may?“ Mi

huh... ..

. ‘3
5..
.
at..."

new. .‘n

Var.

 

 

3 LIBRARY
05 Michig. State
University

 

 

 

This is to certify that the
dissertation entitled

PHYLOGEOGRAPHY AND MANAGEMENT OF SNOW,
ROSS’S, CANADA, AND CACKLING GEESE

presented by

RAINY INMAN SHOREY

has been accepted towards fulfillment
of the requirements for the

PhD. degree in Fisheries and Wildlife
Ecology, Evolutionary Biology
and Behavior

 

 

//;. X2

 

Major Professor’s Signature
5 — Z 31- o i’

 

Date

MSU is an Affinnative Action/Equal Opportunity Institution

 

PHYLOGEOGRAPHY AND MANAGEMENT OF SNOW, ROSS’S,
CANADA, AND CACKLIN G GEESE

By

Rainy Inman Shorey

A DISSERTATION
Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of
DOCTOR OF PHILOSOPHY
Department of Fisheries and Wildlife

2005

ABSTRACT

PHYLOGEOGRAPHY AND MANAGEMENT OF SNOW, ROSS’S,
CANADA, AND CACKLING GEESE

By
Rainy Inman Shorey

Identification of demographically and reproductively isolated populations is an
important aspect of resource management and conservation. This is especially crucial for
highly migratory species like waterfowl, as populations utilize several areas throughout
the year, often co-occurring with individuals from other populations, and because rates of
natality and survival may vary greatly among populations. Waterfowl species,
subspecies, and management populations have traditionally been defined and managed
based on morphology and plumage characteristics; and based on banding studies
emphasizing commonalities in migratory patterns and fidelity to, and potential gene flow
among, wintering and breeding areas. However, banding and morphometric techniques
focus on contemporary attributes of species ecologies, and provide little information
about historic population demographics. Thus, the management units defined for
waterfowl based on morphology and banding may not be supported by species underlying
population genetic structure.

With the development and greater accessibility of molecular techniques, genetic
surveys can be conducted to determine the degree of genetic variation for currently
defined populations, subspecies, and species of waterfowl. Molecular phylogenies based
on maternally- (mitochondrial DNA) and bi-parentally- (neutral nuclear DNA) inherited

markers can be examined in a spatial context, and be used to assess the relative influence

of historical processes and current behavioral and ecological factors on the spatial genetic
structure revealed for specific waterfowl taxa.

This study focused on two co-distributed species groups of migratory waterfowl
in North America that have similar historical geographic ranges. Populations, subspecies,
and species of these two groups, including snow (Chen caerulescens) and Ross’s geese
(Chen rossii) and cackling (Branta hutchinsii) and Canada geese (Branla canadensis)
were genetically characterized using mitochondrial and microsatellite molecular markers.
These molecular surveys were designed to evaluate the degree of genetic variation at
multiple spatial levels, and identify areas of genetic discordance within and among
species. Hypotheses were then tested regarding the relative importance of causal
historical and contemporary factors that have defined the genetic spatial structure
observed. Finally, a determination was made as to whether the present species,
subspecies, and management unit designations for each group of geese reflected the
underlying spatial structure as seen at the level of genes.

Data collected as part of the phylogenetic investigation were then utilized to
address specific issues of population management in snow geese and within cackling and
Canada geese. For snow geese, information regarding the degree of genetic variation
among Western populations was used to determine relative levels of historical and
contemporary gene flow among populations of conservation concern, and to evaluate
population structuring among breeding and wintering groups of these populations. To
improve harvest management in the state of Michigan, established methods of genetic
stock identification and nationally standardized sample collections were used to estimate

proportional contributions of cackling and Canada geese to annual harvests.

ACKNOWLEDGEMENTS

I would like to express my gratitude to the many wonderful professors at
Michigan State University that have offered me guidance and support throughout my
degree. I thank all the wildlife managers and field biologists from many state, federal,
and provincial agencies who donated their time, funding, sampling efforts, and wealth of
knowledge to this project. I greatly appreciate the unending support provided by my
family and friends during this period of my life. Finally, I am thankful I had the
opportunity to discover my very special friend on the North Slope of Alaska during one

of my many Ph.D. adventures.

iv

TABLE OF CONTENTS

LIST OF TABLES .............................................................................. viii
LIST OF FIGURES .............................................................................. xi
CHAPTER 1

INTRASPECIFIC PHYLOGEOGRAPHIES OF SNOW, ROSS’S, CANADA, AND
CACKLING GEESE: DEFINING THE SPATIAL STRUCTURE OF BIOLOGICAL

DIVERSITY ......................................................................................... 1
Introduction .................................................................................. 1
Utility of Genetic Markers ....................................................... 3
Purpose of Project .......................................................................... 6
Background ................................................................................ 10
Defining Spatial Genetic Structure ............................................ 10
Arctic-Nesting Waterfowl ...................................................... 12
Snow and Ross’s Geese ......................................................... 15
Cackling and Canada Geese .................................................... 22
Outline of Chapters ....................................................................... 28
Appendix 1 ................................................................................. 33
Literature Cited ........................................................................... 35
CHAPTER 2
CACKLING GEESE AND CANADA GEESE PHYLOGEOGRAPHY .................. 47
Introduction ................................................................................ 47
Methods .................................................................................... 55
Sample Collection ............................................................... 55
Characterization of Microsatellite Loci ....................................... 57
Characterization of MtDNA ................................................... 59
Gene Diversity and Population Differentiation .............................. 60
MtDNA Sequence Data ......................................................... 62
Phylogeographic Analyses ...................................................... 62
Results ...................................................................................... 63
Genetic Variation ................................................................ 63
Phylogenetic and Population Relationships .................................. 64
Discussion ................................................................................. 73
Management Implications ............................................................... 82
Literature Cited ........................................................................... 86
CHAPTER 3
PHYLOGEOGRAPHY OF SNOW GEESE AND ROSS’S GEESE ...................... 92
Introduction ................................................................................ 92
Methods .................................................................................... 100
Sample Collection ............................................................... 100

Characterization of Microsatellite Loci ..................................... 102

Characterization of MtDNA .................................................. 104
Gene Diversity and Population Structuring ................................. 104
Results .................................................................................... 107
Genetic Variation ............................................................... 107
Phylogenetic and Population Relationships ................................. 108
Discussion ................................................................................ 115
Literature Cited .......................................................................... 126
CHAPTER 4
A NOVEL METHOD FOR CANADA GOOSE HARVEST DERIVATION USING
GENETIC ANALYSIS OF TAIL FEATHERS ............................................. 133
Introduction .............................................................................. l 3 3
Methods ................................................................................... 135
Sample Collection .............................................................. 135
Genetic Analyses ............................................................... 137
Statistical Analyses ............................................................ 138
Results .................................................................................... 139
Discussion ................................................................................ 1 40
Management Implications .............................................................. 143
Literature Cited .......................................................................... 145
CHAPTER 5

COMPOSITION OF CANADA AND CACKLING GOOSE HARVESTS IN
MICHIGAN 1998-2002: A GENETIC ANALYSIS OF USFWS WATERF OWL PARTS

SURVEY SAMPLES ........................................................................... 148
Introduction .............................................................................. 148
Methods ................................................................................... 1 52

Sample Collection .............................................................. 152
Genetic Analyses ............................................................... 153
Harvest Derivations ............................................................ 155
Results .................................................................................... 159
Baseline Population Differentiation .......................................... 159
Harvest Derivations ............................................................ 159
Discussion ................................................................................ 169
Management Implications .............................................................. 176
Literature Cited .......................................................................... 179
CHAPTER 6

A COMPARATIVE ANALYSIS OF WATERFOWL HARVEST DERIVATION
TECHNIQUES: ASSESSING MICHIGAN’S CANADA GOOSE AND CACKLIN G
GOOSE HARVEST USING BAND RETURNS, MORPHOMETRICS, AND

GENETICS ....................................................................................... 1 83
Introduction .............................................................................. 1 83
Methods ................................................................................... 1 85

Band Return Analyses ......................................................... 185

vi

Morphometric Analyses ....................................................... 187

Genetic Analyses ............................................................... 190
Variograrns ...................................................................... 193
Variance in Harvest Estimates ................................................ 194
Results .................................................................................... 197
Band Recovery Analyses ...................................................... 198
Morphometric Analyses ....................................................... 198
Genetic Analyses ............................................................... 203
Discussion ................................................................................ 204
Management Implications .............................................................. 208
Literature Cited .......................................................................... 210
CHAPTER 7
DISPERSAL AND GENE FLOW AMONG WRANGEL ISLAND AND BANKS
ISLAND LESSER SNOW GEESE ............................................................ 215
Introduction .............................................................................. 21 5
Methods ................................................................................... 220
Sample Collection .............................................................. 220
Analysis of Microsatellite Loci ............................................... 221
Analysis of MtDNA ............................................................ 222
Estimates of Gene Diversity and Degree of Population Structuring 223
Estimates of Rates of Migration Among Populations ..................... 224
Results .................................................................................... 225
Discussion ................................................................................ 230
Literature Cited .......................................................................... 238

vii

LIST OF TABLES

CHAPTER 1
Table 1. Life history attributes of snow and Ross's geese and Canada and cackling geese
that may influence spatial genetic structure (Ely and Scribner 1994) ....................... 14
CHAPTER 2

Table 1. Measures of genetic diversity for Canada geese (Branta canadensis) and
cackling geese (Branta hutchinsii) populations ................................................ 58

Table 2. Estimates of hierarchical F-statistics (Weir and Cockerham 1984) for
populations of cackling geese and Canada geese based on eight microsatellite loci.
Variance partitioning: f = alleles within individuals; F = among individuals within the
total population; 0 = among populations. Statistically significant (P < 0.01) values over
all loci are marked with an asterisk ............................................................... 65

Table 3. Pairwise estimates of inter-population variance in allele frequencies (mean 0p
across all loci). Bold values were insignificant after Bonferonni correction for multiple
loci (Rice 1989). Population abbreviations are as listed in Table 1 ......................... 66

Table 4. Hierarchical estimates of variance based on mtDNA clade frequencies. Variance
partitioning: among populations within groups ((Dsc), among populations among groups
((DST), and among groups for multiple group analyses ((DCT). Statistically values (P <
0.05 for mtDNA; P < 0.006 for nDNA after corrections for mulitple loci) are marked
with an asterisk .................................................................................... 68

CHAPTER 3

Table 1. Genetic diversity measures estimated for snow geese and Ross's geese

populations. The species of each population sampled is designated as S = snow = Ross's
geese or R geese. The color phase of each population sample is designated as W = white
or B = blue ........................................................................................ 102

Table 2. Estimates of hierarchical F-statistics (Weir and Cockerham 1984) for
populations of snow geese and Ross's geese based on nine microsatellite loci. Variance
partitioning: f = alleles within individuals; F = among individuals within the total
population; 0 = among populations. Statistically significant (P < 0.01) values over all loci
are marked with an asterisk ...................................................................... 109

viii

Table 3. Pairwise estimates of inter-population variance in allele frequencies (mean 0p
across all loci) for blue- and white-phase lesser snow geese (C. c. caerulescens), greater
snow geese (C. c. atlantica), and Ross's geese (C. rossii). Bold values were statistically
significant after Bonferonni correction for multiple loci (Rice 1989). Population
abbreviations are as listed in Table 1 ........................................................... 110

Table 4. Hierarchical estimates of variance based on mtDNA clade frequencies. Variance
partitioning: among populations within groups (<Dsc), among groups ((DCT), among
populations for non-hierarchical analyses and among populations among groups for
hierarchical analyses ((1)37). Statistically significant values (P < 0.05 for mtDNA; P <
0.006 for nDNA after corrections for mulitple loci) are marked with an asterisk ........ 1 14

CHAPTER 5

Table 1. Mean proportional contributions (iSD) of geese from the Mississippi Valley
Population (MVP), Southern James Bay Population (SJBP), Michigan Mississippi
Flyway Giant Population (MI-MF GP), and Tall Grass Prairie Population (TGPP) to
early (Sept 1-15) and regular season (Sept 16-Dec 31) harvests in MVP and SJBP Goose
Harvest Zones in Michigan fiom 1998-2002. Harvest estimates and standard deviations
(SD) were calculated over 1000 bootstrap replications .................................... 168

CHAPTER 6

Table 1. Comparison of harvest estimates (%) of adult cackling (B. h. hutchinsii) and
Canada (B. c. interior and B. c. maxima) geese based on band recovery, morphometric,
and genetic derivation techniques for the regular hunt season in Michigan from 1998-
2002. Genetic estimates include standard deviation in parentheses based on 1000
bootstrap replicates. Sample sizes are listed below estimates .............................. 201

Table 2. Comparison of harvest estimates (%) of adult cackling (B. h. hutchinsii) and
Canada (B. c. interior and B. c. maxima) geese based on genetic and morphometric
derivation techniques for the early hunt season in Saginaw Bay from 2000-2002.
Morphometric harvest estimates were calculated using three different interior and two
giant culmen reference lengths-estimates listed in this table represent the range of results
from these six morphometric derivations. Genetic estimates include standard deviation in

parentheses based on 1000 bootstrap replicates. Samples sizes are listed below estimates.
....................................................................................................... 202

CHAPTER 7

Table 1. Lesser snow geese allele frequencies and F -statistics for nine biparental and one
maternally inherited marker from three populations of lesser snow geese breeding within

ix

the West Arctic region. Populations include north and south wintering Wrangel Island
and Banks Island .................................................................................. 226

Table 2. Above diagonal: Pairwise F51 comparisonsal among lesser snow geese
populations based on 9 bi-parentally inherited microsatellite loci. Below diagonal:
Pairwise Phisr comparisons among lesser snow geese populations based on maternally
inherited mtDNA. P-values for each pairwise comparison are indicated below estimates
in parentheses. Significant P-values are marked with an asterisk ......................... 229

LIST OF FIGURES

CHAPTER 1

Figure 1. Breeding and wintering distributions and management population boundaries
for snow geese (Chen caerulescens) and Ross’ geese (Chen rossii) .......................... 8

Figure 2. Breeding and wintering distributions and migratory Flyways for cackling geese
(Branta hutchinsii) and Canada geese (Branta canadensis) .................................... 9

Figure 3. Breeding grounds for snow geese (Chen caerulescens) and Ross’s geese (C.
rossii) during the Last Glacial of the Pleistocene (Ploeger 1968). Refuge breeding areas
for snow geese and Ross’s geese included northeast Siberia and Bering Sea area (A): C.
c. caerulescens (white color phase); the Canadian Arctic Archipelago (B): C. c.
caerulescens (blue color phase) and C. rossii; and west Greenland (C): C. c.

atlanticus ........................................................................................... 16

Figure 4. Breeding grounds for Canada geese (Branta canadensis) and cackling geese
(B. hutchinsii) during the Last Glacial of the Pleistocene (Ploeger 1968). Refuge
breeding areas for Canada geese included the Eastern Southern Group (A): B. c.
canadensis, B. c. interior, B. c. maxima, B. c. moflitti, B. c. parvipes; and the Western
Southern Group (B): B. c. fulva, B. c. occidentalis. Refuge breeding areas for cackling
geese included the Aleutian Group (C): B. h. leucopareia; and the Northern Group (D):
B. h. minima, B. h. taverneri, B. h. hutchinsii. Subspecies designations follow Banks et
a1. (2004). Stippled area represents the breeding range of Canada geese and cackling
geese as defined by Ploeger (1968) in the late 1960’s ......................................... 23

CHAPTER 2

Figure 1. Breeding and wintering distributions and migratory F lyways for cackling geese
(Branta hutchinsii) and Canada geese (Branta canadensis). Locations of Pleistocene
glacial refugia proposed by Ploeger (1968) are indicated .................................... 51

Figure 2. Sampling locations for subspecies of cackling geese (letters) and Canada geese
(numbers) populations: A = Semidi Island (Branta hutchinsii leucoparia), B = Buldir
Island (B. h. leucoparia), C = Yukon—Kuskokwim Delta (B. h. minima), D = North Slope
(B. h. taverneri), E = Queen Maud Gulf Bird Sanctuary (B. h. hutchinsii, Short Grass
Prairie Population), F = Baffin Island (B. h. hutchinsii, Tall Grass Prairie Population); 1 =
Fairbanks (B. canadensis parvipes), 2 = Anchorage (B. c. parvipes), 3 = Copper River
Delta (B. c. occidentalis), 4 = Admiralty Island (B. c. fulva), 5 = Washington (B. c.
mofiitti), 6 = Nebraska Sandhills (B. c. moflitti), 7 = Churchill (B. c. interior, Eastern
Prairie Population), 8 = Hudson and James Bays (B. c. interior, Mississippi Valley
Population), 9 = Akimiski Island and southern James Bay (B. c. interior, Southern James

xi

Bay Population), 10 = West Ungava Peninsula (B. c. interior), 11 = East Ungava
Peninsula (B. c. interior), 12 = Greenland (B. c. interior), 13 = Newfoundland (B. c.
canadensis, North Atlantic Population), 14 = Labrador (B. c. canadensis, North Atlantic
Population), 15 = Ontario (B. c. maxima), 16 = southeast Michigan (B. c. maxima)...... 56

Figure 3. Neighbor joining tree describing overall genetic similarities among populations
of cackling and Canada geese based on Cavalli-Sforza and Edwards (1967) chord
distances across eight microsatellite loci. Bootstrap support (>50% out of 1000
replicates) for the consensus microsatellite tree is shown to the right of branches or
identified by an arrow. Population abbreviations are as in Table 1 ......................... 70

Figure 4. Mitochondrial DNA cladogram based on maximum likelihood analysis
describing evolutionary relationships among 26 haplotypes of Canada geese resolved on
the basis of mtDNA control region sequence data. Bootstrap values based on 200
replicates exceeding 50% and population locations for haplotypes are included .......... 72

Figure 5. Mitochondrial DNA cladogram based on maximum likelihood analysis
describing evolutionary relationships among 27 haplotypes of cackling geese resolved on
the basis of mtDNA control region sequence data. Bootstrap values based on 200
replicates exceeding 50% and population locations for haplotypes are included .......... 74

CHAPTER 3

Figure l. Breeding and wintering distributions and wintering / management population
boundaries for snow geese (Chen caerulescens) and Ross’ geese (Chen rossii). Locations
of Pleistocene glacial refugia proposed by Ploeger (1968) are indicated ................... 97

Figure 2. Sampling locations for snow geese and Ross’s geese breeding populations: 1 =
Wrangel Island, Russia; 2 = North Slope, Alaska; 3 = Howe Island; 4 = Banks Island; 5 =
Queen Maud Gulf Bird Sanctuary; 6 = Southharnpton Island; 7 = West Hudson Bay; 8 =
LaPerouse Bay; 9 = Cape Henrietta Maria; 10 = Akimiski Island; 11 = Baffin Island; 12
= Bylot Island. Sampling site not pictured: Kolyma River Delta, Japan. White phase
lesser snow geese (Chen caerulescens caerulescens) were collected at locations 1-11 and
the Kolyma River Delta. Blue phase lesser snow geese were collected at locations 5, 7, 9,
10, and 11. Greater snow geese (C. c. atlantica) were collected at location 12. Ross’s
geese (C. rossii) were collected at locations 5, 6, 7, 9, and 11 ............................. 101

Figure 3. Neighbor joining tree describing overall genetic similarities among populations
of snow and Ross’s geese based on Cavalli-Sforza and Edwards (1967) chord distances
across nine microsatellite loci. Bootstrap support (>50% out of 1000 replicates) for the
consensus microsatellite tree is indicated. Population abbreviations are as in

Table 1 ............................................................................................. 1 12

xii

CHAPTER 4

Figure 1. Mean proportional contributions (:hSD) of subspecies (Branta canadensis
maxima, B. c. interior, and B. hutchinsii hutchinsii) of Canada geese to the 1998-99
season harvest in Michigan ...................................................................... 141

CHAPTER 5

Figure 1. Four local goose management units (GMUs) designed to attract migrating
waterfowl are located in the southern Lower Peninsula (SLP) of Michigan, including
Muskegon County Wastewater (M) and Allegan (A) in the Mississippi Valley Population
(MVP) harvest management zone, and Saginaw (S) and Tuscola/Huron (HT) within the
Southern J arnes Bay Population (SJ BP) harvest management zone. Counties within the
MVP zone are white, while counties within the SJBP zone are shaded. GMUs within
these zones are designated by black rectangles ............................................... 150

Figure 2. Distribution of cackling geese and Canada geese harvest samples collected in
Michigan through the U. S. Fish and Wildlife Service Waterfowl Parts Survey. Shaded
counties indicate at least one sample was collected within that county during the annual
harvest season referenced below each map ................................................... 154

Figure 3. Variograms for the spatial distribution of annual harvest samples of cackling
geese and Canada geese collected in Michigan from 1998-2002 Statewide, and within the
Southern James Bay (SJBP) Harvest Zone .................................................... 161

Figure 4. Mean proprotional contributions (iSD) of Branta hutchinsii hutchinsii, B.
canadensis interior, and B. c. maxima to annual statewide Michigan harvests from 1998—
2002. Annual harvests include samples collected September 1 — December 31 during the
year indicated, as well as samples collected in January of the following year (e.g., 1998
includes September 1 — December 31, 1998 and January 1 — 31, 1999). Sample sizes for
each year: 1998 (n = 471), 1999 (406), 2000 (324), 2001 (486), 2002 (585) ............ 162

Figure 5. Mean proportional contributions (iSD) of Branta hutchinsii hutchinsii, B.
canadensis interior, and B. c. maxima to early (September 1 - 15), regular (September 16
— December 31), regular + late (September 16 — January 31), and late (January 1 — 31;
2001 only) seasonal harvests from 1998-2002 in Michigan. Sample sizes for seasonal
harvests in each year: early season 1998 (n = 255), 1999 (195), 2000 (170), 2001 (255),
2002 (261); regular season 1998 (194), 1999 (196), 2000 (143), 2001 (193), 2002 (310);
regular + late seasons 1998 (206), 1999 (207), 2000 (154), 2001 (231), 2002 (323); late
season 2001 (38) .................................................................................. 163

Figure 6. Mean proportional contributions (iSD) of Branta hutchinsii hutchinsii, B.
canadensis interior, and B. c. maxima to each annual hunting season in Michigan from
1998-2002 during six discrete time periods: A (September 1-7), B (September 8—15), C
(September 16-22), D (September 23-29), E (September 30-October 6), and F (October

xiii

7-December 31). Sample sizes for each hunting period are listed by date parenthetically
for each year ....................................................................................... 165

Figure 7. Mean proportional contributions (:tSD) of Canada geese from the Tall Grass
Prairie Population (TGPP), Mississippi Valley Population (MVP), Southern James Bay
Population (SJBP), and Michigan-Mississippi Flyway Giant Population (MI-MFGP) to
annual harvests (1998-2002) in Michigan within the MVP Goose Management Unit
(GMU) and the SJBP GMU. Sample sizes for each year: Mississippi Valley Population
Zone - 1998 (n = 284), 1999 (276), 2000 (194), 2001 (202), 2002 (327); Southern James
Bay Population Zone — 1998 (174), 1999 (127), 2000 (134), 2001 (291), 2002 (255)
....................................................................................................... 167

Figure 8. Mean proprotional contributions (:tSD) of Branta hutchinsii hutchinsii, B.
canadensis interior, and B. c. maxima to regular season harvests (September 16 —
December 31) in Michigan’s four local Goose Management Units (GMU) including
Huron/Tuscola, Saginaw, Muskegon, and Allegan (HTSMA); and Michigan’s non-GMU
counties statewide and in the Southern Lower Peninsula (SLP) from 1998-2002. Sample
sizes for each year: Statewide non-GMU — 1998 (n = 148), 1999 (145), 2000 (115), 2001
(152), 2002 (277); SLP non-GMU - 1998 (57), 1999 (50), 2000 (65), 2001 (116), 2002
(163); HTSMA GMU — 1998 (51), 1999 (55), 2000 (31), 2001 (44), 2002 (30)......... 170

CHAPTER 6

Figure 1. Omnidirectional variograms for the spatial distribution of two collections of
regular season harvest samples of cackling geese and Canada geese in Michigan from
1998-2002. Harvest samples used for morphology-based analyses were collected through
the Michigan Department of Natural Resources, and harvest samples used for genetic-
based analyses were collected through the United States Fish and Wildlife Service
Waterfowl Parts Survey .......................................................................... 199

CHAPTER 7

Figure 1. Breeding (circles) and wintering (squares) populations of lesser snow geese
from Wrangel Island and Banks Island. Fall migratory pathways south to wintering
grounds are marked by solid arrows, and spring migratory routes north to breeding areas
are marked by dashed arrows. Approximately equal proportions of geese breeding on
Wrangel Island winter in the British Columbia / Washington region (NW1) and in
California’s Central Valley (SW1). Lesser snow geese from Banks Island (BI) winter in
the Central Valley (75% of the population) with the SWI population ................... 217

xiv

CHAPTER 1: INTRASPECIFIC PHYLOGEOGRAPHIES OF SNOW, ROSS’S,
CANADA, AND CACKLING GEESE: DEFINING THE SPATIAL STRUCTURE OF

BIOLOGICAL DIVERSITY

INTRODUCTION

Identification of demographically and reproductively isolated populations is an
important aspect of resource management and conservation (Lande 1988). This is
especially crucial for highly migratory species like waterfowl, as populations utilize
several areas throughout the year, often co-occurring with individuals from other
populations, and because rates of natality and survival may vary greatly among
populations. Waterfowl species in North America vary in a number of characteristics that
greatly influence the focus and implementation of management regulations for individual
species. For instance, some species have populations that are very abundant and have
wide geographic distributions, such as temperate nesting Canada geese (Branta
canadensis; Ankney 1996, Rusch et al. 1996b) and Mid-Continent lesser snow geese
(Chen caerulescens; North American Arctic Goose Conference 2001). Other species,
such as spectacled eiders (Somateriafisheri; Scribner et al. 2001) and harlequin ducks
(Histionicus histionicus; Lanctot et al. 1999) have populations that are threatened or
declining and have limited distributions. Ecological traits of species also vary, such as
migration patterns, levels of fidelity to breeding and wintering sites, and breeding
behaviors including timing of pairing and strength of pairbond (Batt et al. 1992, Ely and
Scribner 1994). Thus, successful management plans must consider the ecological traits

and life history characteristics of each species.

Waterfowl management programs attempt to sustain a desirable level of harvest.
while ensuring the long-term viability of contributing breeding populations and
maintaining current levels of biological diversity within species. Challenges to waterfowl
management include: 1) increases in numerical abundance for certain populations or
species, while others have remained stable or have declined, and 2) recent geographic
range expansions of nesting, molting, and wintering areas associated with growing
numbers of waterfowl in certain populations. In addition, most waterfowl are harvested
in locations distant from breeding grounds in flocks composed of several breeding
populations (and possibly from different subspecies and species) from different
geographic locations.

Species, subspecies, and management populations have traditionally been defined
by morphology and plumage characteristics, and banding studies focused on migratory
patterns and population distributions (Munro and Kimball 1982, Moser and Rolley 1990,
Rusch et al. 1996a). However, banding and morphometric techniques focus on
contemporary attributes of species ecologies, and provide little information about historic
population demographics. Additionally, morphological traits can vary as a function of an
individual’s age (Thompson et al. 1999) and environmental factors such as climate and
available food resources (Leafloor et al. 1998), while banding requires a large number of
individuals are marked and resighted through observation or harvest to assess patterns of
migration and population dispersal. Given the limitations of morphometrics and banding
techniques, and the lack of historic population information provided by these data, an
important question arises: Are the management units defined for waterfowl based on

morphology and banding supported by the species underlying genetic structure?

The degree to which populations of waterfowl are spatially structured, and the
utility of neutral genetic markers to describe gametic affinities among populations, varies
among species. Populations can be influenced by historical vicariance and past
dispersion events, and by current levels of gene flow, breeding systems, and ecological
and behavioral characteristics (Avise et al. 1997, Avise 1998). Biogeographic theory
postulates that species distributions and intraspecific relationships are the result of
vicariance events and allopatric speciation, followed by dispersal from historically
fragmented ranges (Wiley 1980, Endler 1982). Explanations for observations of levels of
population structuring based upon measures of genetic divergence between waterfowl
populations should incorporate information regarding separation times (e. g., times to
mean common ancestors; Nielsen 1998) and locations of historical regions of geographic
discordance (Thorpe et al. 1995). In addition, population structuring of waterfowl
species may be influenced at regional scales by a variety of life history traits, including
the timing and location of pair bonding, the degree (and sex-specificity) of breeding and
wintering site philopatry, breeding behavior, and migration route fidelity (Anderson et al.

1992, Ely and Scribner 1994).

Utility of Genetic Markers

Biological diversity within waterfowl species, and the spatial distribution of
populations, subspecies, and species has traditionally been defined by morphology and
plumage characteristics (Bellrose 1980, Moser and Rolley 1990, Alisauskas 1998b,
Giroux et a1. 2001, Pearce et al. 2003), and based on banding studies emphasizing

commonalities in migratory patterns and fidelity to, and potential gene flow among,

wintering and breeding areas (Raveling 1978, Hines et al. 1999, Kerbes et al. 1999,
Drake and Alisauskas 2005). The development of new technologies, including stable
isotope analysis of feathers (Caccamise et al. 2000), and molecular analysis techniques
(e. g. polymerase chain reaction based sequencing of mitochondrial DNA, and neutral
nuclear DNA markers), has offered additional insight as to how biological diversity may
be spatially allocated among populations of waterfowl (Shields and Wilson 1987, Quinn
et al. 1991, Quinn 1992, Baker 1998, Ruokonen et al. 2000, Scrinber et al. 2001, Paxinos
et al. 2002, Scribner et al. 2003a).

Current molecular technologies have several advantages over other methods that
have historically been used to delineate populations, subspecies, and species, and to
quantitatively test hypotheses about the underlying causes of the spatial structure
described. Unlike morphological delineations based on size and plumage characteristics
which can vary with age and nutritional conditions, genetic markers are heritable and are
maintained throughout an individuals lifetime. Morphological characteristics may also
vary geographically, leading to different interpretations about the number of subspecies
or races within a species (Larsson and Forslund 1991, Leafloor and Rusch 1997, Leafloor
et al. 1998, Alisauskas 1998b). For instance, Canada geese have been defined as having
between 8 subspecies (Palmer 1976) and 83 subspecies (Hanson 1997). An individual’s
DNA cannot be lost like a band, collar, or radio-transmitter (Samuel et al. 1990,
Campbell and Becker 1991, Wiebe et al. 2000, Samuel et al. 2001). Virtually any
cellular tissue can be used as a source for DNA. Sampling techniques can be non-

destructive, and often very little tissue or blood is required to characterize the individual.

Most importantly, molecular markers allow statistical inferences to be made based upon
measures of relatedness among groups, without direct observations of movements.

With the development and greater accessibility of molecular techniques (e. g.
polymerase chain reaction based sequencing of mitochondrial DNA, and neutral nuclear
DNA markers), molecular phylogenies can now be constructed (Avise et al. 1987, Avise
1992, Shields and Cotter 1998, Scribner et al. 2003a). Intra- and inter-specific
phylogenies examined in a spatial context, can be used to evaluate the relative importance
of historical events and current ecologic attributes in shaping phylogeographic observed
patterns (Thorpe et al. 1995, Avise 1998, Bematchez and Wilson 1998, Holder et al.
1999). Comparative analyses of phylogeographic structure across multiple species can be
used to identify commonalities and differences between species relative to movements,
life history and behavior, and locations and times of separations during Pleistocene or
earlier vicariant events (Ploeger 1968, Avise and Walker 1998, Klicka and link 1997).

Microsatellites have several features that make them useful for evolutionary and
ecological applications. These markers are loci which consist of a variable number of
tandem repeat sequences of DNA. Typically, microsatellites consist of 10-50 copies of a
repeat sequence l-lO base pairs long. Microsatellite loci are codominant and are
inherited in a Mendelian fashion, with one allele derived from each parent. Their
abundance throughout the genome, high level of variance, and lack of physical linkage
allow for statistically powerful calculations of allele frequencies (Scribner and Pearce
2000). Estimates of allele frequency allow assessments of levels of gene flow, and define

relationships among individuals, populations, subspecies and species.

Mitochondrial (mt) DNA is maternally inherited and transmitted without
recombination, and can be reliably used to reconstruct population genealogies and species
phylogenies. Unique mtDNA sequences called haplotypes tend to evolve faster than
nuclear DNA. Genetic differences accumulate between populations over short
evolutionary time periods (Randi 2000). To identify individual haplotypes, DNA
sequencing is utilized to determine actual base pair structure in targeted regions of the
mtDNA genome. Mitochondrial DNA haplotypes can be used to construct haplotype
trees that illustrate areas of genetic discordance within species. Hypotheses about the
historical and contemporary factors underlying the patterns of genetic discordance may
then be developed and tested based on haplotype trees branch lengths and topology

(Templeton 1998).

PURPOSE OF PROJECT

The focus of this project is two co-distributed species groups of migratory
waterfowl in North America, including two currently recognized species within each
group. The first group is composed of snow geese (Chen caerulescens) and Ross’s geese
(Chen rossii). Current recognized taxonomy of snow and Ross’s geese include the two
species and two subspecies of snow goose, the lesser (C. c. caerulescens) and the greater
(C. c. atlantica) (American Omithologists’ Union 1998, Ryder and Alisauskas 1995,
Mowbray et al. 2000). In addition, the lesser snow goose subspecies is dimorphic, having
recognized light (white) and dark (blue) color morphs. Lesser snow goose color morphs
are controlled by a single gene locus where the dark allele is incompletely dominant to

the light (Mundy et al. 2004). Snow and Ross’s geese populations are currently managed

by species, subspecies, and in mixed groups based on common breeding and wintering
areas (Ryder and Alisauskas 1995, Mowbray et al. 2000). There are four major
management groups for these geese including the Western Arctic Population, the West
Central Flyway Population, the Mid-Continent Population, and the Eastern Population
(Figure 1).

The second group investigated includes cackling geese (Branta hutchinsii) and
Canada geese (Branta canadensis). Four subspecies, B. h. hutchinsii, B. h. leucopareia,
B. h. taverneri, and B. h. minima, are included as part of the cackling geese species. The
Canada goose species is composed of subspecies B. c. canadensis, B. c. interior, B. c.
occidentalis, B. c. fulva, B. c. maxima, B. c. moflitti, B. c. parvipes (Banks et al. 2003).
Until 2003, the four subspecies of cackling geese were categorized as part of the Canada
goose species group and referred to as “small-bodied subspecies” (Delcour 1956).
Several genetic studies of mitochondrial DNA (VanWagner and Baker 1986, Shields and
Wilson 1987, Quinn et al. 1991, Scribner et al. 2003a) verified previous subspecific
taxonomic delineations based on difference in vocalizations, nesting habits, habitat,
timing of migration, as well as color and size. The subspecies were historically treated as
a single species, but actually constitute two species (Banks et al. 2003). Cackling geese
and Canada geese are currently managed in North America on several levels including
species, subspecies, by affiliation to one of the four migratory F lyways (Figure 2), and by
management populations (n = 19 for cackling and Canada geese combined; Appendix 1;
Dickson 2000, Moser and Caswell 2004).

The first objective of this project was to genetically characterize populations,

subspecies, and species of snow and Ross’s geese and cackling and Canada geese using

 

breeding-snow and Ross’s geese
D breeding-snow
I wintering-snow and Ross’ .
E wintering-snow ff
. ’ A I” ,-
Eastern Population ‘r" f
' ' ' Mid-Continent Population N” - ) L\./‘\,_./l] ,_,,
—— West Central Flyway “
- - -- Western Arctic Population
Figure 1. Breeding and wintering distributions and management population boundaries
for snow geese (Chen caerulescens) and Ross’s geese (Chen rossii).

   
       

f
I

breeding-cackling ..

if; breeding-Canada 1.

D breeding and wintering-Canada \

I wintering-cackling and Canada ‘ z; I g
: : (“1" :
E :z (A 2’ Cir“ I I
.. . , I j i i.

.,..
VI
.3;

Pacific E Central 'MississippiEAtlantic
Flyway Flyway Flyway Flyway

Figure 2. Breeding and wintering distributions and migratory Flyways for cackling geese
(Branta hutchinsii) and Canada geese (Branta canadensis).

mitochondrial and microsatellite molecular markers. These molecular surveys were
designed to evaluate the degree of genetic variation at multiple spatial levels, and identify
areas of genetic discordance within and among species. The second objective is to test
hypotheses regarding the relative importance of causal historical and contemporary
factors that have defined the genetic spatial structure observed. The third objective was
to determine whether the present species, subspecies, and management unit designations
for each group of geese reflect the underlying spatial structure as seen at the level of
genes. Data collected as part of the first three objectives will be utilized to address
specific issues of population management in snow and Ross’s geese and within cackling
and Canada geese. Specifically, the final objective of this project was to use information
of spatial genetic affinities among populations to estimate gene flow and to determine the

composition of admixed groups during periods of harvest.

BACKGROUND

Defining Spatial Genetic Structure

Spatial patterns of genetic diversity within and among species are defined by
historical and contemporary processes (Avise et al. 1987, Avise 1998). Past vicariant
events, such as restrictions of populations within isolated refugia during Pleistocene or
Pliocene glacial periods and subsequent range expansion during interglacial periods,
influence phylogenetic structure within and among species (Avise and Walker 1998, link
1996, Klicka and link 1997, Voelker 1999). Smaller historic effective population sizes
during and following glacial events likely increase levels of genetic drift within

populations, and increase the variance in allele or haplotype frequency among

10

populations (Avise et al. 1988, Chesser et al. 1993, Chesser and Baker 1996). Drift and
levels ofgenetic divergence will increase in a time-dependent fashion (Slatkin 1987,
Arbogast et al. 2002).

Ecology and life history characteristics of a species including mating and
breeding behaviors, fidelity to natal areas and breeding and wintering sites, and migratory
behaviors can also have a profound effect on the degree of genetic drift and patterns of
gene flow, and the resulting spatial patterns of genetic structuring (Chesser 1991a, Ely
and Scribner 1994, Pearce et al. 2000). Genetic drift will decrease and gene flow will
increase as more individuals interbreed across populations within a species (Chesser
1991b). High natal dispersal and low breeding site fidelity in one or both sexes will
increase levels of gene flow among populations and decrease the degree of spatial genetic
structure among populations within species (Greenwood 1980, 1987, Chesser 1991b).
Additionally, pair bond formations that occur when individuals from different
populations are sympatric will potentially increase gene flow among populations (Cooke
et al. 1988, Robertson and Cooke 1999). Behaviors which discourage interbreeding
among individual from different populations will increase genetic structuring between
populations (Cooke et al. 198 8, Chesser 1991a, Petrie and Kempenaers 1998, Miyatake
and Shimizu 1999).

North American avian species including songbirds and Neotropical migrants
(link 1996, Kimura et al. 2002), shorebirds (Wenink et al. 1994), wading birds (Rhymer
et al. 2001), and migratory waterfowl (Avise et al. 1992, Quinn 1992, Scribner et al.
2001, Scribner et al. 2003a, Pearce et al. 2004, Peters et al. 2005, see review in Avise and

Walker 1998) exhibit a variety of phylogeographic patterns (link 1997) ranging from

11

nearly panmictic species (Lanctot et al. 1999, Pearce et al. 2004) to species with strongly
subdivided subspecies (Kimura et al. 2002) or populations (Wenink et al. 1994). Some
phenotypically variable species, like the red-winged blackbird (Agelaius phoeniceus) and
common grackle (Quiscalus quiscula), were found to lack spatial genetic structure (Ball
et al. 1988, link 1996). Other species show evidence of population genetic structure,
such as the fox sparrow (Passerella iliaca; link 1996), dunlin (C alidris alpina; Wenink
et al. 1994).

Species that do not have congruent phylogeographic structure may have not been
historically co-distributed, and may have inhabited different refuges during historic
isolating events (link 1997). In addition, there is a correlation between elapsed time
since sharing a common ancestor and current levels of spatial genetic structure within and
among species. For instance, red-winged blackbirds may have only recently colonized
their current range through dispersal from one panmictic historic population, while the
fox sparrow likely experienced historical isolating events that genetically diverged

populations prior to recent post-glacial dispersals (link 1996).

Arctic-Nesting Waterfowl

Patterns of spatial genetic structure vary greatly among co-distributed species of
Arctic-nesting waterfowl. For instance, western and eastern populations of king eiders
(Somateria spectabilis) are genetically homogeneous (Pearce et al. 2004), while
spectacled eiders (S. fischeri) are characterized by significant genetic variation among
their three primary nesting populations (Scribner et al. 2001). Species such as the

northern pintail (Anas acuta; Cronin et al. 1996) and mallard (Anas platyrhynchus; Avise

12

et al. 1990) appear to be nearly panmictic over extensive geographic areas. In contrast,
brant (Branta bernicla; Shields 1990) and common eiders (S. mollissima; Sonsthagen et
al. 2005) have significant phylogenetically distinct haplotype groups that exhibit a
geographic orientation reflecting past vicariance events. Differences in the degree of
genetic divergence evident in co—distributed taxa points to the varying importance of
historical factors and contemporary levels of gene flow in shaping genetic structure
within different species (Avise and Walker 1998).

Within of waterfowl, cackling and Canada geese, and snow and Ross’ geese
exhibit several characteristics that make them useful groups for intraspecific phylogenetic
analyses. First, they are independent, but co-distributed taxa that have similar historical
geographic ranges (Ploeger 1968). This allows us to examine the degree of population
structuring based on hypothetical glacial refugia and post-glacial dispersal. Secondly,
cackling and Canada geese and snow and Ross’s geese differ in a number of life history
characteristics which may impact levels of genetic drifi and gene flow among
populations, and influence genetic variation at different spatial scales (Table 1; Ely and
Scribner 1994).

There has been growing concern over the management of cackling and Canada
geese, and snow and Ross’s geese. Over the past few decades, both groups have seen
tremendous increases in numerical abundance for certain populations (or subspecies),
while other populations have remained stable or have declined (Ankney 1996, Pacific
Flyway Council 1997, Alisauskas 1998a, Kelley et al. 1998, Kerbes et al. 1999, North
American Arctic Goose Conference 2001). Geographic ranges of nesting, molting. and

wintering areas have also recently increased in conjunction with growing numbers of

13

.833 case a 3:85 n @0359.

 

 

EOE 32 03.2.5, £83568 Bahama. czar? beasfwsam 385-.“ 385$ $523 :3ch
>904 :wE wcobm cox—«EBB 3:28 wcobm wctamZBEB Remand *vomflni mumox \ 30cm
maaeoafim .285 8:232 caseema 832$ Essa was“; .6 E05 9%: an seam
“Em
45552 as»; .3: Biz 332 beam was: ”535 .52
L85

 

ASE septum Ea bmo 0508: 385%
33% cocosmfi BE 35 89% maze—o8 USN guano can omoow mumom BS... Beam me 3:533 beam: £5 4 £an

14

geese (Bateman et al. 1988, Malecki and Trost 1998). Management directed to control
expansions of overabundant populations of geese, while protecting others, emphasizes the
necessity of using techniques that can differentiate populations, subspecies, and species
that may be contributing to mixed migratory or wintering flocks. Canada geese have also
become more prominent in urban areas, and are now considered over abundant in more
than 100 urban areas in 37 states (Mowbray et al. 2002). Increasing mixing of resident
Canada geese with migrant Canada and cackling geese during migration and on winter
areas has resulted in uncertainty of racial composition of harvests that has created
management problems, and potentially threatening the viability of numerically depressed

populations of both species across their North American ranges (Mowbray et al. 2002).

Snow and Ross’s Geese
Historical Processes

Climatic changes over time have likely had a large effect on the phylogeographic
structure of snow geese and Ross’s geese, as both species currently breed in high latitude
regions of North America that were recolonized from refugia in Beringia and other ice-
free areas north of major glaciers (Figure 3). Snow geese and Ross’s geese likely utilized
these high Arctic refuges during the last Pleistocene glacial event (26,000-18,000 ybp;
Ploeger 1968), and may have been isolated in separate refugia in east Siberia and the
Bering Sea area (snow geese), and in the Western Canadian Arctic Archipelago (Ross’s
geese) during the previous glacial maximum (150,000-130,000 ybp). Within the snow
geese complex, greater snow geese may have occupied refuge breeding grounds in

western Greenland during the last glacial, and were likely geographically isolated from

15

 

 

 

 

 

 

 

 

 

Figure 3. Breeding grounds for snow geese (Chen caerulescens) and Ross’s geese (C.
rossii) during the Last Glacial of the Pleistocene (Ploeger 1968). Refuge breeding areas
for snow geese and Ross’s geese included northeast Siberia and Bering Sea area (A): C.
c. caerulescens (white color phase); the Canadian Arctic Archipelago (B): C. c.
caerulescens (blue color phase) and C. rossii; and west Greenland (C): C. c. atlanticus.

lesser snow geese populations (Ploeger 1968). Blue-phase and white-phase lesser snow
geese may have also been allopatric during this period, with blue-phase geese surviving
in the Canadian Arctic Archipelago and white-phase geese inhabiting refuges farther west
in Siberia and the Bering Sea region (Ploeger 1968, Cooke et al. 1988). Ploeger (1968)
speculated that speciation occurred when snow and Ross’s geese ancestors were isolated

between the E. Siberia, Bering Sea area and the Canadian Arctic Archipelago, leading to

16

western and eastern populations from which lesser snow and Ross’ geese subsequently

arose.

Breeding Areas

About 95 percent of all Ross’ geese breed in the central Canadian Arctic in the
Queen Maud Gulf Migratory Bird Sanctuary (Kerbes 1994). Smaller numbers of Ross’
also breed in other areas of the central Arctic including the west and south coasts of
Hudson Bay, and on Southampton and Baffin Islands. Few Ross’ geese are found
breeding in the western Arctic on Banks Island and the north coast of Alaska (Ryder and
Alisauskas 1995, Figure 1). Ross’ geese nest in colonies interspersed with lesser snow
geese, which breeds at very similar latitudes. Like Ross’ geese, breeding populations of
lesser snow geese are greatest in the central Canadian Arctic, and Ross’s and snow geese
breeding populations in this area are collectively termed the “mid-continent population”.
Lesser snow geese are more widely distributed across central and western Arctic breeding
areas than Ross’s geese, and are found at higher latitudes than Ross’s geese. Greater
snow geese breed in the high eastern Canadian Arctic, with the largest colony found on
Bylot Island, and fewer numbers breeding on coastal areas of surrounding islands. Small
numbers. of lesser snow geese are found breeding in conjunction with greater snow geese

colonies (Mowbray et al. 2000).

W intering Areas

The Central Valley of California is the main wintering area for Ross’s, though

increasing numbers of these geese are wintering in Arkansas, Louisiana, New Mexico,

17

Texas, and the north-central highlands of Mexico (Turner et al. 1994, Figure 1). Lesser
snow geese have a more extensive wintering range than Ross’ geese, and are found
within multiple wintering areas in Oregon, California, New Mexico, Oklahoma, Texas,
numerous states along the Mississippi River in the Central and Mississippi Flyways,
north and west Gulf Coast areas, and the highlands of Mexico (Figure 1). The largest
numbers of lesser snow geese winter in coastal areas of Texas, Louisiana, and Mississippi
(Mowbray et al. 2000, US. Fish and Wildlife Service 2001). Greater snow geese winter
with small numbers of lesser snow geese in areas along the Atlantic coast from

Massachusetts to South Carolina (Reed and Chagnon 1987, Cooke et al. 1995, Figure 1).

Population Size

In the early 19005, the Ross’ goose was considered a rare species, and hunting
was prohibited in 1931 in an attempt to increase population sizes. Since that time, Ross’
geese have increased from 2,000-3,000 birds to a current population of 700,000 (Ryder
and Alisauskas 1995). Both Ross’ geese and lesser snow geese have recently taken
advantage of previously unavailable sources of food (agricultural plants and waste grain)
on their wintering grounds and migration routes. Numbers of lesser snow geese have
increased as much as 9 percent per year in some mid-continent populations, and current
estimates are between 6 and 7 million birds (Mowbray et al. 2000). As lesser snow geese
populations have increased, established colonies have expanded and new colonies have
been established in western and central Arctic regions. Several breeding colonies have
appeared south of Queen Maud Gulf (Alisauskas and Boyd 1994) and along Hudson Bay

(Cooke et al. 1995) since 1950, and a colony was established on the North Slope of

18

Alaska during the early 19803 (Johnson 1995). Like Ross’ geese, greater snow geese
were estimated at 2,000-3,000 birds in the early 19005 (Hill and Frederick 1997). Greater
snow geese have followed similar increasing trends as lesser snow geese, reaching

750,000 individuals in the late 1990’s (Reed et al. 1998).

Ecology and Behavior

Female snow and Ross’s geese exhibit a high degree of natal- and breeding site-
fidelity compared to males, who are more likely to disperse (Rockwell and Cooke 1977,
Geramita and Cooke 1982, Cooke 1987, Anderson et al. 1992, Drake and Alisauskas
2005). Males generally follow females to areas near their natal breeding colony after
pairing on wintering grounds or during spring migration (Cooke et al. 1975, Greenwood
1980, Mowbray et al. 2000). Observers of lesser snow geese at LaPerouse Bay have
documented up to twice as many adult females (75.8%) as males (33.7%) returning to a
site in subsequent years (Cooke and Sulzbach 1978, Cooke et al. 1982), and fewer than
1% of males banded as goslings were resighted at their natal colony (Cooke et al. 1975).
High levels of male mediated dispersal may considerably increase gene flow among
breeding colonies of snow and Ross’s geese (Rockwell and Barrowclough 1987).

Wintering site fidelity has been noted in several populations of snow and Ross’s
geese (Robertson and Cooke 1999). Winter affiliations of lesser snow geese nesting in
the western Arctic have been examined using neck collar observations and degree of
facial plumage staining acquired on winter feeding grounds (Baranyuk et al. 1999, Hines
et al. 1999). In a banding study of Wrangel Island nesting lesser snow geese, Baranyuk

et al. (1999) determined that 90% of marked birds with dark facial staining returned to

19

suspected wintering grounds in British Columbia and Washington; while 86% of marker
birds with little or no facial staining returned to suspected wintering grounds in
California. High rates of wintering ground fidelity (96-98%) were also estimated by
Williams et al. (2005) based on banding observations for Wrangel Island and Banks
Island geese.

For snow and Ross’ geese, initial pairing occurs on the wintering grounds or
during spring migration among 2 to 3-year-old-birds (Bellrose 1980, Cooke 1987, Cooke
2001, Ryder and Alisauskas 1995). Thus, pairs are formed when many regional
populations are mixed, prior to their return to arctic breeding sites. Snow and Ross’s
geese exhibit long-term monogamy with strong pair bond stability (Anderson et al. 1992).
Pair bonds last for life, but individuals will commonly pair and mate with a new
individual if the partner dies (Prevett 1972, Bellrose 1980, Cooke et al. 1981).

In the fall and spring, lesser snow geese migrate primarily in the Mississippi and
Pacific Flyways, and utilize the Central and Atlantic Flyways to a lesser extent. Ross’
geese migrate along similar routes as lesser snow geese in the Mississippi and Pacific
Flyways (Bellrose 1980). Greater snow geese are somewhat more isolated, as they travel
down the Atlantic coast to the Carolinas in the fall, and return by the same route in the
spring (Bellrose 1980, Reed et al. 1998). Some populations of snow and Ross’ geese
utilize more than one Flyway. For instance, approximately 80% of lesser snow geese
from the Western Canadian Arctic and Wrangel Island migrate along the Pacific F lyway
to California, while the remainder followed a path along the Central F lyway (Armstrong

et al. 1999). Ross’ geese from Queen Maud Island in the Central Canadian Arctic have

20

been documented migrating down the Missippi F lyway to the Gulf Coast, and along the

Pacific F lyway to the California Coast (Alisauskas 1998a).

