"vmv'v-rw-‘wu_m'

wry-'-

u

2......

g V
. .4. :mnfififrgfti

utthwhfi:

5.?

rt 5:;
a‘ ’

fl..- .r.

L a If. .1

LV..;.I¢:..§:::
.32.:EK :
A c. .. 12::

5.17:.5-3 .. 3.5. :

r

.. ..
4.34 .

a.

3.;
. .21
r

3.

I: 1‘335‘

2...

2‘...

«I Lu“?
:duflrtva J.
I

.1
.9. fashr x

. ,
d z o
I.
a 3.... W. ‘
EV“: x... V.
\ «4.21 . . ‘0’. FM...
V P I I. .h

s A
IN".
ME «:3 .
5.1.. :3... x
.33.. a. ..
492...; n u I.
l

x. .. 2:

3L: hi3...

{.mmmfipzaé

Law- 3.2%...- r..
. .... Ha},

0:!

5235!: i.
31..“ 2.2..
l“! Iii-it...

' I ,7 .65}...

V (t
hing“. hi....¢....u:
Maggi: at...
is.».d.fi.»..u..x.é 4..
»!tflflp1§2.nfll.

34'.
Ia‘tznfihpsikxax .

. 4‘ {-0.8 [.33-1?
§)3nlhflvxtfi.rflufl¢{‘!
i . 7ngh.fix 5.1 u

. 4“]
01.. :6... «(55: \ .t
i£1.t5).i5 3.
“Ell"...ll ..
56": ll ..

.110 ii:

J. . ;

n I . 7"
I1T’thh J‘l \l. 4

3. .. 1% ‘ i; .

23.10.. .« £52.53»: 5

5b.; . )1}...' .A 2“}.Iyt vo-

lltuiizzt. 5:3, 2.“

. 1"” D

I in u w A
.‘l! 1:3,;1‘11? o
t. 51 L‘syisru . SI 2 ,
‘ A. \a‘rta

 

This is to certify that the
thesis entitled

Evaluating the Reproductive Success and Long-Term
Survival Potential of Common Terns (Sterna
hirundo) in the St. Mary's River, MIchigan

presented by

Nicole Elizabeth Lamp

has been accepted towards fulfillment
of the requirements for

Master degree in FiSh. & Wildl.

 

AéQVIL€A4f745iigilggrngo{:r

fiajor professor

Date 4/24/02

0-7639 MS U i: an Affirmative Action/Equal Opportunity Institution

 

 

._ —-=--—--———-u- .—-——~—.._..~_.

 

LIBRARY
Michigan State
University

 

 

 

PLACE IN RETURN BOX to remove this checkout from your record.
TO AVOID FINES return on or before date due.
MAY BE RECALLED with earlier due date if requested.

 

DATE DUE DATE DUE ' DATE DUE
O 3 2
J U L20 06 42004

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

6/01 cJCIRCIDateDuepGScpJS

 

....—-—————-—~ .—

EVALUATING THE REPRODUCTIVE SUCCESS AND LONG-TERM
SURVIVAL POTENTIAL OF COMMON TERNS (S T ERNA HIRUNDO) IN THE
ST. MARY’S RIVER, MICHIGAN

By

Nicole E. Lamp

A THESIS

Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of

MASTER OF SCIENCE

Department of Fisheries and Wildlife

2002

ABSTRACT
EVALUATING THE REPRODUCTIVE SUCCESS AND LONG-TERM SURVIVAL
POTENTIAL OF COMMON TERNS (STERNA HIR UNDO) IN THE ST. MARY’S
RIVER, MICHIGAN
By

Nicole E. Lamp

The common tern (Sterna hirundo) is listed as a Threatened species in Michigan
and is declining due to predation, vegetation encroachment, and loss of suitable breeding
habitat due to high water levels. I investigated the relationship between reproductive
success and vegetation cover on three nesting sites in the St. Mary’s River: Lime Island,
Andrews Reef, and Harbor Island Reef. On natural islands (e. g., Andrews and Harbor
Island Reefs), common terns have better nesting success in low to moderate amounts (20-
40%) of vegetation cover but on human-made sites (i.e., Lime Island), they have better
nesting success in moderate amounts (40-50%) of total vegetation cover. Nesting sites
used by common terns during years of low water levels (< 174.8 m) supported less
vegetation cover (28.8%) and had greater egg survival (67.9%) than nesting sites used
during years of high water levels (> 175.8 m; 66.1% total vegetation cover, 35.9% egg
survival). A population viability analysis (PVA) conducted using VORTEX indicates
that the St. Mary’s River population is declining at a rate ranging from 6.3% - 10% per
year (100-year probability of extinction = 70.4% - 99.9%). The most significant factors
affecting the population’s long-term persistence are juvenile and sub-adult survival.
Management strategies Should be directed at increasing juvenile and sub-adult survivals

and manipulating vegetation cover to levels in which reproductive success is the greatest.

