
the sin link function for estimating weekly survival estimates by individual study sites due to
numerical precision errors produced when calculating parameter estimates and standard errors
using a full model design matrix in conjunction with the logit link function. We maintained a
full model design matrix and the logit link function to determine maximum likelihood estimates
15

of female survival when estimating weekly survival across all sites for all individuals and when
testing the effects of individual covariates on survival. We developed our candidate list of a–
priori models including the effects of female age, body condition, and period within the breeding
season (Table 1.2). We compiled the list largely based on information gathered from previous
studies (see Devries et al. 2003, Hoekman et al. 2006a, Kaminski et al. 2013) as well as
information collected throughout the course of the study. Our a–priori candidate model list
included all additive combinations of covariates and a 2–way interaction between the covariates
of female age and body condition. We hypothesized that older females (ASY) may be more
experienced and arrive on breeding areas in better condition than younger (SY) females,
therefore we felt establishing an interaction between the two covariates made biological sense.
We used Akaike’s Information Criterion adjusted for small sample size (AICc; Hurvich and Tsai
1995) to select among competing models. Finally, we used AICc weights to assess the relative
likelihood of the candidate list of models (Burnham and Anderson 1998).
We classified 3 time periods as (1) weeks 1–6 following initial marking, (2) weeks 7–11
following initial marking, (3) weeks 12–18 following initial marking and representing the
conclusion of the study period. The 3 periods corresponded to when the nesting activities of the
greatest proportion of females were increasing, at peak, and decreasing (Fig. 1.3). We examined
the initiation and termination dates for each individual female that recorded a nesting attempt and
compared the proportion of females known to be actively nesting throughout the 18-week
breeding season. Multiple extrinsic factors may have influenced the peak of nesting activities
across all 9 sites; however, the peak of nesting activity occurred between weeks 7–11 across all
sites.

16

Proportion of hens nesting

1.0
0.9
0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0.0
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18
Week after initiation of marking

Figure 1.3 Proportion of 481 radio-marked female mallards nesting during the18-week breeding
season from 15 March to 18 July grouped by increasing nesting activity, peak nesting activity,
and decreasing nesting activity across 9 sites in the Great Lakes region, 2001–2003.

17

Results
We trapped 536 individual females and fit them with VHF implant radio transmitters (Table 1.2).
Two- hundred-thirty individuals were classified as ASY females, 303 as SY, and 3 individuals
were of unknown age (Table 1.2).We included individual females with unknown age
classifications in the overall survival estimates; however, we removed those individuals from the
sample when testing the effects of female-specific characteristics on survival that included age.
We categorized 49 individuals that were recorded on site for < 14 days as migrants and removed
those individuals from further consideration. Additionally, we removed females from the
analysis that died within 7 days of being radio-marked (n = 3). Few individuals (n = 123) were
monitored as late as 18–weeks after the onset of monitoring the earliest marked individuals.

18

Table 1.2 Number of individual female mallards monitored (Transmitters) by study site, age (after second year = ASY, second year =
SY, or unknown age = UN), number of individuals with 1 nesting attempt (Nest), individuals with no recorded nesting attempts
(Nonnest), number of marked individuals on study site for 14 days (Migrants), and number of individual mortalities across all sites
for the Great Lakes region, 2001–2003.
Study Site
Transmitters
ASY SY UN Nest Non-nest Migrants
Mortalities
MI01
59
25
34
0
47
6
6
11
OH01
57
32
25
0
45
11
1
13
WI01
60
30
29
1
45
10
5
11
IN02
60
32
27
1
47
7
6
7
MI02
60
24
35
1
39
13
8
10
WI02
60
25
35
0
45
12
3
14
MI03
60
19
41
0
45
11
4
9
OH03
60
24
36
0
41
13
6
7
WI03
60
19
41
0
38
12
10
12
Total
536
230 303 3
392
95
49
94

19

Survival by site ranged from 0.624 (SE = 0.092) at WI03 to 0.847 (SE = 0.054) at IN02, while
survival across all sites for all individuals was 0.758 (SE = 0.023). We initiated survival
estimates by individual site when ≥ 10 individuals were radio-marked and concluded the
breeding season survival estimates upon the week of 12 July to remain consistent with week 18
of the overall study period length (Table 1.3). Ninety-four individuals suffered mortalities
throughout the 18-week breeding season. Fifty percent (n = 47) of deaths were the result of
mammalian depredation, 15% (n = 14) resulted from avian depredation, 7% (n = 7) were killed
by farm equipment (e.g., hay cutting machinery), and 28% (n = 26) were classified as unknown
(Table 1.4). One-hundred twenty-three individual females were known to have survived through
the week 18 sampling interval of the breeding season. We right censored 267 females in the
analysis due to a combination of radio failures (n = 35), and emigration from the study sites.

20

Table 1.3 Female mallard breeding season survival estimates (Ŝ) and associated standard errors
(SE) by study site and year. Survival estimates were initiated when 10 individuals were radiomarked at each site and concluded mid–July across the Great Lake region, 2001–2003.
Site Year Weeks
Ŝ
SE
MI01 2001
16
0.7491 0.0664
OH01 2001
17
0.7294 0.0649
WI01 2001
15
0.8003 0.0565
MI02 2002
17
0.7267 0.0748
IN02 2002
18
0.8471 0.0537
WI02 2002
15
0.726
0.069
MI03 2003
16
0.7884 0.0663
OH03 2003
16
0.7838 0.0785
WI03 2003
15
0.6244 0.0918
Mean
18
0.7573 0.0227

21

Table 1.4 Mortality sources of radio-marked female mallards during the breeding season by
study site in the Great Lakes region, 2001–2003.
Farm
Study Area
Mammalian
Avian
Unknown Total
Equipment
IN02
3
0
0
4
7
MI01
5
2
1
3
11
MI02
8
0
2
0
10
MI03
4
3
1
1
9
OH01
7
1
2
3
13
OH03
2
2
0
3
7
WI01
5
3
0
3
11
WI02
8
2
0
4
14
WI03
5
1
1
5
12
Total
47
14
7
26
94

22

Differences in survival among female age (β = 0.284, SE = 0.216, 95% CI Lower = -0.140,
Upper = 0.708) and body condition (β = 0.002, SE = 0.003, 95% CI Lower =-0.003, Upper =
0.007) were not significant. The principle component analysis for corrected body mass indicated
that ASY females had greater mass ( = 1104.95 g, SE = 2.808) than did SY females ( =
1092.39 g, SE = 2.331). We examined the effect of year to year variability (Fig. 1.4) and the
effect of state (Fig. 1.5) on female survival throughout the breeding season and found no
significant differences. Models that included site received little support (AICc weight < 0.001)
and differences among sites were not significant (Table. 1.5).

23

1
Breeding Season Survival
Probability (95% CI)

0.9
0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0
2001
2002
2003
Figure 1.4 Breeding season survival estimates of 484 female mallards by year derived from
known-fate models in Program MARK in the Great Lakes region, 2001–2003.

24

1
Breeding Season Survival
Probability (95% CI)

0.9

0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0

MI Sites
OH Sites
WI Sites
Figure 1.5 Breeding season survival estimates of 484 female mallards by state (e.g., MI =
Michigan, OH = Ohio, WI = Wisconsin) derived from known–fate models in Program MARK in
the Great Lakes region, 2001–2003. The IN02 study site survival estimate was included within
the OH Sites.

25

Table 1.5 Known-fate candidate model results examining the effects of female-specific characteristics on female mallard survival (n =
484) during the breeding season in the Great Lakes region, 2001–2003. Models are ranked based on Akaike’s Information Criterion
corrected for sample size (AICc), the difference from the highest ranked model (∆AICc), and the model weight.
Modela
Parameters
AICc
∆AICc
Model Weight
{S(Nesting Period)}
3
866.4664
0
0.31786
{S(Nesting Period + Age)}
4
866.8938
0.4274
0.2567
{S(Nesting Period + Age* Condition)}
6
867.5294
1.063
0.18682
{S(Nesting Period + Condition)}
4
868.1486
1.6822
0.13707
{S(Nesting Period + Age+ Condition)}
5
868.7819
2.3155
0.09987
{S(Nesting Period + Site)}
11
878.4243
11.9579
0.0008
{S(Nesting Period + Site+ Age)}
12
879.2238
12.7574
0.00054
{S(Nesting Period + Site+ Condition)}
12
880.2316
13.7652
0.00033
{S(.)}
1
902.0776
35.6112
0
{S(Age)}
2
902.3274
35.861
0
{S(Age* Condition)}
4
903.139
36.6726
0
{S(Condition)}
2
903.5702
37.1038
0
{S(Age+ Condition)}
3
904.096
37.6296
0
{S(Site)}
9
912.5018
46.0354
0
{S(Site+ Age)}
10
913.1673
46.7009
0
{S(Site+ Condition)}
10
914.2381
47.7717
0