Genetic Studies

Mitochondrial markers employed to examine the genetic characteristics of snow
and Ross’ geese have been used in multiple studies to investigate the level of genetic
structuring within and among these species (Avise et al. 1992, Quinn et a1. 1992,
Weckstein et al. 2002). Like cackling and Canada geese, two major clades of mtDNA
haplotypes exist among white geese (Avise et al. 1992). However, mtDNA haplotypes
within each white goose clade were found to be widespread among populations of snow
and Ross’ geese in all previous studies (Avise et al. 1992, Quinn 1992, Weckstein et al.
2002). Based on the wide distribution of haplotypes, Avise et al. (1992) concluded that
snow and Ross’ geese had no current phylogeographic structure, and hypothesized clade
differences were a result of secondary introgression among the two species. Quinn
(1992) argued that mtDNA sequences within one of the two haplotype clades were
concordant with the geographic location of eastern and western populations of snow
geese. Ten years later, Weckstein et al. (2002) added sampling locations and individuals
to Quinn’s dataset and concluded the level of mtDNA variation was consistent with the
hypothesis that sharing of two mtDNA haplotype lineages between snow and Ross’ geese
resulted from secondary introgression, or hybridization (Avise et al. 1992). Weckstein et
al. (2002) further stated that population structure found within one haplotype cluster
supported the notion of past allopatry between blue and white phase lesser snow geese

(Cooke et al. 1988). The slight discrepancies among these studies may be clarified by

21

increasing number of snow and Ross’ populations, and number of individuals, sampled
for phylogeographic analyses. In addition, biparentally inherited microsatellite markers
could be used in conjunction with maternally inherited mtDNA to investigate the
influence of contemporary verses historical influences on the genetic structure of these

two species.

Cackling and Canada Geese
Historical Processes

Based on species ecology, nesting habits, and availability of unglaciated breeding
habitat available during the Pleistocene, Ploeger (1968) identified four major breeding
areas for cackling and Canada geese (Figure 4). He hypothesized that Canada geese
inhabited refugia south of the major ice sheets, while cackling geese were found north of
the glaciers in the Aleutian Islands and Canadian Arctic Archipelago. Canada geese were
split between the “Western Southern Group” nested in the Pacific coastal region south of
the Cordilleran ice sheet and included current subspecies B. c. fulva and B. c.
occidentalis; and an “Eastern Southern Group” composed of current subspecies B. c.
canadensis, B. c. interior, B. c. maxima, B. c. moffitti, and B. c. parvipes nested south of
the Cordilleran and Laurentide ice sheets east of the Cascade and Sierra Nevada
Mountains. Cackling geese were also said to inhabit two refugia including the “Aleutian
Group” where B. h. leucopareia nested; and the “Northern Group” of the Canadian
Archipelago which included the three remaining recognized subspecies, B. h. minima, B.

h. taverneri, and B. h. hutchinsii. During the last glacial event, each of these four groups

22

was likely isolated, although interrnixing between subspecies within a group was still

possible (Ploeger 1968).

 

 

 

 

 

 

 

Figure 4. Breeding grounds for Canada geese (Branta canadensis) and cackling geese
(B. hutchinsii) during the Last Glacial of the Pleistocene (Ploeger 1968). Refuge
breeding areas for Canada geese included the Eastern Southern Group (A): B. c.
canadensis, B. c. interior, B. c. maxima, B. c. moffitti, B. c. parvipes; and the Western
Southern Group (B): B. c. fulva, B. c. occidentalis. Refuge breeding areas for cackling
geese included the Aleutian Group (C): B. h. Ieucopareia; and the Northern Group (D):
B. h. minima, B. h. taverneri, B. h. hutchinsii. Subspecies designations follow Banks et
al. (2004). Stippled area represents the breeding range of Canada geese and cackling
geese as defined by Ploeger (1968) in the late 1960’s.

23

Breeding Areas

The breeding ranges for the four subspecies of Cackling geese are generally
located at higher latitudes than the breeding ranges of the seven subspecies of Canada
geese. Subspecies distributions within both species are somewhat isolated, with only
nearest east-west neighbor subspecies overlapping a given subspecies breeding range.
Cackling geese breed from the Aleutian Islands and coastal Alaska east across coastal
areas of northern Canada and west Hudson Bay, and on Victoria, Southampton, and
Baffin Islands (Mowbray et al. 2002, Banks et al. 2003, Figure 2). Canada geese breed
over an extremely wide range of habitats throughout temperate and arctic regions of
North American including central and southeast Alaska, many areas of northern Canada
from Yukon Territory to Newfoundland and Labrador, west Greenland, and south into the
northern and central areas of the United States (Bellrose 1980, Mowbray et al. 2002,
Figure 2). The eleven subspecies recognized as cackling and Canada geese are also
currently defined under 19 management units across North America (Mowbray et al.

2002, Dickson 2000, Appendix 1).

W intering Areas

The wintering range of cackling geese includes southeast coastal Alaska, southern
British Columbia, southern Washington state and northern Oregon, and the Central
Valley of California for the western breeding populations; and southwest Texas, northern
Mexico, and the western Gulf coast for cackling geese breeding in the central and eastern

Arctic (Bellrose 1980, Banks et al. 2003, Figure 2). Wintering areas for Canada geese

24

include most of the United States, northern highlands of Mexico, and the Atlantic and

Gulf coasts of Mexico (Bellrose 1980, Mowbray et al. 2002, Figure 2).

Population Size

Canada and cackling geese have generally increased in North America since the
mid-century, resulting in current estimates of approximately 4.5 million for Canada geese
and 600,000 for cackling geese (Mowbray et al. 2002). However, population indices are
not available for some subspecies (i.e., B. c. parvipes and B. c. fulva), some subspecies
are not well surveyed in portions of their range, and some local populations and
subspecies (B. h. minima) have declined during this period (Drut and Trost 2001,
Mowbray et al. 2002). Populations of local Canada geese breeding in the northern United
States have rapidly increased in recent years, with some groups losing their migratory

habits and becoming established as year-round residents.

Ecology and Behavior

Cackling and Canada geese exhibit strong female and male fidelity to breeding
grounds, (Greenwood 1980, Lessells 1985, Sjoberg and Sjoberg 1998). Both Lessells
(1985) and MacInnes and Lieff (1968) documented return rates to breeding grounds (50-
60%) to be approximately equal for males and females. In contrast, rates of natal fidelity
are female—biased, and much lower (3-10%) than rates of breeding ground fidelity
(MacInnes and Lieff 1968, Surrendi 1970).

Wintering site fidelity has been documented for several subspecies of cackling

and Canada geese. Raveling (1979) estimated winter site fidelity for Canada geese (B. c.

25

maxima) to be 80.0% for male and 76.7% for female yearlings, and 69.2% for male and
48.9% for female 2-year-olds. Similar proportions of marked cackling and Canada geese
(juveniles-68%, adults-73%) returned to California wintering areas over a two year study
(Johnson and Raveling 1988). In a study of eastern US. wintering areas for Canada
geese populations, Hestbeck et al. (1991) found winter fidelity rates of B. c. canadensis to
be 71.0% for mid-Atlantic states, 88.9% near the Chesapeake Bay, and 56.2% for the
Carolinas.

Cackling and Canada geese may pair as yearlings, but are more likely to first pair
and breed when they are in their second year (Craighead and Stockstad 1964, Raveling
1981). Initial pair bonds in cackling and Canada are formed in the spring during
northward migrations or after the geese have returned to breeding sites (MacInnes 1966,
Hanson 1997). Both cackling and Canada geese exhibit long-term monogamy (Anderson
et al. 1992, Mowbray et al. 2002), and pairs will generally remain together through
multiple breeding seasons, unless one of the pair dies (Bellrose 1980). Studies have
illustrated, however, that pair bond stability is variable among subspecies of cackling and
Canada geese (Johnson and Raveling 1988, Raveling 1988, Ely and Scribner).

Cackling and Canada geese show strong affiliation to particular migration routes,
resulting in highly localized populations (Craven and Rusch 1983, Tacha et al. 1991,
Didiuk and Caswell 1998, Gill et al. 1998). Their geographic migration patterns are so
consistent that wildlife managers and biologists often refer to the recognized population
groups that utilize each of the four Flyways in North America (Figure 2; Malecki and
Trost 1998). Migration distances between breeding and wintering areas varies greatly,

with individuals from Arctic and sub-Arctic areas of Canada and Alaska traveling the

26

greatest distances to midlatitude and southern areas of the United States. Populations
nesting in southern Canada and northern states of the U. S. migrate shorter distances
(Mowbray et al. 2002). Some populations of Canada geese winter within their breeding
area, or have lost their migratory habit altogether (Mowbray et al. 2002). Northward
molt-migrations have been documented for non-breeding Canada geese from populations

in the northern US. and southern Canada (Abraham et al. 1999, Luukkonen et al. 2004).

Genetic Studies

Molecular studies of cackling and Canada geese have addressed subspecies
distinctions and phylogeography of these two sister species as one taxonomic group
previously recognized as one species, Canada geese (Delcour 1956, Bellrose 1980). Past
projects focused on mitochondrial (mt) DNA illustrated large differences in haplotypes
resolved for “small-bodied” Canada geese subspecies (now known as cackling geese) as
compared to haplotypes found in “large-bodied” Canada geese subspecies (Shields and
Wilson 1987, Van Wagner and Baker 1990, Quinn et al. 1991, Baker 1998, Shields and
Cotter 1998). Although earlier studies suggested Canada geese subspecies could be
unambiguously distinguished on the basis of haplotype, it is now recognized that
common haplotypes are shared across multiple geographic locales and subspecies within
the small-bodied clade (cackling geese) and within the large-bodied clade (Canada
geese). No known mtDNA haplotypes are shared between cackling geese and Canada
geese (Scribner et al. 2003a).

An investigation of the mtDNA and nuclear DNA of subspecies of cackling and

Canada geese in western North America revealed considerable spatial genetic structure at

27

macro- and micro-geographic scales. In this region, differences between cackling and
Canada geese were most influenced by historical population fragmentation, while
isolation by distance and long-distance colonization events were the main factors driving
genetic differences among subspecies within each species (Scribner et al. 2003a).
Scribner et al. (2003b, 2003c) have completed several genetic-based investigations of
management populations of cackling and Canada geese in the Mississippi and Atlantic
Flyways, but no comprehensive, species-wide phylogeographic study has yet been

conducted for cackling and Canada geese in North America.

OUTLINE OF CHAPTERS

Chapters 2 and 3 describe the comprehensive species-wide phylogenetic analyses
of Canada geese and cackling geese, and snow geese and Ross’s geese within North
America. By utilizing molecular markers which differ in their mode of inheritance and
evolutionary rates, 1 survey the degree of genetic variation at the levels of populations,
subspecies, and species within Canada and cackling geese and within snow and Ross’s
geese. I assess the relative influence of historical processes and current behavioral and
ecological factors on the spatial genetic structure revealed for both groups of taxa.
Current population, subspecies, and species designations are then evaluated for snow and
Ross’s geese and for cackling and Canada geese based on the degree of phylogenetic
structure resolved. These chapters build upon previous genetic studies for both
taxonomic groups by expanding the number of individuals sampled, including sampling

locations from all subspecies and major breeding populations within each species, and by

28

characterizing subspecies and populations sampled using both microsatellite and mtDNA
markers.

Chapters 4, 5, and 6 utilize the genetics data collected as part of the phylogenetic
analyses to address questions related to the harvest management of cackling and Canada
geese. It has recently been demonstrated that the degree of genetic population structuring
among localized groups of cackling and Canada geese can be used successfully for
purposes of harvest derivation (Pearce et al. 2000, Scribner et al. 2003b). Potentially
contributing populations and samples from harvest mixtures can be characterized using
nuclear microsatellite loci, and compared using mixed stock analyses (Pella and Milner
1987, Smouse et al. 1990, Xu et al 1994) in order to estimate the proportional
contributions of management populations, subspecies, and species in harvest mixtures.
Two species (Richardson’s cackling geese, and interior and giant subspecies of Canada
geese) and four management populations [Tall Grass Prairie Population (TGPP), MVP,
SJBP, and Michigan’s Mississippi F lyway Giant Population (MI-MFGP)] of geese
potentially contribute to mixed annual harvests in the state of Michigan. Harvest policies
developed by the Michigan Department of Natural Resources seek to target the large and
rapidly growing population of resident giant Canada geese (Branta canadensis maxima)
and avoid harvest of migrant interior (B. 0. interior) Canada geese and cackling geese (B.
hutchinsii hutchinsii) (Soulliere et al. 1988, Luukkonen and Soulliere 2004).

Chapter 4 focuses on the joint use of standardized parts collections and genetic-
based analyses for harvest derivations. I describe how statewide harvest samples
collected through the United States Fish and Wildlife Service’s Waterfowl Parts Survey

can be used in conjunction with maximum likelihood methods to accurately estimate

29

proportional harvests of populations of cackling and Canada geese. Previous harvest
studies in the state (Scribner et al. 2003b) collected harvested geese from a few private
hunting areas and from state game area check stations (areas managed to attract
migratory cackling and Canada geese). By utilizing the broad-based and systematic
sampling provided by the Parts Survey, and by increasing the number of breeding
populations in the baseline, 1 provide more representative statewide measures of the
racial composition of goose harvest in Michigan as compared to previous harvest studies.

In Chapter 5, I utilize established methods of genetic stock identification and
nationally standardized sample collections (as described in Chapter 4) to estimate
proportional contributions of three subspecies (cackling geese, and interior and giant
Canada geese) and four management populations (TGPP, MVP, SJBP, MI-MFGP) of
geese to consecutive annual harvests over the period l998-2002 in Michigan. The
Michigan Department of Natural Resources has implemented special early and late goose
seasons, and has shifted the opening date of the regular season from early October to
mid-September in order to target a large and rapidly growing population of resident giant
Canada geese and avoid harvest of migrant interior Canada geese and cackling geese
(Soulliere et al. 1988, Luukkonen and Soulliere 2004). Thus, harvest derivations are
conducted for multiple geographic and temporal breakdowns of statewide harvests, in
order to evaluate harvest regulations in Michigan including special seasons and harvest
zones.

Chapter 6 is a comparative analysis of harvest of cackling and Canada geese in
Michigan estimated using three different derivation techniques. Results of the genetic-

based harvest derivations from Chapter 5 will be compared and contrasted to harvest

30

estimates derived from morphological measures and band-return data for the same five
year period in Michigan. An additional comparative analysis of a special early season
harvest from 2000 -— 2002 will be conducted using genetic and morphometric derivation
techniques. A discussion of the efficacy of using each derivation technique at different
spatial and temporal scales of harvest will follow.

Chapter 7 is focused on the management concerns surrounding the lesser snow
goose breeding population located on Wrangel Island, Russia. In contrast to the Mid-
Continent Population of lesser snow geese, and the Banks Island colony in the western
Arctic, the Wrangel Island population has experienced considerable declines over the past
35 years (Kerbes et al. 1999). A better understanding of the population structure among
breeding and wintering western arctic snow geese in terms of accurately defining
appropriate management units, identifying demographic constraints on population
dynamics, and determining the level of interaction and gene flow among populations, is
important for the conservation and management of the Wrangel Island population.

Lesser snow geese on Wrangel Island form one of two large breeding colonies in
the western arctic, and is the only remaining snow goose colony in Asia (Boyd 1995). In
addition, the Wrangel Island population is composed of two subpopulations which nest in
a single mixed colony, but winter in two separate locales in the Pacific F lyway (Bousfield
and Syroechkovsky 1985). The more northern wintering group is found in Canada and
northern Washington, and is exclusively Wrangel Island birds, while the southern
wintering group in California includes lesser snow geese from Wrangel and Banks
Islands, and Ross’ geese and a small number of lesser snow geese from Queen Maud

Gulf (central arctic) (Ryder and Alisauskas 1995, Mowbray et al. 2000). Further

31

complicating population assessment is the likely exchange of geese from the north and
south wintering Wrangel Island population with the Banks Island population (Hines et al.
1999). The wealth of banding data available for these western Arctic populations
(Armstrong et al 1999, Hines et al. 1999, Samuel et al. 2001), and the current availability
of molecular markers, offers a unique opportunity to examine the degree of population
structuring among the Wrangel Island and Banks Island colonies using multiple
techniques. A genetic analysis of both the north and south wintering populations of
Wrangel Island and the Banks Island population will be conducted using nuclear and
mtDNA markers. Genetics data will be compared to recent mark-resight banding studies
of Wrangel and Banks snow geese, to determine relative levels of historical and
contemporary gene flow among populations, and to evaluate population structuring

among breeding and wintering groups.

32

Appendix 1. Breeding and wintering ranges, migration routes and subspecific composition of 19
management units of cackling (Branta hutchinsii) and Canada (Branta canadensis) geese (adapted from
Appendix 1, Mowbray et al. 2002).

 

North Atlantic (B. c. canadensis): Breeds from w. Greeland west and south to e. Labrador and
Newfoundland; migrates down Labrador coast to Maritirnes and south along New England coast; winters
south along Atlantic Coast to Long Island Sound and in smaller numbers south to Pea Island, North
Carolina.

Atlantic (B. c. interior): Breeds throughout Quebec, with highest breeding concentrations along coast of
Ungava Bay and northeast coast of Hudson Bay; migrates south along east shore of Hudson and James
Bays, across New York and e. Pennsylvania to Eastern Shore of Maryland; winters from New England
south to South Carolina, with highest densities on Delmarva Penninsula.

Atlantic Flyway Resident (B. c. canadensis, B. c. interior, B. c. maxima): Breeds and winters from 5.
Quebec and Maritime Provinces south throughout states of Atlantic Flyway.

Southern James Bay (B. 0. interior): Breeds on Akimiski Island and adjacent coastal lowlands of 5. James
Bay; migrates south from James Bay across central Great Lakes; winters from s. Michigan south to
Mississippi, Alabama, Georgia, and South Carolina.

Mississippi Valley (B. c. interior): Breeds throughout n. Ontario, in coastal lowlands south of Hudson Bay;
migrate from west shore of James Bay and south shore of Hudson Bay south down each side of Lake
Michigan; winters from se. Wisconsin south, east of Mississippi River, to s. Illinois.

Eastern Prarie (B. c. interior): Breeds in Hudson Bay lowlands of ne. Manitoba; migrates from east coast
of Hudson Bay southwest to interlakes region of Manitoba, then souther through w. Minnesota and e.
Dakotas; winters from Minnesota south to Missouri.

Mississippi Flyway (B. c. maxima): Re-established throughout states of Mississippi Flyway; breeds and
winters throughout region.

Western Prairie/Great Plains (B. c. maxima/B. c. moffittr): Breeds throughout Saskatchewan, North and
South Dakota, Nebraska, Kansas, Oklahoma, and Texas; northernmost breeders migrate south to Missouri
River, then along river; winters from Missouri River in South Dakota south to Texas.

Tallgrass Prairie (B. h. hutchinsii, possibly includes some B. c. parvipes): Breeds in n. Canada from Great
Plain of the Koukdjuak on Baffin Island west to Queen Maud Gulf and south to McConnell and Maguse
Rivers on west coast of Hudson Bay; migrates south from staging areas on west coast of Hudson Bay along
a broad corridor down Mississippi River; winters from Louisiana west to ne. Mexico, with a major
concentration in Oklahoma.

Shortgrass Prairie (B. h. hutchinsii and some B. c. parvipes): Breeds from Queen Maud Gulf westward to
Mackenzie River delta, including 5. Queen Victoria Island, and inland at lower densities in n. Alberta;
migrates in separate eastern and western corridors from breeding areas to staging areas in se. Alberta and
sw. Saskatchemwan and then in a single corridor south; winters in se. Colorado, ne. New Mexico, and
panhandle arcs of Oklahoma and Texas.

Hi-line (B. c. moflittr): Breeds in se. Alberta, sw. Saskatchewan, e. Montana, Wyoming, and n.-central

Colorado; migrates just a short distance southward; winters from n.-central Colorado south to central New
Mexico.

33

Appendix 1 (cont’d).

 

Rocky Mountain (B. c. moflitti): Breeds from sw. Alberta south through intennountain regions of w.
Montana, Utah, Idaho, Nevada, Colorado, and Wyoming; migration a complex set of movements from
higher elevations to lower elevations and in some cases a short distance south; winters from Montana south
to 3. California, 5. Nevada, and Arizona.

Pacific (B. c. moflittr): Breeds and winters from 5. British Columbia south, west of the Rockies and ldaho.
w. Montana, Washington, and Oregon to nw. Nevada and n. California.

Dusky (B. c. occidentalis): Breeding restricted to Copper River delta and adjacent areas of se. Alaska;
migrates south along Pacific coast; winters in Willamette River Valley of w. Oregon and along lower
Columbia River in n.-central Oregon and s.-central Washington.

Cackling (B. h. minima): Breeds on Yukon-Kuskokwim Delta of w. Alaska; migrates from staging areas
on shores of Bristol Bay, across Gulf of Alaska, south to Columbia River, Oregon; winters primarily in
Willamette River valley of w. Oregon and lower Columbia River valley of n.-central Oregon and s.-central
Washington, with small numbers (<10%) in central valleys of California.

Aleutian (B. h. Ieucopareia): Re-established and successfully breeds on several fox-free islands in
Aleutians; migrates from staging areas in Aleutians south across Gulf of Alaska to n. California coast;
winters primarily in San Joaquin Valley of central California, with small wintering concentrations along
Oregon coast.

Pacific Flyway Lesser (B. c. parvipes): Breeds throughout much of central interior Alaska; migrates south
from interior of Alaska along Pacific coast; winters in Washington, Oregon, and California.

Tavemer’s (B. h. taverneri): Breeds in w. Alaska and coastal tundra from Yukon-Kuskokwim Delta
around to north slope of Alaska; migrates from coastal staging areas at tip of Alaska Peninsula south over
water in fall, retums north through interior valleys of British Columbia and Yukon Basin in spring; winters
primarily in Willamette River valley of w. Oregon and lower Columbia River valley of n.-central Oregon
and s.-central Washington.

Vancouver (B. c. fulva): Breeds in extreme sw. Alaska and coastal areas of nw. British Columbia; winters
over much of same area, from se. Alaska south along coast of British Columbia; occasionally to Willamette
River valley of w. Oregon and lower Columbia River valley of n.-central Oregon and s.-central
Washington.

34

LITERATURE CITED

Abraham, K. F., J. O. Leafloor, and D. H. Rusch. 1999. Molt migrant Canada geese in
northern Ontario and western James Bay. Journal of Wildlife Management 63:649-655.

Alisauskas, R. T. 1998a. Nutritional ecology and population biology of Ross’ Geese.
Unpubl. Progress report and proposal for continued research. Canadian Wildlife Service
Saskatoon, Saskatchewan, Canada. 27pp.

Alisauskas, R. T. 1998b. Winter range expansion and relationships between landscape
and morphometrics of mid-continent lesser snow geese. Auk 115:851-862.

American Omithologists’ Union. 1998. Check-list of North American birds. 7th edition.
American Ornithologists’ Union, Washington, DC.

Anderson, M. G., J. M. Rhymer, and F. C. Rohwer. 1992. Philopatry, disperal, and the
genetic structure of waterfowl populations. Pages 365-395 in B. Batt, A. D. Afton, M. G.

Anderson, C. D. Ankey, D. H. Johnson, J. A. Kadlec, and G. L. Krapu, eds., Ecology and
management of breeding waterfowl. Univ. of Minnesota Press, Minnesota, MN. 63 5pp.

Ankney, C. D. 1996. Invited paper: an embarrassment of riches: too many geese.
Journal of Wildlife Management 60: 217-223.

Arbogast, B. S., S. V. Edwards, J. Wakeley, P. Beerli, and J. B. Slowinski. 2002.
Estimating divergence times from molecular data on phylogenetic and population genetic
timescales. Annu. Rev. Ecol. Syst. 33:707-740.

Armstrong, W. T., K. M. Meeres, R. H. Kerbes, W. S. Boyd, J. G. Silveira, J. P. Taylor,
and B. Turner. 1999. Routes and timing of migration of lesser snow geese from the
Western Canadian Arctic and Wrangel Island, Russia, 1987-1992. Pages 75-88 in
Kerbes, R. H., K. M. Meeres, and J. E. Hines, eds. Distribution, survival, and numbers of
lesser snow geese of the Western Canadian Arctic and Wrangel Island, Russia.
Occasional paper 98, Canadian Wildlife Service, Ottawa, Ontario, Canada. 120pp.

Avise, J. C. 1992. Molecular population structure and the biogeographic history of a
regional fauna: a case history with lessons for conservation biology. Oikos 63: 62-76.

Avise, J. C. 1998. The history and purview of phylogeography: a personal reflection.
Molecular Ecology 7: 371-379.

Avise, J. C., C. D. Ankney, and W. S. Nelson. 1990. Mitochondrial gene trees and the
evolutionary relationship of mallard and black ducks. Evolution 44: 1109-1119.

Avise, J. C., J. Arnold, R. M. Ball, E. Bermingham, T. Lamb, J. E. Neigel, C. A. Reeb, N.
C. Saunders. 1987. Intraspecific phylogeography: The mitochondrial DNA bridge
between population genetics and systematics. Ann. Rev. Ecol. Syst. 18: 498-522.

35

Avise, J. C., R. M. Ball, and J. Arnold. 1988. Current versus historical population sizes
in vertebrate species with high gene flow: a comparison based on mitochondrial DNA
lineages and inbreeding theory for neutral mutations. Mol. Biol. Evol. 5:331-344.

Avise, J. C., R. T. Alisauskas, W. S. Nelson, and C. D. Ankney. 1992. Matriarchal
population genetic structure in an avian species with female natal philopatry. Evolution
46(4): 1084-1096.

Avise, J. C. and D. Walker. 1998. Pleistocene phylogeographic effects on avian
populations and the speciation process. Proc. R. Soc. Lond. B 265: 457-463.

Baker, A. J. 1998. Identification of Canada goose stocks using restriction analysis of
mitochondrial DNA. Pages 435-443 in D. H. Rusch, M. D. Samuel, D. D. Humburg, and
B. D. Sullivan, eds., Biology and management of Canada geese. Proc. Int. Canada Goose
Symp., Milwaukee, WI.

Ball, R. M., Jr., F. C. James, S. Freeman, E. Bermingham, and J. C. Avise. 1988.
Phylogeographic population structure of red-winged blackbirds assessed by
mitochondrial DNA. Proceedings of the National Academy of Sciences 85:1558-1562.

Banks, R. C., C. Cicero, J. L. Dunn, A. W. Kratter, P. C. Rasmussen, J. V. Remsen, Jr., J.
D. Rising, and D. F. Stotz. 2003. Forty-fifth supplement to the American ornithologists’
union check-list of North American birds. The Auk 121 :985-995.

Baranyuk, V. V., J. E. Hines, and E. V. Syroechkovsky. 1999. Mineral staining of facial
plumage as an indicator of the wintering ground affinities of Wrangel Island lesser snow
geese. Pages 111-114 in Kerbes, R. H., K. M. Meeres, and J. E. Hines, eds. Distribution,
survival, and numbers of lesser snow geese of the Western Canadian Arctic and Wrangel
Island, Russia. Occasional paper 98, Canadian Wildlife Service, Ottawa, Ontario,
Canada. 120pp.

Bateman, H. A., T. Joanen, and C. D. Stutzenbaker. 1988. History and status of
Midcontinent snow geese on their Gulf Coast winter range. Pages 495-515 in M. W.
Weller, ed., Waterfowl in winter. Univ. of Minnesota Press, MN. 624pp.

Batt, B. D. J ., A. D. Afton, M. G. Anderson, C. D. Ankney, D. H. Johnson, J. A. Kadlec,
and G. L. Krapu, eds. 1992. Ecology and Management of Breeding Waterfowl.
University of Minnesota Press, Minneapolis, MN. 63 5pp.

Bellrose, F. C. 1980. Ducks, Geese and Swans of North America. Stackpole Books,
Harrisburg, PA. 540pp.

Bematchez, L. and C. C. Wilson. 1998. Comparative phylogeography of Nearctic and
Palearctic fishes. Molecular Ecology 7: 431-452.

36

Bousfield, M. A. and E. V. Syroenchkovsky. 1985. A review of Soviet research on the
lesser snow goose on Wrangel Island, USSR. Wildfowl 36:13-20.

Boyd, W. S. 1995. Lesser snow geese (Anser c. caerulescens) and American three
square bulrush (Scripus americanus) on the Fraser and Skagit river deltas. Ph.D. Thesis,
Department of Biological Sciences, Simon Fraser University, North Burnaby, British
Columbia, Canada. 154pp.

Caccamise, D. F., L. M. Reed, P. M. Castelli, S. Wainright, and T. C. Nichos. 2000.
Distinguishing migratory and resident Canada geese using stable isotope analysis.
Journal of Wildlife Management 64: 1084-1091.

Campbell, B. H. and E. F. Becker. 1991. Neck collar retention in dusky Canada geese.
Journal of Field Ornithology 62:521-527.

Chesser, R. K. 1991a. Gene diversity and female philopatry. Genetics 127: 437-447.

Chesser, R. K. 1991b. Influence of gene flow and breeding tactics on gene diversity
within populations. Genetics 129:573-583.

Chesser, R. K. and R. J. Baker. 1996. Effective sizes and dynamics of uniparentally and
diparentally inherited genes. Genetics 144:1225-1235.

Chesser, R. K., O. E. Rhodes, Jr., D. W. Sugg, and A. Schnabel. 1993. Effective sizes
for subdivided populations. Genetics 135:1221-1232.

Cooke, F. 1987. Lesser snow goose: A long-term population study. Pages 407-432 in F.
Cooke and P. A. Buckley, eds. Avian genetics Academic Press, NY.

Cooke, F. 2001. Timing of first pairing in Wrangel Island snow geese. Abstract. North
American Arctic Goose Conference. Quebec City, Quebec, Canada.

Cooke, F ., C. D. MacInnes, and J. P. Prevett. 1975. Gene flow between breeding
populations of lesser snow geese. Auk 92:493-510.

Cooke, F., and D. S. Sulzbach. 1978. Mortality, emigration, and separation of mated
snow geese. Journal of Wildlife Management 42: 271-280.

Cooke, F ., K. F. Abraham, J. C. Davies, C. S. Findlay, R. F. Healey, A. Sadura, and R. J.

Seguin. 1982. The LaPerouse Bay snow goose project-a l3-year report. Kingston,
Ontario, Dept. Biology, Queens’ University.

Cooke, F., D. T. Parkin, and R. F. Rockwell. 1988. Evidence of former allopatry of the

two color phases of lesser snow geese (Chen caerulescens caerulescens) Auk 105:467-
479.

37

Cooke, F., R. F. Rockwell, and D. B. Lank. 1995. The snow geese of La Perouse Bay:
natural selection in the wild. Oxford University Press, Oxford, UK.

Craighead, J. J. and S. S. Stockstad. 1964. Breeding age of Canada geese. Journal of
Wildlife Management 28(1): 57-64.

Craven, S. R. and D. H. Rusch. 1983. Winter distribution and affinities of Canada geese
marked on Hudson and James Bays. Journal of Wildlife Management 47(2): 307-319.

Cronin, M. A., J. B. Grand, D. Esler, D. V. Derksen, and K. T. Scribner. 1996. Breeding
populations of northern pintails have similar mitochondrial DNA. Canadian Journal of
Zoology 74:992-999.

Dickson, K. M. 2000. The diversity of Canada Geese. Pages 11-24 in K. M. Dickson,
ed., Towards conservation of the diversity of Canada Geese (Branta canadensis).
Occasional paper 103, Canadian Wildlife Service, Ottawa, Ontario, Canada.

Didiuk, A. B. and F. D. Caswell. 1998. Migration and wintering distribution of Tall
Grass Prairie Canada geese (abstract). Page 262 in D. H. Rusch, M. D. Samuel, D. D.
Humburg, and B. D. Sullivan, eds., Biology and management of Canada geese. Proc. Int.

Canada Goose Symp., Milwaukee, WI.

Drake, K. L., and R. T. Alisauskas. 2005. Multistate modeling of movement and
philopatry of breeding Ross’s geese. Animal Biodiversity and Conservation: In Press.

Drut, M. S., and R. E. Trost. 2001. Annual summary of goose population monitoring in

the Pacific Flyway, 2000-2001. US. Fish and Wildlife Service, Division of Migratory
Bird Management, Portland, OR.

Ely, C. and K. T. Scribner. 1994. Genetic diversity in arctic-nesting geese: implications
for management and conservation. Trans. No. Amer. Wildl. Nat. Res. Conf. 59:91-110.

Enler, J. A. 1982. Problems in distinguishing historical form ecological factors in
biogeography. Amer. Zool. 22: 441-452.

Geramita, J. M., and F. Cooke. 1982. Evidence that fidelity to natal breeding colony is
not absolute in female snow geese. Canadian Journal of Zoology 60:2051-2056.

Gill, R. E., C. A. Babcock, C. M. Handel, W. 1. Butler, Jr., and D. G. Raveling. 1998.
Migration, site fidelity, and use of fall staging grounds in Alaska by cackling Canada
geese (abstract). Page 263 in D. H. Rusch, M. D. Samuel, D. D. Humburg, and B. D.
Sullivan, eds., Biology and management of Canada geese. Proc. Int. Canada Goose
Symp., Milwaukee, WI.

38

Giroux, J ., J. Lefebvre, L. Belanger, J. Rodrigue, and S. LaPointe. 2001. Establishment
of a breeding population of Canada geese in southern Quebec. The Canadian Field
Naturalist 115:75-81.

Greenwood, P. J. 1980. Mating systems, philopatry and dispersal in birds and mammals.
Animal Behavior 28: 1140-1162.

Greenwood, P. J. 1987. Philopatry and optimal outbreeding in birds. Pages 207-222 in
F. C. Cooke and P. A. Buckley, eds., Avian genetics Academic Press, Inc., NY.

Hanson, H. C. 1997. The Giant Canada Goose. Illinois Natural History Survey, IL.
252pp.

Hestbeck, J. B., J. D. Nichols, and R. A. Malecki. 1991. Estimates of movement and site
fidelity using mark-resight data of wintering Canada geese. Ecology 72: 523-533.

Hill, M. R. J. and R. B. Frederick. 1997. Winter movements and habitat use by greater
snow geese. Journal of Wildlife Management 61(4):]213-1221.

Hines, J. E. , V. V. Baranyuk, B. Turner, W. S. Boyd, J. G. Silveira, J. P. Taylor, S. J.
Barry, K. M. Meeres, R. H. Kerbes, and W. T. Armstrong. 1999. Autumn and winter
distributions of lesser snow geese from the Western Canadian Arctic and Wrangel Island,
Russia, 1953-1992. Pages 39-74 in Kerbes, R. H., K. M. Meeres, and J. E. Hines, eds.
Distribution, survival, and numbers of lesser snow geese of the Western Canadian Arctic
and Wrangel Island, Russia. Occasional paper 98, Canadian Wildlife Service, Ottawa,
Ontario, Canada. 120pp.

Johnson, S. R. 1996. Staging and wintering areas of snow geese nesting on Howe
Island, Alaska. Arctic: 49: 86-93.

Johnson, J. and D. G. Raveling. 1988. Weak family associations in cackling geese
during winter: Effects of body size and food resources on goose social organization.
Pages 71-89 in M. W. Weller, ed., Waterfowl in winter. Univ. of Minnesota Press, MN.
624pp.

Kelley, J. R., Jr., D. F. Cithamer, and K. A. Wilkins. 1998. Waterfowl population status,
1998. US. Fish and Wildlife Service Publication. 3 lpp.

Kerbes, R. H. 1994. Colonies and numbers of Ross’ geese and lesser snow geese in the
Queen Maud Gulf Migratory Bird Sanctuary. Occasional paper 81, Canadian Wildlife
Service, Ottawa, Ontario, Canada. 47pp.

Kerbes, R. H., K. M. Meeres, and J. E. Hines, eds. 1999. Distribution, survival, and
numbers of lesser snow geese of the Western Canadian Arctic and Wrangel Island,
Russia. Occasional paper 98, Canadian Wildlife Service, Ottawa, Ontario, Canada.
120pp.

39

Kimura, M., S. M. Clegg, I. J. Lovette, K. R. Holder, D. J. Girrnan, B. Mila, P. Wade,
and T. B. Smith. 2002. Phylogeographical approaches to assessing demographic
connectivity between breeding and overwintering regions in a Nearctic-Neotropical
wabler (Wilsonia pusilla). Molecular Ecology 11:1605-1616.

Klicka, J. and R. M. Zink. 1997. The importance of recent ice ages in speciation: A
failed paradigm. Science 277: 1666-1669.

Kuznetsov, S. B., V. V. Baranyuk, and J. Y. Takekawa. 1998. Genetic differentiation
between wintering populations of lesser snow geese nesting on Wrangel Island. Auk
115(4): 1053-1057.

Lanctot, R., B. Goatcher, K. Scribner, S. Talbot, B. Pierson, D. Esler, and D.
Zwiefelhoffer. 1999. Harlequin duck recovery from the Exxon Valdez oil spill: a
population genetics perspective. The Auk 116:781-791.

Lande, R. 1988. Genetics and demography in biological conservation. Science
241:1455-1460.

Larsson, K. and P. Forslund. 1991. Environmentally induced morphological variation in
the barnacle goose, Branta leucopsis. Journal of Evolutionary Biology 4:619-636.

Leafloor, J. O. and D. H. Rusch. 1997. Clinal size variation in Canada geese affects
morphometric discrimination techniques. Journal of Wildlife Management 61:183-190.

Leafloor, J. O., C. D. Ankney, and D. H. Rusch. 1998. Environmental effects on body
size of Canada geese. Auk 115:26-33.

Lessells, C. M. 1985. Natal and breeding dispersal of Canada geese, Branta canadensis.
Ibis 127231-41.

Luukkonen, D. R., H. H. Prince, R. C. Mykut. 2004. Harvest of molt migrant giant
Canada geese from southern Michigan. Page 178 in T. J. Moser, R. D. Lien, K. C.
VerCauteren, K. F. Abraham, D. E. Andersen, J. G. Bruggink, J. M. Coluccy, D. A.
Graber, J. O. Leafloor, D. R. Luukkonen, and R. E. Trost editors. Proceedings of the
2003 International Canada Goose Symposium, Madison, Wisconsin, USA.

Luukkonen, D. R. and G. J. Soulliere. 2004. Change in harvest composition resulting
from mid-September opening of Michigan’s regular goose season, 1998-2002. Michigan
Department of Natural Resources - Wildlife Division, Lansing, Michigan, USA.

MacInnes, C. D. 1966. Population behavior of eastern Arctic Canada geese. Journal of
Wildlife Management 30:536-553.

40

MacInnes, C. D. and B. C. Lieff. 1968. Individual behavior and composition of a local
population of Canada geese. Pages 93-101 in R. L. Hine and C. Schoenfeld, eds. Canada
Goose Management, Dembar Educational Research Services, Madison, Wisconsin.

Malecki, R. A. and R. E. Trost. 1998. Status of breeding Canada geese in North
America. Pages 3-8 in D. H. Rusch, M. D. Samuel, D. D. Humburg, and B. D. Sullivan,

eds., Biology and management of Canada geese. Proc. Int. Canada Goose Symp.,
Milwaukee, WI.

Miyatake, T. and T. Shimizu. 1999. Genetic correlations between life-history and
behavioral traits can cause reproductive isolation. Evolution 53:201-208.

Moser, T. J. and F. D. Caswell. 2004. Long-terrn indices of Canada goose status and
management. Pages 123-120 in T. J. Moser, R. D. Lien, K. C. VerCauteren, K. F.
Abraham, D. E. Andersen, J. G. Bruggink, J. M. Coluccy, D. A. Graber, J. O. Leafloor,
D. R. Luukkonen, and R. E. Trost editors. Proceedings of the 2003 International Canada
Goose Symposium,Madison, Wisconsin, USA.

Moser, T. J. and R. E. Rolley. 1990. Morphometric discrimination of giant and interior
Canada geese of the Mississippi F lyway. Wildlife Society Bulletin 18:381-388.

Mowbray, T. B., F. Cooke, and B. Ganter. 2000. Snow Goose (Chen caerulescens). In
The Birds of North America, No. 514, A. Poole and F. Gill, eds. The Academy of
Natural Sciences, Philadelphia, and The American Ornithologists’ Union, Washington,
DC.

Mowbray, T. B., C. R. Ely, J. S. Sedinger, and R. T. Trost. 2002. Canada Goose (Branta
canadensis). In The Birds of North America, No. 682, A. Poole and F. Gill, eds. The
Academy of Natural Sciences, Philadelphia, and The American Ornithologists’ Union,
Washington, DC.

Mundy, N. I., N. S Badcock, T. Hart, K. Scribner, K. Janssen, and N. J. Nadeau. 2004.
Conserved genetic basis of a quantitative plumage trait involved in mate choice. Science
303: 1870-1873.

Nielsen, R. 1998. Maximum likelihood estimation of population divergence times and
population phylogenies under the infinite sites model. T‘heroretical Population Biology
53: 143-151.

North American Arctic Goose Conference. 2001. Workshop: The management of over-
abundant white goose populations: where do we stand and where do we go. Quebec City,
Quebec, Canada.

Owen, M. 1980. Wild Geese of the World. B. T. Batsford Ltd., London. 236pp.

41

Pacific Flyway Council. 1997. Pacific F lyway management plan for the dusky Canada
goose. Dusky Canada Goose Subcommitte, Pacific Flyway Study Committee, Portland,
OR, USA.

Palmer, R. 8., Ed. 1976. Handbook of North American Birds, Vol. 2. Yale University
Press, New Haven, CT, USA.

Paxinos, E. E., H. F. James, S. L. Olson, M. D. Sorenson, J. Jackson, R. C. Fleischer.
2002. MtDNA from fossils reveals a radiation of Hawaiian geese recently derived from
the Canada goose (Branta canadensis). Proceedings of the National Academy of

Sciences 99: 1399-1404.

Pearce, J. M., B. J. Pierson, S. L. Talbot, D. V. Derksen, D. Kraege, and K. T. Scribner.
2000. A genetic evaluation of morphology used to identify harvested Canada geese.
Journal of Wildlife Management 64(3): 863-874.

Pearce, J. M. and K. S. Bollinger. 2003. Morphological traits of Pacific Flyway Canada
geese as an aid to subspecies identification and management. Journal of Field
Ornithology 74:357-369.

Pearce, J. M., S. L. Talbot, B. J. Pierson, M. R. Petersen, K. T. Scribner, D. L. Dickson,
and A. Mosbech. 2004. Lack of spatial genetic structure among nesting and wintering
king eiders. The Condor 106:229-240.

Pella, J. J. and G. B. Milner. 1987. Use of genetic marks in stock composition analysis.
Pages 247-276 in N. Ryman and F. M. Utter, eds. Population genetics and fishery
management. Univ. Wash. Press, Seattle, WA, USA.

Peters, J. L., K. G. McCraken, Y. N. Zhuravlev, Y. Lu, R. E. Wilson, K. P. Johnson, and
D. E. Omland. 2005. Phylogenetics of wigeons and allies (Anatidae: Anas): the
importance of sampling multiple loci and multiple individuals. Molecular Phylogenetics

and Evolution 35:209-224.

Petrie, M. and B. Kempenaers. 1998. Extra-pair paternity in birds: explaining variation
between species and populations. TREE 13(2): 52-58.

Pierson, B. J., J. M. Pearce, and S. L. Talbot. 2000. Molecular genetic status of Aleutian
Canada geese from Buldir and the Semidi Islands, Alaska. Condor 102: 172-180.

Prevett, J. P. 1972. Family behavior and age-dependent breeding biology of the blue
goose, Chen caerulescens. Ph.D. dissert., Univ. of Western Ontario, London, ON.
192pp.

Quinn, T. W., G. F. Shields, and A. C. Wilson. 1991. Affinities of the Hawaiian goose
based on two types of mitochondrial DNA data. Auk 108:585-593.

42

Quinn, T. W. 1992. The genetic legacy of Mother Goose-phylogeographic patterns of
lesser snow goose Chen caerulescens caerulescens maternal lineages. Molecular
Evolution 1:105-117.

Randi, E. 2000. Mitochondrial DNA. Pages 136-167 in Baker, A. J., eds., Molecular
Methods in Ecology. Blackwell Science Ltd., Oxford, England. 337pp.

Raveling, D. G. 1978. Dynamics of distribution of Canada geese in winter.
Transactions of the North American Wildlife and Natural Resources Conference 43: 206-
225.

Raveling, D. G. 1979. Traditional use of migration and winter roost sites by Canada
geese. Journal of Wildlife Management 53: 1088-1103.

Raveling, D. G. 1981. Survival, experience, and age in relation to breeding success of
Canada geese. Journal of Wildlife Management 45(4): 817-829.

Raveling, D. G. 1988. Mate retention in giant Canada geese. Can. J. Zool. 66: 2766-
2768.

Reed, A. and A. P. Chagnon. 1987. Greater snow geese on Bylot Island, Northwest
Territories, 1983. Journal of Wildlife Management 51:128-131.

Reed, A., J. F., Giroux, and G. Gauthier. 1998. Population size, productivity, harvest
and distribution. Pages 5-31 in B. D. J. Batt, editor. The greater snow goose: report of
the Arctic goose habitat working group. Arctic Goose Joint Venture Special Publication.
US. Fish and Wildlife Service, Washington, DC, USA. and Canadian Wildlife Service,
Ottawa, Ontario, Canada.

Robertson, G. J. and F. Cooke. 1999. Winter Philopatry in Migratory Waterfowl. Auk
116(1): 20-34.

Rockwell, R. F. and G. F. Barrowclough. 1987. Gene flow and the genetic structure of
populations. Pages 223-255 in F. Cooke and P. A. Buckley, editors. Avian genetics: a
population and ecological approach. Academic Press, London.

Ruokonen, M., L. Kvist, and J. Lumme. 2000. Close relatedness between mitochondrial
DNA from seven Anser goose species. Journal of Evolutionary Biology 13:532-540.

Rusch, D. H., F. D. Caswell, M. M. Gillespie, and J. O. Leafloor. 1996a. Research
contributions to management of Canada geese in the Mississippi Flyway. Transactions of
the North American Wildlife and Natural Resources Conference 61:437-449.

Rusch, D. H., M. M. Gillespie, H. G. Lumsden, K. F. Abraham, and A. D. Didiuk. 1998.

Distribution and derivation of the Canada goose harvest in the Mississippi Flyway
(abstract). Page 159 in D. H. Rusch, M. D. Samuel, D. D. Humberg, and B. D. Sullivan,

43

eds. Biology and management of Canada geese. Proceedings of the International Canada
Goose Symposium. Milwaukee, Wisconsin.

Rusch, D. H., J. C. Wood, and G. G. Zenner. 1996b. The dilemma of giant Canada
goose management. Pages 72-78 in J. T. Ratti, eds. Proceedings of the Seventh
International Waterfowl Symposium, Memphis, Tennesee, USA.

Ryder, J. P. and R. T. Alisauskas. 1995. Ross’ Goose (Chen rossii). In The Birds of
North America, No. 162, A. Poole and F. Gill, eds. The Academy of Natural Sciences,
Philadelphia, and The American Omithologists’ Union, Washington, DC.

Rhymer, J. M., M. G. Fain, J. E. Austin, D. H. Johnson, and C. Krajewski. 2001.
Mitochondrial phylogeography, subspecific taxonomy, and conservation genetics of
sandhill crans (Grus canadensis; Aves: Gruidae). Conservation Genetics 2:203-218.

Samuel, M. D., N.T. Weiss, D. H. Rusch, S. R. Craven, R. E. Trost, and F. D. Caswell.
1990. Neck-band retention for Canada geese in the Mississippi flyway. Journal of
Wildlife Management 54:612-621 .

Samuel, M. D., D. R. Goldberg, A. E. Smith, V. V. Baranyuk, and E. G. Cooch. 2001.
Neckband retention for lesser snow geese in the western Arctic. Journal of Wildlife
Management 65:797-807.

Scribner, K. T. and J. M. Pearce. 2000. Microsatellites: Evolutionary and
methodological background and empirical applications at individual, population,
phylogenetic levels in Baker, A. J., eds. Molecular Methods in Ecology. Blackwell
Science Ltd., Oxford, England. 337pp.

Scribner, K. T., M. R. Petersen, R. L. Fields, S. L. Talbot, J. M. Pearce, and R. K.
Chesser. 2001. Sex-biased gene flow in spectacled eiders (Anatidae): inferences from
molecular markers with contrasting modes of inheritance. Evolution 55:2105-2115.

Scribner, K. T., S. L. Talbot, J. M. Pearce, B. J. Pierson, K. S. Bollinger, and D. V.
Derksen. 2003a. Phylogeography of Canada geese (Branta canadensis) in western North
America. The Auk 120:889-907.

Scribner, K. T., J. A. Warrillow, J. O. Leafloor, H. H. Prince, R. L. Inman, D. R.
Luukkonen, and C. S. Flegel. 2003b. A genetic method for determining racial
composition of Canada goose harvests. Journal of Wildlife Management 67: 122-135.

Scribner, K. T., R. A. Malecki, B. D. J. Batt, R. L. Inman, S. Libants, and H. H. Prince.
2003c. Identification of source population for Greenland Canada geese: genetic
assessment of a recent colonization. Condor 105:771-782.

Shields, F. F. 1990. Analysis of mitochondrial DNA of Pacific black brant (Branta
bernicla nigricans). Auk 107:620-623.

44

Shields, G. F. and J. P. Cotter. 1998. Phylogenies of North American geese: The
mitochondrial DNA record. Pages 405-411 in D. H. Rusch, M. D. Samuel, D. D.
Humburg, and B. D. Sullivan, eds. Biology and management of Canada geese. Proc. Int.

Canada Goose Symp., Milwaukee, WI.

Shields, G. F. and A. C. Wilson. 1987. Subspecies of the Canada goose (Branta
canadensis have distinct mitochondrial DNA’s. Evolution 41(3): 662-666.

Sjoberg, K. and G. Sjoberg. 1998. Natal and breeding dispersal in a migratory
population of Canada geese in Sweden. Pages 361-366 in D. H. Rusch, M. D. Samuel, D.
D. Humburg, and B. D. Sullivan, eds. Biology and management of Canada geese. Proc.
Int. Canada Goose Symp., Milwaukee, WI.

Slatkin, M. 1987. Gene flow and the geographic structure of natural populations.
Science 236:787-792.

Smouse, P. E., R. S. Waples, and J. A. Tworek. 1990. A genetic mixture analysis for use
with incomplete source population data. Canadian Journal of Fisheries and Aquatic
Sciences 47: 620-634.

Sonsthagen, S. A., S. L. Talbot, R. B. Lanctot, K. T. Scribner, and K. G. McCracken.
2005. Population genetic structure of common eiders (Somateria mollissima) breeding in
the Beaufort Sea, Alaska. In review.

Soulliere, G. J., G. F. Martz, and W. C. Johnson. 1988. An evaluation of Michigan’s
special goose seasons and characteristics of the harvest, 1982-87. Michigan Department
of Natural Resources-Wildlife Division, Lansing. Report Number 3078.

Surrendi, D. C. 1970. The mortality, behavior, and homing of transplanted juvenile
Canada geese. Journal of Wildlife Management 34: 719-733.

Templeton, A. R. 1998. Nested clade analyses of phylogeographic data: testing
hypotheses about gene flow and population history. Molecular Ecology 7: 381-397.

Thompson, J. E., M. R. Hill, J. Merindino, M. Todd, and C. D. Ankney. 1999.
Improving use of morphometric discrimination to identify Canada goose subspecies.
Wildlife Society Bulletin 27: 274-280.

Thorpe, R. S., A. Malhotra, H. Black, J. C. Daltry, and W. Wuster. 1995. Relating
geographic pattern to phylogenetic process. Phil. Trans. R. Soc. Lond. B 61-68.

Turner, B., R. Tomlinson, R. Leyva, and P. Dominguez. 1994. Wintering populations of

lesser snow geese and Ross’ geese in the northern highlands of Mexico, 1988-1990.
Occasional paper 84, Canadian Wildlife Service, Ottawa, Ontario, Canada. 47pp.

45

United States Fish and Wildlife Service. 2001. Draft environmental impact statement:
light goose management. U. S. Fish and Wildlife Service, Department of the Interior,
Washington, D. C., USA.

Van Wagner, C. E. and A. J. Baker. 1990. Association between mitochondrial DNA and
morphological evolution in Canada Geese. Journal of Molecular Evolution 31:373-382.

Voelker, G. 1999. Dispersal, vicariance, and clocks: historical biogeography and
speciation in a cosmopolitan passerine genus (Anthus: Motacillidae). Evolution 53(5):
1 536-1 552.

Wenink, P. W., A. J. Baker, and M. G. J. Tilanus. 1994. Mitochondrial control-region
sequences in two shorebird species, the tumstone and the dunlin, and their utility in
population genetic studies. Mol. Biol. Evol. 11:22-31.

Wiebe, M. 0., J. E. Hines, and G. J. Robertson. 2000. Collar retention of Canada geese
and greater white-fronted geese from the western Canadian Arctic. Journal of Field
Ornithology 71 :531-540.

Wiley, E. O. 1980. Phylogenetic systematics and vicariance biogeography. Systematic
Botany 5(2): 194-220.