26

Our top ranking model provided support for female survival being adversely affected when the
greatest proportion of females are at the peak of nesting activity (Table 1.5). The weekly
survival estimates for period 1 was 0.997 (SE = 0.001), for period 2 it was 0.971 (SE = 0.004),
and for period 3 it was 0.981 (SE = 0.004). Additive affects of female age and body condition
included with nesting period were among the top ranking models (ΔAICc ≤ 2.3). Breeding
incidence among SY females was 0.769 and 0.857 for ASY females.
Discussion
Breeding season survival estimates are available for mallard populations throughout North
America; however to our knowledge our study is the first to estimate breeding season survival
for female mallards in the Great Lakes region. Our 18-week breeding season survival estimates
for female mallards in the Great Lakes region ( = 0.75, range = 0.62–0.85) (Table 2) were
similar to estimates for female mallards in New Brunswick, Canada ( = 0.80, 95% CI 0.60–
0.95; Petrie et al. 2000), the Prairie Pothole Region ( = 0.76, range 0.63–0.84; Devries et al.
2003), southern Ontario ( = 0.75, range 0.65–0.84; Hoekman et al. 2006a), and central New
York ( = 0.78, range 0.75–0.81; Kaminski et al. 2013). Estimates derived throughout other
regions of the mallard breeding range in North America in the aforementioned studies are
comparable to ours based on similar study designs, trapping techniques, and radio transmitter
types that were used. The use of abdominal implant transmitters likely minimized radio effects
on breeding season behavior and ultimately the survival and reproductive efforts of females
(Rotella et al. 1993, Paquette et al. 1997). Earlier studies estimating female mallard survival
using external transmitters may have resulted in biased survival estimates due to having adverse
effects on reproductive efforts of females (Pietz et al. 1993, Paquette et al. 1997).

27

We found no evidence to suggest that age (ASY vs. SY) had an effect on the survival of female
mallards in the Great Lakes region (β = 0.284, SE = 0.216, 95% CI Lower = -0.139 Upper =
0.708). Previous studies have suggested that older females exhibit a greater nesting effort, and
therefore may be more susceptible to being killed (Cowardin et al 1985, Reynolds et al. 1995).
Additionally, our data suggested that breeding efforts between ASY and SY females were
relatively similar, with 203 SY females recording ≥ 1 nesting attempt and 186 ASY females
recording ≥ 1 nesting attempt. However, nearly twice as many SY individuals recorded no
nesting attempt (n = 61) when compared to ASY females (n = 31). Our trapping methods were
likely biased towards capturing breeding females with established territories and due to the
difficulties associated with locating females initiating nests and nesting attempts that may have
been missed prior to the onset of marking individuals (McPherson et al. 2003), therefore, it is
plausible that we overestimated the number of SY and ASY females that failed to record a
nesting attempt. Our results partially support the alternate hypothesis that SY females
experience higher annual survival by forgoing breeding efforts altogether or due to higher
nonbreeding survival (Dufour and Clark 2002). In their assessments of female mallard survival
during the breeding season, Cowardin et al. (1985) and Hoekman et al. (2006a) found no support
for large age related differences in survival between ASY and SY females. Conversely,
Reynolds et al. (1995) and Devries et al. (2003) found that SY females had slightly higher
breeding season survival than did ASY females.
Our results suggest that body condition of female mallards upon arriving to breeding areas had
no significant effect on breeding season survival. Previous studies have assessed the importance
of body condition to survival during periods of the annual cycle outside of the breeding season
(Reinecke et al. 1987, Bergan and Smith 1993, Jeske et al. 1994), but to our knowledge none
28

have assessed the relationship of body condition on the survival of female mallards during the
breeding season. Arnold et al. (2012) modeled the effect of body mass on breeding season
survival of female mallards and found that survival was negatively correlated to body mass for
brood-rearing females trapped on the nest, but not for those that were decoy trapped; however,
body mass had no effect on survival during the prenesting, incubation, and postnesting stages of
the breeding season. Devries et al. (2008) found that female mallards in better condition had
increased nesting propensity, earlier nest initiation dates and larger clutch sizes. We never tested
the effects of female body condition on these reproductive measures, however further analysis
may be informative for Great Lakes region managers seeking to increase mallard productivity by
increasing the quality of habitats available to mallards during spring migration.
Our results indicated a strong relationship between a mid-season decline in survival and
increased proportions of nesting females similar to those found by Sargeant et al. (1984) and
Devries et al. (2003) who reported that females are at a greater risk of being killed while actively
nesting than at any other time during the breeding season. Sixty-four of the 94 mortalities
recorded occurred within weeks 7-11 during the peak nesting period. During the study only 34
percent of the females that died (n = 31) were recovered on or near the nest. Additional females
were recorded nesting up to the date of mortality and may have been killed while actively
nesting, however, the inability to locate the carcass and transmitter immediately after the death of
the female or within close proximity of the nest site limited our ability to make that conclusion.
Furthermore, we believe that this number may be biased low due to the lack of ability to detect
nests during early laying (McPherson et al. 2003). Conversely, increased mortality during the
peak of the nesting period could be related to prey-switching behavior of common aerial and
terrestrial predators and affected less by the specific breeding activity (Arnold et al. 2012).
29

Throughout the study 5 females were killed during the brood-rearing season, and 2 individuals
were killed while taking a nest break. This would suggest that although a small proportion of
individuals were killed not actively attending a nest, certain predators are proficient at capturing
and killing females away from nests.
Our results suggested no significant difference in survival across study sites, despite there being
great biological variation in survival rates between the WI03 site (Ŝ = 0.624, SE = 0.092) and the
IN02 site (Ŝ = 0.847, SE = 0.054). The survival estimate at the WI03 site was greatly influenced
by the sample size which was drastically reduced throughout the breeding season following the
radio-marking of individuals. Ten radio-marked individuals experienced radio failures, and we
classified 10 additional individuals as migrants that were radio-marked and on site <14 days. In
addition, we right-censored 31 individuals prior to the conclusion of the 18-week breeding
season, thus reducing the number of individuals at risk across all occasions and influencing the
survival estimate at the WI03 site. Our inability to detect a significant difference in survival
across the additional sites may be due to the similarities in the land cover types utilized by
breeding mallards within the sites. Previous studies have suggested that an increase in forested
habitat types suitable to avian predators may increase their abundance and therefore increase
mortality rates of waterfowl (Sargeant et al. 1993, Losito et al. 1995, and Hoekman et al. 2006a).
We selected our study sites to represent the variation of land cover types found within the Great
Lakes region by selecting study sites with varying percentages of forest and agriculture cover
types at the study area scale, yet we were unable to detect any significant effect on breeding
season survival. Mallards use a variety of semi–permanent and permanent wetland basins in the
Great Lakes region (Merindino et al. 1995), and these wetland basins attract and support
populations of mink (Neovison vison); (Arnold and Fritzell 1990), which are proficient at killing
30

female mallards away from the nest (Sargeant et al. 1973, Eberhardt and Sargeant 1977). Mink
were identified explicitly as the cause of female mallard death only once, however mink were
likely responsible for additional mortalities suffered when females were off the nest. Three
females were recorded being killed by farm equipment (i.e., hay cutting machinery) while
actively attending a nest. Hoekman et al. (2006a) found that female mortality increased with the
onset of haying activities, and in regions of North America where hay cover types dominate the
landscape, the effects of haying activities may have far greater impacts on local populations of
breeding mallards than found on our study sites.
Management Implications
Breeding season survival estimates for female mallards across the Great Lakes region appear to
be relatively consistent with previous estimates throughout mallard breeding ranges in North
America. Similarities in breeding season survival probabilities across regions suggest that
variations among landscape level cover types may have little effect on this vital rate at large
scales and that female mallards appear to be generalists (e.g., adaptable) when selecting habitats
to meet their breeding season requirements. Our results indicated that a majority of the mortality
suffered by female mallards during the breeding season in the Great Lakes region occurred
during the period when the greatest proportion of females are actively nesting and resulted
mostly from avian and mammalian predators. Additional research is needed to determine if the
effects of land cover types surrounding the nest site and high use areas within individual home
ranges influence survival in the Great Lakes region; however, efforts to increase female survival
through improving nesting cover may benefit a suite of vital rates affecting mallard productivity
and should remain an important emphasis for Great Lakes managers. Few females were killed
by farm equipment attending nests in hayland cover; however, efforts to delay haying or altering
31

hay machine equipment to include flushing bars may provide benefits to local mallard
populations where high densities of breeding mallards and haying operations are prevalent. We
failed to detect any significant effect of body condition on survival. Additionally, we failed to
detect a relationship between age and survival thus removing the need to incorporate an age
effect on breeding season survival in demographic models for breeding mallards in the Great
Lakes region.

32

LITERATURE CITED

33

LITERATURE CITED
Alisauskas, R. T., and C. D. Ankeny. 1987. Age-related variation in the nutrient reserves of
breeding American coots (Fulica americana). Canadian Journal of Zoology 65:2417–
2420.
Ankney, C. D., and A. D. Afton. 1988. Bioenergetics of breeding Northern Shovelers: diet,
nutrient reserves, clutch size, and incubation. Condor 90:459–472.
Ankney, C. D., and R. T. Alisauskas. 1991. Nutrient reserve dynamics and diet of breeding
female Gadwalls. Condor 93:799–810.
Anderson, D. R., and C. J. Henny. 1972. Population ecology of the mallard. I. U.S. Fish and
Wildlife Service Resource Publication 105, Washington, D.C., USA.
Arnold T. W., E. A. Roche, J. H. Devries, and D. W. Howerter. 2012. Costs of reproduction in
breeding female mallards: predation risk during incubation drives annual mortality.
Avian Conservation and Ecology 7(1):1.
Arnold, T. W., and E. K. Fritzell. 1990. Habitat use by male mink in relation to wetland
characteristics and avian prey abundance. Canadian Journal of Zoology 68:2205–2208.
Baldassarre, G. A., and E. G. Bolen. 1994. Waterfowl ecology and management. John Wiley
and Sons, Inc., New York, New York, USA.
Bergan, J. F., and L. M. Smith. 1993. Survival rates of female Mallards wintering in the Playa
Lakes region. Journal of Wildlife Management 57:570–577.
Brasher, M. G., T. W. Arnold, J. H. Devries, and R. M. Kaminski. 2006. Breeding season
survival of male and female mallards in Canada’s prairie-parklands. Journal of Wildlife
Management 70:805–811.
Burnham, K. P., and D. R. Anderson. 1998. Model selection and inference: a practical
information–theoretic approach. Springer–Verlag, New York, New York, USA.
Coluccy, J. M., T. Yerkes, R. Simpson, J. W. Simpson, L. Armstrong, and J. I. Davis. 2008.
Population dynamics and sensitivity analyses of breeding mallards in the Great Lakes
states. Journal of Wildlife Management 72:1181–1187.
Cowardin, L. M., D. S. Gilmer, and C. W. Shaiffer. 1985. Mallard recruitment in the
agricultural environment of North Dakota. Wildlife Monographs 92:1–37.
Davis, J. I. 2008. Mallard nesting ecology in the Great Lakes. Thesis, Humboldt State
University, Arcata, California, USA.

34

Devries, J. H., J. J. Citta, M. S. Lindberg, D. W. Howerter, and M. G. Anderson. 2003.
Breeding-season survival of mallard females in the prairie pothole region of Canada.
Journal of Wildlife Management 67:551–563.
Devries, J. H., R. W. Brook, D. W. Howerter, and M. G. Anderson. 2008. Effects of spring
body condition and age on reproduction in mallards. The Auk 125:618–628.
Dufour, K. W., and R. G. Clark. 2002. Differential of yearling and adult female mallards and its
relation to breeding habitat conditions. The Condor 104:297–308.
Eberhardt, L. E., and A. B. Sargeant. 1977. Mink predation on prairie marshes during the
waterfowl breeding season. Pages 33–43 in R. L. Phillips and C. Jonkel, editors.
Proceedings of the 1975 Predator Symposium, Montana Forest Conservation Experiment
Station, University of Montana, Missoula, USA.
Gloutney, M. L., R. G. Clark, A. D. Afton, and G. L. Huff. 1993. The timing of nest searches
for upland nesting waterfowl. Journal of Wildlife Management 57:597–601.
Heusmann, H. W. 1991. The history and status of the mallard in the Atlantic flyway. Wildlife
Society Bulletin 19:14–22.
Hoekman, S. T., L. S. Mills, D. W. Howerter, J. H. Devries, and I. J. Ball. 2002. Sensitivity
analyses of the life cycle of mid-continent mallards. Journal of Wildlife Management
66:883–900.
Hoekman, S. T., T. S. Gabor, R. Maher, H. R. Murkin, and M. S. Lindberg. 2006a.
Demographics of breeding female mallards in Southern Ontario, Canada. Journal of
Wildlife Management 70:111–120.
Hoekman, S. T., T. S. Gabor, M. J. Petrie, R. Maher, H. R. Murkin, and M. S. Lindberg. 2006b.
Population dynamics of mallards breeding in agricultural environments in eastern
Canada. Journal of Wildlife Management 70:121–128.
Hurvich, C. M., and C.–L. Tsai. 1995. Model selection for extended quasi–likelihood models in
small samples. Biometrics 51:1077–1084.
Jeske, C. W., M. R. Szymczak, D. R. Anderson, J. K. Ringelman, and J. A. Armstrong. 1994.
Relationship of body condition to survival of Mallards in San Luis Valley, Colorado.
Journal of Wildlife Management 58:787–793.
Johnson, D. H., and A. B. Sargeant. 1977. Impact of red fox on sex ratio of mallards. U.S. Fish
and Wildlife Service Wildlife Research Report 6.
Johnson, D. H., D. W. Sparling, and L. M. Cowardin. 1987. A model of the productivity of the
mallard duck. Ecological Modeling 38:257–276.

35

Johnson, D. H., J. D. Nichols, and M. D. Schwartz. 1992. Population dynamics of breeding
waterfowl. Pages 446–485 in B. D. J. Batt, A. D. Afton, M. G. Anderson, C. D. Ankney,
D. H. Johnson, J. A. Kadlec, and G. L. Krapu, editors. Ecology and management of
breeding waterfowl. University of Minnesota Press, Minneapolis, USA.
Kaminski, M. R., G. A. Baldassarre, J. B. Davis, E. R. Wengert, and R. M. Kaminski. 2013.
Mallard survival and nesting ecology in the lower Great Lakes region, New York.
Journal of Wildlife Management 37:778–786.
Kenward, R. E. 1987. Wildlife radio-tagging: equipment, field techniques, and data analysis.
Academic Press, San Diego, California, USA.
Korshgen, C. E., S. J. Maxson, and V. B. Kuechle. 1984. Evaluation of implanted transmitters
in ducks. Journal of Wildlife Management 48:982–987.
Krapu, G. L., and H. A. Doty. 1979. Age–related aspects of mallard reproduction. Wildfowl
30:35–39
Krapu, G. L., D. H. Anderson, and C. W. Dane. 1979. Age determination of mallards. Journal
of Wildlife Management 43:384–393.
Loos, E. R., and F. C. Rohwer. 2004. Laying–stage nest attendance and onset of incubation in
prairie nesting ducks. Auk 121:587–599.
Losito, M. P., G. A. Baldassarre, and J. H. Smith. 1995. Reproduction and survival of female
mallards in the St. Lawrence River valley, New York. Journal of Wildlife Management
59:23–30.
McCullagh, P., and J. A. Nelder. 1989. Generalizes linear models. Second edition. Chapman
and Hall, New York, New York, USA.
McPherson, R. J., T. W. Arnold, L. M. Armstrong, and C. J. Schwarz. 2003. Estimating the nest–
success rate and the number of nests initiated by radiomarked mallards. Journal of
Wildlife Management 67:843–851.
Merindino, M. T., G. B. McCullough, and N. R. North. 1995. Wetland availability and use by
breeding waterfowl in Southern Ontario. Journal of Wildlife Management 59:527–532.
Munro, R. E., and C. F. Kimball. 1982. Population ecology of the mallard. U.S. Fish and
Wildlife Service Resource Publication 147, Washington, D.C., USA.
Paquette, G. A., J. H. Devries, R. B. Emery, D. W. Howerter, B. L. Joynt, and T. P. Sankowski.
1997. Effects of transmitters on reproduction and survival of wild mallards. Journal of
Wildlife Management 61:953–961.