Williams, C. K., M. D. Samuel, V. V. Baranyuk, E. G. Cooch, D. Kraege. 2005. Winter
fidelity and apparent survival of lesser snow goose populations in the Pacific F lyway.
Unpublished United States Geological Survey manuscript.

Xu, 8., C. Kobak, and P. Smouse. 1994. Constrained least squares estimation of mixed
population stock composition from mtDNA haplotype frequency data. Canadian Journal
of Fisheries and Aquatic Sciences 51: 396-404.

Zink, R. M. 1996. Comparative phylogeography in North American birds. Evolution
50:308-317.

Zink, R. M. 1997. Phylogenetic studies of North American birds. Pages 301-325 in D.

P. Mindell, Editor. Avian Molecular Evolution and Systematics. Academic Press, New
York, New York.

46

CHAPTER 2: CACKLING GEESE AND CANADA GEESE PHYLOGEOGRAPHY

INTRODUCTION

Spatial patterns of genetic diversity within and among species can be influenced
by historical and contemporary processes (Avise et al. 1987, Avise 1998). Past vicariant
events, such as restrictions of populations within isolated refugia during Pleistocene or
Pliocene glacial periods and subsequent range expansion during interglacial periods, have
likely contributed greatly to the phylogeographic structure of species and to cladogenic
events (Avise and Walker 1998, Zink 1996, Klicka and Zink 1997, Voelker 1999). Other
historical factors that may influence the genetic structure of species include size of
populations during glacial and interglacial periods, length of time populations
experienced geographic and reproductive isolation, and length of time since population
separation (Avise et al. 1998, Emerson et al. 2001, Stewart and Lister 2001). Longer
separation times and smaller effective population sizes in isolated populations increase
the probability populations will accrue differences in gene frequencies and develop novel
gene mutations (Avise et al. 1988). Historical population expansions occurring in a
single, punctuated event, and with a low number of founding individuals, could
appreciably change allele and haplotype frequencies in the founding population as
compared to the population of origin. Alternatively, if expansion events included a large
number of individuals over a prolonged time period, then gene frequencies may be
similar among populations GVIiIa 2000, Hewitt 2001).

Ecology and life history characteristics that may affect species genetic structure

include timing and location of pair bond formation, degree and sex-specificity of

47

philopatry, level of gene flow among contemporary breeding and/or wintering
populations, and current sizes of populations (Greenwood 1980, Chesser 1991a, Chesser
1991b, Ely and Scribner 1994, Miyatake and Shimizu 1999). High natal and breeding
site philopatry, and pair bond formation during periods when breeding populations are
isolated, will tend to restrict gene flow and increase genetic structure among populations
(Rockwell and Barrowclough 1987). Fidelity to wintering sites may increase among
population genetic variation if populations are isolated in wintering areas, or homogenize
gene frequencies if populations are admixed in wintering areas and pair bond formation
occurs on these wintering sites (Robertson and Cooke 1999). Gene flow that is sex-
biased may lead to differences in the degree of among population genetic variation
revealed by maternally and bi-parentally inherited markers (Chesser 1991a, Scribner et
al. 2001). Information on the degree of genetic discordance among mitochondrial (mt)
DNA haplotypes, and the frequencies of shared and novel haplotypes and nuclear DNA
alleles provides sources of inference to past events and species ecology. Large
fundamental divisions in the topology of an intraspecific phylogeny reflect historical
evolutionary splits within a taxa, while shallower molecular topologies are evidence of
more recent population subdivisions (Avise 1992). The relative importance of ecological
and historical factors in shaping spatial genetic structure is likely to be taxon specific.
Previous studies have shown that North American species exhibit a variety of
phylogeographic patterns ranging from near panmictic [e.g., harlequin ducks
(Histrionicus histrionicus) Lanctot et al. 1999; king eiders (Somateria spectabilis) Pearce
et al. 2004] to species with strongly subdivided subspecies [Wilson’s warblers (Wilsonia

pusilla), Kimura et al. 2002) or population genetic structure within species [dunlins

48

(Caldris alpine), Wenink et al. 1994]. Phylogenetic investigations have been conducted
for a wide variety of North American avian species (see review in Avise and Walker
1998), including songbirds and Neotropical migrants (Zink 1996, Kimura et al. 2002),
shorebirds (Wenink et a1. 1994), wading birds (Rhymer et al. 2001), and migratory
waterfowl (Avise et a1. 1992, Quinn 1992, Scribner et al. 2001, Scribner et al. 2003a,
Pearce et al. 2004, Peters et al. 2005). Patterns of spatial genetic structure vary greatly
among co-distributed avian species (Zink 1996, Avise and Walker 1998). Some species,
like the song sparrow (Melospiza melodia), red-winged blackbird (Agelaius phoeniceus;
Zink 1996), tumstone (Arenaria interpres; Wenink et al. 1994), and king cider
(Somateria spectabilis; Pearce et a1. 2004) exhibit little genetic differentiation across
there geographic ranges, likely as a result of recent population expansions from a single
glacial refugia and high levels of among population gene flow subsequent to the last
Pleistocene glacial period. Significant intraspecies genetic variation consistent with
population divergence in multiple glacial refugia has been found in fox sparrows
(Passerella iliaca; Zink 1996), Wilson’s warblers (Wilsonia pusilla; Kimura et al. 2002),
rock ptarrnigans (Lagopus mutus; Holder et al. 1999), dunlins (Caldris alpine; Wenink et
al. 1994), and Canada geese (Branta canadensis; Scribner et al. 2003a). In some species
genetic population structure is attributed strong natal and breeding philopatry (e. g.,
dunlin, Wenink et al. 1994) or to high fidelity to wintering sites [e.g., snow geese (Chen
caerulescens) and Ross’s geese (Chen rossii), see Chapter 6]. Thus, differences in the
degree of genetic divergence evident in co-distributed North American avian species can
result from both historical processes and species life history characteristics which

influence contemporary levels of gene flow.

49

Two species of Arctic-nesting geese, Cackling geese (Branta hutchinsii) and
Canada geese (Branta canadensis) (considered one species, B. canadensis, with 11
recognized subspecies until 2003; Banks et al. 2003), have highly variable phenotypes,
ecology, and behavioral characteristics (Delacour 1954, Palmer 1976, Bellrose 1980,
Owen 1980). Subspecies of cackling and Canada geese were originally defined on the
basis of morphology and plumage characteristics (Delacour 1954, American
Ornithologists’ Union 1957). As these characters vary geographically, as few as 8
(Palmer 1976) and as many as 83 (Hanson 1997) subspecies have been recognized for
cackling and Canada geese. Current subspecies include B. h. hutchinsii, B. h. minima, B.
h. taverneri, and B. h. leucoparia for cackling geese; and B. c. maxima, B. c. moffitti, B.
c. interior, B. c. canadensis, B. c. parvipes, B. c. occidentalis, and B. c. fulva for Canada
geese (Banks et al. 2003). Breeding populations of cackling geese are primarily coastal,
and are located on the Aleutian and Semidi Islands off the Alaskan coast, northern and
western Alaska, coastal areas of northern Canada, and Southampton and Baffin Islands
(Figure 1). Cackling geese winter from British Columbia south to California and east to
north Mexico and western Louisiana (Mowbray et al. 2002). Canada geese breeding
populations are found throughout Canada and the northern United States, and winter from
the southern part of the breeding range through most of the United States and into
Mexico (Figure 1; Mowbray et al. 2002).

Cackling geese and Canada geese are currently managed in North America by
species, subspecies, by affiliation to one of the four migratory Flyways (Figure 1), and by
management populations for cackling and Canada geese combined; Dickson 2000, Moser

and Caswell 2004). Population affiliation to migratory F lyways and delineation of

50

   
     

    
    

breeding-cackling geese
ézié breeding-Canada geese

D breeding and wintering-Canada geese
I wintering-cackling and Canada geese

Locations of proposed g :3“: _
glacial refugia : I , In?) if I /
Pacific Central :MississippiEAtlantic ‘4 }

Flyway Flyway Flyway Flyway

Figure 1. Breeding and wintering distributions and migratory Flyways for cackling geese
(Branta hutchinsii) and Canada geese (Branta canadensis). Locations of Pleistocene
glacial refugia proposed by Ploeger (1968) are indicated.

51

management populations has been determined by numerous migratory and breeding
surveys, banding studies, and harvest data over the past 50 years (Moser and Caswell
2004). Early genetic studies based on mitochondrial DNA (mtDNA) supported current
species and subspecies classifications (Shields and Wilson 1987, VanWagner and Baker
1990, Baker 1998). Cackling geese and Canada geese were found to differ by an average
of 2.0% sequence divergence, and subspecies were characterized by distinct mtDNA
haplotypes. A more recent study (Scribner et al. 2003a) estimated the sequence
divergence between species at 14.4%, consistent with the high degree of interspecies
genetic differentiation documented in previous studies. In contrast to earlier research,
Scribner et al. (2003) found shared mtDNA haplotypes in high frequencies among
subspecies of cackling geese and among subspecies of Canada geese. Differences in
study results for subspecies may have been due to much larger sample sizes analyzed per
subspecies in the work of Scribner et al. (2003a). Additionally, previous studies relied on
characterization of mtDNA through restriction fragment analysis, which resolved
sequence variation over a larger portion of the mtDNA molecule than in the Scribner et
al. (2003) study, where a portion of the mtDNA genome was directly sequenced.

For Arctic-nesting waterfowl, current breeding areas and migratory pathways
between breeding and wintering areas have evolved over relatively recent timescales
(Ploeger 1968). High levels of genetic divergence documented previously between
cackling geese and Canada geese is indicative of a fundamental geographic split that may
have occurred during Pleistocene glacial events (Ploeger 1968). Based on species
ecology, nesting habits, and availability of unglaciated breeding habitat available during

the Pleistocene, Ploeger (1968) identified four major breeding areas for cackling and

52

Canada geese. He hypothesized that Canada geese inhabited refugia south of the major
ice sheets, while cackling geese were found north of the glaciers in the Aleutian Islands
and Canadian Arctic Archipelago. Canada geese were split between the “Western
Southern Group” nested in the Pacific coastal region south of the Cordilleran ice sheet
and included current subspecies B. c. firlva and B. c. occidentalis; and an “Eastern
Southern Group” composed of current subspecies B. c. canadensis, B. c. interior, B. c.
maxima, B. c. moflitti, and B. c. parvipes nested south of the Cordilleran and Laurentide
ice sheets east of the Cascade and Sierra Nevada Mountains. Cackling geese were also
said to inhabit two refugia including the “Aleutian Group” where B. h. Ieucopareia
nested; and the “Northern Group” of the Canadian Archipelago which included the three
remaining extant subspecies, B. h. minima, B. h. taverneri, and B. h. hutchinsii. Evidence
of the biogeographic isolation of cackling and Canada geese in the four proposed glacial
refugia would be reflected in a deep phylogenetic split between species and between
subspecies groups associated with separate refugia.

In addition to historical separations based on glacial refirgia, long-term family
associations and strong male and female breeding site fidelity exhibited by cackling and
Canada geese (Anderson et al. 1992, Ely and Scribner 1994, Mowbray et al. 2002) may
reproductively isolate breeding populations and influence current partitioning of genetic
variation. Initial pair bonds in cackling and Canada geese are formed in the 'spring during
northward migrations or after the geese have returned to breeding sites (MacInnes 1966,
Hanson 1997), and pairs exhibit long-term monogamy (Bellrose 1980). Significant

spatial genetic structure among breeding populations may result from breeding site

53

fidelity and formation of pair bonds on breeding grounds, even though populations share
migratory pathways and wintering areas (Ely and Scribner 1994, Scribner et al. 2003a).

Cackling and Canada geese show strong affiliation to particular migration routes,
resulting in highly localized populations (Craven and Rusch 1983, Tacha et al. 1991,
Didiuk and Caswell 1998, Gill et al. 1998). Their geographic migration patterns are so
consistent that wildlife managers and biologists often refer to the recognized population
groups that utilize each of the four migratory Flyways in North America (Figure l;
Malecki and Trost 1998). Populations that share migratory routes and common wintering
areas within Flyways may experience a higher degree of among population gene flow
than populations from different F lyways. Gene flow occurring among populations within
the same F lyway via exchange of individuals during migration and wintering periods may
increase the genetic similarity of breeding populations within a Flyway, and increase the
spatial genetic structure among Flyway groups.

Both cackling and Canada geese are well-studied species (Mowbray et al. 2002)
that are highly managed to maintain viable breeding populations for sustained harvests.
The wealth of survey, banding, and harvest data have helped to develop the current
administrative F lyway and management population definitions (Figure 1, Moser and
Caswell 2004), and offer information on the life history characteristics of these species on
which hypotheses of contemporary levels of gene flow may be based. We evaluate the
phylogenetic structure within and among cackling geese and Canada geese due to historic
geographic isolations in proposed glacial refugia and current levels of gene flow at
population and F lyway scales by utilizing molecular markers which differ in their mode

of inheritance and evolutionary rates. We build upon previous studies by sampling

54

breeding populations of cackling and Canada geese across their entire continental
distribution. We genetically characterizing a large number of individuals from each
location using bi-parentally inherited microsatellite loci and mtDNA sequence data from
the maternally inherited mitochondrial control region. As all eleven subspecies and all
four Flyway groups of cackling and Canada geese are represented in our analyses, we can
estimate the degree of spatial genetic structure between species and among subspecies of
cackling and Canada geese based on proposed glacial refugia, among F lyway groups of
both species, and among breeding populations of throughout the continental range of
cackling and Canada geese. Genetics data is used to characterize and interpret the
phylogenetic structure of species, subspecies, and populations of geese utilizing each of
the four migratory Flyways, and to predict the relative importance of historical and

contemporary factors influencing phylogenetic structure.

METHODS

Sample Collection

Samples were collected from 6 breeding populations of cackling geese (Branta
hutchinsii) and 16 breeding populations of Canada geese (B. canadensis) (Figure 2),
representing each of the four subspecies of cackling geese and seven subspecies of
Canada geese recognized in North America (Bellrose 1980, Banks et al. 2003). Geese
were guided into catch nets using a helicopter or by walking during brood rearing, when
geese were flightless (Cooch 1953, Tim and Bromley 1976). Blood or a blood quill
(growing feather) was sampled from breeding adults or goslings at each location.

Samples were placed into individual tubes containing high-salt buffer and stored at

55

 

Figure 2. Sampling locations for subspecies of cackling geese (letters) and Canada geese
(numbers) populations: A = Semidi Island (Branta hutchinsii leucoparia), B = Buldir
Island (B. h. leucoparia), C = Yukon—Kuskokwim Delta (B. h. minima), D = North Slope
(B. h. taverneri), E = Queen Maud Gulf Bird Sanctuary (B. h. hutchinsii, Short Grass
Prairie Population), F = Baffm Island (B. h. hutchinsii, Tall Grass Prairie Population); 1 =
Fairbanks (B. canadensis parvipes), 2 = Anchorage (B. c. parvipes), 3 = Copper River
Delta (B. c. occidentalis), 4 = Admiralty Island (B. c. fulva), 5 = Washington (B. c.
moflitti), 6 = Nebraska Sandhills (B. c. moflitti), 7 = Churchill (B. c. interior, Eastern
Prairie Population), 8 = Hudson and James Bays (B. c. interior, Mississippi Valley
Population), 9 = Akimiski Island and southern James Bay (B. c. interior, Southern James
Bay Population), 10 = West Ungava Peninsula (B. c. interior), 11 = East Ungava
Peninsula (B. c. interior), 12 = Greenland (B. c. interior), 13 = Newfoundland (B. c.
canadensis, North Atlantic Population), 14 = Labrador (B. c. canadensis, North Atlantic
Population), 15 = Ontario (B. c. maxima), 16 = southeast Michigan (B. c. maxima).

56

ambient temperatures in the field until frozen in the laboratory. Cackling geese were
collected from Semidi Islands and Buldir Islands representing B. h. Ieucopareia, the
Kashunuk River on the Yukon-Kuskokwim Delta representing B. h. minima, Prudhoe
Bay of the North Slope of Alaska representing B. h. taverneri, and Queen Maud Gulf
Bird Sanctuary and Baffin Island representing the Short Grass Prairie Population and Tall
Grass Prairie Population of B. h. hutchinsii, respectively (Figure 2). Canada geese were
collected from the Tanana River‘near Fairbanks and Cook Inlet near Anchorage, Alaska
representing B. c. parvipes, the Copper River Delta representing B. c. occidentalis,
Admiralty Island representing B. c. fulva, Washington state and the Sandhills of Nebraska
representing B. c. moflitti, Churchill, Manitoba representing the Eastern Prairie
Population of B. c. interior, southwest Hudson Bay and northwest James Bay
representing the Mississippi Valley Population of B. c. interior, Akimiski Island and
southern James Bay representing the Southern James Bay Population of B. c. interior,
Greenland representing B. 0. interior, western and eastern Ungava Peninsula representing
the Atlantic Population of B. c. interior, Newfoundland and Labrador representing the
North Atlantic Population of B. c. canadensis, and near Toronto, Ontario, and southeast
Michigan representing B. c. maxima (Figure 2). Population names for B. h. hutchinsii, B.
c. canadensis, and B. c. interior sampling locations are as defined in Dickson (2000).

Sample sizes for each location are listed in Table 1.
Characterization of Microsatellite Loci

DNA was extracted from all samples using DNeasy extraction kits (Qiagen Inc.,

CA). Twenty nuclear microsatellite loci were initially screened for allelic variation.

57

Table 1. Measures of genetic diversity for Canada geese (Branta canadensis) and
cackling geese (Branta hutchinsii) populations.

 

 

Breeding Location Species Population“ n A H0 Hc F18
Canada Geese

Southeast Michigan B. c. maxima MISE 54 7.00 0.670 0.694 0.038
Ontario B. c. maxima ONGI 44 7.13 0.662 0.723 0084"
Hudson and James B. c. interior MVP 100 9.13 0.689 0.729 0056*
Ba s

Huflson and James B. c. interior SJBP 116 9.00 0.709 0.751 0.056“
Ba 5

Wrist Ungava Peninsula B. c. interior APUW 50 8.00 0.700 0.738 0.052
East Ungava Peninsula B. c. interior APUE 50 8.00 0.692 0.724 0.044
Churchill, Manitoba B. c. interior EPP 50 8.13 0.667 0.724 0.079*
Greenland B. 0. interior GRLD 17 6.13 0.720 0.734 0.021
Labrador B. c. canadensis NAPL 51 7.25 0.687 0.705 0.025
Newfoundland B. c. canadensis NAPN 40 6.25 0.605 0.691 0127*
Nebraska Sandhills B. c. moflitti NEMF 49 6.88 0.676 0.694 0.027
Washington B. c. moflitti WASH 18 6.13 0.747 0.755 0.010
Fairbanks, Alaska B. c. parvipes FAIR 20 6.38 0.778 0.737 -0.057
Anchorage, Alaska B. c. parvipes ANCH 45 7.25 0.703 0.741 0.052
Copper River Delta B. c. occidentalis CRD 51 7.63 0.685 0.736 0.070
Admiralty Island 8. c. fulva ADMR 43 7.25 0.654 0.708 0.077
Cackling Geese

Baffin Island B. h. hutchinsii TGPP 86 8.88 0.677 0.734 0078*
Queen Maud Gulf B. h. hutchinsii SGPP 50 8.50 0.703 0.740 0.051
North Slope, Alaska B. h. taverneri NSLP 39 7.63 0.703 0.724 0.030
Yukon-Kuskokwim B. h. minima YKD 38 6.50 0.608 0.692 0.128
Delta

Buldir Island B. h. leucoparia BULD 37 6.00 0.644 0.694 0.073
Semidi Island B. h. leucoparia SEMD 28 4.38 0.626 0.590 -0.061

 

n = sample size per population; A = mean number of alleles over 8 loci; H, = observed heterozygosity;

He = expected heterozygosity; F15 = correlation of genes within individuals within populations with
significant values (P < 0.006 after correction for multiple loci; represented by an asterisk) indicating

possible inbreeding.

aPopulation abbreviations for samples from Scribner et al. (2003): FAIR = IN-AK, WASH = WA,

ANCH = SC-AK, CRD = CRD, ADMR = SE-AK, NSLP = NS, YKD = YKD, BULD = BUL,

SEMD = SEM.

58

Eight bi-parentally inherited loci proved to polymorphic in one or more breeding
populations and were used for subsequent analyses. Loci used included Bcaul, Bcau7,
Bcau9, Bean] 1, Hhiul (Buchholtz et al. 1998); TTUCGl, TTUCGS, (Cathey et al.
1998); and CR-G (A. Baker unpubl. data). Each locus was amplified using polymerase
chain reaction (PCR) in 25 ul reaction volumes, including 100-150 ng DNA, 10-25 pmol
of each primer, PCR buffer (10 mM Tris-HCI pH 8.3, 50 mM KCl, 100 ug/mL gelatin,
0.01%NP-40, 0.01% Triton-X 100), 0.5 U of AmpliTaq DNA Polymerase (Perkin-
Elrner), and 100-200 uM dNTPs. Forward primers of each locus-specific primer pair
were labeled with either Hex or Fluorescein by the manufacturer (IDT Technologies,
Inc.). Thermocycler conditions included a denaturing step of 94 °C for 2 min, followed
by 30-35 cycles of 94 °C for l min, annealing temperature for 1 min [49 °C (Bcau7), 51
°C (Hhiul), 54 °C (TTUCG-S), 56 °C (Bcaul, Bcau9, CR-G), 58 °C (Bcaul l), 60 °C
(TTUCG-l)], and 72 °C for 1 min. Products were visualized using a F MBIO II laser
scanner (Hitachi Software Engineering Co.) after electrophoresis on denaturing 6%
acrylarnide gels. Genotypes were scored based on 20 base-pair standards and reference

samples of known allelic size.

Characterization of MtDNA

A 143-bp fragment of the 5’ end of the mitochondrial DNA control region was
amplified using primers and conditions described in Pierson et al. (2000) and Pearce et al.
(2000). These primers were designed to recognize sites flanking the hypervariable
portion of the control region (3’ end of domain; Baker and Marshall 1997), and

additionally, to amplify and sequence only mitochondrial DNA sequences and not

59

nuclear DNA sequences originating from transposed mtDNA (Sorenson and Fleischer
1996). Approximately 50-100 ng DNA was used for the initial mtDNA amplification
with primers L78 and H493 and the PCR protocol of Kocher et al. (1989). Thermocycler
conditions included initial denaturation step of 94 °C for 2 min, followed by 40 cycles of
94 °C for 45 s, annealing at 60 °C for l min, 72 °C for 1 min, and extending at 72 °C for
7 min. Amplified PCR products were cleaned with QIA-quick spin column kits (Qiagen
Inc., CA), and sequenced using SequiTherrn Excel DNA sequencing kits (Epicentre, Inc.,
Madison, Wisconsin) by following product protocols for use of fluorescently labeled

primers.

Gene Diversity and Population Differentiation

For each population, observed genotype frequencies for each of the eight
microsatellite loci were tested for departure from Hardy-Weinberg expectations were
implemented in the program FSTAT (Goudet 2001). Tests for genotypic linkage
disequilibrium (a measure of independence across loci within a population) were
performed as described in Goudet et al. (1996) using FSTAT. Estimates of measures of
genetic diversity including number of alleles per locus (A), observed (Ho) and expected
(HE) heterozygosity were calculated using the program The Excel Microsatellite Toolkit
(Park 2001).

The F STAT program was used to estimate degree of spatial heterogeneity in gene
frequency within and among cackling and Canada geese populations using hierarchical
F -statistics (Weir and Cockerham 1984, Weir 1996) at three levels: (1) among

individuals within populations (f), (2) among individuals within the total population (F),

60

and (3) among populations (0). Significance of F -statistics was based on 95% confidence
intervals determined by bootstrapping across loci. Confidence intervals that included
zero were considered non-significant. Pair-wise estimates of population F ST were used as
summary measures of inter-population variance in allele frequency. Significance of pair-
wise interpopulation differentiation was determined using the exact G-test (Goudet et al.
1996) in F STAT, as the G-test is more powerful than exact F gT-estimator tests for diploid
populations (Goudet et al. 1996, Petit et a1. 2001). For tests of Hardy-Weinberg, gametic
disequilibrium, and F -statistics, nominal significance levels (alpha) were adjusted to
account for multiple testing using sequential Bonferroni corrections (Rice 1989).

Genetic affinities among populations of cackling and Canada geese were assessed
using several approaches. First, microsatellite allele frequencies across all eight loci were
used to estimate Cavalli-Sforza and Edwards (1967) chord distances among all 24
populations of cackling and Canada geese. A multilocus population neighbor-j oining tree
based on interpopulation distances was then constructed using the program PHYLIP
(version 3.6; Felsenstein 1993). Chord distances have been shown to produce robust tree
topologies (Takezaki and Nei 1996). Statistical significance of branches was based on
1000 bootstrapped population trees constructed using PHYLIP.

A hierarchial analysis of molecular variance (AMOVA; Excoffier et al. 1992) was
used to partition variance in mtDNA haplotype frequencies and nDNA allele frequencies
among cackling geese and Canada geese populations. Estimates of variance among
populations within groups ((Dsc), among groups (dim) and among populations among

groups ((DST) were derived using AMOVA in the program ARLEQUIN (version 2.0,

61

Schneider et al. 2000). Groups were defined by species, subspecies, and migratory

Flyway affiliation for hierarchical analyses.

MtDNA Sequence Data

Mitochondrial DNA sequences were obtained from 680 samples across all 11
subspecies and 22 breeding locales. Sequences were aligned manually, and haplotype
designations were based on at least one base-pair substitution or insertion-deletion events.
Additional haplotypes have been described previously (A-Z, Scribner et al. 2003a; Sl-

523, Scribner et al. 2003b) for cackling geese and Canada geese in North America.

Phylogeographic Analyses

We used MODELTEST version 3.06 (Posada and Crandall 1998) to find the best
available model of DNA evolution based on hierarchical likelihood-ratio tests. The
model of Tamura and Nei (1993) that incorporates information on the shape parameter of
the gamma distribution (G = 0.3362) provided the best fit to the data, and was used in
tree reconstructions for analyses including all cackling geese and Canada geese
populations. The best available model of DNA substitution when species were analyzed
separately was the HKY (Hasegawa et al. 1985) incorporating information on the shape
parameter of the gamma distribution (G = 0.0640) for Canada geese populations; and the
HKY incorporating information on the proportion of invariable sites (I = 0.5168) for
cackling geese populations. Phylogenetic analyses were conducted using PAUP*4.0b10
(Swofford 2000). We used maximum-likelihood and parsimony methods. Heuristic tree

searches were conducted for each analysis, with 20 and 100 random additions of taxa for

62

maximum-likelihood and parsimony analyses, respectively, each followed by tree
bisection-reconnection topological rearrangements. Robustness of nodes was assessed
using tree reconstructions of bootstrap-resampled data (1000 replicates) under parsimony
criteria, and (200 replicates) for maximum-likelihood criteria. We used sequence from
lesser snow goose (Chen caerulescens caerulescens) from a homologous portion of the

control region (GenBank accession number S70800; Quinn 1992) as an outgroup.

RESULTS

Genetic Variation

Allelic variation for the nine nuclear microsatellite loci ranged from 5 to 23
alleles . Both cackling geese and Canada geese populations were characterized by high
levels of microsatellite diversity as evidenced by estimates of observed and expected
heterozygosity and allelic diversity (Table 1). The mean number of alleles over the eight
loci was slightly higher for Canada geese populations (6.13 — 9.13) as compared to
populations of cackling geese (4.38 - 8.88). No loci deviated from Hardy-Weinberg
expectations, and no evidence of linkage disequilibrium was observed. Inbreeding
coefficients (F15) were low and non-significant for all populations of cackling geese
except for the population from Baffin Island (F15 = 0.078; Table 1). Significant
inbreeding coefficients (heterozygote deficiency) were observed for five of the sixteen
populations of Canada geese (Table 1). Three of these populations were B. c. interior
geese breeding along Hudson and James Bays. Heterozygote deficiency in these
populations may be a result of sampling within family groups, rather than an indication of

population inbreeding.

63

Phylogenetic and Population Relationships

Estimates of spatial variation based on the eight biparental microsatellite loci
were significant for individuals within populations (f), among populations (0) and for
individuals across the total population (F) of cackling geese and Canada geese combined
(Table 2). Estimates of spatial variance were also significant for all three hierarchical
levels of analysis when Canada geese populations and cackling geese populations were
analyzed separately (Table 2).

Pairwise estimates of variance among populations (0p) were highly significant
between populations of cackling geese and populations of Canada geese. Among
population differences in allele frequency were greater among populations from different
species (mean 0p = 0.062) than among populations within species (mean 0p = 0.055 for
cackling geese population comparisons; mean 0p = 0.032 for Canada geese population
comparisons) when all 22 populations were included. Allele frequencies were
significantly different among all pairs of Canada geese populations except between two
populations of B. c. interior breeding on the Ungava Peninsula (APUW and APUE), and
between APUW and the B. c. canadensis population from Labrador (N APL; Table 3).
Within cackling geese, allele frequencies were not significantly different between the two
B. h. hutchinsii populations (TGPP and SGPP), and between SGPP and B. h. minima
from the Yukon-Kuskokwim Delta (YKD). Allele frequencies differed significantly
among all other population comparisons for cackling geese (Table 3). Pairwise
differences in allele frequencies were much greater between populations from different

proposed glacial refugia (mean 0p = 0.080) than between populations from within the

64

Table 2. Estimates of hierarchical F-statistics (Weir and Cockerham

1984) for populations of cackling geese and Canada geese based on

eight microsatellite loci. Variance partitioning: f = alleles within individuals;

F = among individuals within the total population; 0 = among populations.
Statistically significant (P < 0.01) values over all loci are marked with an asterisk.

 

 

Locus f F 0
All Populations (n = 22)

Bcaul 0.035 0.069 0.035
Bcau7 0.077 0.104 0.029
Bcau9 0.057 0.104 0.049
Bcaull -0.011 0.057 0.067
I-Ihiul 0.004 0.033 0.028
CR-G 0.004 0.041 0.037
TTUCG-l 0.114 0.164 0.057
TTUCG-S 0.127 0.147 0.023
All Loci 0.090* 0.039* 0053*
Canada Populations (n = 16)

Bcaul 0.039 0.065 0.026
Bcau7 0.075 0.093 0.02
Bcau9 0.052 0.075 0.025
Bean] 1 -0.004 0.035 0.038
Hhiul -0.018 0.01 0.028
CR-G -0.01 1 0.024 0.035
TTUCG-l 0.144 0.169 0.029
TTUCG-S 0.141 0.155 0.017
All Loci 0.054* 0.079* 0.027*
Cackling Geese Populations (n = 6)

Bcapl 0.024 0.025 0.028
Bcau7 0.087 0.126 0.043
Bcau9 0.072 0.146 0.079
Bcaul 1 -0.033 0.062 0.093
Hhiul 0.082 0.029 0.011
CR-G 0.053 0.091 0.041
TTUCG-l 0.011 0.043 0.032
TTUCG-S 0.087 0.117 0.033
All Loci 0.053* 0.095* 0.044*

 

65

 

 

o.o.o GMU
ovoo mvoo IOZ<
vmoo mmoo ”moo y=<m
wood nmoo Nwoo ovoo Im<3
wvoo mvoo oooo omoo wmoo mEmZ
mmoo goo goo mmoo mooo mvoo 7E<Z
wmoo omoo goo vmoo wmoo mvoo wood 1:22
_mo.o Nmoo ”moo Goo wood wmoo mmoo zoo DAM—O
o _ o.o hmoo vmoo mooo Zoo .86 God 3 o.o omoo mom
omoo omoo oooo o _ o.o Nmoo goo o _ o.o hooo Nvoo m.o.o mDm<
mmoo wmoo Nmoo mmoo ovoo nmoo m_o.o wood mmoo o_o.o coo... >23:
mmoo mmoo ovoo N_o.o omoo ”mod n _ o.o oooo vmoo wooo Eoo _ _o.o mmnm
omoo _mo.o omoo m _ o.o Vmoo hmoo o _ o.o m _ o.o mmoo wood S o.o Eoo wood a;
nmoo vmoo oooo mmoo wmoo mvoo nmoo mmoo goo Boo Boo Boo omoo zoo HOZO
mmoo mmoo oooo _No.o ovoo ovoo vmoo mmoo oooo _~o.o N_o.o m _ o.o NNoo v _ o.o o _ o.o MES
MED< _ DMD IOZ< Mam Im<3 @952 2522 #22 0:0 mom mDm< 32% mmam m>2 520

Sc“ 38:32.68 BRSBE .o .m 3&9: .o .m engorged» .o .m .8285 .o .m 32.385

.o .m .o .m «Hm

 

 

A2 .1. 5 macaw—zoom 830 £550

 

A 053. 5 Bow: we 8m 32338an nova—soon— .AowE 8% 62 29:2: .8.“ couooboo Economcom Sam “Bahama:
203 Big Eom $8— :a $88 .5 525 momocosooa 22? E 8532, :ouawaooéoofi .«o woerumo 8333 .m 2an

66

Table 3 (cont’d).

 

Cackling Geese Populations (n = 6)

 

 

 

 

B. h. B. h. B. h.
hutchinsii taverneri minima B. h. leucoparia
SGPP I NSLP I YKD | BULD SEMD
TGPP 0.000 0.019 0.031 0.041 0.099
SGPP 0.01 1 0.021 0.035 0.095
NSLP 0.042 0.045 0.087
YKD 0.064 0.171
BULD 0.064

 

same refugia (mean 0p = 0.027) for cackling geese. In contrast, pairwise comparisons of
allele frequencies for Canada geese were similar for populations within one refugia
(mean 0p = 0.031) and between separate refugia (mean 0p = 0.034). Pairwise population
comparisons of allele frequencies were also similar between populations within the same
Flyway and between populations from different Flyways for Canada geese (within
Flyway mean 0p = 0.028; between Flyway mean 0p = 0.03 3). Cackling geese exhibited
more variation in allele frequencies between populations within a F lyway (mean 0p =
0.068) than between populations in different Flyways (mean 0p = 0.033).

Microsatellite allele frequencies and mitochondrial haplotype frequencies differed
significantly between goose species (mtDNA (DCT = 0.947, nDNA (DCT = 0.020) and
among populations within species (Table 4). Hierarchical analysis of cackling and
Canada geese populations revealed variation in haplotype frequencies was much greater
between species ((DCT = 0.947) than among populations within a species ((Dsc = 0.288).
The variability of haplotype frequencies was greater among populations within proposed

glacial refugia than between refiJgia for cackling and Canada geese. In contrast, allele

67

2.22 {:22 .330 dam... .32.? ”82338.. .335 2.5.2 M38 322 .620

45:2 ”22538 $35 323% ram mzmznmceasaé 5;; .828 £23 .55 .55. .96 £29.. ”82530.. as»; 058....

.58 £3 ”22838.. 5.3»: aaéfiggfiu ”82mm .33 .9; .332 ”823%.. as»: 2:83

 

 

 

 

.83 .m _3 .23 .33 $3 .83 a 2 823:8.“ Ea 868.8385. 58.0 .850
.83 m _3- .33 .23 $3 .83 v o 22338; Ea 86835 33... 538 3206
823.8; \o 328339 a .3 883.33%
.83 .33 .23 .33 .83 .83 4 E 823.3%.“ 35 SEE 32.2 ”380 «850
.83 33- .23 . 3 33 S3 .33 N o 82.238 Ea 33»: mace... ”amuse 3208
825on \o Emnomsob o no $933 bSEMEN
*wmoo mood *vmoo *oomo oSo .386 N 3 macro—zoom o3 Ewéom w=o8< ”330 «9280
.33 .23 .83 .33 $3 .83 N o 25:22.8 Ea 230m mace... 5.80 3:80
8:3on \o 3283.89 3 no szkmx 383$
* goo *omoo ._. _mo.o Road .336 *wwmo m mm 8090 3280 o3 880 wazxomu 595%“
Potato; \o anomsob o .3 88QO
39313.. 1552823
*mmoo a: a: *mwmo a: a: _ E mcooflsoom 880 «330 wooE<
u.Booo a: a: *mmmo a: a: _ o macaw—zoom 880 35.30 wcofi<
momfimfi. 335355-52
.59 Eve Owe Ewe Hoe Owe mama—Ohm whom mumeGOU
<29. <298 oo o oo a

2383 3 53> woo—BE 2.» £02 29:38 So mcouooboo Bad <75: 8o oooo v R 3298 So moo v 5 82?
Quowmufim A83 $9328 95% 0322: Moo 3:on mecca o3 .fimev 3:on macaw mcouawaoo macaw A89 3:on

£53, macaw—zoom mecca ”mficofltwo oo§tm> .8653on own—o <ZeE no woman ougcg mo 83858 Row—898m: .v 033.

68

frequency differences were greater between cackling geese refugia than among
populations within refugia, indicative of substantial variation between populations of B.
c. leucoparia and the other three subspecies. When populations and subspecies of
Canada geese were divided by migratory Flyway affinities, estimates of haplotype and
allele frequency variance were greater among populations within Flyway groups, than
between F lyways (Table 4). A hierarchical analysis of cackling geese among F lyways
revealed similar results. Hierarchical estimates of variance based on haplotype and allele
frequencies were greater among populations within subspecies than between subspecies
for both cackling geese and Canada geese (Table 4).

Populations of cackling geese and Canada geese generally clustered by species
within the consensus neighbor-joining tree based on chord distances derived from bi-
parental loci (Figure 3). One population of Canada geese (B. c. parvipes, ANCH) fell
within the cackling geese cluster, although statistical support for node placement was
low. Significant genetic differentiation between the two populations of B. h. leucoparia
and other cackling geese populations was evident based on the tree topology (95%
bootstrap support). Within Canada geese, populations of B. c. canadensis clustered
together with low support, but other subspecies with multiple populations were
intermixed (Figure 3). No significant phylogeographic pattern was evident among
Canada geese populations or cackling geese populations breeding in close geographic
proximity (Figure 3).

Fifty-three haplotypes were characterized across 680 individuals from six
populations of cackling geese and sixteen populations of Canada geese on the basis of

143 bp of mtDNA sequence data. Two major mtDNA haplotype clades representing

69

SEMD

 

BULD ANCH
953
LP
ADMR
YKD
7 57‘ CRD
TGPP
SGPP ‘63
MISE
24 “6
MP NEMF
646 . EPP
'L (a "V”
APUW . -
' SJBPM
SJBPA ONGI
GRLD
WASH

Figure 3. Neighbor joining tree describing overall genetic similarities among populations
of cackling and Canada geese based on Cavalli-Sforza and Edwards (1967) chord
distances across eight microsatellite loci. Bootstrap support (>50% out of 1000
replicates) for the consensus microsatellite tree is shown to the right of branches or
identified by an arrow. Population abbreviations are as in Table 1.

7O

cackling geese and Canada geese were strongly supported (98-100%) for both maximum
likelihood and maximum parsimony analyses. The majority of haplotypes within each
species clade were unresolved. Thirty-three of 53 haplotypes (62%) were found in a
single sampling locale. Private alleles occurred in frequencies of 0.00 to 0.40 in Canada
geese populations. Similar frequencies of private haplotypes were observed among
Canada geese populations putatively from the “Western Southern” refugia (mean = 0.10)
and Eastern Southern refugia (mean = 0.08). Private haplotypes were found in the
highest frequencies within the Pacific Flyway (mean frequency = 0.16). Frequencies of
private haplotypes in the remaining Flyways were 0.02 in the Central Flyway, 0.07 in the
Mississippi F lyway, and 0.04 in the Atlantic F lyway. Haplotype A was the most
common haplotype in Canada geese, and was found in six of seven subspecies and 15 of
16 populations (not found in B. c. occidentalis from CRD; Figure 4). Population
frequencies of haplotype A ranged from a low of 0.00 in the CRD population to a high of
1.00 in the FAIR population. Haplotype A was much more common on average in
populations from the proposed “Eastern Southern” refugia (mean = 0.63) than in
populations from the “Western Southern” refugia (mean = 0.14). Mean frequency of the
A haplotype was greatest for populations in the Atlantic Flyway (0.72), followed by 0.58
in the Central F lyway, 0.49 in the Pacific F lyway, and 0.41 in the Mississippi F lyway.
Only one statistically supported haplotype clade was resolved within the Canada goose
mtDNA haplotype tree, which included two private haplotypes (I and J) from the CRD

(B. c. occidentalis) and ANCH (B. c. parvipes) populations.

71

NEMF
MISE, NEMF

R1
821

MISE, ONGI
MVP, SJBP 32°

MVP S19

MISE, 1
ONGI, S 8
GRLD

S17

MISE
ONGI, MVP

S16
NAPN

S15
MISE, ONGI,
NAPL

$8

NAPN
S7

MVP, SJBP

NAPN APUE86

APUW, APUE, GRLD $5

MISE
R

34

NAPL

3

 

S3

1

CRD

EPP
R4

100

ANCH

MISE, ONGI
MVP, SJBP
APUW, APUE
EPP, GRLD
NAPN, NAPL
NEMF, WASH

A SGPP,ANCH

FAIR, ADMR
B ANCH, CRD

C ADMR,
CRD

D WASH

GANCH

H NAPL,

R ANCH

Y MISE

S1
32 SJBP, NAPL, NEMF

MISE, ONGI, MVP, SJBP

APl 1W APl IF NAPI

MVP, NAPL

Figure 4. Mitochondrial DNA cladogram based on maximum likelihood analysis
describing evolutionary relationships among 26 haplotypes of Canada geese resolved on
the basis of mtDNA control region sequence data. Bootstrap values based on 200
replicates exceeding 50% and population locations for haplotypes are included.

72

For cackling geese, the most common haplotype (L) was found in all four
subspecies and all six populations (Figure 5). Frequencies of the L haplotype in cackling
geese populations were 20% for BULD and 79% for SEMD representing B. h.
leucoparia, 55% for B. h. minima from YKD, 35% for B. h. taverneri from NSLP, and
34% for SGPP and 43% for TGPP representing B. h. hutchinsii. Seventeen of 25
cackling geese haplotypes (71%) were found in a single population, and nine haplotypes
were observed only in geese from Baffin Island (TGPP; Figure 5). No private haplotypes
were found in the NSLP population. Frequencies of private haplotypes in the remaining
cackling geese populations were 40% for BULD, 7% for SEMD, 18% for YKD, 37% for
SGPP, and 43% for TGPP. Frequencies of the L haplotype were similar between
cackling geese populations from the proposed “Aleutian” refugia (mean = 0.50) and
populations from the “Northern” refugia (mean = 0.39), as were frequencies of private
haplotypes (mean = 0.24 in both refugia groups). Frequencies of the L haplotype were
also similar between populations in the Pacific Flyway (mean = 0.45) and populations in
the Central/Mississippi F lyway (mean = 0.39). The mean frequency of private
haplotypes was twice as large in the Central/Mississippi Flyway populations (mean =

0.40) as in the Pacific Flyway populations (mean = 0.16).

DISCUSSION
Canada geese and cackling geese were considered large- and small-bodied forms
of one species complex (Branta canadensis) until 2003 (Banks et a1. 2003). Past genetic
studies of geese based on mtDNA analyses (Shields and Wilson 1987, Van Wagner and

Baker 1990, Baker 1998, Scribner et al. 2003a) supported the split of the Canada goose

73

TGPP TGPP

 

 

 

YKD 39 S10 s11 TGPP
z
SGPP,YKD u 812 TGPP
TGPP,YKDS S13 TGPP
s14
BULD TGPP
SEMD Q
522 TGPP
TGPP p
BULD R5
SGPP
o
SEMD R7
SGPP
BULD N
R8
BULD M 54 SGPP
SEMD 50
L
SGPP, TGPP
NLSP, YKD F
SGPP, NSLP
SGPP
SGPP R6
7
TGPP
$23
TGPP

Figure 5. Mitochondrial DNA cladogram based on maximum likelihood analysis
describing evolutionary relationships among 27 haplotypes of cackling geese resolved on
the basis of mtDNA control region sequence data. Bootstrap values based on 200
replicates exceeding 50% and population locations for haplotypes are included.

74

species complex into two species, as evidenced by high levels of sequence divergence
between large- and small-bodied forms (14%, Scribner et al. 2003a). It has even been
suggested that the two newly recognized species of Canada geese and cackling geese

have closer evolutionary ties to the Hawaiian goose and barnacle goose, respectively,

rather than being each other’s closest genetic relative (Paxinos et al. 2002).

The high degree of genetic divergence revealed in the above studies may be a
result of species isolation that occurred during the Pleistocene. Based on species
ecology, nesting habits, and availability of unglaciated breeding habitat available during
the Pleistocene, Ploeger (1968) identified four major breeding areas for cackling and
Canada geese. He hypothesized that Canada geese inhabited refugia south of the major
ice sheets, while cackling geese were found north of the glaciers in the Aleutian Islands
and Canadian Arctic Archipelago. Canada geese were split between the “Western
Southern Group” nested in the Pacific coastal region south of the Cordilleran ice sheet
and included current subspecies B. c. film and B. c. occidentalis; and an “Eastern
Southern Group” composed of current subspecies B. c. canadensis, B. c. interior, B. c.
maxima, B. c. moffitti, and B. c. parvipes nested south of the Cordilleran and Laurentide
ice sheets east of the Cascade and Sierra Nevada Mountains. Cackling geese were also
hypothesized to inhabit two refugia including the “Aleutian Group” where B. h.
Ieucopareia nested; and the “Northern Group” of the Canadian Archipelago which
included the three remaining extant subspecies, B. h. minima, B. h. taverneri, and B. h.
hutchinsii.

Our findings of variation in mtDNA and nDNA based on analyses of molecular

variance (Tables 2, 3 and 4), genetic distances among populations based on allele

75

frequencies (Figure 3) and haplotypes (Figures 3, 4, and 5) support a fundamental
evolutionary split between Canada geese and cackling geese. Thus, our data is consistent
with the findings of previous genetic studies that documented large genetic differences
between species. Hierachical analyses of mtDNA haplotype variance (Table 4) and the
deep phylogenetic split between species in our haplotype tree support Ploeger’s (1968)
hypotheses that ancestors of cackling and Canada geese were isolated in separate
Pleistocene refugia.

Subspecies distinctions, however, were not well supported in either species group
on the basis of our analyses. Topologies within the mtDNA haplotype tree for both
species were weakly supported, and demonstrated a lack of phylogenetic structure based
on subspecies classifications. Results of AMOVA for mtDNA indicated that higher
levels of genetic variance were apportioned among populations within subspecies, than
among subspecies within cackling and Canada geese (Table 4). Among population
variation in haplotype frequencies was also greater than between group variance for
hierarchical contrasts of proposed glacial refugia groups and Flyway groups (Table 4).
Ancestral haplotypes within cackling geese (L) and Canada geese (A) were inferred
based on interior orientation within haplotype clades (Figures 4 and 5), high frequency of
occurrence, and broad geographic dispersion. For Canada geese, frequencies of the A
haplotype varied considerably between groups of populations from the “Western
Southern” and “Eastern Southern” glacial refugia proposed by Ploeger (1968), though the
occurrence of private haplotypes was similar between both refugial groups. In contrast,
populations of cackling geese putatively from the “Aleutian” refugia and “Northern”

refugia had similar frequencies of the L haplotype, and similar frequencies of private

76

haplotypes. The large difference in frequency of the ancestral A haplotype between the
two refugial groups of Canada geese, and the lack of the A haplotype in one of two
populations from the “Western Southern Group” (CRD), suggest that Canada geese may
have been separated in two isolated refugia during the Pleistocene, but that separation
times were not long enough for complete lineage sorting to occur. The widespread
occurrence of the L haplotype in similar frequencies within populations of cackling geese
from the two proposed refugia argues that cackling geese may have resided together in
one refugia, or that historical gene flow occurred between the two refugial groups.

In a genetic survey of cackling and Canada geese in western North America,
Scribner et al. (2003a) also found a lack of evidence to support the existence of two
refugia for cackling geese. Their conclusions were based on the widespread occurrence
of the ancestral haplotype L in cackling geese subspecies putatively from the two
different refugia. However, they stated genetic affinities between populations of Canada
geese sampled from interior Alaska and Washington State were supportive of Ploeger’s
hypothesized “Eastern Southern Group”, one of two proposed Pleistocene refugia for
Canada geese. Based on our suite of microsatellite loci we found significant differences
between populations in these regions (FAIR, WASH) and populations did not cluster
together within the neighbor-j oining tree based on affinities in microsatellite allele
frequency (Figure 3). However, our data on the frequencies of the ancestral haplotype A
within the refugial groups support the hypothesis that two isolated refugia may have
existed for Canada geese.

In both cackling geese and Canada geese, we observed a large number of

haplotypes that occurred in only one population, and that differed by only one or two site

77

changes from other haplotypes. Numerous “private” haplotypes (Slatkin and Maddison
1989) suggest that historical founder events and restricted gene flow have led to novel
haplotypes in many populations. Our large samples sizes for each population lend
confidence to frequency estimates of novel haplotypes and ancestral haplotypes for both
cackling and Canada geese. Distribution of private haplotypes among contemporary
populations of Canada geese and cackling geese is likely limited by the high level of
natal area and breeding site philopatry that characterize these species (Anderson et al.
1992, Ely and Scribner 1994, Mowbray et al. 2002).

Previous studies documented that nominal subspecies within cackling and Canada
geese could be unambiguously distinguished based on mtDNA haplotypes (Shields and
Wilson 1987, Van Wagner and Baker 1990, Baker 1998). We did not find evidence of
distinct mitochondrial DNA among subspecies of cackling and Canada geese (Figures 4
and 5). Our results are consistent with a mtDNA survey of cackling and Canada geese in
western North America (Scribner et al. 2003a). Differing results may be attributed to
differences in sample sizes among studies, and how samples were genetically
characterized. Earlier studies utilized small numbers of samples per population, and
additional shared haplotypes may have been undetected when few individuals were
analyzed from each location even if geographic coverage of sampling was broad. Our
study, and the work of Scribner et al. (2003a), had relatively larger sample sizes for all
population collections, which provides greater confidence in inferred phylogeographic
and population relationships (Funk and Omland 2003, Peters et al. 2005). Previous
studies relied on characterization of mtDNA through restriction fiagment analysis, which

resolved sequence variation over a larger portion of the mtDNA molecule than in our

78

study, where a portion of the mtDNA genome was directly sequenced. Resolution among
subspecies may be improved by sequencing a longer region of the mtDNA control region
to increase the number of informative character states. Additionally, sequencing of
mtDNA could be supplemented with sequences from informative nuclear genes, such as
the Z-chromosome—linked chromo-helicase binding protein gene (Peters et al. 2005).
Long-term family associations and strong male and female breeding site fidelity
exhibited by cackling and Canada geese (Anderson et al. 1992, Ely and Scribner 1994,
Mowbray et al. 2002) could reproductively isolate breeding populations and lead to
significant genetic differentiation among breeding populations over time due to genetic
drifi. Pair bond formation of cackling and Canada geese during spring when geese have
returned to breeding grounds (Raveling 1969, Surrendi 1970, Ely and Scribner 1994) may
serve to reinforce genetic isolation among populations. However, populations that share
migratory routes and common wintering areas within Flyways may experience a higher
degree of inter-population gene flow than populations from different F lyways. If a high
level of genetic exchange occurs among populations within Flyways due to individuals
dispersing to new populations during migration or wintering, the degree of spatial genetic
structure within species may be greater among wintering Flyway groups than among
breeding populations (Robertson and Cooke 1999). Hierarchical analyses revealed
greater levels of mtDNA and nDNA genetic variation among breeding populations than
among F lyway groups for both cackling and Canada geese (Table 4). Among population
genetic differentiation was also significantly greater than genetic variability among
subspecies. Our inter-population tree based on allele frequency differences had low

support for phylogenetic structure based on subspecies or Flyways (Figures 3). In

79

addition, mean pairwise population differences in allele frequencies were similar between
populations from different Flyways and between populations in the same Flyway for both
cackling and Canada geese. Our data are consistent with studies that indicate strong
fidelity to breeding sites and formation of pair bonds on breeding grounds likely
increases reproductive isolation among breeding populations of cackling and Canada
geese. The lack of genetic divergence between F lyways suggests that refugial
populations were large in number and time since population separation has been
relatively short on an evolutionary time scale, resulting in no greater genetic differences
between subspecies and migratory F lyway populations as compared to populations within
Flyways. Alternatively, cackling and Canada geese have only been utilizing
contemporary migratory Flyways over relatively short evolutionary periods (i.e.,
subsequent to the last Pleistocene glaciation), and F lyway groups may not have
experienced long enough separation times to develop measurable genetic differences in
gene frequencies.