36

Petrie, M. J., R. D. Drobney, and D. T. Sears. 2000. Mallard and black duck breeding
parameters in New Brunswick: a test of the reproductive rate hypothesis. Journal of
Wildlife Management 64:832–838.
Pietz, P. J., G. L. Krapu, R. J. Greenwood, and J. T. Lokemoen. 1993. Effects of harness
transmitters on behavior and reproduction of wild mallards. Journal of Wildlife
Management 57:696–703
Raftovich, R. V., K. A. Wilkins, S. S Williams, H. L. Spriggs, and K. D. Richkus. 2011.
Migratory bird hunting activity and harvest during the 2009 and 2010 hunting seasons.
U.S. Fish and Wildlife Service, Laurel, Maryland, USA.
Reinecke, K. J., C. W. Shaiffer, and D. Delnicki. 1987. Winter survival of female Mallards in
the Lower Mississippi Valley. Transactions of the North American Wildlife and Natural
Resources Conference 52: 258–263.
Reynolds, R. E., R. J. Blohm, J. D. Nichols, and J. E. Hines. 1995. Spring–summer survival
rates of yearling versus adult mallard females. Journal of Wildlife Management 59:691–
696.
Rotella, J. J., D. W. Howerter, T. P. Sankowski, and J. H. Devries. 1993. Nesting efforts by
wild mallards with 3 types of radio transmitters. Journal of Wildlife Management
57:690–695.
Sargeant, A. B., G. A. Swanson, and H. A. Doty. 1973. Selective predation by mink, Mustela
vison, on waterfowl. American Midland Naturalist 89:208–214.
Sargeant, A. B., S. H. Allen, and R. T. Eberhardt. 1984. Red fox predation on breeding ducks in
midcontinent North America. Wildlife Monographs 89:1–41.
Sargeant, A. B., R. J. Greenwood, M. A. Sovada, and T. L. Schaffer. 1993. Distribution and
abundance of predators that effect duck production–Prairie Pothole Region. U.S. Fish
and Wildlife Service, Resource Publication 194.
Sharp, D. E., and J. T. Lokemoen. 1987. A decoy trap for breeding season mallards in North
Dakota. Journal of Wildlife Management 51:711–715.
Simpson, J. W. 2005. Mallard duckling survival in the Great Lakes region. Thesis, University of
Guelph, Guelph, Ontario, Canada.
U.S. Geological Survey National Land Cover Data [ONLINE]. 2003.
http://landcover.usgs.gov/natllandcover.asp. (28 July 2005).
White, G. C., and K. P. Burnham. 1999. Program MARK: survival estimation from populations
of marked animals. Bird Study 46 (supplement):120–138.

37

White, G. C., and R. A. Garrott. 1990. Analysis of wildlife radio-tracking data. Academic
Press, New York, New York, USA.
Zuwerink, D. A. 2001. Changes in the derivation of mallard harvests from the northern U.S.
and Canada, 1966–1998. Thesis, Ohio State University, Columbus, USA.

38

CHAPTER 2
Effects of Cover Types on Female Mallard Survival during the Breeding Season in the
Great Lakes Region
Abstract
Understanding the effects of land cover types on important vital rates is critical to understanding
the population dynamics of a species and informing conservation planning efforts focused on
influencing populations through habitat management. However, information on the effects of
land cover types on vital rates is often lacking, particularly for waterfowl species. Here we
evaluated female mallard (Anas platyrhynchos) survival during the breeding season in relation to
variation in land cover types throughout the Great Lakes region. We attached Very High
Frequency (VHF) implant transmitters and telemetry-tracked 282 female mallards during the
breeding season over a 3-year period at 9 study sites. We estimated home range size via the
creation of utilization distributions (UDs) and quantified the core area of these UDs. We utilized
known-fate models in Program MARK to test the effects of home range size, land cover type
within core areas, and near distances to various land cover types as they relate to breeding season
survival. The top-ranking model indicated that the probability of female mallard survival during
the breeding season decreased as the proportion of forested land cover increased within the core
areas (β= –1.740, SE= 0.787, 95% CI Lower –3.282 Upper –0.197). No additional upland or
wetland land cover types significantly affected breeding season survival suggesting that breeding
season survival may be more closely associated with environmental factors that we did not
measure (e.g., weather, predator abundance) or are affected by land cover types at larger or
smaller scales.

39

Introduction
Habitat management for waterfowl survival and fecundity is a multi-million dollar industry in
the United States. The success of such initiatives hinges on the efficacy of the management
techniques for improving waterfowl vital rates during critical periods of the annual cycle, such as
the breeding season. Previous studies have suggested that vital rates such as nest success, brood
survival and female survival during the breeding season are the most influential to the population
dynamics of waterfowl species (Sargeant et al. 1984, Hoekman et al. 2002, 2006a, Coluccy et al.
2008). Thus, quantifying the relationship between waterfowl vital rates and land cover types is
integral to effectively manage waterfowl populations (Greenwood et al. 1987, Rotella and Ratti
1992). Such information can guide habitat conservation programs with the objective of
increasing waterfowl populations (Cowardin et al. 1995, Ball 1996, Reynolds et al. 1996).
Survival of female mallards during the breeding season is the subject of tremendous scientific
attention in the upper Midwest of the United States (Cowardin et al. 1985, Dufour and Clark
2002, Devries et al. 2003, Brasher et al. 2006, Arnold et al. 2012) and elsewhere (Losito et al.
1995, Hoekman et al. 2006b, Kaminski et al. 2013). However, few studies have evaluated the
effects of land cover types on female mallard survival (for exceptions see Devries et al. 2003,
Mack and Clark 2006, Howerter et al. 2014) and it remains an important information need for the
Great Lakes region (Soulliere et al. 2007). Moreover, Great Lakes mallards appear to be
relatively distinct from other populations and have become increasingly important to regional
waterfowl harvest (Anderson and Henny 1972, Munro and Kimball 1982, Zuwerink 2001,).
Previous studies suggested that female mallards are at greatest risk of mortality during peak
nesting periods within the breeding season and while actively nesting (Sargeant et al. 1984,

40

Devries et al. 2003, Brasher et al. 2006, Arnold et al. 2012). Areas within an individual home
range that receive the greatest intensities of use are commonly referred to as core areas
(Kaufman 1962, Powell et al. 1997). Core areas often encompass nest sites and other
ecologically important resources that individuals utilize to meet their seasonal requirements
(Burt 1943, Kaufmann 1962, Ford 1983). Therefore, evaluating the effects of land cover types
found within core areas and near core areas of female mallards may provide the greatest insight
into land cover types that influence survival.
From 2012–2014, we analyzed daily location estimates and individual female breeding season
attribute data collected from the Great Lakes region in 2001–2003. Our objectives were to
evaluate if female mallard breeding season survival in the Great Lakes region is affected by 1)
home range size, 2) number of core areas, 3) proportion of critical land cover types within core
areas, and 4) the distance to land cover types such as agriculture, forest, and wetlands.
Study Area
We examined the breeding season land cover requirements of female mallards at 9 different
study sites throughout the Great Lakes region (Fig. 2.1). We selected the 9 study sites to
represent the range of land cover variation typically found in the Great Lakes (for additional
information see Simpson 2005, Davis 2008). The study sites incorporated varying percentages
of forest land cover types and agriculture land cover types to represent the variability in cover
types typically found within the Great Lakes region of the United States (Table 2.1). We utilized
National Land Cover Data (NLCD; USGS 2003) with 30–m resolution to quantify percentages
of cash crops, dairy pasture, and deciduous forest. We used ERDAS IMAGINE 8.4 GIS
software to attribute 30 x 30–m grid cells of the study sites with the proportion of the

41

aforementioned cover types within a 2 x 2 km area. We narrowed the selection of study sites
based on varying combinations of land uses. We selected study sites that had wetland densities
high enough to support breeding populations of mallards and selected these sites over other
potential sites of similar land cover types with lower wetland densities. We considered logistical
concerns during the final selection of the study sites so that housing existed for the field staff
along with a series of adequate road networks to maximize access within the study area.

42

Figure 2.1 Locations of the 9 study sites within the Great Lakes region where we collected data
for research on female mallard survival during the breeding season, 2001–2003. (OH01 = Port
Clinton, OH; MI01 =Riverdale, MI; WI01 = Ripon, WI; OH02 = Stueben County, IN; MI02 =
Battle Creek, MI; WI02 = Shiocton, WI; OH03 = Akron, OH; MI03 = Big Rapids, MI; WI03 =
New Richmond, WI).

43

Table 2.1 Primary land cover types for the Great Lakes region and associated proportion (%) within those categories created for the
initial selection of study sites in the Great Lakes region, 2001–2003. Study Site corresponds to the state (i.e., MI = Michigan, OH =
Ohio, WI = Wisconsin, IN = Indiana) and year (2001–2003).
Proportion of
Land Use Category
Region
Study Site
Low Forest-High Agriculture
28.4%
MI01,OH01, WI01, WI02
Intermediate Forest- High Agriculture
14.7%
IN02, WI03
Intermediate Forest- Intermediate Agriculture
12.0%
MI02, OH03
High Forest- Intermediate Agriculture
8.3%
MI03
High Forest- Low Agriculture
7.0%
Intermediate Forest Low Agriculture
6.0%
Low Forest- Intermediate Agriculture
2.0%
Low Forest- Low Agriculture
urban
High Forest- High Agriculture
n/a

44

Methods
Capture and Monitoring Techniques
We attempted to capture and instrument 60 female mallards at each of the 9 sites with VHF
telemetry implant transmitters. We selected decoy trap site locations by scouting wetland basins
sufficient in size to hold breeding pairs and occasionally selected wetland basins that pairs were
actively using. In certain cases, we trapped several paired hens on the same wetland basin
throughout the spring. Additionally, land ownership often limited our ability to randomly
distribute trap site locations across the study sites. We trapped females using conventional decoy
traps from March to early April (Sharp and Lokemoen 1987). We banded individuals with
standard United States Geological Survey leg bands and implanted females with a 22–g
abdominal transmitter (Advanced Telemetry Systems, Isanti, MN) using surgical procedures
described by Korschegen et al. (1984). We held radio-marked females for one hour following
surgical procedures and released at the trap site. We banded and released males that were
captured in pairs to avoid disrupting pair bonds.
We tracked radio-marked mallards and estimated locations of individuals 1–6 times daily using
truck–mounted null–array systems (Kenward 1987). We attempted to acquire location estimates
once daily for all individuals and as many as 6 times daily for individuals who were part of an
initial home range study in 2001–2003 (B. Reichus, Oregon State University, unpublished data).
We utilized hand–held tracking methods when mallard locations could not be obtained from the
vehicle due to location of road networks and to verify nesting attempts of breeding females. We
conducted aerial surveys in an effort to locate missing females that may have emigrated from the
study site or attempted nesting efforts outside of previously established locations within the

45

study site. In all cases we identified telemetry locations via triangulation using Locate II
software (Pacer, Truro, Nova Scotia, Canada).
We monitored all females from the onset of marking (mid–March to early April) to mid–July and
two individuals through August. Brood survival was a primary emphasis of the original study;
therefore, we monitored females up to 24 weeks to incorporate brood-rearing activities. Previous
studies assessing breeding season survival have varied in length, due to temporal and spatial
variation associated with the arrival of females on the breeding grounds (see Devries et al. 2003,
Hoekman et al. 2006a.). We truncated the length of the study at 18 weeks because it represented
a time period when a majority of females emigrated after failed breeding attempts or we ceased
monitoring of individuals seen unpaired and flocked and exhibiting non-breeding behavior. This
behavior is consistent with the onset of molt migration. Additionally, no mortalities were
recorded beyond week 16.
Utilization Distribution Estimation
We estimated utilization distributions (UDs) (i.e., home ranges) and core areas for radio-marked
female mallards with ≥ 30 independent locations (Seaman et al. 1999). Due to variation in
sampling intensity among our population of mallards (variation between 1 and 6 locations daily),
we randomly selected one estimated location per individual per day to avoid bias. We developed
UDs in Program R (R Version 2.15.1,www.cran.r–project.org, accessed 29 April 2014) using the
95% isopleth with fixed kernel methods and a plug in matrix bandwidth selection to delineate the
extent of these home ranges. (ks package accessed 29 April 2014). The output was an evaluation
grid with the intensity of mallard land cover use within the home range represented by the Zdimension at a 30 m resolution. We converted these grids to raster in ArcMap 10.1

46

(Environmental System Research Institute, Redlands, CA) expressing the density estimate as
percentiles of land cover use. In this case, the 1st percentile corresponded to the area of the home
range that was used most intensely by each mallard whereas the 95th percentile corresponded to
the area that was used most infrequently. We calculated the size of each 95% UD in hectares to
test against the probability of surviving the breeding season.
Core Area Estimation
Core areas within the home range have historically been arbitrarily identified by the 50%
isopleth (Laver and Kelly 2008). These arbitrary decision rules will often result in creating core
areas if one does not exist, potentially causing problems in studies of habitat and resource
selection. Rather core areas should be delineated by separating intensely used areas that relate to
the ecology of the animal (Powell at al. 1997, Shivik and Gese 2000, Wilson et al. 2010). To
quantitatively delineate the core areas of each mallard home range we employed a Bayesian
approach in R (Wilson et al. 2010). This approach utilizes functions within libraries splancs
(Rowlingson and Diggle 1993), spatstat (Baddeley and Turner 2005), adehabitat (Calenge 2006),
MASS (Venables and Ripley 2002), and MCMC pack (Martin and Quinn 2006) to assess the
spatial distribution of location estimates against a Completely Spatially Random (CSR) point
pattern (Ripley 1976). We iterated this process 10,000 times in a Monte Carlo simulation for
each female to test for departures from a CSR pattern. The resulting model converged at the
density percentile that defined the UDs core area.
Land Cover Types
We developed a geographic information system (GIS) to portray the land cover types associated
with each study site. We hand digitized the land cover polygons in 2001–2003 from imagery
47

collected as a part of the NLCD (USGS 2003). We ground–truthed the land cover types at each
study site within 0.8 km of all female and brood locations. We defined upland cover types as
grassland, agriculture, forestland, scrubland, and other. We combined hayland, pastureland, and
cropland into a single agriculture land cover layer. We incorporated residential and other
developed areas into a single land cover classification named other and we removed these from
consideration in our GIS. We incorporated individual wetland basins classified as either
emergent, forested, scrub-shrub, or other wetlands according to the methods of Cowardin et al.
(1979) and combined them into a single wetland layer.
We created individual GIS shapefile layers in Arc Map containing the land cover types:
grassland, wetland, forest, shrubland, and agriculture. We intersected each of the land cover
types with each individual core area isopleth, and calculated the area of each land cover type
within the core area and estimated the proportion of each cover type. We identified agriculture
(range 0.0–1.0) as an important cover type because females mallards have been shown to have
reduced survival when nesting in hay fields (Hoekman et al. 2006a) and other commodity types
included within the agriculture land cover type may provide less than adequate cover for nesting
females, thereby increasing vulnerability to predators (Greenwood et al. 1995, Reynolds et al.
2001, Stephens et al. 2005). We considered forestland (range 0.0–0.9) important because it has
been shown to provide adequate roost locations and habitat for aerial predators of upland nesting
waterfowl (Losito et al. 1995, Hoekman et al. 2006a). Wetland (range 0.0–1.0) and grassland
cover types (range 0.0–1.0) made up a majority of the land cover types utilized by the nesting
females in our study (Davis 2008). Increased proportions of grassland cover and wetland cover
may impede predator movements thus increasing survival probability (Duebbert 1969). Lastly,
we incorporated shrubland cover types (range 0.0–0.8) because increased shrubland cover types
48

within individual home ranges have been shown to increase nest success of mallards (Mack and
Clark 2006).
We also considered proximity metrics based on the distance (in meters) from the center of the
core area within individual home ranges to the nearest agriculture, forest, and wetland cover
types. In cases where mallards had multiple core home ranges we calculated distance from
center points of each core area and utilized the mean distance from each land cover type. We
also considered the number of core areas per individual (range 1–5) as a covariate to evaluate
whether an increase in the number of core areas could be associated with increased nesting effort
and therefore potentially influencing survival probability during the breeding season (Reynolds
et al. 1995).
Known-fate Models
We developed known-fate models in Program Mark (White and Burnham 1999) to calculate the
maximum likelihood estimates of female mallard survival at weekly intervals and throughout the
18–week study period in relation to the land cover types developed in our GIS (White and
Garrott 1990). As the response is binary (0 = alive and 1 = dead), we fit models using a binomial
distribution of errors and a logit link function (McCullagh and Nelder 1989). Our list of 19
candidate models included only additive combinations of land cover covariates and we had no
biological justification to include interactions between any of the additional covariates under
assessment. We used an information theoretic approach for model selection [Akaike’s
Information Criterion adjusting for small sample size (AICc; Hurvich and Tsai 1995)] and
ranked models e using AICc weights (AICcω; Burnham and Anderson 1998).

49

Results
We analyzed the effects of proportion of land cover types within the core areas, near distances of
agriculture, forest, and wetland cover types from the center of the core areas, number of core
areas, and home range size against breeding season survival of 282 female mallards at 9 different
sites across 4 different states from 2001–2003. Thirty-seven individuals died throughout the
course of the 18-week study period. We recorded ≥1 nest attempt for 265 individuals over the
course of the 18-week period, and recorded no nesting attempts for the remaining 18 individuals.
Our best approximating model indicated that proportion of forest cover type within the core area
had the greatest influence on breeding season survival and received more than twice as much
support as the next competing model (Table 2.2). The results of our top model (β = –1.740, 95%
CI Lower = –3.282, Upper = –0.197) suggested that survival decreased as the percent
composition of forested cover within the core area increased (Fig. 2.2). However, the difference
between survival over the 18–week period between the null model (0.824) and our top ranking
model (0.829) fell within 2 ΔAICc (Table 2.2). The mean proportion of forested cover within
core areas by site ranged from 0.02–0.18, ( = 0.09, SE = 0.010) (Table 2.3; mean home range
size and mean area for each cover type are provided in Table 2.4 and Table 2.5, respectively).
The remaining candidate models showed minimal effect on survival (i.e. AICc W <0.10). The
effects of home range size and number of core areas per individual female had widely
overlapping 95% confidence intervals and were not significant.