One subspecies, B. c. leucoparia, was well resolved based on nuclear genetic
distances illustrated in the microsatellite neighbor-joining tree. The island populations of
BULD and SEMD were distinguished from other populations and subspecies with 95%
bootstrap support. A similar topology was not supported in the cackling geese mtDNA
tree, as these populations have multiple shared haplotypes including the ancestral
haplotype (L) found in all populations and subspecies of cackling geese (Figure 5). The
B. h. leucoparia subspecies of cackling geese was extirpated from most of its historical
range after the introduction of non-native predators for fur production fi'om 1830 to 1930,

and was listed as an endangered in 1967 (U SFWS 1991). The genetic divergence

80

between B. h. leucoparia and other subspecies of cackling geese is likely a result of the
severe population bottleneck that occurred in the mid 20th century. Our tests for excess
heterozygosity indicating BULD had likely experienced a recent population bottleneck,
and low levels of allelic diversity of the BULD and SEMD populations relative to other
cackling geese (Table 1), is consistent with the findings of a previous genetic study on B.
h. leucoparia (Pierson et al. 2000).

Some Arctic-nesting avian species have deeply diverged phylogenetic groups that
are geographically widespread. In an investigation of the evolutionary relationship
between mallard ducks (A nas platyrhynchos) and American black ducks (A nas rubripes),
two fundamental, phylogenetically distinct mtDNA clades were described that were
widely distributed in mallards from two geographically isolated breeding populations
(Manitoba and California; Avise et al. 1990). The authors argued that the geographic
uniformity suggested either recent colonization of Manitoba by birds from California
after the last glacial retreat, or that recent gene flow had occurred between the two
populations. A similar pattern of mtDNA variation was found in lesser snow geese and
Ross’s geese across their breeding ranges (Quinn 1992, Avise et al. 1992, Weckstein
2002, see Chapter 6). Two fundamental clades of mtDNA haplotypes suggestive of past
vicariance and divergence were found in across the species breeding ranges, arguing for
recent population expansion and high levels of contemporary gene flow.

Some Arctic species exhibit homogenized gene frequencies among nesting and
wintering populations (harlequin duck, Lanctot et al. 1999; king eider, Pearce et al.
2004). For harlequin ducks, similar gene frequencies over a wide geography was

hypothesized to be a result of the cumulative effects of low levels of gene flow over long

81

time periods, low levels of juvenile gene flow, or episodic dispersal attributed to habitat
alteration. The genetic panmixia and lack of spatial structure in king eiders was
attributed to historical gene flow and recent population growth following Pleistocene
deglaciation (Pearce et al. 2004).

Cackling and Canada geese exhibit a fundamental genetic split between species
related to their biogeographic isolation during the Pleistocene, and there is some evidence
of a historical divergence between Canada geese restricted to two different glacial
refugia. Additionally, our data revealed moderate levels of genetic variation among
breeding populations and F lyway groups that have likely accrued due to strong male and
female breeding site fidelity and pair bond formation that occurs when breeding
populations are largely isolated from other populations. The spatial genetic structure of
cackling and Canada geese is similar to phylogeographic structuring in dunlin (Wenink et
al. 1994, 1996), which exhibit a strongly subdivided population genetic structure. Like
cackling and Canada geese, the genetic variation in the dunlin is attributed to Pleistocene
vicariant events combined with strong natal philopatry that has restricted contemporary

gene flow among populations.

MANAGEMENT IMPLICATIONS
Management of cackling geese and Canada geese has traditionally been based on
subspecies designations and management populations defined on the basis of migratory
Flyway. Population and subspecies classifications were determined on the basis of
morphological characteristics. There are not, however, definitive plumage characteristics

and morphometric standards for all populations and subspecies of cackling and Canada

82

geese. In addition, morphologies can be subject to environmentally induced variation
(Leafloor et al. 1998) and age-related variation (Thompson et al. 1999). Genetic markers
can be used to document spatial patterns in allele or haplotype frequency within species.
and determine if currently recognized management units adequately describe groups that
are presumed to be demographically independent.

Genetic characterizations of cackling and Canada geese provide valuable
information on past and present population structure that may be incorporated into
management plans and conservation policies for these two species. Our data support
previous findings of a fundamental genetic split between cackling geese and Canada
geese species. Within species, our analyses revealed evidence of incomplete lineage
sorting illustrated by the widespread distributions of ancestral haplotypes within cackling
and within Canada geese. We also found indications that contemporary gene flow among
breeding populations and Flyways may be restricted by breeding behaviors and strong
fidelity to breeding grounds.

Current management efforts of cackling and Canada geese are focused largely on
management populations defined by wildlife agencies based on breeding ground
afiliation and wintering ground distribution (Dickson 2000). These populations do not
correspond precisely to subspecies or species. Some subspecies are split into multiple
management populations (i.e., B. c. interior in AP, SJBP, MVP, EPP, and WPP
populations), while other management populations may combine several subspecies or
species into a single unit (i.e., SGPP population of B. c. parvipes and B. h. hutchinsii; see
Appendix 1 in Chapter 1 for a summary of management populations). Large amounts of

funds from federal, provincial, and state agencies are provided annually to support

83

research and regulations that attempt to maintain the diversity and viability of these
management populations of cackling and Canada geese (Schmidt 2004). Results of our
genetic survey suggest that there is much less genetic variability among breeding
populations and subspecies than between species. Based on these results, we would
recommend that future management efforts and monies be less focused on previously
defined management populations (Dickson 2000) and more focused on identifying and
maintaining the diversity within each species.

Both cackling and Canada geese are managed as game species, and the majority
of harvest of these groups occurs on wintering grounds where populations and subspecies
are admixed. Accurate discrimination among subspecies and populations based on
morphology and band observations may be difficult when individuals are mixed in
migratory or wintering groups (Pearce and Bollinger 2003, Scribner et al. 2003b).
However, there is significant genetic structure for cackling and Canada geese at both
macro- and micro-geographic scales that could be utilized in a management context. For
example, genetic analysis techniques were recently combined with satellite telemetry to
infer the geographic origin and racial composition of Canada geese in newly colonized
habitats in Greenland (Scribner et al. 2003c). The authors used likelihood-based
assignment tests based on multilocus genotypes, and observational data on seasonal
movements, to determine that Canada geese in Greenland originated from the Atlantic
Population of geese breeding near southern Ungava Bay in Quebec, Canada. It has also
been demonstrated that genetic data provides a viable alternative to estimate proportional
contributions of subspecies and populations to harvest (Pearce et al. 2000, Inman et al.

2003, Scribner et al. 2003b) at multiple temporal and spatial scales. In Chapter 5, we

84

provide an example of harvest derivations based on genetics data for seasonal cackling
and Canada geese harvests in the State of Michigan from 1998-2002. This chapter is
followed by a comparative analysis of cackling and Canada geese harvest estimates
derived using three different techniques including genetic-based analyses, band recovery

data, and morphological measurements (see Chapter 6).

85

LITERATURE CITED

Avise, J. C. 1998. The history and purview of phylogeography: a personal reflection.
Molecular Ecology 7: 371-379.

Avise, J. C., R. T. Alisauskas, W. S. Nelson, and C. D. Ankney. 1992. Matriarchal
population genetic structure in an avian species with female natal philopatry. Evolution
46(4):1084-1096.

Avise, J. C., C. D. Ankney, and W. S. Nelson. 1990. Mitochondrial gene trees and the
evolutionary relations of mallard and black ducks. Evolution 44:1109-1119.

Avise, J. C., J. Arnold, R. M. Ball, E. Bermingham, T. Lamb, J. E. Neigel, C. A. Reeb, N.
C. Saunders. 1987. Intraspecific phylogeography: The mitochondrial DNA bridge
between population genetics and systematics. Ann. Rev. Ecol. Syst. 18: 498-522.

Avise, J. C. and D. Walker. 1998. Pleistocene phylogeographic effects on avian
populations and the speciation process. Proc. R. Soc. Lond. B 265: 457-463.

Anderson, M. G., J. M. Rhymer, and F. C. Rohwer. 1992. Philopatry, disperal, and the
genetic structure of waterfowl populations. Pages 365-395 in B. Batt, A. D. Afton, M. G.
Anderson, C. D. Ankey, D. H. Johnson, J. A. Kadlec, and G. L. Krapu, eds., Ecology and
management of breeding waterfowl. Univ. of Minnesota Press, Minnesota, MN. 63 5pp.

Baker, A. J ., and H. D. Marshall. 1997. Mitochondrial control region sequences as tools
for understanding evolution. Pages 51-82 in D. P. Mindell, Editor. Avian Molecular
Evolution and Systematics. Academic Press, New York, NY.

Banks, R. C., C. Cicero, J. L. Dunn, A. W. Kratter, P. C. Rasmussen, J. V. Remsen, Jr., J.
D. Rising, and D. F. Stotz. 2003. F orty-fifih supplement to the American ornithologists’
union check-list of North American birds. The Auk 121:985-995.

Bellrose, F. C. 1980. Ducks, geese, and swans of North America. Second edition.
Stackpole, Harrisburg, Pennsylvania, USA.

Buchholz, W., J. M. Pearce, B. J. Pierson, and K. T. Scribner. 1998. Dinucleotide repeat
polymorphisms in waterfowl (family Anatidae): characterization of a sex-linked (Z-
specific) and 14 bi-parentally inherited loci. Animal Genetics 29:323-325.

Cathey, J. C., J. A. DeWoody, and L. M. Smith. 1998. Microsatellite markers in Canada
geese (Branta canadensis). Journal of Heredity 89: 173-174.

Cavalli-Sforza, L. L. and A. W. F. Edwards. 1967. Phylogenetic analysis: models and
estimation procedures. Evolution 21:550-570.

86

Chesser, R. K. 1991. Influence of gene flow and breeding structure on gene diversity
within populations. Genetics 129:573-583.

Cooch, F. G. 1953. Techniques for mass capture of flightless blue and lesser snow
geese. Journal of Wildlife Management 17:460-465.

Comuet, J. M. and G. Luikart. 1996. Description and power analysis of two tests for
detecting recent population bottlenecks from allele frequency data. Genetics 144 :2001-
2014.

Delacour, J. 1954. The Waterfowl of the World. Country Life Limited, London, UK.

Dickson, K. M. 2000. The diversity of Canada Geese. Pages 11-24 in K. M. Dickson,
Editor. Towards conservation of the diversity of Canada Geese (Branta canadensis).
Occasional paper 103, Canadian Wildlife Service, Ottawa, Ontario, Canada.

Ely, C., and K. T. Scribner. 1994. Genetic diversity in arctic-nesting geese: implications
for management and conservation. Transactions of the North American Wildlife Natural
Resources Conference 59:91 -1 10.

Excoffrer, L., P. E. Smouse, and J. M. Quattro. 1992. Analysis of molecular variance
inferred from metric distances among DNA haplotypes: application to human
mitochondrial DNA restriction data. Genetics 131:479-491.

Felsenstein, J. 1993. PHYLIP (Phylogeny Inference Package), version 3.5c.
Department of Genetics, University of Washington, Seattle.

Funk, D. J. and K. E. Omland. 2003. Species-level paraphyly and polyphyly: frequency,
causes and consequencies with insights from animal mitochondrial DNA. Annu. Rev.
Ecol. Evol. Syst. 22:397-422.

Goudet, J. 2001. FSTAT, a program to estimate and test gene diversities and fixation
indices (version 2.9.3). Available from http://www.unil.ch/izea/softwares/fstat.html.
Updated from Goudet (1995).

Goudet, J, M. Raymond, T. Demeeus, F. Rousset. 1996. Testing differentiation in
diploid populations. Genetics 144: 1933-1940.

Greenwood, P. J. 1980. Mating systems, philopatry and dispersal in birds and mammals.
Animal Behavior 28: 1140-1162.

Greenwood, P. J. 1987. Philopatry and optimal outbreeding in birds. Pages 207-222 in
F. C. Cooke and P. A. Buckley, eds., Avian genetics Academic Press, Inc., NY.

Hanson, H. C. 1997. The Giant Canada Goose. Illinois Natural History Survey, IL.
252pp.

87

Hasegawa, M., H. Kishino, and T. Yano. 1985. Dating the human-ape splitting by a
molecular clock of mitochondrial DNA. Journal of Molecular Evolution 22:160-174.

Inman, R. L., K. T. Scribner, H. H. Prince, J. A. Warrillow, D. R. Luukkonen, P. I.
Padding. 2003. A novel method for Canada goose harvest derivation using genetic
analysis of tail feathers. Wildlife Society Bulletin 31:1126-1131.

Kimura, M., S. M. Clegg, I. J. Lovette, K. R. Holder, D. J. Girrnan, B. Mila, P. Wade,
and T. B. Smith. 2002. Phylogeographical approaches to assessing demographic
connectivity between breeding and overwintering regions in a Nearctic-Neotropical
wabler (Wilsonia pusilla). Molecular Ecology 11:1605-1616.

Kimura, M. and J. F. Crow. 1964. The number of alleles that can be maintained in a
finite population. Genetics 49:725-738.

Klicka, J. and R. M. Zink. 1997. The importance of recent ice ages in speciation: A
failed paradigm. Science 277:1666-1669.

Lanctot, R., B. Goatcher, K. Scribner, S. Talbot, B. Pierson, D. Esler, and D.
Zwiefelhoffer. 1999. Harlequin duck recovery from the Exxon Valdez oil spill: a
population genetics perspective. The Auk 116:781-791.

Leafloor, J. O., C. D. Ankney, and D. H. Rusch. 1998. Environmental effects on body
size of Canada geese. Auk 115:26-33.

MacInnes, C. D. 1966. Population behavior of eastern Arctic Canada geese. Journal of
Wildlife Management 30:536-553.

Moser, T. J. and F. D. Caswell. 2004. Long-term indices of Canada goose status and
management. Pages 123-120 in T. J. Moser, R. D. Lien, K. C. VerCauteren, K. F.
Abraham, D. E. Andersen, J. G. Bruggink, J. M. Coluccy, D. A. Graber, J. O. Leafloor,
D. R. Luukkonen, and R. E. Trost editors. Proceedings of the 2003 International Canada
Goose Symposium,Madison, Wisconsin, USA.

Ohta, T. and M. Kimura. 1973. A model of mutation appropriate to estimate the number
of electrophoretically detectable alleles in a finite population. Genet. Res. Camb. 22:201-
204.

Owen, M. 1980. Wild Geese of the World. B. T. Batsford Ltd., London. 236pp.

Palmer, R. 8., Ed. 1976. Handbook of North American Birds, Vol. 2. Yale University
Press, New Haven, CT, USA.

Park, S. D. E. 2001, Trypanotolerance in west African cattle and the population genetic
effects of selection. Ph.D. Thesis. In Prep. University of Dublin.

88

Paxinos, E. E., H. F. James, S. L. Olson, M. D. Sorenson, J. Jackson, R. C. Fleischer.
2002. MtDNA from fossils reveals a radiation of Hawaiian geese recently derived from

the Canada goose (Branta canadensis). Proceedings of the National Academy of
Sciences 99:1399-1404.

Pearce, J. M., B. J. Pierson, S. L. Talbot, D. V. Derksen, D. Kraege, and K. T. Scribner.
2000. A genetic evaluation of morphology used to identify hunter-harvested Canada
Geese. Journal of Wildlife Management 64:863-874.

Pearce, J. M. and K. S. Bollinger. 2003. Morphological traits of Pacific Flyway Canada

geese as an aid to subspecies identification and management. Journal of Field
Ornithology 74:357-369.

Pearce, J. M., S. L. Talbot, B. J. Pierson, M. R. Petersen, K. T. Scribner, D. L. Dickson,
and A. Mosbech. 2004. Lack of spatial genetic structure among nesting and wintering
king eiders. The Condor 106:229-240.

Peters, J. L., K. G. McCraken, Y. N. Zhuravlev, Y. Lu, R. E. Wilson, K. P. Johnson, and
D. E. Omland. 2005. Phylogenetics of wigeons and allies (Anatidae: Anas): the
importance of sampling multiple loci and multiple individuals. Molecular Phylogenetics
and Evolution 35:209-224.

Pierson, B. J ., J. M. Pearce, S. L. Talbot, G. F. Shields, and K. T. Scribner. 2000.
Molecular genetic status of Aleutian Canada Geese from Buldir and Semidi Islands,
Alaska. Condor 102:172-180.

Quinn, T. W. 1992. The genetic legacy of Mother Goose-phylogeographic patterns of
lesser snow goose Chen caerulescens caerulescens maternal lineages. Molecular
Evolution 12105-117.

Raveling, D. G. 1969. Social classes of Canada geese in winter. Journal of Wildlife
Management 33:304-318.

Robertson, G. J. and F. Cooke. 1999. Winter Philopatry in Migratory Waterfowl. Auk
116(1): 20-34.

Rhymer, J. M., M. G. Fain, J. E. Austin, D. H. Johnson, and C. Krajewski. 2001.
Mitochondrial phylogeography, subspecific taxonomy, and conservation genetics of
sandhill crans (Grus canadensis; Aves: Gruidae). Conservation Genetics 2:203-218.

Schmidt, P. R. 2004. Canada geese in North America: past success and future
challenges. Pages 8-11 in T. J. Moser, R. D. Lien, K. C. VerCauteren, K. F. Abraham, D.
E. Andersen, J. G. Bruggink, J. M. Coluccy, D. A. Graber, J. O. Leafloor, D. R.
Luukkonen, and R. E. Trost editors. Proceedings of the 2003 International Canada Goose
Symposium,Madison, Wisconsin, USA.

89

Schneider, 8., D. Roessli, and L. Excoffier. 2000. ARLEQUIN, version 2.0: A Software
for Population Genetic Data Analysis. Genetics and Biometry Laboratory, University of
Geneva, Geneva, Switzerland.

Scribner, K. T., M. R. Petersen, R. L. Fields, S. L. Talbot, J. M. Pearce, and R. K.
Chesser. 2001. Sex-biased gene flow in spectacled eiders (Anatidae): inferences from
molecular markers with contrasting modes of inheritance. Evolution 55:2105-2115.

Scribner, K. T., S. L. Talbot, J. M. Pearce, B. J. Pierson, K. S. Bollinger, and D. V.
Derksen. 2003a. Phylogeography of Canada geese (Branta canadensis) in western North
America. The Auk 120:889-907.

Scribner, K. T., J. A. Warrillow, J. O. Leafloor, H. H. Prince, R. L. Inman, D. R.
Luukkonen, and C. S. Flegel. 2003b. A genetic method for determining racial
composition of Canada goose harvests. Journal of Wildlife Management 67:122-135.

Scribner, K. T., R. A. Malecki, B. D. Batt, R. L. Inman, S. Libants, and H. H. Prince.
2003c. Identification of source population for Greenland Canada geese: genetic
assessment of a recent colonization. The Condor 105:771-782.

Shields, G. F. and A. C. Wilson. 1987. Subspecies of the Canada goose (Branta
canadensis have distinct mitochondrial DNA’s. Evolution 41(3): 662-666.

Shriver, M. D., L. J in, R. Chakraborty, and E. Boerwinkle. 1993. VNTR allele
frequency distributions under the stepwise mutation model a computer simulation
approach. Genetics 134:983-993.

Sorenson, M. D., and R. C. Fleischer. 1996. Multiple independent transposition of
mitochondrial DNA control region sequences to the nucleus. Proceedings of the National
Academy of Sciences 93:15239-15243.

Surrendi, D. C. 1970. The mortality, behavior, and homing of transplanted juvenile
Canada geese. Journal of Wildlife Management 34:719-733.

Takezaki, N. and M. Nei. 1996. Genetic distances and reconstruction of phylogenetic
trees from microsatellite DNA. Genetics 144:389-399.

Thompson, J. E., M. R. Hill, J. Merindino, M. Todd, and C. D. Ankney. 1999.
Improving use of morphometric discrimination to identify Canada goose subspecies.
Wildlife Society Bulletin 27:274-280.

Tirnm, D. E. and R. G. Bromley. 1976. Driving Canada geese by helicopter. Wildlife
Society Bulletin 4: 1 80-181.

90

United States Fish and Wildlife Service (U SFWS). 1991. Aleutian Canada goose
recovery plan. US. Fish and Wildlife Service Report, Anchorage, Alaska.

Valdes, A. M., M. Slatkin, and N. B. Freiner. 1993. Allele frequencies at microsatellite
loci: the stepwise mutation model revisited. Genetics 133: 737-749.

Van Wagner, C. E. and A. J. Baker. 1990. Association between mitochondrial DNA and
morphological evolution in Canada Geese. Journal of Molecular Evolution 31:373-382.

Voelker, G. 1999. Dispersal, vicariance, and clocks: historical biogeography and
speciation in a cosmopolitan passerine genus (Anthus: Motacillidae). Evolution 53(5):
1536-1552.

Weir, B. S. 1996. Genetic Data Analysis 11. Sinauer and Associates, Sunderland.

Weir, B. S. and C. C. Cockerham. 1984. Estimating F -statistics for the analysis of
population structure. Evolution 38:1358-1370.

Wenink, P. W., A. J. Baker, H. Rosner, and M. G. J. Tilanus. 1996. Global
mitochondrial DNA phylogeography of the Holarctic breeding dunlins (Caldris alpina).
Evolution 50:318-330.

Wenink, P. W., A. J. Baker, and M. G. J. Tilanus. 1994. Mitochondrial control-region
sequences in two shorebird species, the tumstone and the dunlin, and their utility in
population genetic studies. Mol. Biol. Evol. 11:22-31.

Zink, R. M. 1996. Comparative phylogeography in North American birds. Evolution
50:308-317.

91

CHAPTER 3: PHYLOGEOGRAPHY OF SNOW GEESE AND ROSS’S GEESE

INTRODUCTION

Genetic variation within and among species can be influenced by both historical
events and the ecology of the species (Avise et al. 1987, Avise 1998). Historical factors
that may influence the spatial genetic structure of species include population isolation
during glacial events, population expansions and dispersals during interglacial periods,
size of populations during glacial and interglacial periods, length of time populations
experienced geographic and reproductive isolation, and length of time since population
separation (Avise et al. 1998, Emerson et al. 2001, Stewart and Lister 2001). Longer
separation times and smaller effective population sizes in isolated populations increase
the probability populations will accrue differences in gene frequencies and develop novel
gene mutations (Avise et al. 1988). Historical population expansions occurring in a
single, punctuated event, and with a low number of founding individuals, could
appreciably change allele and haplotype frequencies in the founding population as
compared to the population of origin. Alternatively, if expansion events included a large
number of individuals over a prolonged time period, then gene frequencies may be
similar among populations (Mila 2000, Hewitt 2001).

Ecology and life history characteristics that may affect species spatial genetic
structure include timing and location of pair bond formation, degree and sex-specificity
of philopatry, levels of gene flow among contemporary breeding and/or wintering
populations, and current population size (Greenwood 1980, Chesser 1991a, Chesser

1991b, Ely and Scribner 1994, Miyatake and Shimizu 1999). High natal and breeding

92

site philopatry, and pair bond formation during periods when breeding populations are
isolated, will tend to restrict gene flow and increase genetic structure among populations
(Rockwell and Barrowclough 1987). Fidelity to wintering sites may increase among
population genetic variation if populations are isolated in wintering areas, or homogenize
gene frequencies across populations if populations are admixed in wintering areas and
pair bond formation occurs on these wintering sites (Robertson and Cooke 1999). Gene
flow that is sex-biased may lead to differences in the degree of among population genetic
variation revealed by maternally and bi-parentally inherited markers (Chesser 1991 a,
Scribner et al. 2001). Information on the degree of genetic discordance among
mitochondrial (mt) DNA haplotypes, and the frequencies of shared and novel haplotypes
and nuclear DNA alleles provides sources of inference to past events and species
ecology. Deeper subdivisions in an intraspecific phylogeny reflect historical
evolutionary splits within a taxa, while shallower molecular separations are evidence of
more recent population subdivisions (Avise 1992).

Phylogenetic investigations have been conducted for a wide variety of North
American avian species (see review in Avise and Walker 1998), including songbirds and
Neotropical migrants (Zink 1996, Kimura et al. 2002), shorebirds (Wenink et al. 1994),
wading birds (Rhymer et al. 2001), and migratory waterfowl (Avise et al. 1992, Quinn
1992, Scribner et al. 2001, Scribner et al. 2003, Pearce et al. 2004, Peters et al. 2005).
Patterns of spatial genetic structure vary greatly among co-distributed avian species (Zink
1996, Avise and Walker 1998). Some species, like the song sparrow (Melospiza
melodia), red-winged blackbird (Agelaius phoeniceus; Zink 1996), tumstone (Arenaria

interpres; Wenink et al. 1994), and king eider (Somateria spectabilis; Pearce et al. 2004)

93

exhibit little genetic differentiation across their geographic ranges, likely as a result of
recent population expansions from a single glacial refugia, and high levels of gene flow
among populations subsequent to the last Pleistocene glacial period. Significant levels of
intraspecifrc genetic variation consistent with population divergence in multiple glacial
refugia has been found in fox sparrows (Passerella iliaca; Zink 1996), Wilson’s warblers
(Wilsonia pusilla; Kimura et al. 2002), rock ptarrnigans (Lagopus mums; Holder et al.
1999), dunlins (Caldris alpine; Wenink et al. 1993), and Canada geese (Branta
canadensis; Scribner et al. 2003). In several of these species (dunlin and Canada goose)
genetic population structure was also attributed strong natal and breeding philopatry
(Wenink et al. 1993; see Chapter 2 for Canada geese). Thus, differences in the degree of
genetic divergence evident in co-distributed North American avian species can result
from both historical processes and species life history characteristics which influence
contemporary levels of gene flow.

Climatic changes over time have likely had a large effect on the phylogeographic
structure of snow geese (Chen caerulescens) and Ross’s geese (Chen rossii), as both
species currently breed in high latitude regions of North America that were recolonized
from refugia in Beringia and other ice-free areas north of major glaciers. Snow geese and
Ross’s geese likely utilized these high Arctic refuges during the last Pleistocene glacial
event (26,000-18,000 ybp; Ploeger 1968), and may have been isolated in separate refirgia
in east Siberia and the Bering Sea area (snow geese), and in the Western Canadian Arctic
Archipelago (Ross’s geese) during the previous glacial maximum (150,000-130,000 ybp).
Within the snow geese complex, greater snow geese (C. c. atlantica) may have occupied

refuge breeding grounds in western Greenland during the last glacial, and were likely

94

geographically isolated from lesser snow geese (C. c. caerulescens) populations (Ploeger
1968). Blue-phase and white-phase lesser snow geese may have also been allopatric
during this period (18,000—26,000 ybp), with blue-phase geese surviving in the Canadian
Arctic Archipelago and white-phase geese inhabiting refuges farther west in Siberia and
the Bering Sea region (Ploeger 1968, Cooke et al. 1988).

In addition to historical vicariant events, the phylogeographic structure of snow
and Ross’s geese has likely been influenced by current levels of gene flow among
populations, degree of philopatry to nesting and wintering sites, breeding behavior, and
recent range expansions and colonizations of new breeding and wintering grounds
(Greenwood and Harvey 1982, Chesser 1991, Anderson et al. 1992, Ely and Scribner
1994, Robertson and Cooke 1999). Female snow and Ross’s geese exhibit a high degree
of natal- and breeding site- fidelity compared to males, who are more likely to disperse
(Rockwell and Cooke 1977, Geramita and Cooke 1982, Cooke 1987, Anderson et al.
1992, Drake and Alisauskas 2005). Significant spatial genetic structure among breeding
populations may result from strong female fidelity, even though populations share
migratory pathways and wintering areas (Ely and Scribner 1994, Scribner et al. 2003a).
Philopatry to wintering grounds may also be critical in defining population structure
among snow and Ross’s geese (Cooke et al. 1988, Robertson and Cooke 1999), as pair
bond formation occurs principally on the wintering grounds where breeding populations
are mixed (Prevett 1972, Cooke 1987, Ryder and Alisauskas 1995, Ganter et al. 2005).
Fidelity to isolated wintering grounds by one or both sexes and associated young,
followed by pair formation, may lead to spatial genetic structure among geese from

different wintering regions rather than nesting areas. Currently, there are four major

95

management groups for snow and Ross’s geese based on their affinities to regional
wintering areas including the Western Arctic Population, the West Central Flyway
Population, the Mid-Continent Population, and the Eastern Population (Figure 1).
Exponential grth of lesser snow geese populations nesting in the Mid-
Continent region (Ankney 1996, Batt 1997, Abraham et al. 1999) and Western Arctic
(Johnson 1995, Johnson 1996, Kerbes et al. 1999), and greater snow geese nesting in the
Eastern Arctic (Ankney 1996) over the past 30 years have led to expansion of population
boundaries in both breeding and wintering areas (Alisauskas 1998). Ross’s geese have
experienced similar range expansions (Fredrick and Johnson 1983) and colonizations of
new breeding areas (Prevett and Johnson 1977, Alisauskas and Boyd 1994) during the
same period. Before 1955, the total population of Ross’s geese (6000 individuals) were
though to be confined to a narrow and well defined area in the Queen Maud Gulf region
(Dzubin 1965, Prevett and MacInnes 1972). An eastward shift of migration pathways
and staging areas were recorded for Ross’s geese by the late 1950’s (Prevett and Johnson
1977, Campbell et al. 1990). Harvest records track the eastward progression of Ross’s
geese in North America. Ross’s geese did not occur in the Central F lyway harvest survey
until 1974, and were absent from harvest surveys in the Mississippi Flyway and Atlantic
Flyway until 1982 and 1996, respectively (Sharp and Moser 1999). Similarly,
populations of blue and white color phases of lesser snow geese may have been allopatric
until the early 20th century (Cooke et al. 1988). Eastward expansions of white geese and
westward expansions of blue geese have been recorded since 1940 (Lewis and Peters

1941, Cooch 1961, Dzubin 1979, Johnson and Troy 1987).

96

 

5'.
I . .'. .' ‘k‘
.’ ~ \". 3’ .
I i ' i': ‘35-“
I ‘ -. ' "-
. ' I I '.
I : a
. I : .~’-- '.
l '3 i K‘-
i “r”- ! "ffl'f‘x
. q I: rag...
! :~~»__—-.-
l -, a; -‘
\ ‘ ‘. .L-é
'\
breeding-snow and Ross’s geese
D breeding-snow geese I
I wintering-snow and Ross’s geese ,E. iv .
E wintering-snow geese ’ off" If ',/
2., z. 5.,
Eastern Population ; Ix ~ 7“. -"' -'.
f) f"

- - - Mid-Continent Population
—— West Central Flyway

-— - —- Western Arctic Population

* Location of Proposed Glacial Refugia

Figure 1. Breeding and wintering distributions and wintering / management population

boundaries for snow geese (Chen caerulescens) and Ross’s geese (Chen rossii).
Locations of Pleistocene glacial refirgia proposed by Ploeger (1968) are indicated.

97

One hypothesis regarding the spatial genetic structure of snow and Ross’s geese is
that present populations are genetically differentiated in a manner reflecting historical
biogeographic isolating events proposed by Ploeger (1968). A second hypothesis is that
contemporary population expansion events have been extensive and may mask any
genetic signature of historical differentiation that may have occurred within and among
species. In the absence of genetic variation attributed to historical processes, snow and
Ross’s geese may still exhibit genetic structure based on population affinities to breeding
and wintering grounds. Several studies have focused on population structuring among
breeding populations of snow geese and Ross’s geese using matemally-inherited mtDNA
data. Avise et al. (1992) found two phylogenetically distinct groups of haplotypes
(designated clades I and II) which co-occurred in lesser snow geese and Ross’s geese
from different regions of their respective breeding and wintering distributions.
Haplotypes of both clades were found in snow and Ross’s geese from multiple
geographic locales, both sexes, and among both color morphs of lesser snow geese.
Clade frequencies were highly variable between snow geese (60% clade I, 40% clade II)
and Ross’s geese (10% clade I, 90% clade 11) species , but were similar among
populations within species. The deep phylogenetic split between clades without
geographic localization was hypothesized to be a result of secondary hybridization and
introgression among snow geese and Ross’s geese from two allopatric populations that
evolved during the Pleistocene, and from which geese subsequently dispersed and mixed
widely (Avise et a1. 1992). Quinn (1992) also observed the presence of two distinct
mtDNA clades distributed across the lesser snow goose subspecies range (Ross’s geese

were not included) without detectable correlation with color-phase or sex of snow geese.

98

Quinn (1992) did, however, find a significant difference in clade frequencies between
geese nesting in the western Arctic as compared to geese nesting in Hudson Bay. Based
on an estimated 6.7% sequence divergence between clade I and clade II lineages, Quinn
(1992) hypothesized that two formerly allopatric populations dispersed across the modern
range of the lesser snow goose, likely during Pleistocene interglacial periods.

Weckstein et al. (2002) reanalyzed Quinn’s (1992) data set including several new
mtDNA sequences from lesser snow and Ross’s geese, and identified two divergent
lineages similar to Quinn’s (1992) findings. Weckstein et al. (2002) also found
significant genetic differences in clade frequencies between western and eastern nesting
lesser snow geese, and concluded sharing of the two divergent haplotype lineages across
species and geographies were consistent with two hybridization episodes between lesser
snow and Ross’s geese. Weckstein et a1. (2002) further stated observed spatial variance
in the frequency of clade II was consistent with Cooke et al.’s (1988) hypothesis that
formerly allopatric populations of blue- and white-phase lesser snow geese had recently
hybridized due to a change in winter feeding habits that caused both morphs to meet and
pair on common wintering grounds. Weckstein et al. (2002) also hypothesized that clade
11 sequences were originally carried by lesser snow geese, while clade I sequences were
originally carried by Ross’s geese.

In each of these studies, conclusions were based on gene trees derived from a
single locus (mtDNA) and relatively few individuals per locale. Our study builds upon
previous work by employing multiple genetic markers (nuclear DNA microsatellites and
mtDNA) that differ in their modes of inheritance and rates of evolution. We have

increased the number of populations included in the analysis, and we sampled a large

99

number of individuals from each location. We examine the variation of bi-parentally
inherited microsatellite loci and site substitutions of mtDNA sequence data from the
maternally inherited mitochondrial cytochrome b gene to estimate the degree of
population structuring between species of snow and Ross’s geese, between subspecies of
snow geese, between color morphs of lesser snow geese, and among breeding and
wintering populations of snow geese and Ross’s geese throughout the geographic range
of both species. Genetics data is used to characterize and interpret the phylogenetic
status of species, subspecies, and color phases of geese, and to predict the relative

importance of historical and contemporary factors influencing population structure.

METHODS

Sample Collection

Samples were collected from breeding populations of lesser snow geese (Chen
caerulescens caerulescens), greater snow geese (C. c. atlantica) and Ross’s geese (C.
rossii) during brood rearing, when geese were flightless. Geese were guided into catch
nets using a helicopter or by walking (Cooch 1953, Tirnm and Bromley 1976). Blood or
a blood quill (growing feather) was sampled from breeding adults or goslings at each
location. Samples were placed into individual tubes containing high-salt buffer and
stored at ambient temperatures in the field until frozen in the laboratory. White phase
lesser snow geese were collected from Kolyma River Delta, Japan, Wrangel Island,
Russia, Alaska’s North Slope, Howe Island, Banks Island, Queen Maud Gulf Bird
Sanctuary, Southampton Island, West Hudson Bay, LaPerouse Bay, Cape Henrietta

Maria, Akimiski Island, and Baffin Island (Figure 2). Blue phase lesser snow geese were

100

 

Figure 2. Sampling locations for snow geese and Ross’s geese breeding populations: 1 =
Wrangel Island, Russia; 2 = North Slope, Alaska; 3 = Howe Island; 4 = Banks Island; 5 =
Queen Maud Gulf Bird Sanctuary; 6 = Southhampton Island; 7 = West Hudson Bay; 8 =
LaPerouse Bay; 9 = Cape Henrietta Maria; 10 = Akimiski Island; 11 = Baffin Island; 12
= Bylot Island. Sampling site not pictured: Kolyma River Delta, Japan. White phase
lesser snow geese (Chen caerulescens caerulescens) were collected at locations 1-11 and
the Kolyma River Delta. Blue phase lesser snow geese were collected at locations 5, 7, 9,
10, and 11. Greater snow geese (C. c. atlantica) were collected at location 12. Ross’s
geese (C. rossii) were collected at locations 5, 6, 7, 9, and 11.

collected concurrently at a portion of these locations including Queen Maud Gulf Bird
Sanctuary, West Hudson Bay, Cape Henrietta Maria, Akimiski Island, and Baffin Island.

Bylot Island was sampled for greater snow geese. Samples of Ross’s geese were

101

collected from Queen Maud Gulf Bird Sanctuary, Southhampton Island, West Hudson
Bay, Cape Henrietta Maria, and Baffrn Island (Figure 2). Sample sizes for each location

are listed in Table 1.

Characterization of Microsatellite Loci

DNA was extracted from all samples using DNeasy extraction kits (Qiagen Inc.,
CA). Twenty nuclear microsatellite loci were initially screened for allelic variation.
Nine bi-parentally inherited loci proved to polymorphic in one or more breeding
populations and were used for subsequent analyses. Loci used included Bcaul , BcauS,
Bcau9, Bean] 1, Hhiul, Hhiu3 (Buchholz et al. 1998); Aalul (Fields and Scribner 1997);
SfiulO (S. Libants, unpubl. data); and CR-G (A. Baker, unpubl. data). Each locus was
amplified using polymerase chain reaction (PCR) in 25 pl reaction volumes, including
100-150 ng DNA, 10-25 pmol of each primer, PCR buffer (10 mM Tris-HCl pH 8.3, 50
mM KCl, 100 ug/mL gelatin, 0.01%NP-40, 0.01% Triton-X 100), 0.5 U of AmpliTaq
DNA Polymerase (Perkin-Elmer), and 100-200 uM dNTPs. Forward primers of each
locus-specific primer pair were labeled with either Hex or F luorescein by the
manufacturer (IDT Technologies, Inc.). Most thermocycler conditions included a
denaturing step of 94 °C for 2 min, followed by 30-35 cycles of 94 °C for 1 min,
annealing temperature for l min [51 °C (I-Ihiul, SfiulO), 56 °C (Bcaul, Bcau9, I-Ihiu3,
CR-G), 58 °C (Bcaul 1), 60 °C (Bcau5)], and 72 °C for l min. Conditions for Aalul
included a denaturing step of 94 °C for 2 min, and 30 cycles of 94 °C for 1 min and 50 °C

for 2 min. Products were visualized using a F MBIO II laser scanner (Hitachi Software

102

Table 1. Genetic diversity measures estimated for snow geese and Ross's geese
populations. The species of each population sampled is designated as S = snow = Ross's
geese or R geese. The color phase of each population sample is designated as W = white

 

 

or B = blue.

Color
Breeding Location Species Phase Population n A Ho Hc F .5 PB
Akimiski Island S B AKBL 58 2.64 0.512 0.530 -0.04 0.23
Baffin Island S B BFBL 53 2.74 0.504 0.487 0.04 0.53
Cape Henrietta S B CHBL 33 2.75 0.503 0.488 0.03 0.32
Maria
West Hudson Bay S B WHBL 53 2.64 0.482 0.470 0.02 0.47
Queen Maud Gulf S B QMBL 15 2.61 0.521 0.508 0.03 0.53
Akimiski Island S W AKWI-I 24 2.68 0.480 0.477 0.01 0.85
Baffin Island S W BFWH 41 2.69 0.521 0.483 0.07 0.53
Cape Henrietta S W CHWH 58 2.38 0.482 0.450 0.07 0.47
Maria
LaPerouse Bay 8 W LPBA 48 2.59 0.503 0.511 -0.02 0.19
South Hampton S W SHWH 54 2.54 0.501 0.479 0.04 0.10
Island
West Hudson Bay S W WHWH 40 2.79 0.510 0.476 0.07 0.68
Queen Maud Gulf S W QMWH 54 2.44 0.493 0.498 -0.01 0.02*
Howe Island S W HOWE 50 2.55 0.508 0.496 0.02 0.50
North Slope, S W NSWH 34 2.67 0.499 0.422 0.16“ 0.53
Alaska
Banks Island S W BKIS 81 2.58 0.493 0.485 0.02 0.85
Wrangel Island (N) S W WRNO 76 2.65 0.503 0.509 -0.01 0.90
Wrangel Island (S) S W WRSO 85 2.62 0.503 0.490 0.03 0.75
Kolyma, Japan S W KOLY 3 l 2.49 0.491 0.470 0.04 0.41
Bylot Island S' W BYIS 61 2.52 0.512 0.522 -0.02 0.10
Baffin Island R W BFRO 22 2.98 0.535 0.519 0.03 0.88
Cape Henrietta R W CHRO 52 2.66 0.516 0.540 -0.05 0.50
Maria
South Hampton R W SHRO 40 2.56 0.533 0.509 0.05 0.55
Island
West Hudson Bay R W WHRO 83 2.79 0.527 0.503 0.05 0.88
Queen Maud Gulf R W QMRO 53 2.78 0.518 0.520 -0.01 0.75

 

n = sample size per population; A = mean allelic richness over 9 loci; Ho = observed heterozygosity; Hc =

expected heterozygosity; F15 = correlation of genes within individuals within populations as a test for

deficit of heterozygotes (P < 0.006 after correction for multiple loci; represented by an asterisk) indicating
possible inbreeding; P3 = P-value from one-tailed tests for heterzygote excess, with significant values (P <

. . . . . a
0.05; represented by an asterisk) indicating a recent population bottleneck. Greater snow geese (Chen
caerulescens atlantica); all other populations of snow geese are lesser snow geese (C. c. caerulescens).

103

Engineering Co.) after electrophoresis on denaturing 6% acrylamide gels. Genotypes

were scored based on 20 base-pair standards and reference samples of known allelic size.

Characterization of MtDNA

Approximately 50-100 ng DNA was used for the initial mtDNA amplification
(1173 bp) with Quinn’s (1992) primers (16775L and 287H-M) and the PCR protocol of
Kocher et al. (1989). Thermocycler conditions included initial denaturation step of 92 °C
for 2 min, followed by 40 cycles of 92 °C for 40 s, annealing at 61 °C for 2 min, 72 °C
for 2 min, and extending at 72 °C for 7 min. A restriction endonuclease, Alul (Quinn
1992), was used to test for the presence or absence of a specific restriction site within the
mtDNA control region amplified for all geese sampled. Quinn (1992) and Avise et al.
(1992) demonstrated this restriction enzyme cleaves the DNA within one group of snow
geese haplotypes (clade I) in this mtDNA region, while leaving the remaining haplotypes
intact (clade II), with sequence divergence between these two mtDNA clades averaging
6.7% (Quinn 1992). Amplified mtDNA was digested with Alul according to
manufacturer protocol (New England Biolabs, Inc.), and visualized using ultraviolet light

after electrophoresis on 1% agarose gels stained with ethidium bromide.

Gene Diversity and Population Structuring

For each population, observed genotype frequencies for each of the nine
microsatellite loci were tested for departure from Hardy-Weinberg expectations (ensures
population genotype frequencies can be estimated based on estimates of allele

frequencies) were implemented in the program FSTAT (Goudet 2001). Tests for

104

genotypic linkage disequilibrium (a measure of independence across loci within a
population) were performed as described in Goudet et al. (1996) using FSTAT. The
FSTAT program was also used to estimate degree of spatial heterogeneity in gene
frequency within and among snow geese populations using hierarchical F -statistics (Weir
and Cockerham 1984, Weir 1996) at three levels: (1) among individuals within
populations (f), (2) among individuals within the total population (F), and (3) among
populations (0). Significance of F -statistics was based on 95% confidence intervals
determined by bootstrapping across loci. Confidence intervals that included zero were
considered non-significant. Pair-wise estimates of population F ST were used as summary
measures of inter-population variance in allele frequency. Significance of pair-wise
interpopulation differentiation was determined using the exact G-test (Goudet et al. 1996)
in FSTAT, as the G-test is more powerful than exact F ST-estimator tests for diploid
populations (Goudet et al. 1996, Petit et al. 2001). For tests of Hardy-Weinberg, gametic
disequilibrium, and F-statistics, nominal significance levels (alpha) were adjusted to
account for multiple testing using sequential Bonferroni corrections (Rice 1989).
Rapidly evolving markers like microsatellites are useful for detecting relatively
recent population bottlenecks that may be indicative of founder events based on a low
number of individuals (Comuet and Luikart 1996). The program BOTTLENECK
(Comuet and Luikart 1996) was used to detect bottlenecks that may have resulted from
natural colonization events associated with the recent range expansions of snow geese
and Ross’s geese populations. BOTTLENECK tests for significant excess in
heterozygosity within each population. This excess is apparent when populations have

experienced a recent genetic bottleneck causing the number of alleles to decrease faster

105

than levels of heterozygosity. A one-tailed Wilcoxon sign-rank test was used to test for
heterozygosity excess under a two-phase mutation model (TPM), as this is the
recommended model for microsatellite data (Comuet and Luikart 1996). The TPM
model was run four times with 80%, 85%, 90%, or 95% of the mutations occurring in one
step, with the remaining proportion of the mutations occurring in multiple steps (variance
= 30).

Microsatellite allele frequencies across all nine loci were used to estimate Cavalli-
Sforza and Edwards (1967) chord distances among all 24 populations of snow and Ross’s
geese. A multilocus population neighbor-joining tree based on interpopulation distances
was constructed using the program PHYLIP (version 3.6; Felsenstein 1993). Chord
distances have been shown to produce robust tree topologies over a range of conditions
(Takezaki and Nei 1996). Statistical significance of tree branches was based on 1000
bootstrapped population trees constructed using PHYLIP.

An analysis of molecular variance (AMOVA; Excoffrer et a1. 1992) was used to
assess among population variation ((DST ) in mtDNA clade frequencies and nDNA allele
frequencies. Hierarchial AMOVA contrasts were used to partition genetic variance
among populations within groups ((Dsc), among groups ((IJCT) and among populations
among groups ($31) in the program ARLEQUIN (version 2.0, Schneider et al. 2000).
Hierarchical AMOVAs were conducted to exam several hypotheses related to the
potential spatial genetic structure of snow and Ross’s geese based on life history
characteristics (Ely and Scribner 1994) and hypothesized Pleistocene refugia (Ploeger
1968) of both species. First, we hypothesized that genetic variation would be greater

between species of snow and Ross’s geese and between subspecies of lesser and greater

106

snow geese than among populations within species and among populations within
subspecies. Secondly, we hypothesized that genetic variation would be greater between
color-phases of lesser snow geese than among populations of the same color-phase.
Finally, we hypothesized that genetic variation may be greater among populations of
snow and Ross’s geese wintering in different regions as compared to populations
wintering in the same area. Hierarchical AMOVAs were conducted for (1) snow geese
compared to Ross’ geese ; (2) lesser snow geese compared to greater snow geese; (3)
white-phase snow geese populations, blue-phase snow geese populations, and Ross’s
geese populations representing three different groups; (4) white-phase snow geese
compared to blue-phase snow geese; and wintering populations of (5) snow and Ross’s
geese, (6) white-phase and blue-phase snow geese, and (7) white-phase snow geese.
Breeding populations were assigned to regional wintering groups based on where the
majority of each breeding population winters annually. Wintering populations included
the Eastern Population represented by greater snow geese from BYIS; the Mid-Continent
Population consisting of AKBL, BFBL, CHBL, WHBL, QMBL, AKWH, BF WH,
CHWH, LPBA, WHWH, and SHWH lesser snow geese populations, and BF RO, CHRO,
SHRO, and WHRO Ross’s geese populations; and the West Central F lyway and Western
Arctic Populations including lesser snow geese from NSWH, HOWE, BKIS, WRNO,

WRSO, and KOLY, and Ross’s geese from QMRO (Figures 1 and 2).

RESULTS

Genetic Variation

107

Allelic variation for the nine nuclear microsatellite loci ranged from 3 to 26
alleles. The number of alleles in each p0pulation was similar for seven of the nine
microsatellite loci. For the remaining two loci, Bcau5 and CR-G, allelic variation was
generally higher in Ross’s geese populations as compared to snow geese populations
(Table 1). Mean allelic richness (mean number of alleles independent of sample size to
allow for comparison among population with different sample sizes; El Mousadik and
Petit 1996, Petit et al. 1998) over the nine loci was similar across all populations (2.38 —
2.98), as were expected (0.422 — 0.540) and observed (0.480 — 0.535) heterozygosity
values (Table 1). Only one locus (SfiulO) exhibited a departure from Hardy-Weinberg
expectations, and no evidence of linkage disequilibrium was observed. Inbreeding
coefficients (F13) were low and non-significant for all populations except for lesser snow
geese sampled from the North Slope of Alaska (F rs = 0.156; Table 1). Tests for
significant excess in heterozygosity indicated the presence of a possible population
bottleneck in only one population of white-phase lesser snow geese (QMWH, PB = 0.019;

Table l).

Phylogenetic and Population Relationships

Estimates of genetic variation based on the nine biparental microsatellite loci
were low for individuals within populations (f = 0.022), but showed a significant
difference in allele frequencies among populations (9 = 0.035) and for individuals across
the total population (F = 0.056) of snow geese and Ross’s geese (Table 2). When snow
geese and Ross’s geese populations were analyzed separately, snow geese exhibited

significant among population differences (0 = 0.008), but Ross’s geese among

108

Table 2. Estimates of hierarchical F-statistics (Weir
and Cockerham 1984) for populations of snow geese
and Ross's geese based on nine microsatellite loci.
Variance partitioning: f = alleles within individuals;
F = among individuals within the total population;

0 = among populations. Statistically significant (P <
0.01) values over all loci are marked with an asterisk.

 

Locus f F 0

 

Snow and Ross's Geese Populations (n = 24)

Aalul 0.002 0.035 0.033
Bcaul 0.016 0.058 0.044
Bcau9 0.002 0.046 0.044
BcauS -0.063 0.01 0.068
CR-G -0.016 0.026 0.041
Hhiul 0.005 0.042 0.037
Hhiu3 0.050 0.081 0.033
SfilOp. 0.102 0.119 0.019
Bean] 1 0.004 0.007 0.003
All Loci 0.022 0056* 0035*
Snow Geese Populations (n = 19)

All Loci 0.023 0.031 0008*
Ross's Geese Populations (n = 5)

All Loci 0.015 0.019 0.004

 

populations variance was not significant. All pairwise estimates of variance among
populations (0p) were highly significant between populations of lesser snow geese and
Ross’s geese (mean 0p = 0.083). Subspecies comparisons between the Bylot Island
population of greater snow geese and all breeding populations of lesser snow geese were
also significant (Table 3). Mean pairwise estimates of variance were greater between
populations of greater and lesser snow geese (mean 0p = 0.016) than among populations
of lesser snow geese (mean 0p = 0.007). Pairwise 0p values between white-phase and

blue-phase geese (mean .0p = 0.007) were on average greater than pairwise values

109

 

=od >40!

n—od m. o.o Omm3
Sod ooo.o ooo.o OZm3
omod omod mood mood mU—m
n—od wood. o—od m—od nmod I3m2
o—od mood. :od o—od Sod ooo.o m30:
m—od wood .ood. mood o—od ooo.o wood I320
o—od wood. o—od o—od Sod ooo.o mood. ooo.o 13:3
n—od ooo.o m—od Sod mmod ooo.o mood ooo.o ooo.o I3Im
o—od mood n—od o—od Sod ooo.o wood hood wood. wood <9:
m—od Sod. mood mood wood ooo.o mood ooo.o _ood mood ooo.o I310
o—od wood o—od _ .od Sod wood mood hood mood. wood wood mood 13mm
~mod mood. m. _ o.o o.od m _o.o wood. ooo.o mood ooo.o wood mood mood. ooo.o I3¥<
omod mood mmod mmod _no.o ooo.o _oo.o mmod _ood _oo.o. Edd wood wood mood .520
m—od ooo.o w—od o—od hmod mood mood =od wood. _ood mood hood o—od .ood mood .523
add .ood o—od o—od mmod hood mood o—od wood wood _ Ed Edd hood hood wood. m—od Ammo
m—od Sod .ood mood o.o.o Sod Qod _ood mood =od wood wood wood wood 2 o.o _ 8d omod 4min
n—od mood .ood. Sod. wood wood Sod ooo.o mood. wood mood mood. .ood mood ooo.o mood wood .ood- 4m¥<

m_>m >402 Om¢3 OZM3 mU—m I3mZ M30: _ 3320 23:3 I3Im <9: 3:0 13mm 3% _ Amos/5 .523 .520 495

 

Beam muons—zoom use? 5883\333.» 3950 803 confisooo “55:80-22 BEBE E33592:
a
5320 830 Beam 833 880 Beam 0:53 880 Beam 02m

 

._ 2an E 36: we 03 223305“? coon—soon awo— ooflw 62 032:8 8o "858:8 Ecobocom Boa Emomfiwwm.
bioumufim 203 mos—g Eom A238 UV 3on Mmmom dam AGNES?» .o .3 88m 268 862w .szmomfizxoao .o UV 08% 265 882

829823 one -093 So coo— :m 388 no 525 momocosooo 233 E 859? zoom—snoododfi do 33:58 3353 .m 2an

110

Table 3 (cont’d).