50

Table 2.2 Model results of candidate models testing the effects of the proportion of individual land cover types within the core area,
distance from the core area to individual land cover types, number of core areas, and home range size on female mallard (n= 282)
survival during the breeding season in the Great Lakes region, 2001–2003.We used known–fate models in Program Mark to evaluate
competing model likelihood using Akaike’s Information Criterion corrected for sample size (AICc), the difference from the highest
ranked model (∆AICc), and the model weight.
Modela
Parameters AICc ∆AICc Model Weight
{S(P_For)}
2
409.81
0
0.28708
{S(.)}
1
411.77
1.96
0.10779
{S(P_Num_Cores)}
2
412.16
2.35
0.0887
{S(P_Wet)}
2
412.86
3.06
0.06231
{S(P_Grass)}
2
413.19
3.38
0.05292
{S(HR)}
2
413.5
3.69
0.04541
{S(D_For)}
2
413.64
3.84
0.04216
{S(D_Wet)}
2
413.72
3.91
0.04057
{S(P_Shrub)}
2
413.76
3.95
0.03989
{S(P_Ag)}
2
413.76
3.95
0.03986
{S(D_Ag)}
2
413.77
3.96
0.03961
{S(D_Ag + P_Wet)}
3
414.71
4.91
0.0247
{S(D_For + P_Wet)}
3
414.77
4.96
0.02407
{S(D_For + P_Grass)}
3
415.09
5.28
0.02045
{S(D_Wet + P_Grass)}
3
415.18
5.37
0.0196
{S(D_Ag + P_Grass)}
3
415.18
5.37
0.01959
{S(D_For + P_Shrub)}
3
415.63
5.82
0.01565
{S(D_Wet + P_Shrub)}
3
415.71
5.91
0.01498
{S(D_Ag + P_Shrub)}
3
415.76
5.95
0.01465
a
+ denotes an additive effect; P_For = proportion forest land cover type; P_Wet = proportion of wetland land cover type; P_Ag =
proportion agriculture land cover type; P_Shrub = proportion of shrubland land cover type; P_Grass = proportion of grassland land
cover type; D_For = distance to forest land cover type from core area; D_Wet = distance to wetland land cover type from core area;
D_Ag = distance to agriculture land cover type from core area; (.) = null model; HR = home range size ha; Num_Cores = Number of
core areas per individual female mallard.
51

0.00
0.05
0.10
0.15
0.20
0.25
0.30
0.35
0.40
0.45
0.50
0.55
0.60
0.65
0.70
0.75
0.80
0.85
0.90
0.95
1.00

Breeding Season Survival

1.00
0.90
0.80
0.70
0.60
0.50
0.40
0.30
0.20
0.10
0.00
Percent Forest Cover

Figure 2.2 Relationship between changes in breeding season survival of female mallards across
18 weeks and percent composition of forested cover within core use areas based on model results
derived from Program MARK in the Great Lakes region, 2001–2003.

52

Table 2.3 Mean proportions of forested cover (P_For), grassland cover (P_Grass), agriculture
cover (P_Ag), shrubland cover (P_Shrub), and wetland cover (P_Wet) within individual core
areas of female mallards (n = 282) across 9 sites during the breeding season in the Great Lakes
region, 2001–2003.
Sites
P_For
P_Grass
P_Ag P_Shrub P_Wet
IN02
0.07
0.27
0.39
0.01
0.16
MI01
0.08
0.13
0.50
0.04
0.15
MI02
0.12
0.10
0.20
0.04
0.35
MI03
0.18
0.22
0.14
0.06
0.35
OH01
0.04
0.07
0.50
0.04
0.27
OH03
0.17
0.28
0.21
0.06
0.24
WI01
0.03
0.11
0.20
0.00
0.59
WI02
0.04
0.17
0.22
0.00
0.47
WI03
0.19
0.32
0.13
0.03
0.31
Total (Mean) 0.09
0.18
0.29
0.03
0.32

53

Table 2.4 Mean home range size (ha) at the 95 percent isopleth (95_Per_HR), mean core area
isopleth (Core_Iso), mean number of core areas (Num_Cores), and the core area size (ha)
(Core_Area) for each individual female (n = 282) across the 9 study sites in the Great Lakes
region, 2001–2003.
Sites
95_Per_HR Core_Iso Num_Cores
Core_Area
IN02
111.83
0.35
1.34
8.95
MI01
180.27
0.35
1.33
10.61
MI02
62.12
0.35
1.52
4.70
MI03
121.30
0.30
1.45
6.59
OH01
172.06
0.42
1.47
16.20
OH03
165.23
0.40
1.25
15.32
WI01
100.46
0.39
1.16
7.19
WI02
132.31
0.36
1.35
12.44
WI03
166.47
0.38
1.11
10.94
Total (Mean)
132.84
0.37
1.34
10.21

54

Table 2.5 Mean area (ha) of individual land cover types within individual core areas of female
mallards (n = 282) across 9 sites during the breeding season in the Great Lakes region, 2001–
2003.
Sites
A_Fora A_Grassb
IN02
0.81
2.53
MI01
1.34
1.24
MI02
0.50
0.41
MI03
1.37
0.99
OH01
1.13
1.09
OH03
3.72
2.87
WI01
0.26
0.96
WI02
0.31
2.30
WI03
1.79
1.81
Total (Mean)
1.12
1.58
a
Area of forested cover types
b
Area of grassland cover types
c
Area of agriculture cover type
d
Area of shrubland cover type
e
Area of wetland cover type

A_Agc
2.57
5.90
1.12
0.96
6.83
4.28
1.72
3.11
2.84
3.28

A_Shrubd
0.07
0.26
0.17
0.17
0.59
0.47
0.00
0.13
0.70
0.25

55

A_Wete
1.05
0.88
1.74
3.01
5.49
3.28
3.66
5.65
3.48
3.10

Discussion
While previous research on mallard vital rates has focused on the effects of habitat level
correlates on female mallard survival in the Canadian Prairie Parklands and the Prairie Pothole
region, this analysis to our knowledge is the first to assess the effects of land cover types within
core areas of a home range on breeding season survival of female mallards in the Great Lakes
region. Our top ranking model indicated that increased percent forested cover type within core
areas decreased survival probability of female mallards during the breeding season.
Additionally, Simpson et al. (2007) found that duckling survival decreased as percent forest
increased within 500–m of the brood locations. It is plausible that increased proportions of forest
cover types within core areas provide adequate habitat for aerial and terrestrial predators of
breeding mallards. However, we suspect that the outcome of this model may have been partially
confounded by the proximity of forest cover types to cover types used by breeding mallards
because water permanency likely restricted conversion of these areas for agricultural or human
development purposes (R. Boyer, personal observation). Additional assessments of this
relationship may be necessary.
Mack and Clark (2006) tested the effects of woodlands and shrubland on nest success and female
mallard survival in the Prairie Pothole Region, and found that nest success increased in
conjunction with increased woodland and shrubland cover, but failed to detect an effect on
female survival. Similarly, Davis (2008) suggested that nest success was higher in primarily
forested landscapes than on landscapes containing high percentages of agriculture. Probability
of survival across the 18–week study period for the top ranking candidate model including
proportion of forest cover was slightly higher, however, minimally so when compared to the null
model although the effect of forestland on survival was negative. To gain a better understanding

56

of the effect on survival, we investigated the effect of the proportion forest cover within core
areas by study site and discovered that it did not have a significant effect on survival across all
sites.
Devries et al. (2003) and Howerter et al. (2014) assessed the effects of habitat related variables
on the survival of female mallards during the breeding season at the study area scale (65

,

and Mack and Clark (2006) tested female survival at the home range scale and study area scale.
Devries et al. (2003) found a relationship between survival of female mallards and increased
wetland densities at the study area scale, but failed to detect any relationships between upland
habitat treatments and survival. Howerter et al. (2014) provided additional analyses of these data
and other data compiled from studies completed in the Canadian Prairie Parklands and the Prairie
Pothole Region and found that survival of female mallards increased with the proportion of
grassland cover in conjunction with wetland habitats at the study area scale. Mack and Clark
(2006) assessed the affects of habitat composition on the survival of female mallards at the home
range and study area scales, but failed to detect a relationship. The Canadian Prairie Parklands
and the Prairie Pothole Region where the aforementioned studies occurred are comprised of large
contiguous blocks of grassland cover types interspersed with high densities of seasonal wetlands
(Reynolds 2000). The land cover types found within the Great Lakes region vary significantly in
comparison to those assessed in the previous studies.
Our previous analysis of breeding season survival tested the effects of study site, year, and state
against survival and found no significant effects. However, we discovered that female survival
was lowest during peak nesting activities (Chapter 1). Thus, we refined the scale at which to
assess the effects of land cover types on survival to the core area which encompasses the spatial
and temporal scale at which females are at greatest risk of mortality (Sargeant et al. 1984,
57

Devries et al. 2003). Arnold et al. (2012) suggested that prey switching behaviors of common
upland nesting waterfowl predators may occur concurrent with the peak of the nesting period,
which may partially explain our inability to detect a strong relationship between most land cover
types and survival. Additionally, our ability to infer the effect of land cover types on survival
was limited by the resolution of our spatial data (i.e., 30 m), thus potentially restricting our
ability to recognize additional relationships with breeding season survival.
We failed to detect any strong relationships between survival and non-forested upland and
wetland cover types. Previous studies suggested that nest survival increases as the abundance of
perennial upland grassland increases (Greenwood et al. 1995, Reynolds et al. 1996). Dense
nesting cover that has been shown to positively influence nest survival and may influence
survival of females during the breeding season. Previous studies have reported mallards nesting
in dense stands of emergent vegetation in wetland habitat (Krapu et al. 1979). Wetland cover
may offer similar advantages to dense stands of upland nesting cover by increasing physical
barriers that reduce the movements and foraging ability of mammalian predators (Duebbert
1969), thus increasing female survival during critical periods of the breeding season. Lastly,
agricultural land can pose various threats to nesting waterfowl, especially those lands actively
used for haying purposes (Hoekman et al. 2006a). Well known avian and mammalian predators
of upland nesting waterfowl commonly use the edges of cover types such as agricultural fields,
wetlands, and forests. The distance from the core area to these specific land cover types that
support predator communities may affect the survival probabilities of female mallards during the
breeding season; however, we were unable to discern any relationship.
Nudds and Ankney (1982) suggested that the home range size of mallards during the breeding
season is greatly influenced by resource availability. Individuals with larger home ranges may
58

have been forced into less than adequate nesting habitat when establishing breeding territories
(Dzubin 1969, Lindstedt et al.1986, Lokemoen et al. 1990). Additionally, breeding females may
need to travel longer distances to meet resource requirements during the breeding season (Mack
et al. 2003). Increased home range size could lead to greater time spent traveling between nest
site locations and foraging areas (Mack and Clark 2006), leading to increased exposure time to
predators therefore increasing the chance of mortality. Our results suggest that home range size
occupied during the breeding season had no significant effect on this vital rate; however, the
metric results alone from our estimation of breeding season home range size for Great Lakes
mallards will be important for regional conservation planning.
Our study design was created to replicate a similar study being conducted by Devries et al.
(2003) in the Prairie Pothole Region and the Canadian Prairie Parklands. Similar to Devries et al
(2003), our study tried to encompass the greatest amount of geographic and landscape variation
across the region to test the effects of various land cover types on the vital rates of mallards in
the Great Lake region. Unfortunately, that decision resulted in the lack of temporal replication
among study sites. Our inability to detect a strong relationship between land cover types and
female survival may have been affected by the inability to replicate the study across multiple
years for the same sites (Howerter et al. 2014). Due to constraints related to land ownership, we
were unable to randomly select trap site locations throughout the study sites, therefore, trap
locations and the nest site locations were often in close proximity to road systems where
wetlands or small bodies of water were of greater accessibility for researchers to locate and trap
breeding females. The resulting similarities in land cover composition within core areas between
individual females may have partially confounded our inabilities to detect any strong effects of
most land cover types on breeding season survival.
59

Management Implications
Female mallard survival during the breeding season is negatively affected by increased
proportions of forested cover types within the core areas. Additionally, mallard duckling
survival in the Great Lakes region which is one of the most important demographic factors
influencing the population dynamics of Great Lakes mallard populations (Coluccy et al. 2008)
decreases as the proportion of forested cover increases within a 500 m area of brood locations
(Simpson et al. 2007). Waterfowl management interests in the Great Lakes region striving to
increase breeding season survival of mallard females and ducklings should focus habitat
management efforts in areas with low densities of forested cover types. We failed to detect
strong relationships between female mallard breeding season survival and additional upland and
wetland land cover types suggesting that female mallard survival in the Great Lakes region may
be affected by land cover factors not assessed as a part of this study or at alternative spatial
scales. Respectively, breeding season survival of females accounts for approximately 1 percent
of the variation in population growth of Great Lakes mallards while nest success and duckling
survival account for nearly half of the variation (Coluccy et al. 2008). Thus, efforts to increase
duckling survival and nest success through the enhancement and restoration of dense nesting
cover and palustrine emergent wetland habitats are likely the best ways to increase mallard
productivity throughout the Great Lakes region and should remain the primary focus of regional
conservation efforts (Simpson et al. 2007, Coluccy et al. 2008, Davis 2008). These management
practices may have limited effects on breeding season survival of females as survival rates
appear to be relatively unaffected by varying proportions of critical land cover types within and
surrounding core areas.

60

LITERATURE CITED

61

LITERATURE CITED
Anderson, D. R., and C. J. Henny. 1972. Population ecology of the mallard. I. U.S. Fish and
Wildlife Service Resource Publication 105, Washington, D.C., USA.
Arnold, T. W., E. A. Roche, J. H. Devries, and D. W. Howerter. 2012. Costs of reproduction in
breeding female mallards: predation risk during incubation drives annual mortality.
Avian Conservation and Ecology 7(1):1.
Baddeley, A., and R. Turner. 2005. Spatstat: an R package for analyzing spatial point patterns.
Journal of Statistical Software 12:1–42.
Ball, I. J. 1996. Managing habitat to enhance avian recruitment. Transactions of the North
American Wildlife and Natural Resource Conference 61:109–117.
Brasher, M. G., T. W. Arnold, J. H. Devries, and R. H. Kaminski. 2006. Breeding–season
survival of male and female mallards in Canada’s prairie parklands. Journal of Wildlife
Management 70:805–811.
Burnham, K. P., and D. R. Anderson. 1998. Model selection and inference: a practical
information–theoretic approach. Springer–Verlag, New York, New York, USA.
Burt, W. H. 1943. Territoriality and home range concepts as applied to mammals. Journal of
Mammology 24:346–352.
Calenge, C. 2006. The package adehabitat for the R software: a tool for the analysis of space
and habitat use by animals. Ecological Modeling 197:516–519.
Coluccy, J. M., T. Yerkes, R. Simpson, J. W. Simpson, L. Armstrong, and J. I. Davis. 2008.
Population dynamics and sensitivity analyses of breeding mallards in the Great Lakes
states. Journal of Wildlife Management 72:1181–1187.
Cowardin, L. M., V. Carter, F. C. Golet, and E. T. LaRoe. 1979. Classification of wetlands and
deepwater habitats of the United States. U.S. Fish and Wildlife Service Pub. FWS/OBS79/31, Washington D.C., 103p.
Cowardin, L. M., D. S. Gilmer, and C. W. Shaiffer. 1985. Mallard recruitment in the
agricultural environment of North Dakota. Wildlife Monographs 92:1–37.
Cowardin, L. M., T. L. Shaffer, and K. M. Kraft. 1995. How much habitat management is
needed to meet Mallard production objectives? Wildlife Society Bulletin 23:48–55.
Davis, J. I. 2008. Mallard nesting ecology in the Great Lakes. Thesis, Humboldt State
University, Arcata, California, USA.