 

b
Ross’s Geese

 

CHRO SHRO WHRO QMRO
BFRO 0.003 0.001 0.006 -0.002
CHRO 0.006 0.0“ 0.009
SHRO 0.002 -0. 003
WHRO 0.000

 

a . .
Greater snow geese represent the Eastern wmtenng
Population.

bBFRO, CHRO, SHRO, and WHRO are part of the Mid-
Continent wintering Population; QMRO is part of the
West Central F lyway and Western Arctic Populations.

between white-phase populations of geese (mean 0p = 0.006), and less than pairwise
values between blue-phase populations of geese (mean 0p = 0.008). Differences in allele
frequencies between blue-phase and white-phase lesser snow geese populations breeding
in the same geographic locale were only significant for the two populations from Queen
Maud Gulf Bird Sanctuary (0p = 0.022). Variation in allele frequency was greater
between populations of snow geese known to winter in different regions (mean 0p =
0.0094) than between populations within the same wintering group (mean 0p = 0.0056;
Table 3). Pairwise comparisons revealed Ross’s geese from Cape Henrietta Maria
(CI-IRO) had significantly different allele frequencies from Ross’s geese in West Hudson
Bay, Queen Maud Gulf Bird Sanctuary, and Southampton Island (Table 3). Allele
frequencies were similar for all other Ross’s geese population pairs.

Populations of snow geese and populations of Ross’s geese formed two distinct
clusters within the consensus neighbor-joining tree based on chord distances derived
from bi-parental loci (Figure 3). Bootstrap support for the tree node separating the two

species was 100%. Significant genetic differentiation between greater snow geese and

Ill

BIS

BFWH WR o WRNO
\BFBL LPBA 685
QMBL ; QMWH
HOW ' AKW H
KOLY

 

CHBL NSWH

WHBL

BYIS

WHRO

CHRO

 

Figure 3. Neighbor joining tree describing overall genetic similarities among populations
of snow and Ross’s geese based on Cavalli-Sforza and Edwards (1967) chord distances
across nine microsatellite loci. Bootstrap support (>50% out of 1000 replicates) for the
consensus microsatellite tree is indicated. Population abbreviations are as in Table 1.

112

lesser snow geese populations was also evident based on the tree topology, although
statistical support for differentiation between snow geese subspecies was less than
between species. The only genetically distinct group within the lesser snow geese cluster
included the BKIS, WRNO, and WRSO breeding populations from the Western Arctic
wintering population (Figure 3). Lesser snow geese populations wintering in other
regions did not cluster together on the neighbor-joining tree, and no significant
phylogeographic pattern was evident among lesser snow geese populations breeding in
close geographic proximity or among lesser snow geese populations of the same color
phase (Figure 3).

Among population differences in mtDNA clade frequencies and nDNA allele
frequencies were significant across all breeding populations of snow geese and Ross’s
geese (mtDNA (DST = 0.093, nDNA (DST = 0.035; Table 4). Among population
differences in clade frequencies were significant for snow geese ((DST = 0.05 8), but not
significant for Ross’s geese ($5; = -0.006), when species were considered separately.
Clade frequencies were equivalent among populations of blue-phase lesser snow geese
((DST = 0.029), but differed significantly among populations of white-phase lesser snow
geese ($51 = 0.059). Hierarchical analyses of variance based on maternally and bi-
parentally inherited DNA frequencies revealed significant levels of genetic variation
among populations within groups ((DSC) when population groups were designated by
species, subspecies, color phase of lesser snow geese, or wintering region (Table 4).
Genetic variation between snow and Ross’s gees species was significant, and
proportionally larger, than variance among populations within species. In contrast,

among population estimates of genetic variance ((Dsc) were greater than among group

113

3:85am 885B aezaaaea zone Eeo £86.sz we $.5an 8: 82B ae

deman— wdaeam wa_._8a_B 2323.. 8 doamfima 80B macaw—anon mEduoam deg—amen. E8mmm 2: AS dad .aeaflaaen— 38550.22 8: Amy macaw—amen
zaBbm 3850 883 tax o:8< E833 8: A: mafia—8: mqaeaw 3:: SB ova—a 80B 08% Mame”. dam Beam Leo maemtaaaaeo 5:2:er wat8£3a

 

 

.23. 255 .83 .83 .86 .83. m 3 223.38 wcassB 880 38m 085223 825

odd wood flood .. _ flod .mod *owod m o. maeoadaem mat8£3 3.80 Beam maeE<

.33 Sod- .085 $8.0 See. .820 m 4m “28238 wsaasB 325 08a 2a 325 825.

8225; \e 388880 e .3 aeceoamek MSLSSE

flood ooo.o- .mood flood w _o.o .mmdd m 2 emee0 Beam omega-02m .0800 Beam cacao-823 aooBaom

.wmod flwod flood flm _ .o *mmod .mwod m wm emee0 38.». .0800 Beam BEE-8am .380 Beam omega-0:3 maeE<

mutate; \e 3883290 3 8 3.30 Beam-\e 8.8a .630

agod mood flood *omod omod- Lood m 2 emeo0 Beam 5380 6800 Beam 883 aoeBaom

.mwod *oflod flood flm: .o .32 .o .. _ mod m wm 880 38¢ .380 Beam 525:5

8223: \e areaeQEe-O a we amoeemaoam Le autumn

1838883

nswood aa aa ooo.o- aa aa _ m maeza—aoem 880 Mame-m maeE<

uemood ma ma omod aa aa _ w Beta—amen 830 Beam guano-cam maeE<

smood aa aa iomod aa aa _ E Sena—age.— omoo0 Beam omega-ea; maeE<

.mood «a ma ammod aa aa _ 2 macaw—amen 380 Beam =< waeE<

.mmod aa aa {mood aa a: g wm maeoflaqem 0800 Named dam Beam =< maeE<

.3an88?an

.56 59 86 $6 .59 86 3:20 maem 888an
<ZDa <75:a me a me u

 

3883 am 5B doe—.88 Ba 902 29:38

So maezeobee Boa <75: .8.“ wood v k ”(.298 So mod v a: 82? aaaeEammm bfieomofim Ahmed membaaa 30208.82

aeo moaeam waeaaa Beta—amen waeaaa dam mama—mam BoanaaoE-aea .8.“ macaw—amen waeaa .fieev moaeam maeEa domed museum
aaEB macaw—amen maeEa ”maEemamvao oeaaia> .momeaoaoeo ova—e SAGE ae woman eeaata> me 838:3 302888: .w 03am.

114

variance estimates ((DCT), for subspecies, lesser snow geese color phase, and wintering

population contrasts (Table 4).

DISCUSSION

Snow geese and Ross’s geese may have been isolated in separate refugia in east
Siberia and the Bering Sea area (snow geese), and in the Western Canadian Arctic
Archipelago (Ross’s geese) during the previous glacial maximum (150,000-130,000 ybp).
Contemporary populations of both species often nest within the same colonies, and are
admixed on wintering grounds during pair formation (Ryder and Alisauskas 1995,
Mowbray et al. 2000) providing opportunity for species introgression. Molecular
genetics data from bi—parentally and maternally inherited markers for snow geese and
Ross’s geese revealed considerable genetic differentiation between these species. The
significant genetic divergence in mtDNA and nDNA frequencies is indicative of a
historic divergence between snow geese and Ross’s geese. Historical genetic differences
between species may be maintained by differences in courtship and breeding behaviors
(Palmer 1976, Prevett and MacInnes 1980, Owen 1980, Cooke et al. 1995) and
morphology (Bellrose 1980, Owen 1980), rather than to geographic isolation on breeding
and wintering grounds. Our data demonstrates that interbreeding between species is
likely infrequent and interspecies genetic exchange is low, although mixed pairs and
hybrids have been known to occur (Trauger et al. 1971, Hatch and Shortt 1976, Cooke et
al. 1995).

Previous studies based on the mtDNA characteristics of snow and Ross’s geese

have offered two competing hypotheses about how the fundamental phylogenetic split

115

between clade I and clade II haplotypes had evolved. One hypothesis proposes that clade
I haplotypes originated in lesser snow geese and clade II haplotypes originated in Ross’s
geese, and the widespread occurrence of both clades in snow and Ross’s geese
p0pulations is a result of dispersal and secondary hybridization between the two species
(Avise et al. 1992, Weckstein et al. 2002). The second hypothesis proposes that the
divergence of clade I and clade II haplotypes occurred prior to speciation of snow and
Ross’s geese, and subsequent divergence in haplotype frequencies between the two
species (Quinn 1992). Our analyses revealed populations of Ross’s geese have 73 — 83%
clade II haplotypes, whereas 18 of the 19 populations of snow geese have relatively
equivalent proportions of clade I and clade II haplotypes or have greater proportions of
clade I haplotypes. The northern wintering population of lesser snow geese breeding on
Wrangel Island, Russia (WRNO) exhibited a much lower frequency of clade I haplotypes
than all other snow geese populations. The small proportion of clade I birds present
within the WRNO population is more consistent with the clade I frequencies documented
for Ross’s geese populations (0.10, Avise et al. 1992; 0.30, Weckstein et al. 2002; 0.18 —
0.27, Shorey unpublished data) than for snow geese populations. WRNO snow geese
have been isolated historically and during contemporary periods from Ross’s geese
populations in North America based on putative Pleistocene refugia locations (Ploeger
1968) and known migratory pathways and wintering sites (Ryder and Alisauskas 1995,
Mowbray et al. 2000). Given this geographic isolation, low frequencies of clade I
haplotypes in the WRNO population similar to Ross’s geese populations suggests that the
evolutionary split between mtDNA clades occurred prior to speciation of snow and

Ross’s geese, supporting Quinn’s (1992) evolutionary hypothesis.

116

Unlike Ross’s geese, greater snow geese have minimal contact with lesser snow
geese on breeding or wintering areas. Greater snow geese breed at higher latitudes in the
northeast Arctic than most lesser snow populations (Reed and Chagnon 1987), and winter
along the Atlantic Coast from Massachusetts to South Carolina in groups composed
exclusively of greater snow geese (V eit and Petersen 1993, Reed et al. 1998). Both
historical and contemporary geographic isolation between subspecies, and morphological
differences in body size (greater snow geese are generally larger than lesser snow geese)
between subspecies may limit potential pairing and gene flow between greater and lesser
snow geese (Cooke et al. 1995, Reed et al. 1998). Genetic differentiation between greater
and lesser subspecies of snow geese was low based on differences mtDNA clade
frequencies, but was significant based on variation in microsatellite allele frequency.
This suggests historic rates of gene flow among subspecies may have been greater than
contemporary levels of genetic exchange. Hierarchical analyses of genetic variance
indicated variation among populations within subspecies was greater than the level of
variation between greater and lesser snow geese. However, our analyses only included
one population of greater snow geese (BYIS), indicating that the higher degree of genetic
variation within subspecies as compared to among subspecies could be attributed to
differences among populations of lesser snow geese. Characterizing additional
populations of greater snow geese, and sequencing mtDNA to determine individual
haplotypes within greater and lesser snow geese may offer greater resolution of potential
genetic variation between subspecies that we were not able to resolve based on clade

frequencies.

117

Before 1955, total numbers of Ross’s geese were estimated at 6000 individuals,
and were though to be confined to a narrow and well defined area in the Queen Maud
Gulf region (Dzubin 1965, Prevett and MacInnes 1972). An eastward shift of migration
pathways and staging areas were recorded for Ross’s geese by the late 1950’s (Prevett
and Johnson 1977, Campbell et al. 1990). Continued growth in the population numbers
and eastward expansion of the species has led to new colonies on the west coast of
Hudson Bay and on Southampton and Baffin Islands (Ryder and Cooke 1973, Prevett and
Johnson 1977, Frederick and Johnson 1983), and migrations to new wintering areas
(Ryder and Alisauskas 1995). Current populations include 1 million breeders in the
Queen Maud Gulf region, 80,000 nesting on the west coast of Hudson Bay, and smaller
(but unknown) numbers of individuals on the islands (R. Alisauskas, personal
communication). Similarity in mtDNA clade frequencies among Ross’s geese breeding
populations support the hypothesis that Ross’s geese were previously one panmictic
group prior to recent population expansions, and is concordant with earlier mtDNA
studies (Avise et al. 1992, Weckstein et al. 1992).

The majority (90%) of Ross’s geese from Queen Maud Gulf (QMRO) winter in
the Central California Valley as part of the Western Arctic Population. The remaining
Ross’s geese from Queen Maud Gulf and the other four populations of Ross’s geese in
our study migrate south with the Mid-Continent Population of snow and Ross’s geese. If
Ross’s geese have strong fidelity to wintering areas, then genetic differences may accrue
as a function of winter fidelity and time since colonization between the QMRO
population and the four other populations wintering in a different region. We found

significant genetic variation in allele frequencies between Ross’s geese nesting in Cape

118

Henrietta Maria (CHRO) and all other Ross’s populations based on pairwise 0p
comparisons and microsatellite chord distances. Differences in chord distances were also
evident between Ross’s geese breeding in west Hudson Bay (WHRO) and the remaining
populations of Ross’s geese. Gene frequencies were similar among QMRO, and Ross’s
geese on Southampton and Baffin Islands. Genetic differentiation between Ross’s geese
nesting in west Hudson Bay and Cape Henrietta Maria coastal populations compared to
the putative ancestral population in Queen Maud Gulf and Southampton and Baffin
Islands populations suggests the WHRO and CHRO populations may have been
colonized by geese from Queen Maud earlier than the island populations. Alternatively,
founding population sizes may have been smaller in WHRO and CHRO than on the
islands, resulting in population bottlenecks that changed allele frequencies in WHRO and
CHRO relative to the Queen Maud population. We found no evidence of population
bottlenecks or lowered allelic diversity in the WHRO and CHRO populations relative to
other Ross’s geese. Thus, the WHRO and CHRO populations were likely colonized
earlier than Southampton and Baffin Islands, and genetic drift has increased genetic
variation between WHRO and CHRO and Ross’s geese from Queen Maud. Future
genetic divergence between the Queen Maud population and the other four recently
established populations may increase over time as a function of wintering fidelity to
different wintering areas.

Banding studies indicate lesser snow geese fidelity to breeding grounds is female-
biased (70-80% annual return), with males more prone to dispersal (SO-66% annual
return) (Cooke and Sulzbach 1978, Cooke and Abraham 1980, Ganter and Cooke 1998).

High male dispersal rates from natal breeding areas increases potential gene flow and

119

decreases spatial genetic structure among breeding populations. Phylogenetic structure at
the level of breeding populations may also be diminished by the formation of breeding
pairs on the wintering grounds or during spring migration when populations are mixed
(Prevett 1972, Cooke 1987, Ganter et al. 2005). Thus, philopatry to wintering grounds
may be more influencial in determining population structure in lesser snow geese than
fidelity to breeding sites (Cooke et al. 1988, Robertson and Cooke 1999). Estimated
annual fidelity to wintering areas varies among wintering regions, and is very high for
male and female lesser snow geese in the Western Arctic wintering population (96-98%
return in subsequent year; Armstrong et al. 1999, Baranyuk et al. 1999, Williams et al.
2005), and relatively lower for lesser snow geese in the Mid-Continent wintering
population (43-72%; Alisauskas 1998).

If winter fidelity is an important factor influencing genetic structure of snow
geese, we would expect a higher degree of genetic variation among wintering populations
than among breeding populations sharing the same wintering grounds. Hierarchical
analyses of mtDNA and nDNA variance indicated genetic differences were greater
among breeding populations within a wintering group than among wintering groups. The
hierarchical analysis calculated the proportional variance from with a “bottom-up”
approach, beginning with the genetic variation among individuals. Both the mtDNA
control region and microsatellite loci utilized for this analysis are highly polymorphic,
and the high level of variation among individuals may make it difficult to resolve
variation among the top hierarchical contrasts (i.e., among wintering groups). Pairwise
comparisons of population 0p values revealed that mean pairwise population differences

were twice as large between lesser snow geese from different wintering areas as

120

compared lesser snow geese from the same wintering population. In addition, one
wintering population was statistically supported among lesser snow geese in the
population distance tree. The cluster was composed of three populations (WRNO,
WRSO, BKIS) representing the majority of lesser snow geese contributing to the Western
Arctic wintering population. There is also evidence that lesser snow geese from Japan
(KOLY) which are reproductively isolated from all other snow geese populations during
breeding periods, but share wintering grounds in the Central California Valley with other
snow and Ross’s geese populations (M. Samuel, personal communication), have similar
gene frequencies as most North American populations. Additionally, populations of
lesser snow geese nesting in one colony on Wrangel Island, Russia, but wintering in two
geographically distinct areas (WRNO geese winter in southern British Columbia and
northern Washington state and WRSO geese winter in the Central Valley of California),
exhibit significant differences in allele and haplotype frequencies. Thus, fidelity to
wintering areas is likely a more important influence on the genetic structure of snow
geese than fidelity to geographically separated breeding sites. Contemporary wintering
areas have likely only been utilized since the retreat of the last Pleistocene glacier, and
wintering populations may not have been separated long enough to generate large genetic
differences among all wintering groups. It is possible that genetic variation among
wintering populations may continue to increase over time as a function of restricted gene
flow among populations from different wintering areas.

The two color morphs of lesser snow geese were considered separate species until
1983 when studies revealed that the two forms were a single interbreeding, dimorphic

species (Am. Ornithol. Union 1983) with the color dimorphism controlled by a single

121

gene, the melanocortin-l receptor (Mundy et al. 2004). Historic data suggests blue-phase
and white-phase lesser snow geese may have been isolated in two separate refugia during
the last Pleistocene glacial period (18,000-26,000 ybp), with blue-phase geese surviving
in the Canadian Arctic Archipelago and white-phase geese inhabiting refuges farther west
in Siberia and the Bering Sea region (Ploeger 1968, Cooke et al. 1988). Color phases
may have remained allopatric until the early 20th century (Cooke et al. 1988). Prior to the
1920’s, 99% of the geese nesting in western colonies of Hudson Bay were white and 99%
of geese within eastern colonies of Hudson and James Bays were blue (Barnston 1860,
Sanders 1917). During this period, mixing of blue and white snow geese on major
wintering grounds was extremely rare (Cooke et a1. 1988). Eastward expansions of white
geese and westward expansions of blue geese that have occurred since 1940 (Lewis and
Peters 1941, Cooch 1961, Dzubin 1979, Johnson and Troy 1987), and increasing numbers
of lesser snow geese across the continent (Ankney 1996, Batt 1997, Abraham et al. 1999)
have increased opportunities for genetic exchange between blue and white snow geese.
Lack of genetic differences among blue and white lesser snow geese populations in
similar geographic locales indicate that the length of time blue and white snow geese may
have been historically isolated was not long enough for significant genetic variation to
develop between the two color morphs. In addition, mixed-mating between the color
morphs during more recent evolutionary periods may have homogenized gene
frequencies among blue and white geese wintering and breeding in similar regions.

Few avian species have deeply diverged phylogenetic groups which are
geographically widespread as in snow and Ross’s geese (Avise and Walker 1998).

Several other species of Arctic-nesting birds exhibit significant genetic divergences

122

attributed to historical events that display a strong geographic orientation (dunlin,
Wenink et a1 1993; spectacled eider, Scribner et a1. 2001; Canda geese; Scribner et al.
2003). Canada geese, for example, were shown to have a large genetic divergence
between two mtDNA haplotype clades (14% sequence divergence), which was related to
the division of large-bodied and small-bodied forms of geese within the species (Scribner
et al. 2003; see also Chapter 2). This significant phylogenetic split was attributed to
isolation of the two forms in separate glacial refugia during the Pleistocene. In contrast,
some Arctic species exhibit homogenized gene frequencies among nesting and wintering
populations [harlequin duck (Histrionicus histrionicus), Lanctot et al. 1999; king eider,
Pearce et al. 2004]. For instance, king eiders are genetically panmictic across the species
range, and lack of spatial structure was attributed to historical gene flow and recent
population growth following Pleistocene deglaciation (Pearce et al. 2004). Our genetics
data support the hypothesis of a fundamental genetic split between snow and Ross’s
geese species, but propose that speciation occurred subsequent to evolutionary
divergence between the two, geographically widespread mtDNA haplotype clades.

We used multiple genetic markers which differed in mode of inheritance and rate
of evolution to resolve historical signatures of genetic variation and detect contemporary
changes in genetic divergence within and among snow and Ross’s geese species. We
investigated two hypotheses regarding the spatial genetic structure of snow and Ross’s
geese. Our first hypothesis proposed that the present populations of snow and Ross’s
geese are genetically differentiated in a manner reflecting historical biogeographic
isolating events proposed by Ploeger (1968). We found significant phylogenetic

divergences among species based on mtDNA clade fi'equencies and allele frequencies,

123

supporting Ploeger’s hypothesis that snow and Ross’s geese were isolated in separate
glacial refugia during the Pleistocene. We also found evidence of genetic differentiation
between greater and lesser snow geese, although the degree of genetic variation between
subspecies was much less than between species. The lack of significant variation
between subspecies based on mtDNA may indicate that greater snow geese were not
completely isolated in the northeastern Arctic glacial refugia proposed by Ploeger (1968),
or the period of subspecies separation was not long enough to accrue significant
differences in gene frequencies. Alternatively, we may not have been able to resolve
subspecies differences based on only one population of greater snow geese and without
sequencing haplotypes within mtDNA clades.

Our second hypothesis stated that contemporary population expansion events
within snow and Ross’s geese have been extensive, and any genetic signature of
historical differentiation that may have occurred within and among species may be
masked by these expansions. Alternatively, snow and Ross’s geese may still exhibit
genetic structure based on population affinities to breeding and wintering grounds,
despite expansions that may obscure genetic variation attributed to historical processes
(Ely and Scribner 1994, Robertson and Cooke 1999). We were able to resolve historic
divergence at the level of species, but not at the subspecies level or among color phases
of lesser snow geese. Recent widespread expansions of snow geese populations may
have diluted past genetic divergence between subspecies and between color phases.
However, it is also possible that separation times between subspecies and between blue
and white snow geese may have been too short to allow for significant genetic variation

to accrue between groups. We provide evidence that the current population genetic

124

structure of snow and Ross’s geese is strongly influenced by population affiliation to
wintering regions, and hypothesize wintering populations may continue to diverge
genetically over time.

Both historical vicariant events and recent demographic changes, including
population expansions and colonization of new nesting and wintering sites, have greatly
influenced contemporary levels of genetic differentiation among populations of snow and
Ross’s geese. By utilizing maternally and bi-parentally inherited genetic markers, our
study provides inferences about the relative importance of historical and ecological
factors that have shaped the spatial genetic structure of snow and Ross’ geese across their
species ranges. In previous studies (Quinn 1992, Avise et al. 1992, Weckstein et al.
2002), conclusions were based on gene trees derived from a single locus (mtDNA),
relatively few individuals per locale, and a limited number of populations within each
species. By increasing the number of populations included in the analysis, and sampling
a large number of individuals from each location, we were able to examine hypothesis
regarding the genetic differentiation of among species of snow and Ross’s geese, between
subspecies of snow geese, between color morphs of lesser snow geese, and among
breeding and wintering populations of snow geese and Ross’s geese throughout the

geographic range of both species.

125

LITERATURE CITED

Abraham, K. F., J. O. Leafloor, and H. G. Lumsden. 1999. Establishment and growth of
the lesser snow goose, Chen caerulescens caerulescens, nesting colony on Akimiski
Island, James Bay, Northwest Territories. Canadian Field-Naturalist 113:245-250.

Alisauskas, R. T. 1998. Winter range expansion and relationships between landscape
and morphometerics of midcontinent Lesser snow geese. Auk 115:851-862.

Alisauskas, R. T. and H. Boyd. 1994. Previously unrecorded colonies of Ross’s and
lesser snow geese in the Queen Maud Gulf Bird Sanctuary. Arctic 47:69-73.

American Ornithologists’ Union. 1983. Check-list of North American birds. 6’h Edition.
American Omithologists’ Union, Washington, D. C.

Anderson, M. G., J. M. Rhymer, and F. C. Rowher. 1992. Philopatry, dispersal, and the
genetic structure of waterfowl populations. Pages 365-3 85 in B. D. Batt, A. D. Afton, M.
G. Anderson, C. D. Ankney, D. H. Johnson, J. A. Kadlec, and G. I. Krapu, Editors.
Ecology and management of breeding waterfowl. University of Minnesota Press,
Minneapolis, Minnesota, USA.

Ankney, C. D. 1996. An embarrassment of riches: too many geese. Journal of Wildlife
Management 60:217-222.

Armstrong, W. T., K. M. Meeres, R. H. Kerbes, W. S. Boyd, J. G. Silveria, J. P. Taylor,
and B. Turner. 1999. Routes and timing of migration of lesser snow geese from the
western Canadian Arctic and Wrangle Island, Russia. Pages 75-88 in Distribution,
survival, and numbers of Lesser Snow Geese of the Western Canadian Arctic and
Wrangel Island, Russia (R. H. Kerbes, K. M. Meeres, J. E. Hines, Eds.). Occasional
Paper No. 98. Canadian Wildlife Service, Ottawa, Ontario.

Avise, J. C. 1992. Molecular population structure and the biogeographic history of a
regional fauna: a case history with lessons for conservation biology. Oikos 63:62-76.

Avise, J. C., J. Arnold, R. M. Ball, E. Bermingham, T. Lamb, J. E. Neigel, C. A. Reeb,
and N. C. Saunders. 1987. Intraspecific phylogeography: the mitochondrial DNA bridge

between population genetics and systematics. Annual Review of Ecology and
Systematics 18:489-522.

Avise, J. C., R. T. Alisauskas, W. S. Nelson, and C. D. Ankney. 1992. Matriarchal
population structure in an avian species with female natal philopatry. Evolution 46:1084-
1096.

Avise, J. C., R. M. Ball, and J. Arnold. 1988. Current versus historical population sizes

in vertebrate species with high gene flow: a comparison based on mitochondrial DNA
lineages and inbreeding theory for neutral mutations. Mol. Biol. Evol. 5:331-344.

126

Avise, J. C., and D. Walker. 1998. Pleistocene phylogeographic effects on avian
population and the speciation process. Proceedings of the Royal Society, London, Series
B 265:1707-1712

Baranyuk, V. V., J. E. Hines, and E. V. Syroechkovsky. 1999. Mineral staining of facial
pltunage as an indicator of the wintering ground affinities of Wrangel Island lesser snow
geese. Pages 111-114 in Distribution, survival, and numbers of Lesser Snow Geese of
the Western Canadian Arctic and Wrangel Island, Russia (R. H. Kerbes, K. M. Meeres, J.
E. Hines, Eds.). Occasional Paper No. 98. Canadian Wildlife Service, Ottawa, Ontario.

Barnston, G. 1860. Recollections of the swans and geese of Hudson Bay. Ibis 2:253-
259.

Batt, B. D. J. 1997. Arctic ecosystems in peril: report of the Arctic Goose Habitat
Working Group. Arctic Goose Joint Venture Special Publication. US. Fish and Wildlife
Service, Washington, D. C. and Canadian Wildlife Service, Ottawa, Ontario. 120pp.

Bellrose, F. C. 1980. Ducks, Geese and Swans of North America. Stackpole Books,
Harrisburg, PA. 540pp.

Buchholz, W. G., and J. M. Pearce, B. J. Pierson, and K. T. Scribner. 1998.
Dinucleotide repeat polymorphisms in waterfowl (family Anatidae): Characterization of a
sex-linked (Z-specific and 14 bi-parentally inherited loci. Animal Genetics 29:323-325.

Campbell, R. W., N. K. Dawe, I. MacTaggart-Cowan, J. M. Cooper, G. W. Kaiser, and
M. C. McNall. The birds of British Columbia. Royal British Columbia Museum,
Victoria, British Columbia.

Cavalli-Sforza, L. L. and A. W. F. Edwards. 1967. Phylogenetic analysis: models and
estimation procedures. Evolution 21 :550-570.

Chesser, R. K. 1991a. Gene diversity and female philopatry. Genetics 127: 437-447.

Chesser, R. K. 1991b. Influence of gene flow and breeding tactics on gene diversity
within populations. Genetics 129:573-583.

Cooch, F. G. 1953. Techniques for mass capture of flightless blue and lesser snow
geese. Journal of Wildlife Management 17:460-465.

Cooch, F. G. 1961. Ecological aspects of the blue-snow goose complex. Auk 78:72-89.

Cooke, F. 1987. Lesser snow goose: A long-term population study. Pages 407-432 in
Avian genetics (F. Cooke, P. A. Buckley, Eds.). Academic Press, New York.

127

Cooke, F. and K. F. Abraham. 1980. Habitat and locality selection in Lesser Snow
Geese: the role of previous experience. Proc. Int. Ornithol. Congr. 17:998-1004.

Cooke, F., D. T. Parkin, and R. F. Rockwell. 1988. Evidence of former allopatry of the
two color phases of lesser snow geese (Chen caerulescens caerulescens). Auk 105:467-
479.

Cooke, F., R. F. Rockwell, and D. B. Lank. 1995. The snow geese of La Perouse Bay:
natural selection in the wild. Oxford University Press, Oxford, UK.

Cooke, F. and D. S. Sulzbach. 1978. Mortality, emigration and separation of mated
Snow Geese. Journal of Wildlife Management 42:271-280.

Comuet, J. M. and G. Luikart. 1996. Description and power analysis of two tests for
detecting recent population bottlenecks from allele frequency data. Genetics 144 :2001-
2014.

Drake, K. L., and R. T. Alisauskas. 2005. Multistate modeling of movement and
philopatry of breeding Ross’s geese. Animal Biodiversity and Conservation: In Press.

Dzubin, A. 1965. A study of migrating Ross’ geese in western Saskatchewan. Condor
67:5 1 1-534.

Dzubin, A. 1979. Recent increases of blue geese in western North America. Pages 141-
175 in R. L. Jarvis and J. C. Bartonek, Editors. Management and biology of Pacific
F lyway geese. Oregon State University Book Stores, Corvallis, Oregon.

El Mousadik, A. and R. J. Petit. 1996. High level of genetic differentiation for allelic
richness among populations of the argan tree [Argania spinosa (L.) Skeels] endemic to
Morocco. Theoretical Applied Genetics 92:832-839.

Ely, C., and K. T. Scribner. 1994. Genetic diversity in arctic-nesting geese: implications
for management and conservation. Transactions of the North American Wildlife Natural
Resources Conference 59:91-1 10.

Emerson, B. C., E. Paradis, and C. Thebaud. 2001. Revealing the demographic histories
of species using DNA sequences. Trends in Ecology and Evolution 16:707-716.

Excoffier, L., P. E. Smouse, and J. M. Quattro. 1992. Analysis of molecular variance
inferred from metric distances among DNA haplotypes: application to human
mitochondrial DNA restriction data. Genetics 131:479-491.

Felsenstein, J. 1993. PHYLIP (Phylogeny Inference Package), version 3.5c.
Department of Genetics, University of Washington, Seattle.

128

Fields, R. L., and K. T. Scribner. 1997. Isolation and characterization of novel

waterfowl microsatellite loci: cross-species comparisons and research applications.
Molecular Ecology 6: 199-202.

Frederick, R. B. and R. R. Johnson. 1983. Ross’ geese increasing in central North
America. Condor 85:257-258.

Ganter, B., W. S. Boyd, V. V. Baranyuk, and F. Cooke. 2005. First pairing in snow
geese Anser caerulescens: at what age and at what time of year does it occur? Ibis
147:57-66.

Ganter, B. and F. Cooke. 1998. Colonial nesters in deteriorating habitat: site fidelity and
colony dynamics of Lesser Snow Geese. Auk 1152642-652.

Geramita, J. M., and F. Cooke. 1982. Evidence that fidelity to natal breeding colony is
not absolute in female snow geese. Canadian Journal of Zoology 60:2051-2056.

Goudet, J. 2001. FSTAT, a program to estimate and test gene diversities and fixation
indices (version 2.9.3). Available from http://www.uni1.ch/izea/sofiwares/fstat.html.
Updated from Goudet (1995).

Goudet, J, M. Raymond, T. Demeeus, F. Rousset. 1996. Testing differentiation in
diploid populations. Genetics 144:1933-1940.

Greenwood, P. J. 1980. Mating systems, philopatry and dispersal in birds and mammals.
Animal Behavior 28: 1140-1162.

Greenwood, P. J. and P. H. Harvey. 1982. The natal and breeding dispersal of birds.
Annual Review of Ecology and Systematics 13:1-21.

Hatch, D. R. M. , and A. H. Shortt. 1976. Possible intermediate Ross’ goose and snow
goose in Manitoba. Auk 93:391-392.

Hewitt, G. M. 2001. Speciation, hybrid zones and phylogeography — or seeing genes in
space and time. Molecular Ecology 10:537-549.

Johnson, 8. R. 1996. Staging and wintering areas of Snow Geese nesting on Howe
Island, Alaska. Arctic 49:86-93.

Johnson, S. R. 1995. Immigration in a small population of snow geese. Auk 1122731-
736.

Kerbes, R. H, V. V. Baranyuk, and J. E. Hines. 1999. Estimated size of the Western
Canadian Arctic and Wrangel Island Lesser Snow Goose populations on their breeding
and wintering grounds. Pages 25-38 in Distribution, survival, and numbers of Lesser
Snow Geese of the Western Canadian Arctic and Wrangel Island, Russia (R. H. Kerbes,

129

K. M. Meeres, J. E. Hines, Eds.). Occasional Paper No. 98. Canadian Wildlife Service,
Ottawa, Ontario.

Klicka, J., and R. M. Zink. 1997. The importance of recent ice ages in speciation: A
failed paradigm. Science 277:1666—1669.

Kocher, T. D., W. K. Thomas, A. Meyer, S. V. Edwards, S. Péiéio, F. X. Villablanca, and
A. C. Wilson. 1989. Dynamics of mitochondrial DNA evolution in animals:

amplification and sequencing with conserved primers. Proceedings of the National
Academy of Sciences 86:6196-6200.

Lewis, H. F. and H. S. Peters. 1941. Notes on birds of the James Bay region in the
autumn of 1940. Canadian Field-Naturalist 55:111-117.

Mila, B. D. H. Girman, M. Kimura, T. B. Smith. 2000. Genetic evidence for the effect
of a postglacial population expansion on the phylogeography of a North American
songbird. Proceedings of the Royal Society of London, Series B 267:1033-1040.

Miyatake, T. and R. Shimizu. 1999. Genetic correlations between life-history and
behavioral traits can cause reproductive isolation. Evolution 53:201-208.

Mowbray, T. B., F. Cooke, and B. Ganter. 2000. Snow Goose (Chen caerulescens). In
The Birds of North America, No. 514, A. Poole and F. Gill, eds. The Academy of
Natural Sciences, Philadelphia, and The American Omithologists’ Union, Washington,
DC.

Mundy, N. I., N. S Badcock, T. Hart, K. Scribner, K. Janssen, and N. J. Nadeau. 2004.
Conserved genetic basis of a quantitative plumage trait involved in mate choice. Science
303: 1870-1873.

Owen, M. 1980. Wild Geese of the World. B. T. Batsford Ltd., London. 236pp.

Palmer, R. S., Ed. 1976. Handbook of North American Birds, Vol. 2. Yale University
Press, New Haven, CT, USA.

Prevett, J. P., and F. C. Johnson. 1977. Continued eastern expansion of breeding range
of Ross’ goose. Condor 79:121-123.

Prevett, J. P. and C. D. MacInnes. 1972. The number of Ross’ geese in central North
America. Condor 74:431-438.

Prevett, J. P. and C. D. MacInnes. 1980. Family and other social groups in snow geese.
Wildlife Monographs 71 : 1-46.

Petit E., F. Balloux, and J. Goudet. 2001. Sex-biased dispersal in a migratory bat: a
characterization using sex-specific demographic parameters. Evolution 55: 63 5-640.

130

Petit, R. J ., A. El Mousadik, and O. Pons. 1998. Identifying populations for conservation
on the basis of genetic markers. Conservation Biology 12:844-855.

Prevett, J. P. 1972. Family behavior and age-dependent breeding biology of the blue
goose, Anser caerulescens. Ph.D. dissertation, University of Western Ontario, London.

Prevett, J. P. And F. C. Johnson. Continued eastern expansion of breeding range of Ross’
goose. Condor 79:121-123.

Quinn, T. W. 1992. The genetic legacy of mother goose-phylogeographic patterns of
lesser snow goose Chen caerulescens caerulscens maternal lineages. Molecular Ecology
1:105-117.

Reed, A., J-F. Giroux, and G. Gauthier. 1998. Population size, productivity, harvest and
distribution. Pages 5-31 in B. D. J. Batt, Editor. The greater snow goose: report of the
Arctic Goose Habitat Working Group. Arctic Goose Joint Venture Special Publication.
US. Fish and Wildlife Service, Washington, DC. and Canadian Wildlife Service,
Ottawa, Ontario.

Reed, A. and P. Chagnon. 1987. Greater snow geese on Bylot Island, Northwest
Territories, 1983. Journal of Wildlife Management 51:128-131.

Robertson, G. J. and F. Cooke. 1999. Winter Philopatry in Migratory Waterfowl. Auk
116(1): 20-34. '

Rockwell, R. F. and G. F. Barrowclough. 1987. Gene flow and the genetic structure of
populations. Pages 223-255 in F. Cooke and P. A. Buckley, editors. Avian genetics: a
population and ecological approach. Academic Press, London.

Ryder, J. P. and F. Cooke. 1973. Ross’ geese nesting in Manitoba. Auk 90:185-215.

Ryder, J. P. and R. T. Alisauskas. 1995. Ross’ Goose (Chen rossii). In The Birds of
North America, No. 162, A. Poole and F. Gill, eds. The Academy of Natural Sciences,
Philadelphia, and The American Omithologists’ Union, Washington, DC.

Sanders, W. E. 1917. Wild geese at Moose Factory. Auk 34:334.

Schneider, 8., D. Roessli, and L. Excoffier. 2000. ARLEQUIN, version 2.0: A Software
for Population Genetic Data Analysis. Genetics and Biometry Laboratory, University of
Geneva, Geneva, Switzerland.

Scribner, K. T., S. L. Talbot, J. M. Pearce, B. J. Pierson, K. S. Bollinger, and D. V.
Derksen. 2003a. Phylogeography of Canada geese (Branta Canadensis) in western North
America. Auk 120:889-907.

131

Scribner, K. T., J. A. Warrillow, J. O. Leafloor, H. H. Prince, R. L. Inman, D. R.
Luukkonen, and C. S. Flegel. 2003b. A genetic method for determining racial
composition of Canada goose harvests. Journal of Wildlife Management 67:122-135.

Sharp, D. E. and T. J. Moser. 1999. Light geese in the Central Flyway. US. Department
of Interior, Fish and Wildlife Service, Denver, Colorado.

Stewart, J. R. and A. M. Lister. 2001. Cryptic northern refugia and the origins of the
modern biota. Trends in Ecology and Evolution 16:608-612.

Takezaki, N. and M. Nei. 1996. Genetic distances and reconstruction of phylogenetic
trees from microsatellite DNA. Genetics 144:389-399.

Tim, D. E. and R. G. Bromley. 1976. Driving Canada geese by helicopter. Wildlife
Society Bulletin 4: 180-181 .

Trauger, D. L., A. Dzubin, and J. P. Ryder. 1971. White geese intermediates between
Ross’ geese and lesser snow geese. Auk 88:856-875.

Veit, R. and W. Petersen. 1993. Birds of Massachusetts. Massachusetts Audubon
Society, Lincoln, Massachusetts.

Weckstein, J. D., A. D. Afton, R. M. Zink, and R. T. Alisauskas. 2002. Hybridization
and population subdivision within and between Ross’s geese and lesser snow geese: a
molecular perspective. Condor 104:432-436.

Weir, B. S. 1996. Genetic Data Analysis 11. Sinauer and Associates, Sunderland.

Weir, B. S. and C. C. Cockerham. 1984. Estimating F-statistics for the analysis of
population structure. Evolution 38:1358-1370.

Williams, C. K., M. D. Samuel, V. V. Baranyuk, E. G. Cooch, D. Kraege. 2005. Winter

fidelity and apparent survival of lesser snow goose populations in the Pacific Flyway.
Unpublished United States Geological Survey manuscript.

132

CHAPTER 4: A NOVEL METHOD FOR CANADA GOOSE HARVEST

DERIVATION USING GENETIC ANALYSIS OF TAIL FEATHERS

INTRODUCTION

Harvest management for Canada geese (Branta canadensis) attempts to meet
species, population, and resource user goals by maintaining viable breeding populations
that allow for sustained harvests. The success of Canada goose management may be
increased by monitoring harvest composition in areas where birds from multiple breeding
populations or subspecies mix during seasonal hunting periods. To accurately estimate
proportional contributions of populations or subspecies within a mixed flock, the harvest
must be unambiguously identified to population/subspecies of origin. In addition, harvest
samples must be representative of the entire harvest at desired spatial and temporal
scales.

Genetic markers have been recently used to estimate proportional contributions of
Canada goose populations in harvest mixtures at multiple geographical scales (Pearce et
al. 2000, Scribner et al. 2003). Scribner et a1. (2003) were also able to estimate harvest
composition for discrete time periods associated with early, regular, and late hunting
seasons within each of several years and between managed and private lands in close
proximity. However, sample sizes in some of the harvest mixtures analyzed were
relatively small, and all hunter-harvested birds in both earlier studies were collected at a
few private hunting areas and at state game area check stations that were managed for
migrating Canada geese and other waterfowl. Restricted sampling of harvested geese

may be a source of error if harvest derivations are assumed to represent the harvest

133

composition across larger geographic regions (e.g., entire states or Flyways). Thus, a
more broad-based and systematic framework of sampling harvested birds is needed to
increase the accuracy and precision of genetic-based estimates of harvest composition
over large geographic areas. Annual collections of tail fans (feathers) of harvested geese
from hunters participating in the US. Fish and Wildlife Service (USFWS) Waterfowl
Parts Survey (Martin and Carney 1977) may provide sampling coverage necessary to
determine population-specific Canada goose harvest estimates at spatial and temporal
resolutions necessary to answer many research and management questions, and without
the need to conduct additional directed collections.

The sex and age composition of harvest is needed to estimate differential
susceptibility when evaluating harvest regulations (Pospahala et al. 1974, Reynolds and
Sauer 1991). However, determining sex and age may be problematic when complete
specimens or sex organs are not available for inspection, particularly in sexually
monomorphic species like Canada geese (Bellrose 1980, Owen 1980). Tail fans may be
used to discriminate large from small subspecies of Canada geese based on size
morphology, but subspecies determinations may be inaccurate if both sexes are included
in analyses (Tacha et al. 1987). Although Canada geese cannot be sexed from tail fans
alone, the sex of each harvested Canada goose sample can be determined using genetic
markers (e.g., Griffiths et al. 1998). 1

Our objectives were to (1) use nuclear microsatellite loci and likelihood methods
based on mixed stock analyses to estimate proportional contributions of breeding
populations of Richardson’s, interior, and giant Canada geese using tail fans obtained

from US. Fish and Wildlife Service Parts Survey for Michigan’s 1998-99 harvest, and

134

(2) to use genetic markers to determine sex and estimate sex ratios of Canada geese
harvested. We build upon the work of Scribner et al. (2003) by incorporating systematic
Parts Survey collections to expand the geographical coverage of Canada goose harvest
sampling in Michigan, and by providing methodology for molecular determination of sex
that can be accomplished using Parts Survey samples. By increasing the number of
breeding populations in our baseline, and by increasing sample size, we provide more
representative state-wide measures of the racial composition of Canada goose harvests in

Michigan.

METHODS

Sample Collection

Blood was sampled from feather quills taken from pre-fledging goslings collected
according to methods outlined in Scribner et al. (2003). Specific sampling locations
include the baseline breeding populations identified in Scribner et al. (2003), and four
additional breeding populations including Baffin Island, and Michigan’s eastern upper
peninsula, northeast lower peninsula, and northwest lower peninsula. Collections of
interior Canada geese included samples obtained within the Mississippi Valley
Population (MVP) breeding range in both Hudson Bay (n = 100) and James Bay (n = 65)
and within the breeding range of the Southern James Bay Population (SJ BP) on the
mainland (n = 100) and Akimiski Island (n = 83); Richardson’s Canada geese were
collected within the Tall Grass Prairie Population (TGPP) breeding areas on Baffin Island
(n = 86). Giant Canada geese were sampled within breeding areas across Michigan

including the southwest (n = 119), south-central (n = 129), southeast (n = 79), northwest

135

(n = 46), and northeast (n = 30) regions of the lower peninsula, areas surrounding
Saginaw Bay (n = 78), and within western (n = 58) and eastern portions of the upper
peninsula (n = 19).

Canada goose harvest samples were collected during the 1998-99 fall hunting
season in Michigan through the USFWS Waterfowl Parts Survey (Martin and Carney
1977). The Waterfowl Parts Survey was initiated in 1961 in an effort to provide the
USFWS with direct estimates of species, age, and sex composition of annual harvests that
could not be accurately determined from the USFWS Hunter Questionnaire Survey,
which has been used since 1952 for total harvest estimates. The Parts Survey also
facilitates harvest analyses at finer geographical (e. g., state vs. flyway) and temporal
(e. g., daily or weekly vs. entire season) scales than was possible through the
Questionnaire Survey (Martin and Carney 1977, Geissler 1990). Cooperating hunters
were asked to mail the USFWS primary wing tips and tail retrices for each goose
harvested, and samples returned were cataloged by a unique identification number in the
USFWS database associated with date and location of harvest. Biologists then categorize
goose samples returned by species and age (juvenile or adult) (Carney 1964, 1992). The
USFWS attempts to solicit hunter participation to collect samples from each state in
proportion to that state’s contribution to the total flyway harvest.

We obtained hunter-harvested samples of all Canada geese collected in Michigan
during the 1998-99 USFWS Parts Survey (n = 481). Each sample envelope contained 1-
12 tail feathers from geese harvested in 51 out of 83 possible counties in Michigan during

the early (n = 259), regular (n = 201) and late (n = 11) hunting seasons (date range:

136

9/1/98-2/3/99). Samples were presumed to represent a mixture from all sampled baseline

locations.

Genetic Analyses

We extracted DNA from 3 tail fan feathers (when available) for each harvested
Canada goose to maximize quantities of DNA. Use of multiple feathers may lead to
contamination (Pearce et al. 1997) if hunters inadvertently placed feathers from multiple
individuals in each envelop. However, given the high allelic diversity of loci surveyed
(Scribner et al. 2003), we would expect to observe evidence of amplification of >2 alleles
from DNA of contaminated samples, inconsistent with known Mendelian inheritance
(Cathey et al. 1998, Buchholz et al. 1998). Feathers were dissected on a clean glass plate.
The feather rachis was diced into 0.30 cm pieces using a sterile blade that was flamed
between samples to prevent contamination. These pieces were subsequently placed into a
1.5 mL Eppendorf tube, and Qiagen Dneasy kits and protocols (QIAGEN Incorporated,
Valencia, CA, USA) were used to extract DNA. We quantified DNA concentration using
a fluorometer or spectrophotometer, and diluted samples to working concentrations of 50
ng/uL for subsequent analyses.

We used five bi-parentally inherited microsatellite DNA loci (Bcau7, Bcau9,
Bcaul l, and Hhiul [Buchholz et al. 1998], and TTUCG-l [Cathey et al. 1998]) to
estimate allele and genotype frequencies for all baseline breeding populations and
hunter-harvested samples. Loci were amplified using polymerase chain reaction (PCR),
and products were electrophoresed on denaturing 6% polyacrylamide gels and visualized

using a FMBIO II laser scanner (Hitachi Software Engineering, Alameda, California,

137

USA). Resulting genotypes were scored based on 20 base-pair standards and reference
samples of known genotype.

Sex of all hunter-harvested samples was determined using the chromo-helicase-
DNA-binding (CHD) locus (Griffiths et a1. 1998). Primers amplify portions of the CHD
locus on the W and Z sex chromosomes (females are ZW and males are 22) within non-
coding intron regions whose lengths differ between the chromosomes. Thus, males can
be identified by the presence of one band (two introns of the same size), while females

are characterized by two bands (two introns of different sizes).

Statistical Analyses

We conducted mixed stock analyses of the harvest using maximum likelihood
methods (Pella and Milner 1987, Scribner et al. 2003). We sampled all three subspecies
and four managed populations of Canada geese [MVP, SJBP, TGPP, Michigan
Mississippi Flyway Giant Population (MI-MFGP)] likely to contribute to Michigan’s
harvest (Luukkonen and Soulliere 2002). In order to ensure genetic markers used were
independent, tests for genotypic linkage disequilibrium were conducted as described by
Weir (1996) using the program GENEPOP (Raymond and Rousset 1995). For each
breeding population, we used a Fisher’s Exact Test implemented in program GENEPOP
to test observed genotype fiequencies in each of the five microsatellite loci utilized for
deviations from Hardy-Weinberg expectations. In addition, we used simulated harvest
mixtures to estimate accuracy and precision of the harvest mixture composition based on

resampling of baseline populations.

138

We used hierarchical F-statistics (Weir 1996) in the program F-STAT (Goudet
2000) to estimate degree of genetic differentiation among subspecies and among
management populations. Measures of variance in allele frequency among subspecies
and among management populations were summarized as pair-wise estimates of
population Fst, and significance of mean Fst values (across the five microsatellite loci)
was determined by jackknifing procedures (Weir 1996). To account for multiple testing,
sequential Bonferroni procedures (Rice 1989) were used to adjust nominal significance
levels in tests of Hardy-Weinberg, gametic disequilibrium, and F -statistics.

Harvest derivations were calculated using the Statistics Program for Analyzing
Mixtures (Debuvec et al. 2000), which compares the distributions of genotypic
frequencies of each baseline population with the genotypic frequencies observed in
harvest mixtures (see Scribner et al. 2003 for details). Proportional contributions of each
subspecies were calculated as the mean over 500 replicate resamplings (with
replacement) of baseline populations and the harvest mixture, and confidence limits (SE)

for harvest estimates were calculated using Monte Carlo bootstrapping.

RESULTS
Each of the five microsatellite loci were polymorphic (4-13 alleles per locus;
mean heterozygosity per locus 0.4060806; specific data not shown). Tests for each
locus in each of the baseline breeding populations revealed approximately 95% of
individual tests, and cumulative tests across all loci for each baseline population, were
within Hardy-Weinberg expectations (P > 0.05). No linkage was evident for any locus

pair in any population, indicating microsatellite loci used in harvest mixture analyses

139

were independent. We observed no evidence of cross-sample contamination (data not
shown).

Pair-wise comparisons among the thirteen Canada geese breeding areas revealed
significant differences in allele frequencies (mean F5; across all populations = 0.025, P <
0.001). Differences were greater among subspecies (interior, Richardson’s, and giant
Canada geese; mean pair-wise F st = 0.039, P < 0.005) than among management
populations (MVP, SJBP, TGPP, MI-MFGP; mean pair-vvise Fst = 0.033, P < 0.005).
Simulations based on allele frequencies of breeding populations indicated that allocating
a mixed harvest to populations of origin can be accomplished with high accuracy (85-
96%). The majority of misclassifications within the simulations were to other breeding
populations within the same subspecies (interior to interior, giant to giant; specific data
not shown). Therefore, we were able to more accurately and precisely assign the mixed
harvests to subspecies (interior, Richardson's, and giant).

DNA concentrations were sufficient for PCR and analysis for 471 of the 481
(98%) Canada geese sampled through the Parts Survey. Maximum likelihood estimates

(iSE) for the 1998-99 fall harvest of Canada geese in Michigan were 69.2 i 6.2% giant

Canada geese, 26.3 i 5.8% interior Canada geese, and 3.8 i 1.2% Richardson's geese
(Figure 1). Of the 471 tail fan samples used in the sexing analysis, 224 were female and

239 were male, resulting in a harvest ratio of 0.937 females for every 1 male.