62

Devries, J. H., J. J. Citta, M. S. Lindberg, D. W. Howerter, and M. G. Anderson. 2003.
Breeding-season survival of mallard females in the prairie pothole region of Canada.
Journal of Wildlife Management 67:551–563.
Duebbert, H. F. 1969. High nest density and hatching success of ducks on South Dakota CAP
land. Transactions of the North America Wildlife and Natural Resources Conference
34:218–228.
Dufour, K. W., and R. G. Clark. 2002. Differential of yearling and adult female mallards and its
relation to breeding habitat conditions. The Condor 104:297–308.
Dzubin, A. 1969. Comments on carrying capacity of small ponds for ducks and possible effects
of density on mallard production. Pages 138–160 in Saskatoon Wetland Seminar.
Canadian Wildlife Service Report Series, 6, Ottawa.
Ford, R. G. 1983. Home range in a patchy environment: optimal foraging predictions.
American Zoology 23:315–326.
Greenwood, R. J., A. B. Sargeant, D. H. Johnson, L. M. Cowardin, and T. L. Shaffer. 1987.
Mallard nest success and recruitment in prairie Canada. Transactions of the North
American Wildlife and Natural Resource Conference 52:298–309.
Greenwood, R. J., A. B. Sargeant, D. H. Johnson, L. M. Cowardin, and T. L. Shaffer. 1995.
Factors associated with duck nest success in the prairie pothole region of Canada.
Wildlife Monographs 128.
Hoekman, S. T., L. S. Mills, D. W. Howerter, J. H. Devries, and I. J. Ball. 2002. Sensitivity
analyses of the life cycle of mid-continent mallards. Journal of Wildlife Management
66:883–900.
Hoekman, S. T., T. S. Gabor, R. Maher, H. R. Murkin, and M. S. Lindberg. 2006a.
Demographics of breeding female mallards in Southern Ontario, Canada. Journal of
Wildlife Management 70:111–120.
Howerter, D. W., M. G. Anderson, J. H. Devries, B. L. Joynt, L. M. Armstrong, R. B. Emery,
and T. W. Arnold. 2014. Variation in mallard vital rates in Canadian aspen parklands:
The Prairie Habitat Joint Venture assessment. Wildlife Monographs 188:1–37.
Hurvich, C. M., and C.–L. Tsai. 1995. Model selection for extended quasi–likelihood models in
small samples. Biometrics 51:1077–1084.
Kaminski, M. R., G. A. Baldassarre, J. B. Davis, E. R. Wengert, and R. M. Kaminski. 2013.
Mallard survival and nesting ecology in the lower Great Lakes region, New York.
Journal of Wildlife Management 37:778–786.

63

Kaufmann, J. H. 1962. Ecology and social behavior of the coati, Nasua nirica on Barro
Colorado Island Panama. University of California Publications in Zoology 60:95–222.
Kenward, R. E. 1987. Wildlife radio-tagging: equipment, field techniques, and data analysis.
Academic Press, San Diego, California, USA.
Korshgen, C. E., S. J. Maxson, and V. B. Kuechle. 1984. Evaluation of implanted transmitters
in ducks. Journal of Wildlife Management 48:982–987.
Krapu, G. L., D. H. Johnson, and C. W. Dane. 1979. Age determination of mallards. Journal of
Wildlife Management 43:384–393.
Laver, P. N., and M. J. Kelly. 2008. A critical review of home range studies. Journal of
Wildlife Management 72:290–298.
Lindstedt, S. L., B. J. Miller, and S. W. Buskirk. 1986. Home range, time, and body size in
mammals. Ecology 67:413–418.
Lokemoen, J. T., H. F. Duebbert, and D. E. Sharp. 1990. Homing and reproductive habits of
mallards, gadwall, and blue–winged teal. Wildlife Monographs 106.
Losito, M. P., G. A. Baldassarre, and J. H. Smith. 1995. Reproduction and survival of female
mallards in the St. Lawrence River valley, New York. Journal of Wildlife Management
59:23–30.
Mack, G. G., and R. G. Clark. 2006. Home range characteristics, age, body size, and breeding
performance of female mallards (Anas platyrhynchos). Auk 123:467–474.
Mack, G. G., R. G. Clark, and D. W. Howerter. 2003. Size and habitat composition of female
mallard home ranges in the prairie–parkland region of Canada. Canadian Journal of
Zoology 81:1454–1461.
Martin, A. D., and K. M. Quinn. 2006. Applied Bayesian inference in R using MCMCpack. R
News 6:2–7.
McCullagh, P., and J. A. Nelder. 1989. Generalized linear models. Second edition. Chapman
and Hall, New York, New York, USA.
Munro, R. E., and C. F. Kimball. 1982. Population ecology of the mallard. U.S. Fish and
Wildlife Service Resource Publication 147, Washington, D.C., USA.
Nudds, T. D., and Ankney, C. D. 1982. Ecological correlates of territory and home range size of
North American dabbling ducks. Wildfowl 33: 28–62.

64

Powell, R. A., J. W. Zimmerman, and D. E. Seaman. 1997. Ecology and behaviour of North
American black bears: home ranges, habitat, and social organization. Chapman and Hall,
London, United Kingdom.
Reynolds, R. E. 2000. Waterfowl responses to the Conservation Reserve Program in the
Northern Great Plains. Pages 35–43 in W. L. Hohman and D. J. Halloum, editors. A
comprehensive review of farm bill contributions to wildlife conservation, 1985–2000.
U.S. Department of Agriculture, Natural Resources Conservation Service, Wildlife
Habitat Management Institute, Technical Report, Washington, D.C.,USA.
Reynolds, R. E., R. J. Blohm, J. D., Nichols, and J. E. Hines. 1995. Spring-summer survival
rates of yearling versus adult mallard females. Journal of Wildlife Management 57:691–
696.
Reynolds, R. E., D. R. Cohan, and M. J. Johnson. 1996. Using landscape information
approaches to increase duck recruitment in the Prairie Pothole Region. Transactions of
North American Wildlife and Natural Resource Conference 61:86–92.
Reynolds, R. E., T. L. Shaffer, R. W. Renner, W. E. Newton, and B. D. J. Batt. 2001. Impact of
the Conservation Reserve Program on duck recruitment in the U.S. Prairie Pothole
Region. Journal of Wildlife Management 65:765–780.
Ripley, B. D. 1976. The second-order analysis of stationary point processes. Journal of Applied
Probability 13:255–266.
Rotella, J. J., and J. T. Ratti. 1992. Mallard brood survival and wetland habitat conditions in
southwestern Manitoba. Journal of Wildlife Management 56:499–507.
Rowlingson, B. S., and P. J. Diggle. 1993. SPLANCS: spatial point pattern analysis code Splus. Computers and Geosciences 19:627–655.
Sargeant, A. B., S. H. Allen, and R. T. Eberhardt. 1984. Red fox predation on breeding ducks in
midcontinent North America. Wildlife Monographs 89:1–41.
Seaman, D. E., J. J. Millspaugh, B. J. Kernohan, and G. C. Brundige. 1999. Effects of sample
size on kernel home range estimates. Journal of Wildlife Management 63:739–747.
Sharp, D. E., and J. T. Lokemoen. 1987. A decoy trap for breeding season mallards in North
Dakota. Journal of Wildlife Management 51:711–715.
Shivik, J. A., and E. M. Gese. 2000. Territorial significance of home range estimators for
coyotes. Wildlife Society Bulletin 28:940–946.
Simpson, J. W. 2005. Mallard duckling survival in the Great Lakes region. Thesis, University
of Guelph, Guelph, Ontario, Canada.

65

Simpson, J. W., T. J. Yerkes, T. D. Nudds, and B. D. Smith. 2007. Effects of habitat on mallard
duckling survival in the Great Lakes region. Journal of Wildlife Management 71:1885–
1891.
Soulliere, G. J., B. A. Potter, J. M. Coluccy, R. C. Gatti., C. L. Roy, D. R. Luukkonen, P. W.
Brown, and M. W. Eichholz. 2007. Upper Mississippi River and Great Lakes Region
Joint Venture Waterfowl Habitat Conservation Strategy. U.S. Fish and Wildlife Service,
Fort Snelling, Minnesota, USA.
Stephens, S. E., J. J. Rotella, M. S. Lingberg, M. L. Taper, and J. K. Ringleman. 2005. Duck
nest survival in the Missouri Coteau of North Dakota: landscape effects at multiple
spatial scales. Ecological Applications 2137–2149.
U.S. Geological Survey National Land Cover Data [ONLINE]. 2003.
http://landcover.usgs.gov/natllandcover.asp. (28 July 2005).
Venables, W. N., and B. D. Ripley. 2002. Modern applied statistics with S. Springer, New
York, New York, USA.
White, G. C., and R. A. Garrott. 1990. Analysis of wildlife radio-tracking data. Academic
Press, New York, New York, USA.
White, G. C., and K. P. Burnham. 1999. Program MARK: survival estimation from populations
of marked animals. Bird Study 46 (supplement):120–138.
Wilson, R. R., M. B. Hooten, B. N. Strobel, and J. A. Shivik. 2010. Accounting for individuals,
uncertainty, and multiscale clustering in core area estimation. Journal of Wildlife
Management 74: 1343–1352.
Zuwerink, D. A. 2001. Changes in the derivation of mallard harvests from the northern U.S.
and Canada, 1966–1998. Thesis, Ohio State University, Columbus, USA.

66