DISCUSSION
Management agencies use harvest derivations to determine if current hunting

regulations meet species and population goals for sustainable harvests while maintaining

140

)---i
O
I

 

 

c:
.2
E 08
‘2 .
8
+5; 0.6
E
E 0.4
Te
a
.2
t: 0.2
o
a.
9.
‘3: 0.0
B.c. maxima B.c. interior B.c. hutchinsii
Subspecies

Figure 1. Mean proportional contributions (iSD) of subspecies (Branta canadensis
maxima, B. c. interior, and B. hutchinsii hutchinsii) of Canada geese to the l998-99
season harvest in Michigan.

diversity and viability of breeding populations. For harvest derivation techniques to be
reliable and widely applicable, harvest sampling regimes must enable managers to
accurately estimate harvest by subspecies and population. Use of genetic markers and
mixed stock analyses allows for accurate discrimination among subspecies and
populations of Canada geese contributing to harvests. As problems of differentiating

Canada goose populations and subspecies are not unique to Michigan or the Mississippi

141

F lyway (Rusch et al. 1995), the derivation techniques employed in Michigan’s Canada
goose harvest could be applied to many areas of North America where multiple
subspecies or populations co-occur during harvest.

Long-term, adaptive management of Canada geese necessitates that harvest
derivation techniques facilitate collections of sufficient data to provide evaluations of the
harvest at spatial and temporal scales relevant to management objectives. Incorporating a
comprehensive and systematic sampling methodology such as the USFWS Waterfowl
Parts Survey reduces sampling bias, and Canada geese harvest estimates can be derived at
desired spatial and temporal scales. As the derivation techniques employed in
Michigan’s Canada goose harvest allow for multi-scale analyses, these techniques could
be used to evaluate annual harvest regulations, including harvest quotas, harvest zones,
and seasonal harvest periods.

The sex ratio of Canada geese harvested in Michigan during the 1998-99 was
nearly 1:1. Multiple studies have documented nearly even proportions of sexes among
several age groups of Canada geese (Hanson and Smith 1950, Sherwood 1965, Vaught
and Kirsch 1966, see Bellrose 1980 for review). An even sex ratio would be expected
given both sexes actively participate in parental care of offspring, family groups oflen
migrate together (Raveling 1988, Raveling et al. 2000), and hunting mortality is similar
among males and females (Bellrose 1980, but see Irnber 1968). This is in contrast to
sexually dimorphic ducks, where females incur higher mortality during nesting, brood
rearing, and hunting, leading to a 2:1 sex ratio (males:females) in harvests of many
species (Bellrose et al. 1961, Bellrose 1980). Hunter selectivity for one sex in dimorphic

species may also account for skewed sex ratios in ducks.

142

MANAGEMENT IMPLICATIONS

Increasing the accuracy of harvest derivations across multiple spatial scales and
time periods of relevance to managers will improve evaluations of the impacts that
alternative harvest regimes have on co-occurring populations of Canada geese. Such
knowledge is essential when management objectives differ among populations. By
combining samples from the USFWS Parts Survey with genetic analyses, we increased
spatial coverage of state-wide harvest samples, reducing sampling bias encountered when
harvest derivations estimated from samples obtained from a few selected harvest
locations were extrapolated to other non-sampled areas or time periods. If sufficient
genetic differences exist among breeding populations, and if sufficient samples are
available, genetic-based harvest derivation techniques could be applied in other states,
and for other migratory species at many temporal and geographic scales. Molecular
methods of sex determination would eliminate gender-based biases inherent in

morphologically-based methods of subspecific or population classification.

ACKNOWLEDGEMENTS
Funding for this study was provided by the Michigan Department of Natural
Resources (MDNR) and through the Partnership for Ecosystem Research and
Management (PERM) Program between the MDNR and the Department of Fisheries and
Wildlife at Michigan State University. Baseline samples were collected by many
individuals associated with the MDNR, including B. Odum, S. Schmidt, T. Cooley, J.

Gormley, J. Lerg, A. Karr, B. Voyle, T. Gearman, and J. Dastack. Additional baseline

143

samples were obtained by K. Abraham, D. Caswell, and J. Leafloor. We also thank the
Michigan State University students who assisted with baseline sample collection
including J. Niewoonder, M. Monfils, L. Briggs, M. Wasilco, R. White, A. Bump, L.
Gay, E. Butler, J. Lemson, D Gaebel, and J. Murphy. Hunter-harvested samples were
collected by the US. Fish and Wildlife Service. We acknowledge J. Martz and C.

Bennett, Jr., MDNR, for their assistance throughout the project.

144

LITERATURE CITED

Bellrose, F. C., T. G. Scott, A. S. Hawkins, and J. B. Low. 1961. Sex ratios and age
ratios in North American Ducks. Illinois Natural History Survey Bulletin 27:391-474.
Bellrose, F. C. 1980. Ducks, geese, and swans of North America. Second edition.
Stackpole, Harrisburg, Pennsylvania, USA.

Buchholz, W., J. M. Pearce, B. J. Pierson, and K. T. Scribner. 1998. Dinucleotide repeat
polymorphisms in waterfowl (family Anatidae): characterization of a sex-linked (Z-
specific) and 14 bi-parentally inherited loci. Animal Genetics 292323-325.

Carney, S. M. 1964. Preliminary keys to waterfowl age and sex identification by means
of wing plumage. United States Fish and Wildlife Service Special Scientific Report 82.

Carney, S. M. 1992. Species, age and sex identification of ducks using wing plumage.
United States Fish and Wildlife Service, Department of Interior, Washington, D. C.

Cathey, J. C., J. A. DeWoody, and L. M. Smith. 1998. Microsatellite markers in Canada
geese (Branta canadensis). Journal of Heredity 89:173-174.

Debuvec, E. M., R. B. Gates, M. Masuda, J. Pella, J. H. Reynolds, and L. W. Seeb. 2000.
SPAM (version 3.2): Statistics program for analyzing mixtures. Journal of Heredity
91 :509-510.

Geissler, P. H. 1990. Estimation of confidence intervals for Federal waterfowl harvest
surveys. Journal of Wildlife Management 54:201-205.

Goudet, J. 2000. F STAT, a program to estimate and test gene diversities and fixation
indices (version 2.9.1). Available from http://www.unil.ch/izea/softwares/fstathtml.
Updated from Goudet (1995).

Griffiths, R., M. C. Double, K. Orr, and R. J. G. Dawson. 1998. A DNA test to sex most
birds. Molecular Ecology 7:1071-1076.

Hanson, H. C. and R. H. Smith. 1950. Canada geese of the Mississippi F lyway with
special reference to an Illinois flock. Illinois Natural History Survey Bulletin 25:67-210.

Imber, M. J. 1968. Sex ratios in Canada goose populations. Journal of Wildlife
Management 32:905-920.

Luukkonen, D. R. and G. J. Soulliere. 2002. Second-year evaluation of Michigan’s
experimental early Canada goose hunting season in the Saginaw Bay area. Michigan

Department of Natural Resources - Wildlife Division, Lansing, Michigan, USA.

Martin, C. M. and S. M. Carney. 1977. Population ecology of the mallard: VI. A review
of duck hunting regulations, activity, and success, with special reference to the mallard.

145

United States Fish and Wildlife Service, Resource Publication 130, Washington, D. C.,
USA.

Owen, M. 1980. Wild geese of the world. Their life history and ecology. B. T.
Batsford, London, Great Britain.

Pearce, J. M., R. L. Fields, and K. T. Scribner. 1997. Nest materials as a source of
genetic data for avian ecological studies. Journal of Field Ornithology 68:471-481.

Pearce, J. M., B. J. Pierson, S. L. Talbot, D. V. Derksen, D. Kraege, and K. T. Scribner.
A genetic evaluation of morphology used to identify harvested Canada geese. Journal of
Wildlife Management 64:863-874.

Pella, J. J. and G. B. Milner. 1987. Use of genetic marks in stock composition analysis.
Pages 247-276 in N. Rymand and F. M. Utter, editors. Population genetics and fishery
management. University of Washington, Seattle, Washington, USA.

Pospahala, R. S., D. R. Anderson, and C. J. Henny. 1974. Population ecology of the
mallard. II. Breeding habitat conditions, size of the breeding populations, and

production indices. United States Fish and Wildlife Service Resource, Publication 115,
Washington, D. C., USA.

Raveling, D. G. 1988. Mate retention in giant Canada geese. Canadian Journal of
Zoology 66:2766-2768.

Raveling, D. G., J. S. Sedinger, and D. S. Johnson. 2000. Reproductive success and

survival in relation to experience during the first two years in Canada geese. Condor
102:941-945.

Raymond, M. and R. Rousset. 1995. GENEPOP: (Vers. 1.2): population genetics
software for exact tests and ecumenicism. Journal of Heredity 86:248-249.

Reynolds, R. E. and J. R. Sauer. 1991. Changes in mallard breeding populations in
relation to production and harvest rates. Journal of Wildlife Management 55:483-487.

Rice, W. R. 1989. Analyzing tables of statistical tests. Evolution 43:223-225.

Rusch, D. H., R. E. Malecki, and R. Trost. 1995. Canada geese in North America.

Pages 26-28 in E. T. LaRoe, editor. Our living resources: a report to the nation on
distribution, abundance, and health of US. plants, animals and ecosystems. United States
Department of the Interior, National Biological Service, Washington, DC, USA.

Scribner, K. T., J. A. Warrillow, J. O. Leafloor, H. H. Prince, R. L. Inman, D. R.

Luukkonen, and C. S. F legel. 2003. A genetic method for determining racial
composition of Canada goose harvests. Journal of Wildlife Management 67:122-135.

146

Sherwood, G. A. 1965. Canada geese of the Seney National Wildlife Refuge. U. S. Fish
and Wildlife Service Completion Report, Wildlife Management Study 1 & 2,
Minneapolis, Minnesota, USA.

Tacha, T. C., D. D. Thomburg, and R. A. Williamson. 1987. Bias of tail-fan surveys for
estimating Canada goose age ratios. Wildlife Society Bulletin 15:533-534.

Vaught, R. W. and L. M. Kirsch. 1966. Canada geese of the eastern prairie population,
with special reference to the Swan Lake flock. Missouri Department of Conservation,
Technical Bulletin 3, St. Louis, Missouri, USA.

Weir, B. S. 1996. Genetic Data Analysis II. Sinauer and Associates, Sunderland,
Massachusetts, USA.

147

CHAPTER 5: COMPOSITION OF CANADA AND CACKLING GOOSE HARVESTS
IN MICHIGAN 1998-2002: A GENETIC ANALYSIS OF USFWS WATERFOWL

PARTS SURVEY SAMPLES

INTRODUCTION

The number of states where harvest regulations for cackling geese (Branta
hutchinsii, Banks et al. 2003) and Canada geese (Branta canadensis) include special
seasons and hunting zones is growing in response to requests for increased hunting
opportunities and decreased human-geese conflicts, and the need to target overabundant
non-migratory resident populations while protecting smaller migratory populations
(Lindberg and Malecki 1994, Williams and Johnson 1995, Ankney 1996, Heusmann
1999, Luukkonen and Souillere 2004a). Agencies across North America including the
Michigan Department of Natural Resources have implemented special early and late
goose seasons, and has shifted the opening date of the regular season from early October
to mid-September to target a large and rapidly growing population of resident giant
Canada geese (Branta canadensis maxima) and to minimize harvests of migrant interior
(B. c. interior) Canada geese and cackling geese (B. hutchinsii hutchinsii) (Soulliere et al.
1988, Luukkonen and Soulliere 2004a). Current understanding of migration patterns and
timing of interior Canada geese and cackling geese based on banding studies suggest
these birds remain on their northern breeding grounds until late September (Tacha et al.
1991, Mowbray et al. 2002, Leafloor et al. 2004), and generally do not arrive in major
harvest areas of southern Michigan until early October (Luukkonen and Soulliere 2004a).

Daily limits are more liberal during early and late seasons as compared to the regular

148

season to in an attempt to increase harvest of resident Canada geese during periods of low
migrant abundance (Luukkonen and Soulliere 2004b). As harvest regulations may vary
among seasons, it would be important to determine if arrival times of migrants are similar
among years, or if arrival times vary annually. Derivation techniques that provide
accurate estimates of harvest at fine temporal scales are necessary to evaluate harvests of
resident Canada geese and migrant Canada and cackling geese within and among hunt
seasons. Additionally, it is important to determine if harvest sampling is spatially
uniform among years, or if there is large variation in annual sample collections. If
sampling among years is equivalent, then differences in annual harvests may be attributed
to biological factors (i.e., chronology of migration of interior Canada geese and cackling
geese) that lead to changes in the proportion of migrant and resident geese harvested each
year. If sampling among years is highly variable, then annual harvest differences may be
a result of sampling variation.

In addition to the seasonal changes implemented to increase resident goose
harvests and protect migratory geese, special harvest zones and game management units
(GMUs) have been created. Michigan has been divided into two harvest management
zones; the Mississippi Valley Population (MVP) zone in the Upper Peninsula and
northwest Lower Peninsula and the Southern James Bay Population (SJ BP) zone in the
southeast Lower Peninsula (Figure l). The MVP and SJBP zones were defined based on
long-term banding and radio-telemetry data (Tacha et al. 1991) and range descriptions
(Bellrose 1980) that defined migratory corridors utilized by MVP and SJ BP geese (B. c.
interior). The MVP and SJBP zones have been used to regulate harvests of birds that

breed within geographically defined breeding populations by adjusting season length and

149

 

 

 

 

 

 

 

 

 

 

 

 

      
 

 

 

 

   

SLP

SJBP Zone
/‘ : my

I
>
1

 

 

 

 

 

 

Figure 1. Four local goose management units (GMUs) designed to attract migrating
waterfowl are located in the southern Lower Peninsula (SLP) of Michigan, including
Muskegon County Wastewater (M) and Allegan (A) in the Mississippi Valley Population
(MVP) harvest management zone, and Saginaw (S) and Tuscola/Huron (HT) within the
Southern James Bay Population (SJ BP) harvest management zone. Counties within the
MVP zone are white, while counties within the SJBP zone are shaded. GMUs within
these zones are designated by black rectangles.

timing and daily limits within these zones, and implementing closures in areas where
Federally imposed state quotas have been reached for the annual harvest. GMUs are
smaller public hunting areas within harvest management zones traditionally managed to

attract migrant geese through the supplementation of browse foods and by providing open

150

water refuges (Luukkonen and Soulliere 2004b). The Michigan DNR monitors annual
harvests within GMUs, and the season is closed when state quotas for what are assumed
to migrant geese are achieved. Management prescriptions have assumed that harvests of
MVP geese occurs mostly within the MVP harvest management zone and on GMUs
within that zone, while harvests of SJ BP geese are assumed to occur principally in the
SJBP zone. However, recent studies have indicated that harvest of both MVP and SJ BP
geese may be more widespread in Michigan than previously determined (Scribner et al.
2003, Fritzell and Luukkonen 2004). Thus, there is a need for comprehensive harvest
evaluations across public and private areas within the MVP and SJBP harvest
management zones and within local GMUs.

Inman et a1. (2003) recently demonstrated that genetic analysis of annual
collections of tail fans (feathers) of harvested geese from the US. Fish and Wildlife
Service (U SF WS) Waterfowl Parts Survey (Martin and Carney 1977) provided accurate
discrimination among subspecies and management populations of Canada geese and
cackling geese contributing to Michigan harvests. Through this comprehensive sampling
methodology, hunters from across the state mail the USFWS tail retrices from harvested
geese from public and private hunting areas within both harvest management zones, and
throughout all seasonal harvest periods. The broad temporal and spatial coverage of
these harvest samples combined with genetic-based mixed stock analysis (Pearce et al.
2000, Scribner et al. 2003) can be used to estimate proportional contributions of resident
and migrant geese to special harvest seasons and within local GMUs and harvest

management 201168.

151

Our first objective was to utilize nuclear microsatellite loci and mixed stock
analyses to estimate proportional contributions of three subspecies (cackling geese, and
interior and giant Canada geese) and four management populations [Tall Grass Prairie
Population (TGPP), MVP, SJBP, and Michigan’s Mississippi Flyway Giant Population
(MI-MFGP)] of geese that are potentially harvested during special seasons and within
defined harvest zones. We estimate seasonal and annual variation in population and
subspecies contribution to annual harvests using genetic markers and established mixture
analysis (Pella and Milner 1987, Pella and Masuda 2001) for early, regular and late
seasons, and for six discrete time periods from the beginning of the early season to the
end of the regular season. Additionally, we derive harvest estimates for local GMUs and
surrounding private areas, and for the MVP and SJBP harvest management zone for
annual, early season, and regular season harvest periods. Our second objective was to
test the equality of harvest estimates derived for harvest mixtures collected at different
times or from different harvest zones. We present results based on Monte Carlo
likelihood ratio tests to assess significant changes in harvest composition among the five
years sampled. Our final objective was to determine if bias exists among harvest sample
collections. Variograms, which illustrate the average degree of similarity among number
of individuals sampled as a function of distance separating sampling locales (Rossi et al.
1992), were used to determine if the spatial pattern of harvest samples collected across

the state was consistent among years.

METHODS

Sample Collection

152

Baseline samples were collected from pre-fleging goslings in breeding
populations of cackling geese, and interior and giant Canada geese according to methods
outlined in Scribner et al. (2003), including all sampling locations outlined in Inman et a1.
(2003). A total of 964 individuals were collected from baseline breeding populations of
cackling geese (n = 1 population), and interior (n = 4) and giant (n = 8) Canada geese.
Samples from the goose harvest were collected during the 1998-2002 annual hunting
seasons in Michigan using the United States Fish and Wildlife Service (USFWS)
Waterfowl Parts Survey (Martin and Carney 1977) (see Inman et al. 2003 for Survey
details). We obtained hunter-harvested samples of all geese collected in Michigan during
the l998-99 (n = 471), 1999-2000 (n = 406), 2000-01 (n = 324), 2001-02 (n = 486), and
2002-03 (n = 585) USFWS Parts Surveys (Figure 2). Each sample envelope contained 1-
12 tail feathers, and were identified by date and county of harvest within Michigan.

Samples were presumed to represent a mixture from all sampled baseline locations.

Genetic Analyses

We extracted and quantified DNA from geese tail fan feathers using techniques
detailed in Inman et al. (2003). Five bi-parentally inherited microsatellite DNA loci
(Bcau7, Bcau9, Bcaul l, and Hhiul [Buchholz et al. 1998], and TI'UCG-l [Cathey et al.
1998]) were used to estimate allele and genotype frequencies for all baseline breeding
populations and hunter-harvested samples. Loci were amplified using polymerase chain
reaction (PCR), and products were electrophoresed on denaturing 6% polyacrylamide

gels and visualized using a FMBIO II laser scanner (Hitachi Software Engineering,

153

due :03 323 coo—BSCE cemmom “mots:
15:5 05 waist 5:38 85 5515 880:8 33 29:3 oco H32 8 88:3: 3958 nonagm .xotsm 3am 38583 838m
0.26:3 EB 55 .m .D 05 «3:85 33:32 E 880:8 338% 83.8: 88m «9280 Ea 3on wEEomo mo .83anme .N oSwE

MOON-NOON NGON- FOON

rOONéOON w OOONummm r . mam Tana —.

 

 

154

Alameda, California, USA). Resulting genotypes were scored based on 20 base-pair
standards and reference samples of known genotype.

Sex of all hunter-harvested samples was determined using the chromo-helicase-
DNA-binding (CHD) locus (Griffiths et al. 1998). Males were identified by the presence
of one amplified band (two introns of the same size), while females were characterized

by two bands (two introns of different sizes).

Harvest Derivations

We first used variograms to determine if harvest sampling is spatially uniform
among years, or if there is large spatial variation in annual sample collections for the total
harvest, early season harvest, regular season harvest, and harvest within the MVP and
SJ BP harvest management zones. Similar to spatial covariance and correlation functions,
variograms model the average degree of similarity (e. g., level of variability) among
number of individuals sampled as a function of distance separating sampling locales
(Rossi et al. 1992). Each variogram was constructed as a plot of half the average squared
difference among harvest sample sizes for counties separated by about the same distance
(distance between county pairs). Geographic distances between each pair of Michigan
counties were calculated using x,y coordinates (in meters) from the MIGeoref projection
(MDNR Spatial Data Library 2005). Variograms were constructed using the program
SAS (SAS Institute 2005). Ten distance classes were included in all variograms except
for the SJ BP harvest zone total harvest variogram (7 distance classes), and were
determined by the lag distance (30 km for SJ BP, 40 km for all others) multiplied by the

number of lags, or distance classes. Distance classes included harvest samples from

155

pairwise county comparisons whose geographic distance between counties were within
each distance interval at half the lag distance.

We conducted mixed stock analyses of annual harvests using conditional
maximum likelihood methods (Pella and Milner 1987) within the Statistics Program for
Analyzing Mixtures (SPAM 3.7, Debuvec et al. 2000), which compares the distributions
of genotypic frequencies of each baseline population with the genotypic frequencies
observed in harvest mixtures (see Scribner et al. 2003 for details). We utilized Bayesian
modeling of baseline allele frequency distributions (Pella and Masuda 2001) prior to
maximum likelihood estimation of the harvest to accommodate the possibility of alleles
that are found in mixtures but not present in the baseline. This method is an
improvement over conditional maximum likelihood methods, as absence of an allele from
a particular baseline population sample implies it is only rare and was missed in sampling
rather than assuming it is nonexistent. Proportional contributions of cackling and Canada
geese were calculated as the mean (iSD) over 1000 replicate resamplings (with
replacement) of baseline populations and the harvest mixture, and confidence limits for
harvest estimates were calculated using Monte Carlo bootstrapping.

Harvest derivations were conducted for multiple geographic and temporal
subdivisions of statewide harvests. For each of the five harvest years (1998-2002),
proportional contributions of cackling and Canada geese were estimated for each annual
period, and for early (September 1-15), regular (September 16 — December 31), regular +
late (September 16 — January 31; except 1998 which included samples until February 3),
and late (January 1 — 31; 2001 only) seasons within each annual harvest. Annual

collections of statewide harvest samples were also analyzed within six discrete time

156

periods: September 1 - 7, September 8— 15, September 16— 22, September 23 — 29,
September 30 — October 6, and October 7 - December 31.

Similar statewide harvest derivations were conducted on an annual basis for
samples harvested within Michigan’s MVP and SJBP harvest management zones (Figure
1). Seasonal harvests were analyzed for early and regular seasons only due to the low
number of geese harvested in the late season within both of these areas. Proportional
contributions of each of the four management populations were estimated within the
MVP and SJBP harvest zones for statewide and seasonal derivations.

Four local Goose Management Units (GMU) within the southern Lower Peninsula
(SLP) of Michigan, including Huron/Tuscola, Saginaw, Muskegon Wastewater, and
Allegan, were compared to non-GMU county harvests at the statewide level and within
SLP counties (Figure 1). Analyses were conducted for regular seasons 1998-2002, as
local GMUs were only open to hunting during the regular season in these years. Geese
harvested within Huron, Tuscola, Saginaw, Muskegon, and Allegan counties during the
regular season were defined as GMU samples for this analysis. Proportional
contributions of subspecies were estimated in the regular season for all four local GMUs
combined (HTSMA), non-GMU counties statewide, and non-GMU counties within the
SLP.

Monte Carlo likelihood ratio tests were conducted in SPAM 3.7b (Debuvec et al.
2000) to test the equality of harvest mixtures (i.e., did harvest mixtures collected at
different times or from different spatial locales originate from the same underlying
mixture). This test is more powerful for detecting differences among mixtures than

overlap in the confidence interval estimates for each population contribution from each

157

mixture. The latter technique suffers from inflated Type I error rates arising from the
simultaneous inferences, and inflated Type 11 error rates due to the use of marginal
(region-specific) measures of mixture difference (Reynolds and Templin 2004). The
likelihood ratio compares the likelihood of the observed mixture samples under two
different models: (1) the null model, where each of the M mixture samples comes from a
common mixture, and (2) the alternative model, where each of the M mixture samples
comes from a (possibly) different mixture. The observed likelihood ratio is obtained by
fitting both models and forming the ratio of their likelihoods. This observed likelihood
ratio is then compared to the reference distribution of likelihood ratio expected under the
null model to calculate a P-value. This method assumes all mixture samples were
gathered independently and randomly, all populations contributing to the mixtures are
represented in the baseline, and that the genetic markers used to characterize samples are
in Hardy-Weinberg equilibrium within each contributing population (Reynolds and
Templin 2004).

Likelihood ratio tests of equality among potentially different mixture samples
were conducted for several different spatial and temporal harvest derivations of cackling
and Canada geese. Yearly mixtures (n = 5, 1998-2002 harvests) were tested for equality
within three geographic areas: (1) statewide, (2) within the MVP harvest management
zone, and (3) within the SJ BP harvest management zones. Similarly, yearly mixtures (n
= 5, 1998-2002 harvests) were tested for equality over four time periods of the statewide
harvest: (1) annual harvest, (2) early season, (3) regular season, and (4) regular + late
seasons. For each likelihood ratio test, parametric bootstrapping was used to test the null

hypothesis that the harvest samples collected each year (n = 5) came from a common

158

mixture (n = 5000 simulations). Mixture simulations and model fitting were done in
SPAM 3.7b, and final analysis of simulation results was conducted in Microsoft Excel

(Microsoft Office, Microsoft, Inc., Redmond, WA.).

RESULTS

Baseline Population Differentiation

Inman et al. (2003) previously determined the 5 microsatellite loci used to
characterize baseline populations were independent, polymorphic, and did not deviate
from Hardy-Weinberg expectations. These determinations were made using several tests
within the program GENEPOP (Raymond and Rousset 1995). Hierarchical F -statistics
(Weir 1996) calculated in the program F -STAT (Goudet 2000) revealed significant
differences in allele frequencies among subspecies and among management populations
(Inman et a1. 2003). Inman et al. (2003) demonstrated that allocating mixed harvest
groups to population of origin could be accomplished with high accuracy (SS-96%) based

on allele frequency distributions of baseline breeding populations.

Harvest Derivations

Statewide distributions of goose harvest samples included 54 counties in 1998-99,
56 counties in 1999-2000, 46 counties in 2000-01, 51 counties in 2001-02, and 54
counties in 2002-03 (Figure 2). Annually, the number of counties sampled within MVP
and SJBP harvest zones (Figure 1) were: 1998-99 (35 MVP, 19 SJBP), 1999-2000 (29
MVP, 18 SJBP), 2000-01 (30 MVP, 16 SJBP), 2001-02 (28 MVP, 23 SJBP), and 2002-

03 (31 MVP, 23 SJBP). Numbers of geese sampled annually within MVP harvest zone

159

counties were as follows: 1998-99 (n=284), 1999-2000 (n=276), 2000-01 (n=l94), 2001-
02 (n=202), and 2002-03 (n=327), Annual sample sizes within SJBP harvest zone
counties were: 1998-99 (n=l74), 1999-2000 (n=127), 2000-01 (n=l34), 2001-02
(n=291), and 2002-03 (n=255).

The variogram analysis illustrated that there is less variance in the sample size of
harvested geese among counties in close geographic proximity to one another than among
counties separated by larger distances (Figure 3). The variance in harvest sample size
increases in a curvilinear fashion as a function of intervening distance among counties
sampled. This pattern is consistent across all five years of harvest sampling. Early and
regular season samples showed no marked differences in spatial patterns across years,
and sampling variability within early and regular seasons was similar to statewide
sampling variances (data not shown). Spatial patterns of harvest samples collected within
the MVP harvest zone were similar across years, and sampling variance was consistent
across years within the MVP harvest zone. Number of samples collected within the SJ BP
harvest zone were somewhat more variable (among counties) in 1999 and 2000 than the
remaining three years (Figure 3). Similar patterns in sampling variation across multiple
years and spatial scales indicates the USFWS Waterfowl Parts Survey is consistent across
years. Thus, variation in annual harvest estimates can be attributed to biological

phenomenon rather than inter-annual sampling bias.

Statewide
Total annual harvests (iSD) were composed mainly of giant Canada geese (67.4

i 6.5% - 82.4 i 4.2%), with lesser proportions of interior Canada geese (16.2 i 4.1% -

160

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

l X I
, 4 g __ _ _ w m8 L _ ._ __ 3
l E x ! l
l g . x ‘ ' 2 ’ x o A1998i
l 3, 3 HHJM ———-—.——-w-r- - .19991
1". o 0 x ‘
g . I2ooo
.§ 2 Z a .2001
l 3 x2002
l 1 ‘“ * — M ‘_ _ _ 3
l
I o I ‘ r j
. o 2 4 6 8 1o 12 1
l Lag Distance (40 km units) I
Statewide
l 5 —» ~~-~ - I
l ’ '
i E : i . I 3 i
2 . A1998
g 3 ,__._. __z_..- ,_._ ,- . I . ‘ ’ O 1999
t ‘ ‘ I A
g o o X = '2000
.a 2. .. -.-._* O 4 - .2001
'5 x x x o x x2002
1 — ~ ~ + l
O
X
0 I
o 2 4 6 8 l
Lag Distance (30 km units) 1
SJBP Harvest Zone

Figure 3. Variograms for the spatial distribution of annual harvest samples of cackling
geese and Canada geese collected in Michigan from 1998-2002 Statewide, and within the
Southern James Bay (SJBP) Harvest Zone.

161

30.5 i 6.5%) and cackling geese (0.3 i 0.4% - 3.8 i- 1.5%) (Figure 4). Giant Canada
geese dominated statewide early season harvests, with contributions ranging from 76.2 i
5.8% to 91.0 i 4.6% (Figure 5). Proportional harvests of interior Canada geese increased
greatly from the early to the regular season in all five years. The largest increase (33.7%)
between seasons occurred in 2002, which was also the year of the largest proportional
contribution of interior birds to the regular season harvest. Proportions of cackling geese
increased slightly from the early to the regular season in three of the five years analyzed.
although all estimates were less than 10% (Figure 5). Likelihood ratio tests rejected the
null model that the five yearly harvest mixtures were sampled from a common mixture of
cackling geese and Canada geese for the statewide harvest during annual (P = 0.0096)
and early season (P = 0.0048) harvest periods. In contrast, regular season harvest

mixtures were not significantly different among the five years sampled (P = 0.1745).

 

 

 

 

 

 

 

T T" W " __ I B. h. hutchinsii
. g 1.0 . B. c. interior 1
'g C] B. c. maxima {
g 0.8 FL
‘5
8 0.6 ~
g 0.4 -
E 0.2 ~
0
E 0.0 -

 

 

 

 

 

 

 

 

 

Figure 4. Mean proprotional contributions (iSD) of Branta hutchinsii hutchinsii, B.
canadensis interior, and B. c. maxima to annual statewide Michigan harvests from 1998-
2002. Annual harvests include samples collected September 1 — December 31 during the
year indicated, as well as samples collected in January of the following year (e.g., 1998
includes September 1 — December 31, 1998 and January 1 — 31, 1999). Sample sizes for
each year: 1998 (n = 471), 1999 (406), 2000 (324), 2001 (486), 2002 (585).

162

.39 Sea comma 32 mag 88 .A _ ms 58 5.2V 88 .55 82 .Gou M32 288... as

+ 5&2 x28 88 x8: 58 .5: 88 .Ga: 32 .3: M32 538 5&2 Ass 88 A36 88 .85 88 Ag: 32
Ammm u 5 moo. 2933 base “39» some 5 3833 3:884. no.“ mofim 295m .cmeomE E 83432 Sch 3822 383% Eco
SON :m I H >325: 8m— 98 .2 m Edam I 3 “38835 82 + 33on A _ m uonEoouD I 2 ..onEqumv 5332 A2 - fl “38339
3:3 8 SEES ..o .m was (0.285 3ng38 .m ..u.~§323& 5.23.63: 633% m0 353 EOE—£950 Raomtomoa .802 .m 2sz

 

 

 

 

 

 

 

 

 

moow Sow
331.2%»...1- Imam“..- I .1551 I II a: 33.5%.: 535. Pm .
0
:w
m
. w
to m
o.o W.
L a...»
i
, , r
coon mag 89.
I inner p.53. I IraI I I..1.........§; .I._a.m I .25 er I I

  
    

a;

 

 

 

 

 

 

.3530:

     

'.
o

O
O
uopnquwoo puomodmd

 

25.5.: d d D
3.8:: d .m-
3553: i dI 11.2-11...

 

 

163

Sample sizes during the late season were too low to perform individual season
analyses in four of five annual harvests. For all five years we combined regular and late
seasons for derivations, and found that proportional estimates of all three subspecies were
similar within years between the regular and regular + late seasonal derivations. Unlike
annual regular season harvest estimates, proportional estimates were significantly
different among years for the regular + late season (P = 0.0385) harvest. The harvest
pattern in 2001, the only year the late season was analyzed separately from the regular
season, indicated proportional contribution of interior geese decreased in the late harvest.
while giant Canada geese and cackling geese proportions increased (Figure 5).

Estimated proportions of giant Canada geese harvested statewide decreased after
the early season (i.e., two weeks) in 1998, 2001, and 2002, and decreased after the first
month of harvest in 1999 and 2000 (Figure 6). Proportional contributions of giants
continued to decrease, and interior contributions increased, into early October in 1999
and through the regular hunting season in 2000 and 2001, while estimates for giants and
interiors remained relatively constant after September for 2002. In 1998, estimates of
giant Canada geese harvested increased, and estimates of interiors decreased, from
October through then end of the regular hunting season in December. Proportional
harvest estimates of interior Canada geese were much greater during the first week of the
early season in 1998 and 2000 as compared to the remaining three years. This suggests a
larger number of interiors were present in Michigan during the early season in 1998 and

2000 than in 1999, 2001, and 2002.

164

.89» :08 8a b—mozoficoam 8% .3 com: Em totem means:

88 é 8% oasam .A _ m Basso-h Bose a as .6 Bozo-om Became m gem-m Banana o ANN-2 Eggnog

0 Am T” 38388 m A: Eggnog < ”mecca 2% 328% am wane Sow-M32 82,. sang: a 838 mafia 3:93
some 8 35.385 ..6 .m use {ELSE mmmrmhguo .m ..zEEoS: 2.33.63: SEEN mo Amway 32:55:00 Raomtoaoa :32 .0 oSwE

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

NOON room
If: AQMIIASE .50 63a Esrj “II-=3“. 33m IA-«wm. -...weIo :3; 63: -I

T .1 -, I ,_ 4a.- . 2 a
m m w a
. I _I I--- I to m.
I .I I -_ - - II o.o
-I-| .- I I - - I Ir. IrH . I- ad W

OOON mmmw wmmw

 

 

 

 

.5: .3: .93 .35 fix. 3:2.

. .. . .9-
. V- . \u
M - .v.
.
.I -II I

an: 3:. $3 .33 5; avg. I3: 5:. SE :3 as; aazI

 

 

 

».

   

    

’9
o

 

Ill- ; I I'll lil .

  

 

I l T III III

 

on
d

I
.. V.

o
uonnqmuoo WW6

 

 

 

 

 

 

 

 

l’
'9.
o

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

0

G

I
o

 

 

 

 

 

 

 

165

MVP and SJBP Harvest Zones

Proportional harvest estimates for geese originating from the MVP and SJBP
management populations varied greatly within and among years for designated MVP and
SJBP harvest zones (Figure 7). Birds of MVP origin were harvested in greater
proportions than birds of SJ BP origin within the SJBP harvest zone in four of five years
in the early season and in three years during the regular season (Table 1). Estimated
proportional harvests of geese originating from the MVP population were greater in the
SJBP harvest zone in both seasons in 1999 and 2002. Alternatively, we estimated that
geese originating from the SJBP population had higher estimated harvests in the MVP
harvest zone in early and regular seasons in 1998 and 2001. Overall, proportional
contributions of MVP Canada geese were greater than contributions of SJBP geese to
statewide harvests and in the MVP and SJBP harvest zones in 1999 and 2002 (over all
seasons and total annual harvest), while SJ BP geese were harvested more often than
MVP geese statewide and in both harvest zones in 2001 (Figure 7, Table 1). Proportional
harvests of giant Canada geese were similar in the MVP and SJBP harvest zones within
years, and followed proportional statewide harvest trends among years . Cackling geese
had low proportional harvest estimates in both MVP and SJ BP harvest zones (Figure 7).
Likelihood ratio tests rejected the null model of equality among annual harvests estimates
within the MVP harvest management zone (P = 0.0022). In contrast, annual harvests
estimates within the SJ BP harvest management zone were not significantly different (P =
0.1384). Annual variations in harvest estimates of migrants within the MVP zone can be
linked to variations in proportional contributions of migrants during the early and regular

seasons. In the SJBP zone, fluctuations in the annual harvest of migrants were more

166

$3 Son A88 88 is: 88 AR: 82 .35 $2 I 25 8:238
am seam 605:8 53 88 £3 58 is c 88 .65 $2 5% u 5 32 - 85 8358a 3:; 39335 ”3% :08

8m 35m 038% .220 mam 2: use 529 E5 Eofiowmcaz 880 n52 2: £53, gamma-3:2 E Amoom-wmoc $832 3:93
2 822-:6 assuage :36 :35 39232-5305 c5 55% 8338 am as“. 805:8 .Eté 8:238
>o=m> 3&3?va 530.3 coumgqom 3.82m $80 :3. 05 80h 38m 3930 we Emi acorn—p.580 322:2.er 532 .N. 233m

 

 

 

 

 

 

 

 

 

 
 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

mcoN case-.58.”. 23 $82. EmcSom mCON consaaom >m=m> 5933:).
38 38 88 82 83 88 88 82 32 I
a - aim _ W! a! 2. a a ,2 u
IHIII W WII «a «M I p «d M
II III I I -I-tc m I I to m
I -III II I com 1% I M. 3m
JHI - H IITI mm-I 3 m 1% I I Ii I nousée 3 q
u- ! .383 m
II I I I I, I I o... m I n52. or w
{III I - I I I-I L I I II II I _I I - nae:

 

 

 

 

 

167

 

 

Gm T8 § T5 Ens SE8 Ens

28.8 wood 28.8 83 33.8 ooo.o 25.8 :3 2.8.8 89o RES

288 :3 28.8 83 88.8 83 c: _.8 Rod 2.8.8 ~33 ne.-22-3

28.8 ooo.o 53.8 :8 28.8 83 88.8 58 23.8 33 .58

§ _.8 83 28.8 2 3 83.8 ago 32.8 23 28.8 2 3 8:2 Baum .58
2.8.8 sans cans cans 5W8

28.8 83. 2.8.8 Rod 88.8 m So 28.8 :58 28.8 08¢ .33

28.8 $3 88.8 823 22.8 Sod 28.8 :3 2.8.8 33 882-3

28.8 was 23.8 Rod 2.2.8 :2 398 23 28.8 2.3 .58

88.8 83 8.8.8 558 2:8 83 28.8 N88 $8.8 Rod .5: ram .58
am T5 smus fins 2§u5 a: T8

65.8 23 288 :58 28.8 988 83.8 :3 2.3.8 :38 .53

2.8.8 32 28.8 83 23.8 :28 23.8 ”moo $88 was mung-22

238 83 2: _.8 :3 32.8 $3 68.8 :3 2828 End .58

2: _.8 83 $88 83 28.8 2.3 23.8 :2 28.8 83 .52 5:38 “52
2:. TS 2: T8 32 7.5 SW 7.5 2.28

28.8 ooo.o 28.8 83 28.8 Sod 28.8 «8.0 88.8 wood .53

68.8 .58 28.8 :3 $38 33 $88 32. 28.8 a; “.8222

2.8.8 ~88 83.8 858 23.8 :3 85.8 83 2.8.8 2 3 mm;

88.8 ES 28.8 23. 288 ES 28.8 NS. 38.8 Sad .5: ram .52

E8 seas-mm 68 333mm 88 325mm 88 258mm 88 258mm 8:225 mega 2:6

88 58 88 88 M32

 

accusing 926802 cog 85 8833—8 203 SE 22833 @5853 28 83828 $3.83 $88-on— Sod

fiwfloaz E mocoN 6052 0800 mmmm Ea EVE E 38%: Gm 892 Eomv cemeom 33on 23 Am 72 Eamv ban

9 2.58 823332 DEE sac =fl 2a 28:-=8 82338.2 230 3;; Enema: 5&5: .2528 82338.2
25 8:5. Eofisom 5528 =33?qu >o=e> mqammmmmmaz 05 Sea 08% mo AQmfiv 32:55:00 320E882 :32 .2 03m..-

168

likely a result of fluctuations in the early season harvest, as regular season estimates of

migrant harvest were fairly constant among years (Table 1).

Local GMUs

Migrant interior geese were harvested in greater proportions during the regular
season within local GMUs (HTSMA) than within non-GMU counties in the SLP (Figure
8). Proportional contributions of migrant geese were also greater in non-GMU counties
statewide as compared to non-GMU counties within the SLP. Giant Canada geese had
the highest estimated harvest proportions in all areas and years except for the 2000

regular season local GMU harvest (70.4% i 15.1% B. c. interior, 19.3% i 14.6% B. c.

maxima, 10.3% i 7.0% B. h. hutchinsii).

Sex Determination

Statewide geese sex ratios were slightly male biased in 2001 and 2002.
Approximately equal harvests of the sexes occurred during 1998, 1999, and 2000.
Specific annual ratios of females to males were as follows: 1998-99 (0.937), 1999-2000

(1.020), 2000-01 (0.929), 2001-02 (0.800), 2002-03 (0.807).

DISCUSSION
In response to Michigan’s rapidly growing giant Canada goose population, special
late and early goose seasons were implemented in the late 1970’s and mid 1980’s
respectively (Soulliere et a1. 1988). Michigan’s resident goose population continued to

grow at a rate of approximately 14% per year, and regular harvest seasons were extended

169

 

‘1‘
Statewide i
non-GMU \

‘

 

 

 

    

2001

 

 

 

 

 

 

 

Proportional Contribution

 

 

Figure 8. Mean proprotional contributions (chD) of Branta hutchinsii hutchinsii, B.
canadensis interior, and B. c. maxima to regular season harvests (September 16 —
December 31) in Michigan’s four local Goose Management Units (GMU) including
Huron/Tuscola, Saginaw, Muskegon, and Allegan (HTSMA); and Michigan’s non-GMU
counties statewide and in the Southern Lower Peninsula (SLP) from 1998-2002. Sample
sizes for each year: Statewide non—GMU - 1998 (n = 148), 1999 (145), 2000 (115), 2001
(152), 2002 (277); SLP non-GMU — 1998 (57), 1999 (SO), 2000 (65), 2001 (116), 2002
(163); HTSMA GMU — 1998 (51), 1999 (55), 2000 (31), 2001 (44), 2002 (30).

170

in 1998 to include the latter two weeks of September (Luukkonen and Souillere 2004a).
As these seasonal changes were established in order to target giants and avoid harvest of
migrant cackling and Canada geese, harvest derivations must accurately estimate
contributions of species, subspecies and populations during multiple time periods of
annual hunts. Adding to the temporal complexity of Michigan’s harvests are local GMUs
traditionally managed to attract migrant geese, and regional harvest zones established to
better manage harvest of interior geese originating from the MVP and the SJ BP. We
utilized genetic markers and mixed stock analyses, which allow for accurate
discrimination among species, subspecies and populations of geese contributing to
harvests, to assess special season and hunt zone regulations in Michigan.

Variograms of harvest sampling suggests variance in the spatial pattern of
samples is consistent among years for multiple temporal and geographic scales. Spatial
correlations between sample size variance and pairwise distance between counties
indicate sampling variance decreases as a function of decreasing distance between
counties sampled. Thus, differences among proportional estimates of harvest among
years are not likely a result of spatial or temporal sampling bias among years.
Differences in estimates of proportional contributions to annual harvests were due to
biological phenomenon including arrival timing of migratory interior Canada geese and
cackling geese into Michigan and the duration of their stay in the state. Early spring
weather, such as levels of precipitation, daily mean temperature, and relative humidity
strongly influence the arrival of migratory geese on their northern breeding grounds and
timing of nest initiation (Blokpoel and Gauthier 1980, Wege and Raveling 1983). Nest

initiation date is closely correlated with timing of fall migrations (K. Abraham, personal

17]

communication), with earlier nest initiation leading to earlier fall migrations to staging
areas and southern wintering grounds. Annual inconsistencies in climatic conditions
would cause nest initiation dates to vary among years, and migrations of interior Canada
geese and cackling geese into Michigan would be earlier in some years than others.
Subsequent harvest of migratory geese would increase during the early season in years of
early arrival. Our discrete time analyses indicated there were annual fluctuations in the
initial arrival time of migratory geese. Interior Canada geese began arriving in Michigan
prior to the early season in 1998 and 2000, as evidenced by the greater estimated harvest
of interiors during the first week of the early season during these two years as compared
to 1999, 2001, and 2002.

Among year differences in early and regular season estimates of interior and giant
Canada geese were reflected in the pattern of among year differences in total annual
estimates. Annual variation in timing of migration of interior Canada geese and cackling
geese and subsequent early season harvest could have a large effect on total annual
harvest estimates, as early season bag limits are much more liberal than regular season
limits. However, annual harvests of interior Canada geese are greatest during the regular
season, so changes in harvest regulations during this season may have the most impact on
total take of interior Canada geese. Future harvest regulations may benefit from directed
research focused on how variations in the opening date, duration, and bag limits during
early and regular seasons would effect overall annual harvests of resident and migratory
geese.

Initial arrival date of migratory interior Canada geese may be an important factor

determining residence time in Michigan and susceptibility to harvest throughout the

172

hunting season. However, in years of higher estimated proportional interior harvest
during early seasons, and likely earlier initial arrival of migrant interiors to Michigan,
regular and regular + late season interior harvests were not proportionally greater as
compared to regular and regular + late seasons in other years. Thus, initial arrival date of
migratory interiors may influence susceptibility to harvest in the early season, but may
not strongly affect regular and late season harvest of migrants.

Influence of late season hunts on total annual harvests is likely small given low
harvest levels during J anuary/F ebruary. However, during 2001 when adequate sample
size allowed an analysis of late season harvest alone, higher proportions of the harvest
were comprised of giants compared to the regular season. Thus, migratory interior
Canada geese may exhibit greater proclivity for southerly migration during the winter
months. Special seasons during the winter period could be successful in targeting
resident giants.

The change in subspecies harvest proportions from the regular to the late season
in 2001 may indicate late season sample sizes were too low in the other three years to
resolve late season trends when combined with regular season harvest samples. This lack
of late season resolution is also apparent from likelihood ratio tests results that indicate
that regular + late season harvest mixtures vary annually, while regular season harvest
mixtures have relatively equal contributions of migratory and resident geese among
years. Differences in mid-winter weather conditions among years could lead to different
migration patterns if propensities for migration further south differ between subspecies.

Selectively managing harvests of MVP and SJBP may be difficult using

Michigan’s current harvest management zones because of the broad overlap and annual

173

variation in population-specific harvests within these zones. Harvest mixtures differed
from year to year within the MVP harvest zone, while staying relatively uniform over
time within the SJ BP harvest management zone. In 1999 and 2002, MVP birds
comprised the majority of the interior Canada geese harvest in the MVP and SJBP
harvest zones, while SJ BP birds were harvested in greater proportions in both harvest
zones in 2001. The repeated occurrence of greater proportional harvests of MVP birds
than SJBP birds in the SJBP harvest zone, and greater proportional harvests of SJ BP
birds than MVP birds in the MVP harvest zone, in multiple seasons and years, suggests
the fall distribution of these populations extends well beyond previously defined
migratory corridors in Michigan based on previous banding and radio-transmitter studies
(Tacha et al. 1991) and range descriptions (Bellrose 1980). Concurrent with our findings,
Fritzell and Luukkonen (2004) recently documented that MVP geese were harvested
across a larger proportion of the state of Michigan than previously estimated based on
1980-1989 band recovery data. MVP band recoveries increased in number and
geographic distribution in the Lower Peninsula of Michigan, following an increase in the
amount and distribution of banding on eastern MVP breeding grounds on James Bay in
1990. Scribner et al. (2000, 2003), also documented substantial harvests of MVP and
SJBP geese beyond traditional MVP and SJBP harvest zones in Michigan using genetic-
based derivation techniques. They estimated contributions of MVP geese to harvests in
southeastern Michigan game management areas, and SJ BP harvests at southwestern
Michigan game management areas, were greater than 30% in some years.

If there is greater mixing of migrants originating from the MVP and SJ BP

populations in the southwest Lower Peninsula relative to the northwest Lower Peninsula,

174

then the occurrence of higher proportions of birds originating from the SJ BP population
(relative to birds originating from the MVP population) in the MVP harvest zone in 1998
and 2001 could be explained by relatively lower numbers of counties and individual
harvested geese sampled in the northwest Lower Peninsula in those years as compared to
the number and distribution of samples collected from the southwest Lower Peninsula.
However, estimates of harvested birds originating from the MVP population were not
always greater in the northwest Lower Peninsula as compared to the southwest Lower
Peninsula within each year, and estimated proportions of birds originating from the SJ BP
population were greater in the northwest Lower Peninsula than the southwest Lower
Peninsula in four of five years sampled (data not shown). This indicates that potential
sampling bias in 1999 and 2001 is not the dominate factor influencing estimates of SJ BP
migrants harvested within the MVP harvest zone, and mixing of birds from SJ BP and
MVP populations may not be measurably greater in southwest Michigan relative to
mixing of both populations in the northwest Lower Peninsula.

Local GMUs were traditionally managed to attract interior Canada geese through
provisions of browse foods and open-water refuges (Luukkonen and Souillere 2004a).
Results of GMU harvest derivations indicate these areas continue to draw greater
proportions of interior geese than surrounding non-GMU areas. A comparison of
derivation results from non-GMU statewide harvests to non-GMU estimates in the SLP
revealed that greater numbers of interior Canada geese and cackling geese are harvested
in Michigan’s Upper Peninsula (UP) and northern portion of the Lower Peninsula (N LP)

as compared to the southern portion of the Lower Peninsula in all annual harvests. This

175

suggests an increased vulnerability to harvest and/or increased presence of migrant geese

in the UP and NLP, and lower numbers of resident giant Canada geese in these areas.

MANAGEMENT IMPLICATIONS

Harvest is the most direct way to impact population dynamics of Canada geese
(Williams and Nichols 1990). Michigan has approximately 50,000 goose hunters
(Soulliere and Frawley 2001) that can be utilized as a practical means of controlling large
populations of giant Canada geese. By combining samples from the USFWS Parts
Survey with genetic analyses, we describe results from an extensive study designed to
monitor annual harvest levels including special seasons and within defined management
zones. We expand previous studies that have used genetic methodology to monitor
Michigan cackling geese and Canada geese harvests by incorporating new methods to
assess sampling bias and through the use of likelihood ratio tests that assess the
magnitude of variance in harvest composition among spatial and temporal samples.

Harvest estimates suggest that the number of resident and migratory Canada
geese, as a proportion of the total goose population, varies in different hunt zones of
Michigan and different time periods of the season. Annual and seasonal variation in
proportional harvest estimates of cackling and Canada geese is likely influenced by
several factors. Relative population sizes are likely to fluctuate over time, as is
population productivity. Increased productivity in a given year leads to higher
proportions of juveniles which tend to be more susceptible to harvest than adults.
Variation in annual climatic conditions during spring migration and nesting of migratory

geese will lead to differences in nest initiation dates. Weather in the late summer will

176

also cause fall migrations to shift temporally. Initial arrival dates of migratory cackling
and Canada geese may be an important factor determining early season susceptibility of
migratory geese, and should be considered in future management of early season hunts.

Policies aimed at reducing numbers of giant Canada geese may benefit from
increased hunting pressure in non-GMU areas in southern lower Michigan, especially
during early and late season harvests. Harvest derivations within the MVP and SJBP
harvest zones indicate MVP and SJBP Canada geese likely migrate through both
Michigan harvest zones frequently, making both populations susceptible to harvest in
both zones. Management of interior Canada geese may be improved by redefining
breeding population management units based on genetics information.

We introduce a novel approach for detection of sampling variance as a source of
bias in data. In the absence of spatial heterogeneity of harvest samples among years,
annual variance in harvest proportions can be attributed to biological phenomenon. This
is also useful information for USFWS managers, in that sampling methodology is
targeting a consistent cross-section of Michigan hunters annually.

Annual variation in proportions of giant and interior Canada geese in early and
regular + late season harvests indicate that a single opening date may not be appropriate
to achieve desired targeted harvests. Additional studies of impacts of nest initiation date,
and investigations of late summer weather as it correlates to the initiation of fall
migration of interior Canada geese and cackling geese, are warranted.

Differences in migratory tendencies between resident and migratory geese during
mid-winter could lead to different harvest proportions if late seasons are used as a

management tool. Further research of the effects of weather on migratory behavior of

177

geese is necessary. Managers can use this information to ascertain the impacts of annual
harvest regulations and determine if current regulations are successful in directing
hunters to target giant Canada geese while protecting migratory geese. Adaptive harvest
management of resident Canada geese and migratory Canada and cackling geese will

benefit from the information gained through this analysis and similar future studies.

ACKNOWLEDGEMENTS

The Michigan Department of Natural Resources (DNR) and the Partnership for
Ecosystem Research and Management (PERM) Program between the Michigan DNR and
the Department of Fisheries and Wildlife at Michigan State University provided funding
for this study. Baseline samples were collected by many individuals associated with the
MDNR, Michigan State University, K. Abraham and J. Leafloor from the Ontario
Ministry of Natural Resources, and individuals from the Canadian Wildlife Service and
Environment Canada including D. Caswell and K. Dickson. We thank all participants of
the Mississippi F lyway Wing Bee who assisted in cataloging hunter-harvested samples
collected by the US. Fish and Wildlife Service, especially Barb Lercel Avers and Bill

Scullon of the Michigan DNR, and Dave Rave of the Minnesota DNR.

178

LITERATURE CITED

Abraham, K. F ., J. O. Leafloor, and D. H. Rusch. 1999. Molt migrant Canada geese in
northern Ontario and western James Bay. Journal of Wildlife Management 63:649-655.

Ankney, C. D. 1996. Invited paper: an embaressment of riches: too many geese. Journal
of Wildlife Management 60:217-223.

Banks, R. C., C. Cicero, J. L. Dunn, A. W. Kratter, P. C. Rasmussen, J. V. Remsen, Jr., J.
D. Rising, and D. F. Stotz. 2003. Forty-fifth supplement to the American ornithologists’
union check-list of North American birds. The Auk 121:985-995.

Bellrose, F. C., T. G. Scott, A. S. Hawkins, and J. B. Low. 1961. Sex ratios and age
ratios in North American Ducks. Illinois Natural History Survey Bulletin 27:391-474.

Bellrose, F. C. 1980. Ducks, geese, and swans of North America. Second edition.
Stackpole, Harrisburg, Pennsylvania, USA.

Blokpoel, H. and M. C. Gauthier. 1980. Weather and migration of Canada geese across
southeastern Ontario in spring 1975. Canadian F ield-Naturalist 94:293-299.

Buchholz, W., J. M. Pearce, B. J. Pierson, and K. T. Scribner. 1998. Dinucleotide repeat
polymorphisms in waterfowl (family Anatidae): characterization of a sex-linked (Z-
specific) and 14 bi-parentally inherited loci. Animal Genetics 29:323-325.

Carney, S. M. 1964. Preliminary keys to waterfowl age and sex identification by means
of wing plumage. United States Fish and Wildlife Service Special Scientific Report 82.

Carney, S. M. 1992. Species, age and sex identification of ducks using wing plumage.
United States Fish and Wildlife Service, Department of Interior, Washington, D. C.

Cathey, J. C., J. A. DeWoody, and L. M. Smith. 1998. Microsatellite markers in Canada
geese (Branta canadensis). Journal of Heredity 89:173-174.

Debuvec, E. M., R. B. Gates, M. Masuda, J. Pella, J. H. Reynolds, and L. W. Seeb. 2000.
SPAM (version 3.2): Statistics program for analyzing mixtures. Journal of Heredity
91 :509-510.

Fritzell, Jr., P. A. and D. R. Luukkonen. 2004. Influence of banding distribution on
harvest indices of Mississippi Valley Population Canada geese. Pages 51-59 in T. J.
Moser, R. D. Lien, K. C. VerCauteren, K. F. Abraham, D. E. Andersen, J. G. Bruggink, J.
M. Coluccy, D. A. Graber, J. O. Leafloor, D. R. Luukkonen, and R. E. Trost editors.
Proceedings of the 2003 International Canada Goose Symposium, Madison, Wisconsin,
USA.

179

Geissler, P. H. 1990. Estimation of confidence intervals for Federal waterfowl harvest
surveys. Journal of Wildlife Management 54:201-205.

Goudet, J. 2000. F STAT, a program to estimate and test gene diversities and fixation
indices (version 2.9.1). Available from http://www.unil.ch/izea/softwares/fstat.html.
Updated from Goudet (1995).

Griffiths, R., M. C. Double, K. Orr, and R. J. G. Dawson. 1998. A DNA test to sex most
birds. Molecular Ecology 711071-1076.

Heusmann, H. W. 1999. Special hunting seasons and resident Canada goose
populations. Inman, R. L., K. T. Scribner, H. H. Prince, J. A. Warrillow, D. R.
Luukkonen, P. I. Padding. 2003. A novel method for Canada goose harvest derivation
using genetic analysis of tail feathers. Wildlife Society Bulletin 31:1126-1131.

Leafloor, J. O., K. F. Abraham, F. D. Caswell, K. E. Gamble, R. H. Helm, D. D.
Humburg, J. S. Lawrence, D. R. Luukkonen. R. D. Pritchert, E. L. Warr, and G. G.
Zenner. 2004. Canada goose management in the Mississippi Flyway. Pages 22-36 in T.
J. Moser, R. D. Lien, K. C. VerCauteren, K. F. Abraham, D. E. Andersen, J. G. Bruggink,
J. M. Coluccy, D. A. Graber, J. O. Leafloor, D. R. Luukkonen, and R. E. Trost editors.
Proceedings of the 2003 International Canada Goose Symposium, Madison, Wisconsin,
USA.

Lindberg, M. S. and R. A. Malecki. 1994. Hunting vulnerability of local and migrant
Canada geese in Pennsylvania.

Luukkonen, D. R. and G. J. Soulliere. 2004a. Change in harvest composition resulting
from mid-September opening of Michigan’s regular goose season, 1998-2002. Michigan
Department of Natural Resources - Wildlife Division, Lansing, Michigan, USA.

Luukkonen, D. R. and G. J. Soulliere. 2004b. Evaluation of Michigan’s experimental
early Canada goose hunting season in the Saginaw Bay area. Michigan Department of
Natural Resources - Wildlife Division, Lansing, Michigan, USA.

Luukkonen, D. R. and G. J. Soulliere. 2002. Second-year evaluation of Michigan’s
experimental early Canada goose hunting season in the Saginaw Bay area. Michigan
Department of Natural Resources - Wildlife Division, Lansing, Michigan, USA.

Martin, C. M. and S. M. Carney. 1977. Population ecology of the mallard: VI. A review
of duck hunting regulations, activity, and success, with special reference to the mallard.
United States Fish and Wildlife Service, Resource Publication 130, Washington, D. C.,
USA.

MDNR Spatial Data Library. 2005. Michigan Geographic Framework Counties (v4b).
http://www.mcgi.state.mi.us/mgdl.

180

Mowbray, T. B., C. R. Ely, J. S. Sedinger, and R. T. Trost. 2002. Canada Goose (Branta
canadensis). In The Birds of North America, No. 682, A. Poole and F. Gill, eds. The
Academy of Natural Sciences, Philadelphia, and The American Omithologists’ Union,
Washington, DC, USA.

Pella, J. J. and M. Masuda. 2001. Bayesian methods for analysis of stock mixtures from
genetic characters. Fisheries Bulletin 99:151-167.

Pella, J. J. and G. B. Milner. 1987. Use of genetic marks in stock composition analysis.
Pages 247-276 in N. Rymand and F. M. Utter, editors. Population genetics and fishery
management. University of Washington, Seattle, Washington, USA.

Raymond, M. and R. Rousset. 1995. GENEPOP: (V ers. 1.2): population genetics
software for exact tests and ecumenicism. Journal of Heredity 86:248-249.

Reynolds, J. H. and W. D.Templin. 2004. Comparing mixture estimates by parametric
bootstrapping likelihood ratios. Journal of agricultural, biological, and environmental
statistics 9:57-74.

Rossi, R. E., D. J. Mulla, A. G. Journel, and E. H. Franz. 1992. Geostatistical tools for
modeling and interpreting ecological spatial dependence. Ecological Monographs
62:277-314.

Rusch, D. H., R. E. Malecki, and R. Trost. 1995. Canada geese in North America.

Pages 26-28 in E. T. LaRoe, editor. Our living resources: a report to the nation on
distribution, abundance, and health of US. plants, animals and ecosystems. United States
Department of the Interior, National Biological Service, Washington, DC, USA.

Scribner, K. T., J. A. Warrillow, H. H. Prince, R. L. Inman, D. R. Luukkonen, J. O.
Leafloor, C. S. Flegel. 2000. Genetic analysis of spatial and temporal variation in racial
composition of harvested Canada geese in Michigan. Michigan Department of Natural
Resources—Wildlife Division, Lansing. Final Report.

Scribner, K. T., J. A. Warrillow, J. O. Leafloor, H. H. Prince, R. L. Inman, D. R.
Luukkonen, and C. S. Flegel. 2003. A genetic method for determining racial
composition of Canada goose harvests. Journal of Wildlife Management 67:122-135.

Soulliere, G. J. and B. J. Frawley. 2001. Michigan waterfowl hunter activity and
opinions on regulations, management, and satisfaction, 1998-99. Michigan Department
of Natural Resources—Wildlife Division, Lansing. Report Number 3357.

Soulliere, G. J ., G. F. Martz, and W. C. Johnson. 1988. An evaluation of Michigan’s

special goose seasons and characteristics of the harvest, 1982-87. Michigan Department
of Natural Resources-Wildlife Division, Lansing. Report Number 3078.

181

Tacha, T. C., A. Woolf, W. D. Kirnstra, and K. F. Abraham. 1991. Migration patterns of
the Mississippi Valley Population of Canada geese. Journal of Wildlife Management 55:
94-102.

Wege, M. L. and D. G. Raveling. 1983. Factors influencing the timing, distance, and
path of migrations of Canada geese. Wilson Bulletin 95:209-221.

Weir, B. S. 1996. Genetic Data Analysis 11. Sinauer and Associates, Sunderland,
Massachusetts, USA.

Williams, B. K. and J. D. Nichols. 1990. Modeling and the management of migratory
birds. Natural Resource Modeling 4:273-311.

Williams, B. K. and F. A. Johnson. 1995. Adaptive management and the regulation of
waterfowl harvests. Wildlife Society Bulletin 23:430-436.

182

CHAPTER 6: A COMPARATIVE ANALYSIS OF WATERFOWL HARVEST
DERIVATION TECHNIQUES: ASSESSING MICHIGAN ’S CANADA GOOSE
AND CACKLING GOOSE HARVEST USING BAND RETURNS,

MORPHOMETRICS, AND GENETICS

INTRODUCTION

Harvest management is a key component in the conservation of Canada geese
(Branta canadensis) and cackling geese (Branta hutchinsii), which are continentally
distributed and composed of multiple subspecies and management populations (Bellrose
1980, Banks et al. 2003). In many states and provinces there is a growing need to target
overabundant resident populations of geese while protecting smaller migratory
populations to ensure populations remain viable and harvests are sustainable over time
(Williams and Johnson 1995, Ankney I996). The number of states/provinces where
harvest regulations include special seasons and hunting zones has increased in response
to requests for increased hunting opportunities and decreased human-resident geese
conflicts (Luukkonen and Soulliere 2004).

In response to Michigan’s rapidly growing giant Canada goose (B. c. maxima)
population, Michigan’s Department of Natural Resources (DNR) established a special
early season harvest in the Lower Peninsula of Michigan in 1986. The lS-day season
beginning September 1 was designed to focus harvest on resident and molt-migrant giant
Canada geese, and avoid migratory interior Canada geese (B. 0. interior) and cackling
geese (B. h. hutchinsii) that were assumed to arrive in Michigan later in the hunting

season (Soulliere and Martz I997). Michigan’s DNR also requested and was granted

183

regular goose season opening dates in mid-September beginning in 1998 as a further
attempt to target resident geese during periods of low interior and cackling migration
(Luukkonen and Soulliere 2004). In order to assess if current hunt regulations were
meeting prescribed state and federal management goals, accurate harvest derivation
techniques were needed to assess the proportion of resident giant Canada geese and
migratory interior Canada and cackling geese harvested during experimental early
seasons and during extended regular seasons.

A variety of techniques have been employed by biologists seeking to differentiate
populations or subspecies of Canada and cackling geese contributing to harvests. These
techniques have included marking geese with leg bands and neck collars (Lindberg and
Malecki 1994, Fritzell and Luukkonen 2004) or satellite transmitters (Luukkonen et al.
2004), measuring morphometric characters (Johnson et al. 1979, Moser and Rolley 1990,
Merendino et al. 1994), stable isotope analysis of feather tissue (Caccamise et al. 2000),
and genetic discrimination (Pearce et al. 2000, Scribner et al. 2003, Inman et al. 2003).
No comparative studies have been conducted to investigate the concordance of results
across independent harvest derivation techniques. Michigan is uniquely suited to address
this issue, as band recovery data, morphometric measures, and genetics information has
been simultaneously collected for the regular season statewide goose harvest during a
five year period after implementation of the mid-September season opening in 1998. In
addition, a local study of the morphometric and genetic characteristics of geese harvested
during the experimental early seasons in the Saginaw Bay region was carried out from
2000-2002. Our main objectives were to compare proportional harvest estimates of

interior and giant Canada geese and cackling geese derived from multiple data sources

184

for the regular hunt season statewide and for the early season in the Saginaw Bay area.
We discuss the degree of similarity among estimated harvests derived from different
methods, and the efficacy of using each harvest derivation technique at statewide and
local geographic scales. Our secondary objectives were to assess potential biases in
estimates that may result from assumptions of harvest sampling dispersion, sample sizes,
and harvest derivation calculations. We discuss potential sources of bias for each
derivation technique. Additionally, we use variograms, which illustrate the average
degree of similarity among number of individuals sampled as a function of distance
separating sampling locales (Rossi et al. 1992), to determine if the spatial pattern of

harvest samples collected across the state was consistent among years.

METHODS

Band Return Analyses
Sample Collection

We analyzed the harvest composition of banded Canada geese shot during the
regular hunting season in Michigan. Only direct recoveries of geese (i.e., banded bird
recovered the first hunting season after banding; Munro and Kimball 1982) shot or found
dead were included in the harvest derivation estimates. Counts of band recoveries from
each state or province were corrected for estimates of reporting rates for solicited and
unsolicited bands. Solicited band recoveries are those reported by someone other than
the hunter, such as natural resource agency employees, while unsolicited band recoveries
are reported directly by the hunter. As only limited information regarding differential

reporting rates of solicited verses unsolicited bands on geese is available, reporting rates

185

recorded for banded mallard ducks harvested during the same hunting period were used
as a general estimate of reporting rates for geese recoveries (reporting rate approximately
76%, Royal and Garrettson 2005). To estimate the actual number of banded geese
harvested from each state/province, reporting rates of solicited leg bands were assumed to
be 100%, while reporting rates of unsolicited leg bands or neck collars were assumed to
be 80%. No differential reporting rates for goose body size or population of origin were
included in the band recovery analysis, but could be incorporated in future derivations as
additional data become available. Sexes of geese were determined at the time of banding

by cloacal examination.

Harvest Derivation

Contributions of giant and interior Canada geese and cackling geese to
Michigan’s regular season harvests from 1998-2002 were derived using weighted band
recoveries (see Rusch et al. 1996), similar to methods outlined in Munro and Kimball
(1982) developed to estimate derivation of mallard harvests. Derivation of the harvest
was calculated by using estimates of spring population size and annual production to
estimate the fall flight of each population of geese. These population estimates were then
used to “weight” the number of bands applied to geese in each population, or estimate the
number of geese in a particular population represented by each banded bird that is shot
and reported. The number of band recoveries, multiplied by the respective weight for
each population, provided a weighted harvest estimate. Weighted estimates were then
converted into percentages to reflect the subspecies composition of the harvest within

Michigan (Rusch et al. 1996). Geese without young that were banded on migrant goose

186

(interior and cackling) breeding areas were considered possible molt migrants (Abraham
et al. 1999), and were excluded from derivations. We attempted to exclude molt-migrant
giant Canada geese from our samples of interior Canada geese by excluding geese judged
to be giant Canada geese based on morphological characters at the time of banding, (pers.
comm. Lyle Walton, Ontario Ministry of Natural resources) and by including only adults
banded within the same 10 minute degree block in which goslings were banded on the
same date as adults (obtained from banding records). Annual derivations were calculated
for adult or after hatch-year (AHY) geese: 1998 (n = 61), 1999 (n = 52), 2000 (n = 34),
2001 (n = 59), and 2002 (n = 48). AHY geese were distinguished from juveniles or

hatch-year (HY) geese based on banding records.

Morphometric Analyses
Sample Collection

Samples for this analysis were obtained through requests in the Michigan
Waterfowl Hunting Guide and through news releases and other advertisements (e.g..
postings in DNR field offices, personal communication to hunters). Participating hunters
provided goose heads and tails with cloaca attached to DNR field offices, and samples
were transferred to staff at the Rose Lake Pathology Laboratory for processing. Goose
gender and age class (HY or AHY) were determined by cloacal exam and cuhnen and
skull measurement techniques outlined in Dzubin and Cooch (1992). Mean culmen
lengths of harvested geese were first estimated in each of Michigan’s three hunting zones
for the period 1997-2002. Annual statewide mean culmen length was then estimated by

weighting zone means by the proportion of the Canada goose harvest that occurred in

187

each of Michigan’s three geographic hunting zones. This improved the accuracy of
statewide culmen length estimates by helping to prevent bias due to over- or under-
sampling harvest in a particular zone. A database containing estimates of Michigan’s

Canada goose harvest in each zone was provided by the USFWS for 1998-2002.

Harvest Derivation

Although other populations of interior geese contribute to Michigan’s harvest,
giants and MVP geese account for >90% of the statewide regular season harvest (Fritzell
and Luukkonen 2004). Thus, the contribution of interior and giant Canada geese to
annual regular season harvests was estimated from culmen measurement standards for
Michigan giant and Mississippi Valley Population (MVP) interior geese (Moser and
Rolley 1990) using the techniques described by Trost et al. (1992). Because of the
potential for incomplete development of HY geese during the regular season, we only
used measurements from AHY-aged birds in the analysis: 1998 (n = 142), 1999 (n =
179), 2000 (n = 232), 2001 (n = 164), and 2002 (n = 113). Female geese with cuhnen
measurements 541.3 mm and male geese with measurements 543.0 mm were assumed to
represent cackling geese (K. Bataille, Missouri Dept. of Conservation, personal
communication).

Similar to the statewide morphometric harvest derivation, only AHY geese were
used in the local analysis: 2000 (n = 117), 2001 (n = 126), and 2002 (n = 48). A
limitation of using Moser and Rolley's (1990) morphometric technique to estimate
harvest composition for early season samples from the Saginaw Bay area is that the

proportional contribution of only two populations can be estimated. This makes choosing

188

reference populations somewhat problematic in an area like Saginaw Bay that has more
than two potential populations contributing to harvest (e. g., Michigan giants, molt-
migrant giants from other states and provinces, Northwest James Bay MVP, Akimiski
Island SJBP, mainland SJBP, and Richardson’s cackling geese). Canada goose size
varies among populations (Moser and Rolley 1990), subpopulations (Leafloor and Rusch
1997), and sometimes within populations in response to environmental variation
(Leafloor et al. 1998). Significant differences in the size of Michigan giants has also
been described across different regions of the state (Soulliere et al. 1995). Consistent
with findings of Leafloor and Rusch (1997) with interior and other giant populations, the
largest Michigan giants occur in the southern part of the state. Thus, we needed to
accommodate multiple giant and interior reference populations. This problem was
further exacerbated by the relatively large body size of mainland SJBP, which results in
more overlap in size distributions with giant Canada goose populations (Leafloor and
Rusch 1997).

Culmen measurements were not available for all reference baseline populations of
Canada geese that may contribute to harvests in Saginaw Bay because Leafloor and
Rusch (1997) reported only skull measurements for Northwest James Bay MVP and
mainland SJ BP. Thus, mean culmen length was predicted from mean skull length using
values in the literature (Leafloor and Rusch 1997) and linear regression. Six sets of
means for culmen and skull length were taken from two studies (Moser and Rolley 1990
and Merendino et al. 1994) to develop equations to predict mean culmen length from
mean skull length. Mean culmen length was predicted from skull length for geese from

Northwest James Bay and the Southern James Bay mainland (Leafloor and Rusch 1997).

189

These estimates, and documented mean culmen length for Akimiski Island geese
(Merendino et al. 1994), were used as references to estimate and compare proportional
contribution of interior and giant populations in the early season harvest on the study

area.

Genetic Analyses
Sample Collection

Baseline samples were collected from pre-fleging goslings in breeding
populations of cackling geese, and interior and giant Canada geese according to methods
outlined in Scribner et al. (2003), and included all sampling locations outlined in Inman
et al. (2003). A total of 964 individuals were collected from baseline breeding
populations of cackling geese (n = 1 population), and interior (n = 4), and giant (n = 8)
Canada geese. Statewide goose harvest samples were collected during regular hunting
seasons (September 16 — December 31) in Michigan through the United States Fish and
Wildlife Service (U SF WS) Waterfowl Parts Survey (Martin and Carney 1977) (see
Inman et al. 2003 for Survey details). We obtained hunter-harvested samples of geese
collected in Michigan during regular seasons from 1998-2002. Each sample consisted of
1-12 tail feathers pulled from each goose harvested, and were identified by date and
county of harvest within Michigan. Similar to banding and morphometric investigations,
only samples from AHY geese were used for genetic analysis. Trained waterfowl
biologists identified the age class (AHY or HY) of each harvested sample by
independently aging tail feathers and primary feathers provided by participating hunters

(P. Padding, unpublished USFWS report). As primary wing tips provide a more accurate

190

assessment of age for Canada geese than tail feathers (Tacha et al. 1989), age
classification determined by primary feathers was first used to separate AHY and HY
samples. Remaining samples for which primary feathers were not included by the hunter
were separated into age classes according to tail feather aging.

Early season harvest samples from the Saginaw Bay area were submitted by area
hunters as part of the morphometric analysis of geese described above. We utilized
tongue tissue sampled from these AHY harvested geese for genetic analyses. Tongues
were clipped from geese during head and tail collections at regional DNR field offices.
Tongue samples were placed in individual vials containing a high-salt buffer (100 mM

Tris, pH 8.0, 100 mM EDTA, 0.5% sodium dodecyl sulfate, 50 mM NaCl) and were

frozen at -20°C until analyzed.

Lab Analysis

We extracted and quantified DNA from geese tail fan feathers using techniques
detailed in Inman et al. (2003). DNA was extracted from early season tongue samples
using Qiagen Dneasy kits and protocols (QIAGEN Incorporated, Valencia, CA). Five bi-
parentally inherited microsatellite DNA loci (Bcau7, Bcau9, Bcapl 1, and Hhiul
[Buchholz et al. 1998], and TTUCG-I [Cathey et al. 1998]) were used to estimate allele
and genotype frequencies for all baseline breeding populations and hunter-harvested
samples. Loci were amplified using polymerase chain reaction (PCR), and products were
electrophoresed on denaturing 6% polyacrylamide gels and visualized using a FMBIO II

laser scanner (Hitachi Software Engineering, Alameda, California, USA). Resulting

191

genotypes were scored based on 20 base-pair standards and reference samples of known
genotype.

Sex of all hunter-harvested samples was determined using the chromo-helicase-
DNA-binding (CHD) locus (Griffiths et al. 1998). Males were identified by the presence
of one amplified band (two introns of the same size), while females were characterized
by two bands (two introns of different sizes). For early season samples, sex was

determined by cloacal exam at the time morphometric measurement were taken.

Harvest Derivation

We conducted mixed stock analyses of annual harvests using conditional
maximum likelihood methods (Pella and Milner 1987) within the Statistics Program for
Analyzing Mixtures (SPAM 3.7b, Debuvec et al. 2000), which compares the distributions
of genotypic frequencies of each baseline population with the genotypic frequencies
observed in harvest mixtures (see Scribner et al. 2003 for details). We utilized a pseudo-
Bayes method (Pella and Masuda 2001) to estimate baseline allele frequency
distributions, which are calculated as the average of observed allele frequencies in each
population and the unweighted arithmetic mean of the allele frequencies among baseline
populations at each locus. All mean allele frequency estimates are positive as a result of
this calculation, so absence of an allele from a particular baseline population sample
implies it is only rare and was missed in sampling rather than assuming it is nonexistent.
Proportional contributions of interior and Canada geese and cackling geese were
calculated as the mean (iSD) over 1000 replicate resamplings (with replacement) of

baseline populations and the harvest mixture. Annual harvest derivations were calculated

192

for all AHY geese which included 754 individuals: 1998 (n = 142), 1999 (n = 158), 2000
(n = 111), 2001 (n = 126), and 2002 (n = 217). Harvest derivations for the three early
seasons in the Saginaw Bay area included 274 adult geese: 2000 (n = 102), 2001 (n =
124), and 2002 (n = 48).

Monte Carlo likelihood ratio tests were conducted in SPAM 3.7b (Debuvec et al.
2000) to test the equality of harvest mixtures (Reynolds and Templin 2004; see Chapter 5
Methods for summary of likelihood ratio tests assumptions and design). Likelihood ratio
tests were completed for statewide regular season harvests and early season harvests from
Saginaw Bay. For each likelihood ratio test, parametric bootstrapping was used to test
the null hypothesis that the harvest samples collected each year came from a common
mixture (n = 5000 simulations). Mixture simulations and model fitting were done in
SPAM 3.7b, and final analysis of simulation results was conducted in Microsoft Excel

(Microsoft Office, Microsofi, Inc., Redmond, WA.).

Variograms

We used variograms (Rossi et al. 1992) to test for consistency among annual
spatial distribution patterns of regular season harvest samples for morphometric datasets
and for genetic-based datasets. Each variogram was constructed as a plot of half the
average squared difference among harvest sample sizes for counties separated by about
the same distance (distance between county pairs). Geographic distances between each
pair of Michigan counties were calculated using x,y coordinates (in meters) from the
MIGeoref projection (MDNR Spatial Data Library 2005). Variograms were constructed

using the program SAS (SAS Institute 2005). Ten distance classes were included in each

193

variogram, and were determined by the lag distance (40 km) multiplied by the number of
lags, or distance classes. Distance classes included harvest samples from pairwise county
comparisons whose geographic distance between counties were within each distance

interval :1: half the lag distance.

Variance in Harvest Estimates

Several factors may influence the variance and bias in harvest estimates
regardless of the technique used to derive these estimates. First, all derivation techniques
assume that all populations putatively contributing to the harveSt have been adequately
sampled in terms of numbers of individuals collected and spatial distribution of the
samples. Non-representative sampling of baseline populations (numerical or geographic)
contributing to harvest may increase variance of harvest estimates (Crissey 1955, Dufour
et al. 1993, F ritzell and Luukkonen 2004). A second assumption is that samples collected
from harvest mixtures adequately characterize the true harvest mixture. The accuracy
and precision of harvest derivation estimates could be compromised if harvest samples do
not adequately cover locations where harvest is occurring, or if sample sizes are too low
to include all populations represented in the mixture. Thus, it is important that
derivations incorporate a comprehensive and systematic sampling scheme for collecting
harvested individuals to reduce possible sampling bias (Inman et al. 2003). A third issue
is that all three derivation techniques rely on voluntary contributions of samples from
hunters. There is a potential to increase bias in harvest estimates if there are temporal or
spatial inconsistencies in volunteer contributions. For instance, there may be a tendency

for hunter effort in submitting samples to decrease over the season, submitting more

194

samples early in the harvest and less toward the end of the season as their interest
declines. One example of this bias could arise for genetic samples collected through the
USFWS Waterfowl Parts Survey, as hunters who run out of return envelopes may not
request more even if they harvest more geese. It is also possible that hunters selectively
submit samples based on their impression of how “good” an individual bird may be in
terms of coloration or size. Selectively submitting samples in this manner may occur
more often with morphometic techniques, as whole birds are generally handed to
biologists for sampling. Finally, the accuracy and precision of harvest estimates is highly
dependent on the number of potentially contributing populations, whether all the
potentially contributing populations are known, and the actual composition of the mixture
(Pella and Milner 1987). Based on numerous previous studies of cackling and Canada
geese, we are confident that only three subspecies (B. h. hutchinsii, B. c. interior, B. c.
maxima) are potentially harvested in Michigan during annual hunts, thus minimizing
error in harvest estimates.

Harvest derivation calculations based on band returns for AHY geese include
estimations of the size of spring adult cackling and Canada geese populations. Interior
Canada geese population estimates were based on aerial surveys of stratified, fixed-wing
transects conducted in early spring on breeding grounds, while giant Canada geese
population estimates were based on transect counts from fixed-wing aircraft, helicopters,
or ground checking during nesting periods on breeding grounds. Estimates of the spring
population size of adult cackling geese were based on helicopter transect counts
conducted on Baffin Island breeding grounds in the fall prior to the southward migration

of geese. While interior and giant Canada geese population transect counts would

195

include only AHY birds, fall counts of cackling geese would include both AHY and HY
individuals. Estimates of the spring goose population for each subspecies are likely the
predominant source of variation in adult harvest estimates based on band recoveries. The
precision and accuracy of harvest estimates based on band recoveries may also be
influenced by temporal and geographic differences in banding efforts (F ritzell and
Luukkonen 2004) and band recovery and reporting rates (Conroy and Blandin 1984,
Caswell et al. 1987, Royal and Dubovsky 2001). We attempted to limit variance
attributed to band reporting rates by including rate correction factors for solicited and
unsolicited bands based on previous band reporting studies (Royal and Garrettson 2005).
Sources of variation in harvest estimates based on culmen measurements include
variation in culmen lengths among populations of interior Canada geese (Merendino et al.
1994, Leafloor et al. 1996, Leafloor and Rusch 1997) and among populations of giant
Canada geese contributing to harvests (Moser and Rolley 1990); variation among
individual culmen lengths within populations due to environmental influences (Leafloor
et al. 1998) and age of individuals (Thompson et al. 1999); and measurement error of
harvest samples. Measurement error includes variation in measurements conducted by
different observers (Rasmussen et al. 2001), though this possible error was addressed by
limiting the number of people conducting measurements (2 individuals). Errors in
assigning individual birds to sex and age categories could also introduce variation and
bias into harvest estimates as subspecies classification was based on culmen
measurement standards for AHY geese, and female and male standards differ within a
subspecies. We attempted to limit this type of error by restricting our observers to only

those individuals that were very experienced in sexing and aging techniques.

196

The level of variance and bias associated with genetic-based harvest estimates is
greatly dependent on the amount of genetic divergence among potentially contributing
populations of geese (Pella and Milner 1987, Wood et a1. 1987). Bias in estimates will be
largest when baseline populations that are similar in their genetic characteristics differ
greatly in abundance (Millar 1987). We attempted to limit this bias by employing a suite
of microsatellite markers which have demonstrated differences in allele frequencies
among cackling geese, and interior and Canada geese (Inman et al. 2003). If the true
composition of a harvest mixture is highly skewed toward one population (near 100%).
genetic-based derivations may underestimate harvest contributions of the prevalent
population and overestimate harvest contributions of other putative populations, even
when baseline populations are sufficiently differentiated by a suite of genetic markers
(Marlowe and Busack I995, Scribner et al. 2003). The program SPAM used to calculate
harvest composition accounts for the effects of baseline and mixture sampling variation
by repeatedly drawing samples with replacement from the baseline and mixture groups so
final harvest composition estimates are a mean of multiple likelihood estimates, with
related confidence intervals. Additionally, we addressed the issue of sampling variance
leading to potential bias of genetic-based harvest estimates by having a large number of

samples representing baseline populations and harvest mixtures.

RESULTS
Based on USFWS harvest estimates, Michigan hunters took an estimated 57,200,
50,100, 57,800, 55,600, and 52,400 Canada geese during the 1998, 1999, 2000, 2001, and

2002 regular seasons, respectively. Variograms for regular season, statewide harvests

197

illustrated less variance in the sample size of harvested geese among counties in close
geographic proximity to one another than among counties separated by larger distances
for morphology- and genetic-based analyses (Figure l). The variance in harvest sample
size increased in a curvilinear fashion as a function of increasing distance between
counties sampled. This pattern was consistent across all five years of harvest sampling,
although the range of sampling variance among years was slightly greater for the

morphology sample dataset than the genetic sample dataset.

Band Recovery Analyses

A total of 254 bands were recovered for adult geese from the 1998-2002 regular
hunt seasons. No banded cackling geese were recovered in any of the five harvest
seasons in Michigan. Estimated proportional harvests of interior Canada geese were
greater than estimated proportions of giant Canada geese harvested in 1999 (interiors:
59.3%), 2000 (62.3%), and 2002 (51.7%) (Table I). In 1998 and 2001, estimated
proportions of giant Canada geese harvested were 53.7% and 61.3%, respectively (Table
1). Sex ratios of the banded harvest were male dominated in 1998 (female to male ratio:

0.65), 1999 (0.77), and 2001 (0.83), and female dominated in 2000 (1.19) and 2002 (1.5).

Morphometric Analyses

Based on culmen measurements from both sexes, the proportion of giants in the
Michigan regular season harvest has increased since the shift of the regular season
opening to mid-September which occurred after 1997 (46% giants). Proportional

harvests of giants were estimated to be 61% in 1998, 76% in 1999, 83% in 2000 and 90%

198

 

 

 

 

 

 

 

 

 

 

 

 

 

 

5 ~ - "—1
4 -._ , — - -2 A f 2 ‘ 1
E I ' ? A1998
3 a . ---— - : ' P- : * -
I. O
a e O S ' ' 2000
> S X '
.2 2 .. -__.;__. - t ; Q- 8 J 8 i .2001
5 x x x 9 XZOOZ
1- - H-
O . 3
0 2 4 6 8 1O 12
Lag Distance (40 km units)
Morphology Samples
5 l
l
4 — ——————-w —— --+,
E x ’o‘ x x . ‘ A1998
3 3 -»--— - 2 : W2: ’ if T 3 ' . 1999
.3 3 U . | 2 ‘ g 0 ‘ f uzooo
.; 2 +_ ._.-_-___ -» h" T .2001
g j x2002
1 _ W l L L 2 2 1
0 i . . J
0 2 4 6 8 10 12
Lag Distance (40 km units)

 

 

Genetic Samples

Figure l. Omnidirectional variograms for the spatial distribution of two collections of
regular season harvest samples of cackling geese and Canada geese in Michigan from
1998-2002. Harvest samples used for morphology-based analyses were collected through
the Michigan Department of Natural Resources, and harvest samples used for genetic-

based analyses were collected through the United States Fish and Wildlife Service

Waterfowl Parts Survey.

199

 

 

in 2001 (Table 1). During 2002, the proportion of giants in the harvest declined to 61%,
but remained above the estimate in 1997 and was similar to the 1998 estimate. The
proportion of samples classified as cackling geese was highest in 1997 and lowest in
1999 (Table 1). However, this result is based on small numbers of geese meeting the
culmen length criteria for this subspecies (n = 22 from 1997-2002). Estimated interior
Canada goose harvest decreased each year from 1998 through 2001 while estimated giant
harvest increased (Table 1). Based on morphology samples, estimates of adult sex ratios
for regular season harvests were strongly male biased in 1998 (female to male ratio: 0.53)
and 1999 (0.64). Approximately equal harvests of the sexes occurred during 2000 (1.11),
2001 (1.08), and 2002 (0.95).

Culmen measurements were collected from 291 AHY birds in the Saginaw Bay
area during the three years of early seasons (Table 2). No geese met the culmen
measurement criteria for cackling geese. Estimates of mean adult culmen lengths from
the harvest sample differed from means reported for Michigan giants (Moser and Rolley
1990) by less than 0.5 mm for both sexes and were larger than reference means from
northern Michigan giants (Soulliere et al. 1995).

Estimates of the contribution of giant Canada geese in the early season harvest
ranged from 90 to 100% for females and from 87 to 100% for males depending on
reference populations in the morphometric analysis and sample year (Table 2). The
lowest estimate of the contribution of giants was for male harvest samples in 2000 under
the assumption that all geese harvested were either giants from Michigan only or interiors
from the Southern James Bay mainland. Michigan giants have relatively large culmens

(Moser and Rolley 1990) and geese from the Southern James Bay mainland were the

200

Table 1. Comparison of harvest estimates (%) of adult cackling (B. h. hutchinsii) and
Canada (B. c. interior and B. c. maxima) geese based on band recovery, morphometric,

and genetic derivation techniques for the regular hunt season in Michigan from
1998-2002. Genetic estimates include standard deviation in parentheses based on 1000
bootstrap replicates. Sample sizes are listed below estimates.

 

 

Band Recovery Morphometric Genetic Analysis
1998
Be. interior 46.3 36.3 27.7 (8.1)
Be. maxima 53.7 61.3 68.7 (8.2)
B.h. hutchinsii 0.0 2.4 3.6 (2.5)
(n=61) (n=142) (n=142)
1999
Be. interior 59.3 24.2 24.8 (7.5)
Ba maxima 40.7 75.8 73.5 (7.7)
B.h. hutchinsii 0.0 0.0 1.8 (1.8)
(n = 52) (n = 179) (n = 158)
2000
Be. interior 62.3 16.0 45 (10.2)
B.c. maxima 37.7 83.6 54.2 (10.3)
B.h. hutchinsii 0.0 0.4 0.8 (1.6)
(n=34) (n=232) (n=lll)
2001
BC. interior 38.7 8.6 29.1 (8.8)
RC. maxima 61.3 90.3 69.7 (8.8)
B.h. hutchinsii 0.0 1.1 1.3 (1.9)
(n = 59) (n = 164) (n = 126)
2002
Be. interior 51.7 37.8 49.5 (7.1)
RC. maxima 48.3 61.1 50.4 (7.1)
B.h. hutchinsii 0.0 1.1 0.1 (0.5)
(n=48) (n=113) (n=217)

 

201

Table 2. Comparison of harvest estimates (%) of adult cackling
(B. h. hutchinsii) and Canada (B. c. interior and B. c. maxima)

geese based on genetic and morphometric derivation techniques
for the early hunt season in Saginaw Bay from 2000-2002.
Morphometric harvest estimates were calculated using three
different interior and two giant culmen reference lengths-estirnates
listed in this table represent the range of results from these six
morphometric derivations. Genetic estimates include standard
deviation in parentheses based on 1000 bootstrap replicates.

Samples sizes are listed below estimates.

 

Genetic Analysis

Morphometric

 

2000
B. c. interior

B. c. maxima
B. h. hutchinsii

200 1
B. c. interior

B. c. maxima
B. h. hutchinsii

2002
B. c. interior

B. c. maxima
B. h. hutchinsii

20.6 (8.3)
79.1 (8.6)
0.3 (1.2)
(n = 102)

30.6 (7.9)
69.1 (8.0)
0.3 (0.9)
(n=124)

12.9 (9.4)
86.5 (9.4)
0.6 (1.5)
(n=48)

2-13
87-98
0
(n = 117)

0-3
97-100
(n = 126)
0-7
93-100

(n = 48)

 

largest interiors recorded. Using the largest reference means for interior Canada geese

results in the most conservative estimate of giants in the calculated harvest composition.

Assuming that interior geese migrating through the study area are affiliated with SJBP

Akimiski Island and Southern James Bay mainland in proportion to abundance on the

breeding ground (about 30% of SJBP occur on Akimiski), then a reference size based on

the weighted average of the two SJ BP segments may be the most realistic interior
reference. Using this SJBP weighted average reference, the southern Michigan giant
reference, and the average culmen sizes of samples from adult birds, an estimated 98% of
female and 96% of male geese were giants for the entire 3-year experiment. Sex ratios of
harvested samples collected for morphometric analyses were male biased in 2000

(females to males: 0.67) and 2002 (0.41), and female biased in 2001 (1.17).

Genetic Analyses

Statewide distributions of Canada geese regular season harvest samples varied
among years, and included 40 counties in 1998, 42 counties in 1999, 27 counties in 2000,
33 counties in 2001, and 44 counties in 2002. Regular season harvests of adults were
composed mainly of giant Canada geese (50.4% - 73.5%), with lesser proportions of
interior (24.8% - 49.5%) Canada geese and cackling geese (0.1% - 3.6%) (Table 1).
Estimates of giant harvest were greatest in 2001, but did not indicate harvests of resident
geese steadily increased from 1998-2001 as illustrated by morphometric derivation
results. Contributions of cackling geese to regular season harvests varied among years,
but were always less than 10% of the total harvest (Table 1). Likelihood ratio tests of
regular season harvests indicated that proportional estimates of adult giant and interior
Canada geese and cackling geese harvested annually differed significantly among years
(P = 0.039). Estimates of adult sex ratios for regular season harvests were slightly male
biased in 2001 and 2002. Approximately equal proportions of males and females were
harvested during 1998, 1999, and 2000. Specific annual ratios of females to males were

as follows: 1998 (0.937), 1999 (1.020), 2000 (0.929), 2001 (0.800), 2002 (0.807).

203

Tongues were clipped for genetic-based analyses from 274 of the adult geese
collected for morphometric analyses. Similar to morphometric analyses, proportional
harvest estimates were greater for giant Canada geese than interior Canada geese during
all three years (Table 2). However, proportional harvest estimates of giant Canada geese
based on genetic analyses (69.1 -86.5%) were consistently lower than morphometric-
based estimates. Estimated proportions of interior Canada geese harvested during the
three early seasons were higher than estimates from morphometric derivations, and
ranged between 12.9 — 30.6%. Cackling geese were estimated to compose than 1% of the
early season harvest each year (Table 1). Likelihood ratio tests revealed no significant
differences in proportional harvest estimates of cackling geese and interior and giant
Canada geese among years for genetic derivations (P = 508). Female to male sex ratios
in early season harvests as determined by genetic sexing for 2000 (females to males:
0.54) and 2002 (0.34) samples were similar to sex ratios determined by cloacal
examination in the field. The sex ratio for 2001 was 0.91 , which contrasted with the

morphology-based sex ratio of 1.17 for the same year.

DISCUSSION
All three derivation techniques provide estimates of the harvest composition and
it would be inappropriate to view estimates derived from one type of data as a “check” on
estimates derived from other datasets. However, we have greater confidence in our
harvest derivations when these independently obtained estimates agree with each other.
Regular season harvest estimates of each subspecies were most similar among all three

derivations in 1998 and 2002, and least similar in 2000. In 1999, morphometric and

204

genetic derivations produced equivalent results, while estimates of harvests were more
comparable between band recovery and genetic derivations in 2001.

Harvest estimates may be consistent across techniques in some years and
disparate in others due to inadequate baseline collections. Baseline populations included
samples from relatively small portions of the geographic ranges for interior Canada geese
and cackling geese, as sampling focused on coastal areas and excluded most inland
populations. Giant Canada geese in the baseline were represented by Michigan
populations only. Influxes of birds from populations that were not represented in the
baseline (e. g., interior Canada geese and cackling geese from inland breeding areas, giant
Canada geese from Ontario) may result in harvest estimates that are not concordant
among multiple derivations, especially if the ability to correctly categorize birds from
non-represented populations varies among derivation techniques.

Potential temporal and geographic biases of harvest mixture samples discussed in
the methods may also lead to differences among harvest estimates derived from various
techniques. We may assume that in years of concordant estimates harvest samples
collected independently for each derivation are more likely to represent similar
geographic areas and temporal coverage of the regular season harvest than in years where
estimates disagree among derivation results. It is possible that in years of discordant
estimates, harvest samples for each derivation do not equally represent the harvest
mixture in time and/or space. Temporal and spatial sampling biases may be created by
hunters in terms of selectively submitting samples, or failing to submit samples later in
the season due to hunter “burn out”. Geographic and temporal biases may also be present

in the sampling methodology for each derivation method. However, sampling biases

205

inherent in each technique would not likely change from year to year, and differences
among annual harvest estimates derived from each technique would then be consistent
over the five years analyzed. In addition, spatial patterns of variance for harvest samples
were similar between morphology and genetic collections among years of concordant and
discordant estimates, as indicated by comparative variograms (Figure 1).

Band recovery derivations consistently estimated higher proportional regular
season harvests of interior Canada geese as compared to harvest estimates of interiors
based on morphometric and genetic derivations. Estimates of interiors harvested were
lowest, and harvest estimates of giant Canada geese were highest, in 4 of the 5 years
analyzed based on morphometric derviations as compared to annual results of the other
two techniques. Morphology-based harvest derivations may underestimate contributions
of interior Canada geese, as the morphometric technique employed (Moser and Rolley
1990) is designed to separate only two populations of geese. This technique is difficult to
apply to the multiple populations of interior Canada geese that have measurable
differences in size (Leafloor and Rusch 1997, Leafloor et al. 1998), and likely contribute
to annual harvests in Michigan. Observed and potentially unobserved geographic
variation in the size of interiors may skew harvest estimates of interior Canada geese
based on morphometric derivations.

Regular season harvests of cackling geese were estimated to be less than 5% of
the harvest in all years for all derivations. However, no cackling geese were recorded
during the five regular seasons of band recoveries, indicating no presence of this
subspecies in the harvests. The annual number of individuals sampled through band

recoveries was much less than the annual sample sizes for morphometric and genetic

206

derivations. If cackling geese are rare but present in harvests as indicated by
morphometric and genetic analyses, larger numbers of band returns may be necessary to
detect low levels of these geese.

Early season harvest comparisons utilized the same set of harvest samples for
morphometric and genetic derivations, reducing potential sampling biases between the
two derivations. Both early season derivations resulted in relatively high estimates of
proportions of giant Canada geese as compared to interior Canada geese harvest
estimates, but morphology-based analyses resulted in consistently higher estimates of
giants and lower estimates of interiors than harvest results derived from genetic
characteristics. This suggests that there may be underlying differences in these two
techniques causing bias in one, or both, sets of estimates. Bias may result from
inadequate sampling of interior and/or giant populations used to develop baseline
references for morphometric and genetic derivations, as discussed above for the regular
season harvest. The problem of applying morphometric analysis to mixtures potentially
involving more than two populations is also relevant for early season harvests, although
multiple culmen measurements for interior and Canada geese were used in this analysis
to minimize potential error.

The efficacy of using each of the derivation techniques may vary depending on
the goals of the harvest evaluation. For instance, only adult harvests can be assessed
using morphology-based methods, whereas band recovery and genetic harvest derivations
can also incorporate estimates of juvenile harvests or combined harvests of adults and
juveniles. Although only adults were included in our analyses for comparative purposes,

annual sample collections of geese from regular season harvests in Michigan were

207

composed of 19-3 5% juveniles in genetic datasets, 13-56% juveniles in morphology
datasets, and 67-73% juveniles in band recovery datasets. Early season harvest
collections included 34-44% juveniles. By excluding juveniles, we lose a large number
of harvest samples that may be informative in harvest assessments and potentially skew

estimates of harvest composition for each subspecies.

MANAGEMENT IMPLICATIONS

It is challenging to manage a highly mobile biological resource like cackling
geese and Canada geese that are distributed across heterogeneous, changing
environments, and characterized by complex, only partially recognized
biotic/environmental interactions (Williams and Nichols 1990). Successful management
of these species requires accurate and precise estimates of harvest composition that may
be used to direct future harvest policies. Partially blind management, based on imprecise
and/or biased monitoring programs, will fail to consistently recognize the need for
harvest restrictions, or to respond to harvest opportunities when they arise (Williams et
a1. 1996). It is therefore important to contrast harvest estimates derived by multiple
techniques currently being used to evaluate the harvest of cackling and Canada geese in
order to investigate potential biases associated with each derivation method.

Band recovery, morphometric, and genetic analyses all provide estimates of the
“true” harvest composition. It is difficult, however, to assess the accuracy and precision
of estimates calculated from each derivation technique if only one type of derivation is
used to analyze a harvest mixture. By comparing the results of multiple harvest

derivation techniques applied to goose harvests in Michigan over similar temporal and

208

geographic scales, we were able to assess how potential biases associated with each
derivation technique may influence resulting harvest estimates. Additional study is
needed to determine why estimates among techniques are concordant in some years, and
dissimilar in others, and to fully explain differences in derivation patterns that are
consistent among years (i.e., higher proportional estimates of giant Canada geese

resulting from morphometric derivations) in order to improve harvest monitoring.

ACKNOWLEDGEMENTS

Funding for this study was provided by the Michigan Department of Natural
Resources (MDNR) and through the Partnership for Ecosystem Research and
Management (PERM) Program between the MDNR and the Department of Fisheries and
Wildlife at Michigan State University. Baseline samples were collected by many
individuals associated with the MDNR, Michigan State University, and individuals from
the Canadian Wildlife Service and Environment Canada including K. Abraham, D.
Caswell, K. Dickson, and J. Leafloor. We thank all participants of the Mississippi
F lyway Wing Bee who assisted in cataloging hunter-harvested samples collected by the
US. Fish and Wildlife Service, especially Barb Lercel Avers and Bill Scullon of the

MDNR, and Dave Rave of the Minnesota DNR.

209

LITERATURE CITED

Abraham, K. F., J. O. Leafloor, and D. H. Rusch. 1999. Molt migrant Canada geese in
northern Ontario and western James Bay. Journal of Wildlife Management 63:649-655.

Ankney, C. D. 1996. Invited paper: an embaressment of riches: too many geese. Journal
of Wildlife Management 60:217-223.

Banks, R. C., C. Cicero, J. L. Dunn, A. W. Kratter, P. C. Rasmussen, J. V. Remsen,.Jr., J.
D. Rising, and D. F. Stotz. 2003. Forty-fifth supplement to the American ornithologists’
union check-list of North American birds. The Auk 121:985-995.

Bellrose, F. C., T. G. Scott, A. S. Hawkins, and J. B. Low. 1961. Sex ratios and age
ratios in North American Ducks. Illinois Natural History Survey Bulletin 27:391-474.

Bellrose, F. C. 1980. Ducks, geese, and swans of North America. Second edition.
Stackpole, Harrisburg, Pennsylvania, USA.

Buchholz, W., J. M. Pearce, B. J. Pierson, and K. T. Scribner. 1998. Dinucleotide repeat
polymorphisms in waterfowl (family Anatidae): characterization of a sex-linked (Z-
specific) and 14 bi-parentally inherited loci. Animal Genetics 292323-325.

Caccamise, D. F., L. M. Reed, P. M. Castelli, S. Wainright, and T. C. Nichos. 2000.
Distinguishing migratory and resident Canada geese using stable isotope analysis.
Journal of Wildlife Management 64: 1084-1091.

Caswell, F. D., G. S. Hochbaum, D. J. Nieman, and B. C. Turner. 1987. Temporal and
geographic differences of mallard survival/recovery rates in prairie Canada. Transactions
of the North American Wildlife and Natural Resources Conference 52:284-297.

Cathey, J. C., J. A. DeWoody, and L. M. Smith. 1998. Microsatellite markers in Canada
geese (Branta canadensis). Journal of Heredity 89:173-174.

Conroy, M. J. and W. Blandin. 1984. Geographic and temporal differences in band
reporting rates for American black ducks. Journal of Wildlife Management 48:23-36.

Crissey, W. F. 1955. The use of banding data in determining waterfowl migration and
distribution. Journal of Wildlife Management 19:75-84.

Debuvec, E. M., R. B. Gates, M. Masuda, J. Pella, J. H. Reynolds, and L. W. Seeb. 2000.
SPAM (version 3.2): Statistics program for analyzing mixtures. Journal of Heredity
91 :509-510.

Dufour, K. W., C. D. Ankney, and P. J. Weatherhead. 1993. Nonrepresentative

sampling during waterfowl banding: emphasis on body condition. Journal of Wildlife
Management 57:741-751.

210

Dzubin, A., and E. G. Cooch. 1992. Measurements of geese: general field methods.
California Waterfowl Association, Sacramento, California, USA.

F ritzell, Jr., P. A. and D. R. Luukkonen. 2004. Influence of banding distribution on
harvest indices of Mississippi Valley Population Canada geese. Pages 51-59 in T. J.
Moser, R. D. Lien, K. C. VerCauteren, K. F. Abraham, D. E. Andersen, J. G. Bruggink, J.
M. Coluccy, D. A. Graber, J. O. Leafloor, D. R. Luukkonen, and R. E. Trost editors.
Proceedings of the 2003 International Canada Goose Symposium, Madison, Wisconsin,
USA.

Griffiths, R., M. C. Double, K. Orr, and R. J. G. Dawson. 1998. A DNA test to sex most
birds. Molecular Ecology 7:1071-1076.

Inman, R. L., K. T. Scribner, H. H. Prince, J. A. Warrillow, D. R. Luukkonen, P. I.
Padding. 2003. A novel method for Canada goose harvest derivation using genetic
analysis of tail feathers. Wildlife Society Bulletin 31:1126-1131.

Johnson, D. H., D. E. Tim, and P. F. Springer. 1979. Morphological characteristics of
Canada geese in the Pacific Flyway. Pages 56-76 in RP. Jarvis and J .C. Bartonek,
editors. Management and biology of Pacific flyway geese: a symposium. Oregon State
University Book Stores, Corvallis, Oregon, USA.

Leafloor, J. O., A. E. Smith, and J. C. Wood. 1996. Hunting vulnerability of local and
migrant Canada geese: a comment. Journal of Wildlife Management 60:452-457.

Leafloor, J. O., and D. H. Rusch. 1997. Clinal size variation in Canada geese affects
morphometric discrimination techniques. Journal of Wildlife Management. 61 :183-190.
Leafloor, J. O., C. D. Ankney, and D. H. Rusch. 1998. Environmental effects on body
size of Canada geese. The Auk. 115:26-33.

Lindberg, M. S. and R. A. Malecki. 1994. Hunting vulnerability of local and migrant
Canada geese in Pennsylvania. Journal of Wildlife Management 58:740-747.

Luukkonen, D. R., H. H. Prince, and R. C. Mykut. 2004. Harvest of molt migrant giant
Canada geese from southern Michigan. Pages 178 in T. J. Moser, R. D. Lien, K. C.
VerCauteren, K. F. Abraham, D. E. Andersen, J. G. Bruggink, J. M. Coluccy, D. A.
Graber, J. O. Leafloor, D. R. Luukkonen, and R. E. Trost editors. Proceedings of the
2003 International Canada Goose Symposium, Madison, Wisconsin, USA.

Luukkonen, D. R. and G. J. Soulliere. 2004. Change in harvest composition resulting
from mid-September opening of Michigan’s regular goose season, 1998-2002. Michigan
Department of Natural Resources - Wildlife Division, Lansing, Michigan, USA.

Marlowe, C. and C. Busack. 1995. The effect of decreasing sample size on the precision
of GSI stock composition estimates for chinook salmon (Oncorhynchus tshawytscha):

211

using data from the Washington coastal and Strait of Juan de F uca troll fisheries 1989-90.
Washington Department of Fisheries, Project Report Series No. 2, Olympia, WA, USA.

Martin, C. M. and S. M. Carney. 1977. Population ecology of the mallard: VI. A review
of duck hunting regulations, activity, and success, with special reference to the mallard.
United States Fish and Wildlife Service, Resource Publication 130, Washington, D. C.,
USA.

Millar, R. B. 1987. Maximum likelihood estimation of mixed stock fishery composition.
Canadian Journal of Fisheries and Aquatic Science 44:583-590.

Merendino, M. T., C. D. Ankney, D. G. Dennis, and J. O. Leafloor. I994. Morphometric
discrimination of giant and Akimiski Island Canada geese. Wildlife Society Bulletin.
22:14-19.

Moser T. J ., and R. E. Rolley. 1990. Discrimination of giant and interior Canada geese
of the Mississippi Flyway. Wildlife Society Bulletin. 18:381-388.

Munro, RE. and C.F.Kimball. 1982. Population ecology of the mallard VII.
Distribution

and derivation of the harvest. US Department of the Interior. Resource publication
147. Washington DC, USA.

Pearce, J. M., B. J. Pierson, S. L. Talbot, D. V. Derksen, D. Kraege, and K. T. Scribner.
2000. A genetic evaluation of morphology used to identify harvested Canada geese.
Journal of Wildlife Management 64: 863-874.

Pella, J. J. and M. Masuda. 2001. Bayesian methods for analysis of stock mixtures from
genetic characters. Fisheries Bulletin 99:151-167.

Pella, J. J. and G. B. Milner. 1987. Use of genetic marks in stock composition analysis.
Pages 247-276 in N. Rymand and F. M. Utter, editors. Population genetics and fishery
management. University of Washington, Seattle, Washington, USA.

Rasmussen, P. W., W. E. Wheeler, T. J. Moser, L. E. Vine, B. D. Sullivan, and D. H.
Rusch. 2001. Measurements of Canada goose morphology; sources of error and effects
on classification of subspecies. Journal of Wildlife Management 65:716-725.

Reynolds, J. H. and W. D.Templin. 2004. Comparing mixture estimates by parametric
bootstrapping likelihood ratios. Journal of agricultural, biological, and environmental
statistics 9257-74.

Rossi, R. E., D. J. Mulla, A. G. Journel, and E. H. Franz. 1992. Geostatistical tools for

modeling and interpreting ecological spatial dependence. Ecological Monographs
62:277-314.

212

Royle, J. A. and J. A. Dubovsky. 2001. Modeling spatial variation in waterfowl band-
recovery data. Journal of Wildlife Management 65:726-737.

Royle, J. A. and P. R. Garrettson. 2005. The effect of reward band value on
midcontinent mallard band reporting rates. Journal of Wildlife Management. In press.

Rusch, D. H., F. D. Caswell, M. M. Gillespie, and J. O. Leafloor. 1996. Research
contributions to management of Canada geese in the Mississippi F lyway. Transactions of
the North American Wildlife and Natural Resources Conference 61 :43 7-449.

Scribner, K. T., J. A. Warrillow, J. O. Leafloor, H. H. Prince, R. L. Inman, D. R.
Luukkonen, and C. S. F legel. 2003. A genetic method for determining racial
composition of Canada goose harvests. Journal of Wildlife Management 67:122-135.

Soulliere, G. J ., G. F. Martz, and T. E. Maples. 1995. Morphometric comparison of
resident giant Canada geese in northern and southern Michigan. Michigan Department of
Natural Resources-Wildlife Division, Lansing. Report Number 3225.

Soulliere, G. J ., and G. F. Martz. 1997. Evaluation of experimental late Canada goose
seasons in Michigan: a final report to the U. S. Fish and Wildlife Service and Mississippi
F lyway Council. Michigan Department of Natural Resources-Wildlife Division, Lansing.
Report Number 3263.

Tacha, T. C., D. D. Thomburg, and R. A. Williamson. 1989. Use of wing-tips to
estimate age ratios of Canada geese. Wildlife Society Bulletin 17:146-148.

Thompson, J. E., M. R. Hill, M. T. Meredino, and C. D. Ankney. 1999. Improving use
of morphometric discrimination to identity Canada goose subspecies. Wildlife Society
Bulletin 27:274-280.

Trost, R. E., T. J. Moser, M. S. Lindberg, and P. M. Castelli. 1992. Evaluating the use of
morphological measurements of Canada geese to determine breeding-ground affiliations.
United States Fish and Wildlife Service, Office of Migratory Bird Management Interim
Report.

Williams, B. K. and J. D. Nichols. 1990. Modeling and the management of migratory
birds. Natural Resource Modeling 4:273-311.

Williams, B. K. and F. A. Johnson. 1995. Adaptive management and the regulation of
waterfowl harvests. Wildlife Society Bulletin 23:430-436.

Williams, B. K., F. A. Johnson, and K. Wilkins. 1996. Uncertainty and the adaptive
management of waterfowl harvests. Journal of Wildlife Management 60:223-232.

213

Wood, C. C., and S. McKinnell, T. J. Mulligan, and D. A. Fournier. 1987. Stock
identification with the maximum-likelihood mixture model: sensitivity analysis and
application to complex problems.

214

CHAPTER 7: DISPERSAL AND GENE FLOW AMONG WRANGEL ISLAND AND

BANKS ISLAND LESSER SNOW GEESE

INTRODUCTION

Exponential growth of Mid-Continent lesser snow geese (Chen caerulescens
caerulescens) populations (Ankney 1996, Batt 1997), and some Western populations
(Kerbes et al. 1999) over the past 30 years have led to expansion of population
boundaries in both breeding and wintering areas (Alisauskas 1998). Expansions of lesser
snow geese populations could increase dispersal and potential gene flow among breeding
colonies and/or wintering groups. Current management policies directed to control
overabundant populations through harvest on wintering grounds and migratory pathways
when populations are mixed could significantly impact the diversity and long-term
viability of lesser snow geese if population dynamics among breeding and wintering
groups are not well understood.

The two largest breeding colonies of lesser snow geese in the Western Arctic are
currently experiencing very different population trends. The population nesting on
Wrangel Island (WI) off the northeast coast of Russia has greatly declined in abundance
over the past 35 years. Population numbers have decreased from 150,000 geese in the
late 1960’s to 65,000 in the mid-1990’s. In contrast, the colony on Banks Island, Canada
has grown from 170,000 individuals to 486,000 individuals during the same time period
(Kerbes et al. 1999, Hines et al. 1999b). The WI breeding colony consists of two
wintering groups. The northern group which comprises an estimated 50% of the total

Wrangel population migrates to the Fraser River Delta, British Columbia and the Skagit

215

River Delta, Washington. The southern group migrates 600 km farther south to winter in
California’s Central Valley (Figure l, Kuznetsov et al. 1998, Armstrong et al. 1999). The
southern WI group (SW1) shares its wintering grounds with approximately 75% of the
lesser snow geese population from BI and other smaller Western Canadian Arctic
breeding populations, while the northern WI wintering group (NW1) is isolated from
other breeding populations (Syroechkovsky et al. 1994, Hines et al. 1999a). In the spring,
the NWI population reverses its fall migration pathway, following the Pacific coast north,
while the SWI population follows a more inland route typical of the BI population
wintering in the Pacific and Central Flyways (Bousfield and Syhroechkovsky 1985,
Armstrong et al. 1999). During spring migration the NWI and SWI populations remain
isolated until at least the Yukon-Kuskokwim Delta in Alaska, and may arrive separately
on WI in some years (Ganter et al. 2005).

Identifying the degree of dispersal and population structure among breeding and
wintering groups of NWI, SW1, and BI geese is essential to the successful management
of these populations. Both breeding and wintering patterns of these populations likely
influence the interactions among NW1, SW1, and BI birds and potential gene flow among
populations. Currently the WI population is of conservation concern due to its declining
abundance and because it is the only significant snow goose colony nesting on the Asian
continent (Kuznetsov et al. 1998). In addition, the NWI population is the only group of
lesser snow geese wintering in Canada (Mowbray et al. 2000). Without an understanding
of the potential genetic exchange and population structure among NWI, SW1, and BI
lesser snow geese, the abundance and diversity of the WI population may be negatively

impacted if harvest limits in California are increased in an effort to control the growth of

216

 

Figure I. Breeding (circles) and wintering (squares) populations of lesser snow geese
from Wrangel Island and Banks Island. Fall migratory pathways south to wintering
grounds are marked by solid arrows, and spring migratory routes north to breeding areas
are marked by dashed arrows. Approximately equal proportions of geese breeding on
Wrangel Island winter in the British Columbia / Washington region (NW 1) and in
California’s Central Valley (SWI). Lesser snow geese from Banks Island (BI) winter in
the Central Valley (75% of the population) with the SW1 population.

the BI population. Thus, it is important to quantify gene flow between SW1 and BI
populations potentially occurring during admixture on wintering grounds and spring
migration. Genetic exchange between NW1 and SWI populations could occur as a result

of mixing on breeding grounds, during fall migration, and due to movement of birds

217

between north and south wintering areas. A measure of this gene flow is vital in defrning
management units of lesser snow geese.

The majority of NWI and SWI (90%) lesser snow geese nest in one large colony
along the Tundra River (Kerbes et al. 1999), so there is potential for genetic exchange
during breeding periods. However, colonial nesting snow geese pair mainly on wintering
grounds or during spring migration to breeding grounds (Cooke et al. 1975, Ely and
Scribner 1994, Ganter et al. 2005). Winter or early spring pair bond formation may limit
gene flow between the two groups of WI geese, while gene flow among SW1 geese and
BI geese may be comparatively higher because they winter together in similar regions
and follow similar spring migratory pathways. A previous study of WI population
structure based on neck collar resightings estimated exchange of migrants between SWI
and NWI at approximately 3% per year (Syroechkovsky et al. 1994). In addition to
exchange of migrants, possible gene exchange on breeding grounds due to extra-pair
copulation, intraspecific nest parasitism, and fostering, increases the amount of potential
gene flow between the two WI populations to 9% per generation (Syroechkovsky et al.
1994). This level of exchange is likely too high to allow for appreciable genetic
differences to accrue between the populations. Allozyme analyses of biparentally
inherited blood proteins and esterases from samples of NW1 and SW1 geese found no
statistical differences in allele frequency between populations (Kuznetsov et al. 1998),
supporting conclusions based on banding observations.

Our study of Western Arctic lesser snow geese builds upon previous
investigations by employing multiple genetic markers (nDNA microsatellites and

maternally inherited mtDNA) and sampling a large number of individuals from NWI,

218

SW1 and BI populations, to quantify levels of gene flow among these three populations
and assess historical and contemporary levels of dispersal among breeding and wintering
groups. We are able to compare our indirect measures of dispersal to direct measures of
movement of NWI, SWI, and BI lesser snow geese quantified by Williams et al. (2005)
in a mark-recapture study of geese banded on WI and BI breeding grounds. Geese for the
Williams et al. (2005) study were banding during the same period samples were collected
for our genetic analyses. Estimates of dispersal among the NWI, SW1, and BI
populations could potentially differ between the two techniques as banding methods
measure degree of movement among populations, whereas genetic methods measure gene
flow mediated by immigrants who disperse from source populations and subsequently
breed with individuals from receiving populations. Estimates may also differ because
genetic analyses based on bi-parentally and maternally inherited markers provide indirect
estimates of contemporary and historical rates of dispersal, while banding observations
provide direct estimates of contemporary dispersal only. However, valuable information
about the population dynamics of NW1, SWI, and BI lesser snow geese may be gained
from both direct and indirect measures of dispersal even if resulting estimates vary
between the two techniques. Previous studies have utilized either direct or indirect
measures of dispersal among Western Arctic nesting lesser snow geese (Syroechkovsky
et a1. 1994, Kuznetsov et al. 1998, Armstrong et a1. 1999, Baranyuk et al. 1999), but have
not combined the two techniques to provide a more thorough analyses of past and current

interactions among populations.

219

METHODS
Sample Collection

Samples for genetic analysis were collected in 1994 from flightless lesser snow
geese breeding populations during mid-summer brood drives. Collections were made
from July 19 to July 23 on B1 and from July 25 to August 1 on WI. Geese were
aggregated in mobile corral nets in catch areas using an all-terrain vehicle or on foot on
WI and using helicopter-drive techniques on B1 (Cooch 1953, Tim and Bromley 1976).
Numbers of birds sampled at each location were: NWI (n=7l), SWI (n=79), and BI
(n=79) (Figure I). Blood or a blood quill (growing feather) was sampled from breeding
adults. Samples were placed into individual tubes containing high-salt buffer and stored
at ambient temperatures in the field until frozen in the laboratory.

All geese captured from 1993-1996 on WI and from 1994-1996 on BI were
marked with metal United States Fish and Wildlife Service legbands, and most adults
were fitted with colored plastic neckbands as part of a larger study of populations of
Western Canadian Arctic and Wrangel Island nesting snow geese (Kerbes and Meeres
1999). Neckband color was red with white characters for WI birds, and black with white
characters for BI (Samuel et al. 2001).

On the WI breeding colony, north wintering geese can be reliably distinguished
from south wintering geese by the degree of reddish-staining on their head and face
(Baranyuk and Syroechkovsky 1994, Baranyuk et al. 1999). Geese from northern
wintering grounds forage in tidal marshes, and acquire a red stain due to the mineral salts
in the soils and water of these coastal areas (Hohn 1955, Baranyuk and Syroechkovsky

1994, Baranyuk et al. 1999). Southern wintering geese feed mainly in agricultural fields

220

on waste grain, retaining their white plumage (Pacific Flyway Technical Committee
1992, Baranyuk et al. 1999). Facial staining is associated with a color score from 1-6,
with 1 being completely white and 6 being strongly red. Past neckband observations
have recorded high fidelity (86-90%) of WI geese with face plumage scores 1-3 to
southern wintering areas, and W1 geese with scores of 4-6 to northern wintering areas
(Baranyuk et al. 1999). For our analyses, we included geese scored as 1 or 2 (southern)
and 5 or 6 (northern), and eliminated intermediate-stained (plumage scores of 3 or 4)
geese that had higher likelihoods of being incorrectly classified to wintering regions
(Kuznetsov et al. 1998, Baranyuk et al. 1999, Williams et al. 2005). Face stain score was
recorded by one trained biologist (V. V. Baranyuk) for all individuals sampled for genetic

analysis and all individuals banded (Williams et a1. 2005).

Analysis of Microsatellite Loci

DNA was extracted from all samples using DNeasy extraction kits (Qiagen Inc.,
CA). Twenty nuclear microsatellite loci were initially screened for allelic variation.
Nine bi-parentally inherited loci proved to be polymorphic in one or more breeding
populations and were used for subsequent analyses. Loci used included Bcaul , BcauS,
Bcau9, Bean] 1, Hhiul, Hhiu3 (Buchholz et al. 1998), Aalul (Fields and Scribner 1997),
SfiplO (S. Libants, unpubl. data), and CR-G (A. Baker, unpubl. data). Each locus was
amplified using polymerase chain reaction (PCR) in 25 ul reaction volumes, including
100-150 ng DNA, 10-25 pmol of each primer, PCR buffer (10 mM Tris-HCI pH 8.3, 50
mM KCI, 100 pg/mL gelatin, 0.01%NP-40, 0.01% Triton-X 100), 0.5 U of AmpliTaq

DNA Polymerase (Perkin-Elmer), and 100-200 11M dNTPs. Forward primers of each

221

locus-specific primer pair were labeled with either Hex or F luorescein by the
manufacturer (IDT Technologies, Inc.). Most thermocycler conditions included a
denaturing step of 94 °C for 2 min, followed by 30-35 cycles of 94 °C for 1 min,
annealing temperature for 1 min [51 °C (Hhipl, SfiulO), 56 °C (Bcaul, Bcau9, Hhiu3,
CR-G), 58 °C (Bcaul l), 60 °C (Bcau5)], and 72 °C for 1 min. Conditions for Aalul
included a denaturing step of 94 °C for 2 min, and 30 cycles of 94 °C for 1 min and 50 °C
for 2 min. Products were visualized using a FMBIO II laser scanner (Hitachi Software
Engineering Co.) afier electrophoresis on denaturing 6% acrylamide gels. Genotypes

were scored based on 20 base-pair standards and reference samples of known allelic size.

Analysis of MtDNA

Approximately 50-100 ng DNA was used for the initial mtDNA amplification
(1173 bp) with primers 16775L and 287H-M (Quinn 1992) and PCR protocols of Kocher
et al. (1989). Thermocycler conditions included an initial denaturation step of 92 °C for
2 min, followed by 40 cycles of 92 °C for 40 s, annealing at 61 °C for 2 min, 72 °C for 2
min, and extension at 72 °C for 7 min. Previous studies of snow geese phylogeographic
structure based on control region mtDNA sequences revealed two major mtDNA clades
(clade I and clade II) that differed by an average of 6.7% sequence divergence (Avise et
al. 1992, Quinn 1992, Weckstein et al. 2002). Of the 22 variable sites between clade I
and clade II mtDNA haplotypes, seven differences were fixed between the two clades.
Two of the fixed differences fall within the recognition sequence for the restriction
enzyme AluI (Quinn 1992). This restriction enzyme cleaves the mtDNA of haplotypes

within clade I, but does not cleave the mtDNA of haplotypes within clade II. Restriction

222

digests were performed using Alul to test if the amplified region of mtDNA for was cut
by the enzyme in order to classify individual snow geese as belonging to clade I or clade
II. Amplified mtDNA was digested with Alul according to manufacturer protocol (New
England Biolabs, Inc.) and visualized using ultraviolet light after electrophoresis on 1%

agarose gels and staining with ethidium bromide.

Estimates of Gene Diversity and Degree of Population Structuring

Departure from Hardy-Weinberg expectations was tested using the program
F STAT (Goudet 2001) For each of the nine microsatellite loci in all three snow geese
populations. Tests for genotypic linkage disequilibrium, a measure of independence
across loci within a population, were performed as described in Goudet et al. (1996)
using FSTAT. The F STAT program was also used to estimate degree of spatial
heterogeneity in gene frequency within and among snow geese populations using
hierarchical F -statistics (Weir and Cockerham 1984, Weir 1996) at three levels: (1)
among individuals within populations, (2) among individuals within the total population,
and (3) among populations. Significance of F -statistics were based on 95% confidence
intervals determined by bootstrapping across loci. Confidence intervals that included
zero were considered non-significant. Pair-wise estimates of population F ST were used as
summary measures of inter-population variance in allele frequency. Significance of pair-
wise interpopulation differentiation was determined using the exact G-test (Goudet et al.
1996) in FSTAT, as the G-test is more powerful than exact F ST-estimator tests for diploid
populations (Goudet et al. 1996, Petit et al. 2001). For tests of Hardy-Weinberg, gametic

disequilibrium, and F -statistics, nominal significance levels (alpha) were adjusted to

223

account for multiple testing using sequential Bonferroni corrections (Rice 1989). A
hierarchial analysis of molecular variance (AMOVA; Excoffier et al. 1992) was used to
partition variance in mtDNA clade frequencies (clade I versus clade 11) among snow

geese populations (PhiST).

Estimates of Rates of Migration Among Populations

Coalescent-based methods (Beerli and Felsenstein 2001) were used to estimate
relative measures of 0 (4Ncu, a composite measure of effective population size and
mutation rate) and interpopulation migration rates (4Nm, number of immigrants per
generation) based on microsatellite data. Estimates were made using maximum
likelihood methods based on a stepwise mutation model in program MIGRATE (v1.5,
Beerli 2002). Twenty individuals were randomly selected from each population,
including NWI, SWI, and B1. The goodness of fit of data was evaluated under several
different models of 0 and migration rates. The models tested included: A) Full model
with unrestricted migration among all populations and unrestricted estimates of effective
population size. B) N-dimensional island model assuming equal migration among all
populations and equal effective population sizes. C) Restricted migration model
assurrring equal migration among all populations, but unrestricted estimates of effective
population size. D) Model with unrestricted migration among all populations, but equal
effective population sizes. The full model was run three times with progressively more
extensive search strategies to ensure estimates were not based on convergence within
local maximums. We then ran the full model three times with the following search

parameters: 10 short chains, each with a total of 100,000 geneologies and a sampling

224

increment of 100 geneologies; and 3 long chains which were combined, each with a total
of 1,000,000 geneologies and a sampling increment of 1000 geneologies. The first
10,000 geneologies in each chain were discarded. Initial parameters for each of these
three runs were the resulting migration and 0 estimates from the previous run.
Alternative models were each run once, with initial parameters of migration and 0 set as
the mean of resulting estimates from the three full model runs. Likelihood estimates of
each alternative model were compared to the full model using a likelihood-ratio test to
evaluate goodness-of-fit of each model, and determine which model of gene flow and
population size was statistically supported by the data. Each likelihood-ratio test was
compared to a Chi-square distribution with degrees of freedom equal to the difference in
the number of parameters between the two models being tested (Beerli and Felsenstein

2001).

RESULTS

Eight of the nine microsatellite loci used were polymorphic in all three
populations, with two to seventeen alleles present within each population (Table 1). Each
of the bi-parental loci used did not deviate significantly from Hardy-Weinberg
equilibrium, and were determined to be genetically independent, as no evidence of
gametic disequilibrium was detected for any locus combination in any population.
Hierarchical F -statistics showed no evidence of significant variation among individuals
within populations (f), among individuals within the total population (F), or among

populations (F ST) based on microsatellite loci (Table I).

225

Table 1. Lesser snow geese allele frequencies and F-statistics for nine biparental and one maternally
inherited marker from three populations of lesser snow geese breeding within the West Arctic region.
Populations include north and south wintering Wrangel Island and Banks Island.

 

. .. . a
Varrance partrtronrng

 

 

Alleles
within Among
individuals individuals
Allele/ Banks Wrangel Wrangel within within total Among

Locus mtDNA Clade Island North South populations population populations
Aalul 82 0.196 0.180 0.190 -0.124 -0.126 -0.002

84 0.044 0.073 0.065

86 0.101 0.180 0.119

88 0.456 0.407 0.423

90 0.184 0.127 0.173

92 0.019 0.033 0.030

n 79 75 84
Bcaul 1 12 0.006 0.013 0.000 0.005 0.013 0.008

1 14 0.045 0.013 0.030

1 16 0.064 0.113 0.095

118 0.391 0.253 0.280

120 0.058 0.060 0.042

122 0.122 0.067 0.131

124 0.051 0.160 0.131

126 0.083 0.120 0.131

128 0.071 0.047 0.071

130 0.064 0.087 0.065

132 0.019 0.053 0.012

134 0.026 0.013 0.012

n 78 75 84
Bcau9 98 0.000 0.000 0.006 0.016 0.015 -0.00 I

102 0.411 0.487 0.423

104 0.013 0.013 0.006

106 0.196 0.173 0.131

108 0.070 0.033 0.054

110 0.127 0.127 0.149

112 0.171 0.160 0.232

1 14 0.013 0.007 0.000

n 79 75 84
BcauS 200 1.000 0.987 0.994 -0.005 -0.004 0.001

202 - 0.000 0.013 0.006

n 81 76 84

 

226

Table I (cont’d).

 

. .. . a
Variance partitioning

 

 

Alleles
within Among
individuals individuals
Allele/ Banks Wrangel Wrangel within within total Among
Locus mtDNA Clade Island North South populations population populations
CR-G 164 0.975 0.993 0.994 -0.008 -0.008 0.000
166 0.012 0.000 0.000
168 0.012 0.007 0.006
n 81 76 84
Hhiul 174 0.006 0.000 0.012 -0.006 -0.001 0.005
178 0.006 0.007 0.000
180 0.006 0.000 0.000
182 0.012 0.020 0.018
184 0.000 0.007 0.006
186 0.019 0.020 0.006
188 0.062 0.092 0.071
190 0.031 0.020 0.054
192 0.062 0.066 0.143
194 0.451 0.421 0.333
196 0.105 0.125 0.107
198 0.080 0.072 0.071
200 0.031 0.033 0.030
202 0.080 0.066 0.060
204 0.012 0.020 0.030
206 0.025 0.026 0.042
208 0.006 0.000 0.006
210 0.000 0.007 0.006
212 0.000 0.000 0.006
214 0.006 0.000 0.000
n 81 76 84
Hhiu3 115 0.235 0.237 0.292 0.1 1 1 0.109 -0.002
1 17 0.012 0.020 0.006
121 0.006 0.007 0.000
123 0.235 0.171 0.196
125 0.500 0.553 0.488
127 0.012 0.013 0.018
n 81 76 84

 

227

Table l (cont’d).

. .. . a
Variance partitioning

 

 

Alleles
within Among
individuals individuals
Allele/ Banks Wrangel Wrangel within within total Among
Locus mtDNA Clade Island North South populations population populations
SfiplO 126 0.481 0.507 0.560 0.135 0.136 0.001
128 0.000 0.026 0.000
130 0.519 0.467 0.440
n 81 76 84
Bcaul l 134 0.000 0.000 0.006 -0.057 -0.062 -0.005
138 0.006 0.020 0.018
140 0.883 0.862 0.887
142 0.111 0.118 0.089
n 81 76 84
All Biparental Loci 0.011 0.012 0.001
(N S) (N S) (N 5)
mtDNA clade I 0.568 0.182 0.439 na na 0.098
clade 11 0.432 0.818 0.561 (P < 0.001)
n 74 66 82

 

aF-statistics for biparental loci and maternally inherited mtDNA. Nomenclature is as follows: Alleles
within individuals within populations represented by fi among individuals within total population
represented by F, and among populations represented by F31 and PhiST (Weir and Cockerham 1984,
Excoffier et al. 1992). NS = nonsignificant (i.e. P > 0.05); na indicates that no F -statistic was applicable
for mtDNA clade frequency data.

Pair-wise population differences in allele frequency were summarized across the
nine microsatellite loci as the proportion of total genetic diversity partitioned between
each pair of populations. No pair-wise comparisons (Table 2) were significant, indicating
allele frequencies were similar across populations (mean F st across all populations =
0.0018, P > 0.05). In contrast to results based on microsatellites, mtDNA clade
frequencies were significantly different among populations (Phisr = 0.098, P < 0.001;

Table 1). Notably, there were large differences between northern and southern wintering

228

Table 2. Above diagonal: Pairwise F ST comparisons3 among lesser snow
geese populations based on 9 bi-parentally inherited microsatellite loci.

Below diagonal: Pairwise Phisr comparisonsb among lesser snow geese
populations based on maternally inherited mtDNA. P-values for each
pairwise comparison are indicated below estimates in parentheses.
Significant P-values are marked with an asterisk.

 

Wrangel North Wrangel South Banks Island

Wrangel North --- -0.0008 0.0016
(0.5833) (0.3000)
Wrangel South 0. 1047 --- 0.0029
(0.0000)* (0.3667)
Banks Island 0.1932 0.0047 ---
(0.0000)* (0.0999)

 

aWeir and Cockerham 1984
bExcoffier et a1. 1992

populations of lesser snow geese breeding on WI (pair-wise Phis-r = 0.1047, P < 0.0001),
and between NW1 and BI (pair-wise PhiST = 0.1932, P < 0.0001; Table 2).

In the comparison of competing models of migration and effective population
size, only Model B, the N-dimensional island model [Ln(L) = -565], was found to have
significantly higher likelihood than Model A, the full model [Ln(L) = -678; P < 0.0001].
Models C [Ln(L) = -817] and D [Ln(L) = -767] had lower likelihoods than Model A, and
both likelihood ratio tests were non-significant. Genetic-based estimates of effective
population size and migration rate are based on microsatellite data, which reflect
contemporary evolutionary patterns for each population. This suggests that effective

population sizes and exchange of migrants and potential gene flow is equivalent among

229

NWI, SW1, and BI populations over recent evolutionary periods. These results are
consistent with the lack of differentiation in microsatellite loci frequencies among the
three populations, and the equivalent rates of migration among the three populations
documented by Williams et al. (2005). For the N-dimensional island model, mean
effective population size (0 or 4Neu) was estimated at 0.50 (95% CI: 0.46, 0.56) and
mean migration (4Nm) among the three populations was estimated to be 9.23 (95% CI:

8.72, 9.79) individuals per generation.

DISCUSSION

The NW1 population wintering in British Columbia and Washington State was
highly differentiated genetically from the SWI and BI populations wintering together in
the Central Valley of California as evidenced by significant differences in mtDNA clade
frequencies. Previous studies comparing the mtDNA characterisitics of lesser snow
geese populations, including geese from WI, observed that two distinct mtDNA clades
(6.7% sequence divergence) were distributed across the lesser snow goose subspecies
range without geographic localization (Avise et al. 1992, Quinn 1992, Weckstein et al.
2002). Two different hypotheses were raised to explain the lack of geographic pattern
among the highly divergent mtDNA clades. Quinn (1992) stated that two ancestral lesser
snow goose populations likely occupied different historic refugia during glacial events
(Ploeger 1968) and subsequently dispersed across the modern range of lesser snow geese
during interglacial periods. An implication of this hypothesis is that speciation of snow
geese and Ross’s geese would have been recent, postdating the split in the mtDNA gene

tree, with both species retaining the ancestral polymorphism (Avise et a1. 1992). Both

230

Avise et al. (1992) and Weckstein et al. (2002) supported an alternative hypothesis of
secondary introgression and hybridization between formerly allopatric populations of
snow and Ross’s geese that were isolated during the Pleistocene.

In the previous studies, snow geese with clade I and clade Il haplotypes were
identified at every population surveyed, but the proportions of individuals within each
mtDNA clade differed among populations. Frequencies of clade I haplotypes were 0.37
for WI snow geese and 0.50 for geese from LaPerouse Bay in a study conducted by
Quinn (1992). Weckstein et al. (2002) added to Quinn’s (1992) data by sampling snow
geese from Queen Maud Gulf, where the fiequency of clade I haplotypes was 0.22.
Avise et al. (2002) surveyed snow geese from WI, Queen Maud Gulf, and Anderson
River, where clade I frequencies were 0.40, 0.64, and 0.71, respectively. Similar to these
previous studies, we documented a higher number of WI snow geese with clade II
haplotypes than clade I haplotypes. We estimated the frequency of clade I haplotypes to
be 0.32 for the WI breeding population, 0.18 for the NWI wintering population, and 0.44
for SW1 wintering population. In contrast to WI birds, most geese from BI were
identified as having clade I haplotypes (frequency = 0.57), similar to geese from
LaPerouse Bay (Quinn 1992, Weckstein et al. 2002) and Queen Maud Gulf (Avise et al.
1992). Quinn (1992) and Weckstein et al. (2002) found mtDNA clade frequencies were
significantly different between WI and LaPerouse Bay and between WI and Queen Maud
Gulf snow geese populations. In our study, significant differences in clade frequencies
were documented between NWI and SWI populations and between NWI and BI
populations, but frequencies were similar between SWI and BI populations sharing the

same wintering areas. Avise et al. (1992) reported clade frequencies were statistically

231

similar among the W1, Anderson River, and Queen Maud Gulf breeding populations, but
sample sizes were low for the WI population (n = 10).

The frequency of clade I haplotypes is much lower in the NWI population than
other regional snow geese populations. The small proportion of clade I birds present
within the NWI population is more consistent with the clade I frequencies documented
for Ross’s geese populations (0.10, Avise et al. 1992; 0.30, Weckstein et al. 2002; 0.18 —
0.27, Shorey unpublished data) than for snow geese populations. NWI snow geese have
been isolated historically and during contemporary periods from Ross’s geese
populations in North America based on putative Pleistocene refugia locations (Ploeger
1968) and known migratory pathways and wintering sites (Ryder and Alisauskas 1995.
Mowbray et a1. 2000). Given this geographic isolation, low frequencies of clade I
haplotypes in the NWI population similar to Ross’s geese populations suggests that the
evolutionary split between mtDNA clades occurred prior to speciation of snow and
Ross’s geese, supporting Quinn’s (1992) evolutionary hypothesis.

NW1 and SWI lesser snow geese may have originated in two different glacial
refugia, and retain some genetic signature of their formerly allopatric state. as evidenced
by significant differences in mtDNA clade frequencies. Kuznetsov et al. (1998)
hypothesized that the NWI and SWI populations were once allopatric, and the SWI
population may have originated from geese nesting on the Russian coast near Wrangel
Island known to winter in California. These mainland geese may have joined the WI
colony after being displaced from their coastal breeding area prior to the 19305
(Bousfield and Syroechkovsky 1985). In addition to possible historical differences in

origin between NW1 and SWI geese, these populations may have been largely isolated in

232

different colonies that existed on Wrangel Island until the late 19503 when one of the two
remaining colonies on the island was disrupted by a geologic expedition (Syroechkovsky
and Krechmar 1981, Kuznetsov et al. 1998). Since 1969 the majority of nesting geese on
Wrangel Island have inhabited a single large colony (Bousfield and Syroechkovsky
1985), increasing the potential for mixing of geese from NWI and SWI populations on
the breeding grounds. Lack of nDNA allele frequency differences between the two
populations indicate that genetic differences that accrued during historic isolation have
decayed at a faster rate for microsatellite markers than matemally-inherited markers due
to male-biased gene flow (Cooke and Sulzbach 1978, Cooke and Abraham 1980,
Rockwell and Barrowclough 1987) and a much greater mutation rate for microsatellites
as compared to mtDNA. Contemporary increases in gene flow between NWI and SWI
populations may be due to loss of breeding areas and recent changes in the spatial
structure of the breeding colony on WI.

High rates of fidelity (96-98%) of NW1 lesser snow geese to northern wintering
grounds and SWI and BI populations to southern wintering grounds were estimated by
Williams et al. (2005) based on banding observations and documented previously for
these groups in similar banding studies (Baranyuk et al. 1999, Armstrong et al. 1999).
Rates of winter fidelity were equivalent for males and females from NW1, SWI, and BI
populations, and greater than rates of fidelity to breeding areas for female (70-80%) and
male (SO-66%) lesser snow geese (Cooke and Sulzbach 1978, Cooke and Sulzbach 1978,
Cooke and Abraham 1980, Ganter and Cooke 1998). Based on rates of fidelity to
wintering grounds as compared to rates of fidelity to breeding grounds, we would expect

greater spatial genetic structure among wintering populations than among breeding

233

populations of Western Arctic lesser snow geese. Genetic analyses indicate wintering
populations NWI and SWI are more highly structured than WI and BI breeding
populations of lesser snow geese in the Western Arctic, supporting the thesis of
Robertson and Cooke (1999) that patterns of population structure and gene flow may be
defined by wintering regions rather than fidelity to breeding site.

Both banding data and genetic analyses based on nDNA provide estimates of
contemporary dispersal among populations, but do not provide information about historic
dispersal or gene flow among the populations which may or may not be similar to current
patterns. Alternative markers, such as mtDNA, must be utilized in order to assess how
historic factors may have influenced the exchange of individuals among populations and
shaped current genetic structure. As a matemally-inherited marker, mtDNA also
provides a measure of female-mediated dispersal and gene flow. Williams et al. (2005)
found low (2%) and equal rates of exchange of migrants among NW1, SWI, and BI
populations for both male and female snow geese. Our coalescence-based analysis of
microsatellite data also indicated an equal rate of gene exchange among these three
populations in concordance with Williams et al. (2005) direct observations of dispersal.
In contrast, significant variation in mtDNA clade frequencies among NW1, SWI, and BI
populations suggests that rates of dispersal and subsequent gene flow may not be
equivalent among populations and is male-biased. These results are more reflective of
earlier banding research conducted from 1974-1979 documenting male-biased dispersal
from the SWI population (Syroechkovsky et al. 1994). Two hypotheses may explain the
differential results from our mtDNA analyses as compared to our nDNA analyses and the

banding observations of Williams et al. (2005). If dispersal and gene flow among the

234

three populations are male-biased, but dispersal rates are low for both sexes, Williams et
al. (2005) banding study may not have provided the resolution necessary to detect a
difference in dispersal between males and females. Alternatively, both banding and
genetic studies results may be accurate, and indicate historic dispersal and gene flow
among NW1, SWI, and BI lesser snow geese populations were more restricted than
during evolutionary recent periods.

Degree of face staining has been a reliable indicator of population affiliation for
NWI and SWI lesser snow geese (Kuznetsov et al. 1998, Baranyuk et al. 1999). Banding
studies have documented high fidelity (82 — 90 % return rate in subsequent winters) of
WI geese with little or no face staining to southern wintering areas in Oregon and
California, and W1 geese with dark red staining to northern wintering areas in the F raser-
Skagit region (Baranyuk et a1. 1999, Williams et al. 2005). Past studies conducted to
investigate the relationship between the degree of face staining and genetic population
structuring have found no evidence of genetic differentiation between NW1 and SW1
snow geese based on pairing observations (Syroechkovsky et al. 1994) and
electrophoretic analysis of blood proteins and esterases (Kuznetsov et al. 1998). We also
found a lack of genetic variation between NWI and SWI populations based on
microsatellite allele frequencies. However, mtDNA clade frequencies differed
significantly between the two populations, suggesting face staining is indicative of
genetic isolation and population structuring between geese from the NWI and SWI
wintering regions.

Data from our study illustrates the importance of both past vicariance and ongoing

demographic changes and range expansions within and among populations in shaping the

235

current population genetic structure of Western Arctic lesser snow geese. This study
highlights how the combination of direct and indirect measures of dispersal and genetic
exchange, and the use of multiple genetic markers with different patterns of inheritance,
can provide a more comprehensive picture of demographic patterns through time and aid
in population assessment. Our results suggest there is a low level of male-mediated gene
flow occurring between the NWI and SWI lesser snow geese population, and that current
levels of population and genetic exchange may be greater than during historic periods.
However, significant genetic differences between the NWI and SWI populations indicate
that these populations should be currently managed as separate populations, or
subpopulations, even though most individuals nest within one large breeding colony on
WI. Genetic differences between NW1 and BI snow geese, and the lack of genetic
variation between SWI and BI geese wintering in the same region, could be used to argue
that management units of lesser snow geese should be defined by wintering sites rather
than breeding areas. If BI and SWI birds were managed as one unit, we would caution
that policies directed to control the overabundant BI population through harvest on
wintering grounds when geese from the SWI population are present could significantly
impact the long-term viability of the W1 breeding population. Factors influencing the
decline of the WI population are not well understood, and management plans should
minimize potential harvests of both NWI and SWI geese until further investigations can

be conducted.

ACKNOWLEDGMENTS
We thank the following individuals for assistance in conducting field sampling:

V. V. Baranyuk, D. R. Goldberg, B. Ganter, J. Takekawa, E. G. Cooch. A special thanks

236

to L. Main for extensive laboratory assistance. Financial support was provided by the
United States Geological Survey, Environment Canada, Michigan State University, and

the Arctic Goose Joint Venture.

237

LITERATURE CITED

Alisauskas, R. T. 1998. Winter range expansion and relationships between landscape
and morphometerics of midcontinent Lesser snow geese. Auk 115:851-862.

Ankney, C. D. 1996. An embarrassment of riches: too many geese. Journal of Wildlife
Management 602217-222.

Armstrong, W. T., K. M. Meeres, R. H. Kerbes, W. S. Boyd, J. G. Silveria, J. P. Taylor,
and B. Turner. 1999. Routes and timing of migration of lesser snwo geese fiom the
western Canadian Arctic and Wrangle Island, Russia. Pages 75-88 in Distribution,
survival, and numbers of Lesser Snow Geese of the Western Canadian Arctic and
Wrangel Island, Russia (R. H. Kerbes, K. M. Meeres, J. E. Hines, Eds.). Occasional
Paper No. 98. Canadian Wildlife Service, Ottawa, Ontario.

Avise, J. C., R. T. Alisauskas, W. S. Nelson, and C. D. Ankney. 1992. Matriarchal
population structure in an avian species with female natal philopatry. Evolution 46:1084-
1096.

Baranyuk, V. V. and E. V. Syroechkovsky. 1994. The usage of natural marks in
Wrangel Island snow geese for the population studies. Omithologica 26:45-49 (in
Russian).

Baranyuk, V. V., J. E. Hines, and E. V. Syroechkovsky. 1999. Mineral staining of facial
plumage as an indicator of the wintering ground affinities of Wrangel Island lesser snow
geese. Pages Ill-114 in Distribution, survival, and numbers of Lesser Snow Geese of
the Western Canadian Arctic and Wrangel Island, Russia (R. H. Kerbes, K. M. Meeres, J.
E. Hines, Eds.). Occasional Paper No. 98. Canadian Wildlife Service, Ottawa, Ontario.

Batt, B. D. J. 1997. Arctic ecosystems in peril: report of the Arctic Goose Habitat
Working Group. Arctic Goose Joint Venture Special Publication. US. Fish and Wildlife
Service, Washington, D. C. and Canadian Wildlife Service, Ottawa, Ontario. 120pp.

Beerli, P. 2002. MIGRATE: documentation and program, part of LAMARC. Version
1.5. Free program distributed by the authors over the intemet from
http://evolution.genetics.washington.edu/lamarc.html.

Beerli, P. and J. Felsenstein. 2001. Maximum likelihood estimation of a migration
matrix and effective population sizes in n subpopulations by using a coalescent approach.
PNAS 98:4563-4568.

Bousefield, M. A. and E. V. Syroechkovsky. 1985. A review of Soviet research on the
lesser snow goose on Wrangel Island, USSR. Wildfowl 36:13-20.

238

Buchholz, W. G., and J. M. Pearce, B. J. Pierson, and K. T. Scribner. 1998.
Dinucleotide repeat polymorphisms in waterfowl (family Anatidae): Characterization of a
sex-linked (Z-specific and 14 bi-parentally inherited loci. Animal Genetics 29:323-325.

Cooch, F. G. 1953. Techniques for mass capture of flightless blue and lesser snow
geese. Journal of Wildlife Management 17:460-465.

Cooke, F. and K. F. Abraham. 1980. Habitat and locality selection in Lesser Snow
Geese: the role of previous experience. Proc. Int. Ornithol. Congr. 17:998-1004.

Cooke, F., C. D. MacInnes, and J. P. Prevett. 1975. Gene flow between breeding
populations of Lesser Snow Geese. Auk 93:493-465.

Cooke, F. and D. S. Sulzbach. 1978. Mortality, emigration and separation of mated
Snow Geese. Journal of Wildlife Management 42:271-280.

Ely, C., and K. T. Scribner. 1994. Genetic diversity in arctic-nesting geese: implications
for management and conservation. Transactions of the North American Wildlife Natural
Resources Conference 59:91-1 10.

Excoffier, L., P. E. Smouse, and J. M. Quattro. 1992. Analysis of molecular variance

inferred fiom metric distances among DNA haplotypes: application to human
mitochondrial DNA restriction data. Genetics 131:479-491.

Fields, R. L., and K. T. Scribner. 1997. Isolation and characterization of novel
waterfowl microsatellite loci: cross-species comparisons and research applications.
Molecular Ecology 6: 1 99-202.

Ganter, B., W. S. Boyd, V. V. Baranyuk, and F. Cooke. 2005. First pairing in snow
geese Anser caerulescens: at what age and at what time of year does it occur? Ibis
147:57-66.

Ganter, B. and F. Cooke. 1998. Colonial nesters in deteriorating habitat: site fidelity and
colony dynamics of Lesser Snow Geese. Auk 115:642-652.

Goudet, J. 2001. F STAT, a program to estimate and test gene diversities and fixation
indices (version 2.9.3). Available from http://www.unil.ch/izea/sofiwares/fstathtml.
Updated from Goudet (1995).

Goudet, J, M. Raymond, T. Demeeus, F. Rousset. 1996. Testing differentiation in
diploid populations. Genetics 144:1933-1940.

Hines, J. E, V.V. Baranyuk, B. Turner, W. S. Boyd, J. G. Silveira, J. P. Taylor, S. J.

Barry, K. M. Meeres, R. H. Kerbes, and W. T. Armstrong. 1999a. Autumn and winter
distributions of Lesser Snow Geese from Western Canadian Arctic and Wrangel Island,
Russia, 1953-1992. Pages 39-73 in Distribution, survival, and numbers of Lesser Snow

239

Geese of the Western Canadian Arctic and Wrangel Island, Russia (R. H. Kerbes, K. M.
Meeres, J. E. Hines, Eds.). Occasional Paper No. 98. Canadian Wildlife Service,
Ottawa, Ontario.

Hines, J. E, M. O Wiebe, S. J. Barry, V. V. Baranyuk, J. P. Taylor, R. McKelvey, S. R.
Johnson, and R. H. Kerbes. 1999b. Survival rates of lesser snow geese in the Pacific and
Western Central flyways, 1953-1989. Pages 89-109 in Distribution, survival, and
numbers of Lesser Snow Geese of the Western Canadian Arctic and Wrangel Island,
Russia (R. H. Kerbes, K. M. Meeres, J. E. Hines, Eds.). Occasional Paper No. 98.
Canadian Wildlife Service, Ottawa, Ontario.

Johnson, S. R. 1996. Staging and wintering areas of Snow Geese nesting on Howe
Island, Alaska. Arctic 49:86-93.

Kerbes, R. H. 1987. International Snow Goose Neckbanding Project instructions to
observers for monitoring neckbanded birds. Unpublished report, Canadian Wildlife
Service, Saskatoon, Saskatchewan. lOpp.

Kerbes, R. H, V. V. Baranyuk, and J. E. Hines. 1999. Estimated size of the Western
Canadian Arctic and Wrangel Island Lesser Snow Goose populations on their breeding
and wintering grounds. Pages 25-38 in Distribution, survival, and numbers of Lesser
Snow Geese of the Western Canadian Arctic and Wrangel Island, Russia (R. H. Kerbes,
K. M. Meeres, J. E. Hines, Eds.). Occasional Paper No. 98. Canadian Wildlife Service,
Ottawa, Ontario.

Kerbes, R. H. and K. M. Meeres. 1999. Project overview. Pages 15-24 in Distribution,
survival, and numbers of Lesser Snow Geese of the Western Canadian Arctic and
Wrangel Island, Russia (R. H. Kerbes, K. M. Meeres, J. E. Hines, Eds.). Occasional
Paper No. 98. Canadian Wildlife Service, Ottawa, Ontario.

Kocher, T. D., W. K. Thomas, A. Meyer, S. V. Edwards, 8. P550, F. X. Villablanca, and
A. C. Wilson. 1989. Dynamics of mitochondrial DNA evolution in animals:

amplification and sequencing with conserved primers. Proceedings of the National
Academy of Sciences 86:6196-6200.

Kuzetsov, S. B., V. V. Baranyuk, and J. Y. Takekawa. 1998. Genetic differentiation
between wintering populations of lesser snow geese nesting on Wrangel Island. Auk
115:1053-1057.

Mowbray, T. B., F. Cooke, and B. Ganter. 2000. Snow Goose (Chen caerulescens). In
The Birds of North America, No. 514, A. Poole and F. Gill, eds. The Academy of
Natural Sciences, Philadelphia, and The American Omithologists’ Union, Washington,
DC.

Petit E., F. Balloux, and J. Goudet. 2001. Sex-biased dispersal in a migratory bat: a
characterization using sex-specific demographic parameters. Evolution 55: 63 5-640.

240

Ploeger, P. L. 1968. Geographical differentiation in Arctic Anatidae as a result of
isolation during the last glacial. Ardea 56:1-159.

Posada, D., and K. A. Crandall. 1998. MODELTEST: Testing the model of DNA
substitution. Bioinforrnatics 14:81 7-81 8,

Quinn, T. W. 1992. The genetic legacy of mother goose-phylogeographic patterns of
lesser snow goose Chen caerulescens caerulscens maternal lineages. Molecular Ecology
1:105-117.

Rice, W. R. 1989. Analyzing tables of statistical tests. Evolution 43:223-225.

Robertson, G. J ., and F. Cooke. 1999. Winter philopatry in migratory waterfowl. Auk
116:20-34.

Rockwell, R. F. and G. F. Barrowclough. 1987. Gene flow and the genetic structure of
populations. Pages 223-255 in Avian genetics: a genetic population and ecological
approach (F. Cooke and P. A. Buckley, Eds.). Academic Press, London.

Rockwell, R. F., and F. Cooke. 1977. Gene flow and local adaptation in a colonially
nesting dimorphic bird: the lesser snow goose (Anser caerulescens caerulescens).
American Naturalist 111:91-97.

Samuel, M. D., D. R. Goldberg, A. E. Smith, V. V. Baranyuk, and E. G. Cooch. 2001.
Neckband retention for lesser snow geese in the wester Arctic. Journal of Wildlife
Management 65:797-807.

Scribner, K. T., R. A. Malecki, B. D. J. Batt, R. L. Inman, S. Libants, and H. H. Prince.
2003. Identification of source population for Greenland Canada geese: genetic
assessment of a recent colonization. The Condor 105:771-782.

Swofford, D. L. 2000. PAUP": Phylogenetic Analysis Using Parsimony (*and Other
Methods), version 4. Sinauer Associates, Sunderland, Massachusetts.

Syroechkovsky, E. V., F. Cooke, and W. J. L. Sladden. 1994. Population structure of the
lesser snow geese of Wrangel Island, Russia. Ecoscience 1:311-316.

Syroechkovsky, E. V. and A. V. Krechmar. 1981. The main factors determining the
abundance of Snow Geese. Pages 3-37 in The ecology of mammals and birds of Wrangel
Island. USSR Academy of Sciences, Vladivostok, Russia.

Tim, D. E. and R. G. Bromley. 1976. Driving Canada geese by helicopter. Wildlife
Society Bulletin 4: 180-181 .

241

Williams, C. K., M. D. Samuel, V. V. Baranyuk, E. G. Cooch, D. Kraege. 2005. Winter
fidelity and apparent survival of lesser snow goose populations in the Pacific Flyway.
Unpublished United States Geological Survey manuscript.

Weckstein, J. D., A. D. Afton, R. M. Zink, and R. T. Alisauskas. 2002. Hybridization
and population subdivision within and between Ross’s geese and lesser snow geese: a
molecular perspective. Condor 104:432-436.

Weir, B. S. 1996. Genetic Data Analysis 11. Sinauer and Associates, Sunderland.

Weir, B. S. and C. C. Cockerham. 1984. Estimating F-statistics for the analysis of
population structure. Evolution 38: 1358-1370.

242

11111111111111111