To my mom and dad.
I wouldn’t have gotten this far without your love, support, and sacrifices.

iii

ACKNOWLEDGEMENTS

Funding for this project was provided by the Michigan Department of Natural
Resources (MDNR), Wildlife Division, Natural Heritage Program; Partnerships for
Wildlife; the United States Fish and Wildlife Service (U SFWS); Michigan State
University, Department of Fisheries and Wildlife; and the George J. and Martha C.
Wallace Endowed Scholarship. I am indebted to my committee members for their
contributions to this project. To my major advisor, Dr. Kelly Millenbah, thank you for
your invaluable mentoring, expertise, confidence, and support throughout this project.
To Dr. Scott Winterstein and Dr. Catherine Lindell, thank you for Sharing your
knowledge and offering guidance throughout this project. A Special thanks goes to Ray
Rustem for his ongoing support of the common tern project and to David Best for
allowing us to be subpermittees under his Master Bander permit.

This project could not have been completed without the valuable field assistance
and friendship of Amy Matusz and Jennifer Finger. Other contributors to this project
include MDNR, Forest Management Division, Bill and Norma Mieras, Ted and Barbara
Ludwig, Pete Butchko, Anthony Duffiney, Gary Roloff, Betsy Cook, Dr. Dan Hayes, Dr.
Jack Liu , Fisherman’s Point, and Autumn Larkins. I owe a special thanks to Ali Felix
and Dave Thomas for venturing into the St. Mary’s River and not jumping ship despite
major motor malfunctions. I also owe a special thanks to Tameka Dandridge for her field
assistance, knowledge of small engine repair, and humorous attitude following our
“rescue” by the US. Coast Guard. I extend my deepest gratitude to Team Millenbah for
their support, encouragement, statistics knowledge, laughs, and friendship. Finally, I

would like to thank Drew Kramer for his love, patience, knowledge, and support.

iv

TABLE OF CONTENTS

LIST OF TABLES ......................................................................................................... viii
LIST OF FIGURES .......................................................................................................... xi
THESIS STRUCTURE ..................................................................................................... 1
STUDY AREA ................................................................................................................. 2
LITERATURE CITED ......................................................................................... 9

CHAPTER 1: IMPACTS OF VEGETATION COVER ON COMMON TERN
REPRODUCTIVE SUCCESS IN THE ST. MARY’S RIVER, MICHIGAN

ABSTRACT .......................................................................................................... 10
INTRODUCTION ................................................................................................ 1 2
OBJECTIVES ....................................................................................................... l8
METHODS ........................................................................................................... 19
Nest Checks and Survival Estimates ......................................................... 19
Vegetation Cover ...................................................................................... 21

Data Analysis ............................................................................................ 21
Vegetation Comparisons within Nesting Sites ............................. 21

Vegetation Comparisons among Nesting Sites ............................. 22

Vegetation Cover and Nest Success ............................................. 23

Vegetation Cover and Survival ..................................................... 23

RESULTS ............................................................................................................. 25
Nest Checks and Banding ......................................................................... 25
Vegetation Comparisons within Nesting Sites ......................................... 25

Nests vs. Random Points ............................................................... 25

Successful vs. Unsuccessful Nests ................................................ 28

Vegetation Comparisons among Nesting Sites ......................................... 33
Vegetation Cover and Nest Success ......................................................... 33
Survival Estimates .................................................................................... 36
Vegetation Cover and Survival ................................................................. 40
DISCUSSION AND MANAGEMENT IMPLICATIONS .................................. 42
Nest Checks and Banding ......................................................................... 42
Vegetation Comparisons within Nesting Sites ......................................... 44

Nests vs. Random Points ............................................................... 44

Successful vs. Unsuccessful Nests ................................................ 45

Vegetation Comparisons among Nesting Sites ......................................... 45

Vegetation Cover and Nest Success ......................................................... 46
Survival Estimates .................................................................................... 47
Vegetation Cover and Survival ................................................................. 48
LITERATURE CITED ......................................................................................... 52
APPENDIX 1 ........................................................................................................ 55
APPENDIX 2 ........................................................................................................ 59

CHAPTER 2: IMPACTS OF VEGETATION COVER ON COMMON TERN EGG
SURVIVAL ON HIGH AND LOW WATER NESTING SITES IN THE ST. MARY’S
RIVER AND SAGINAW BAY, MICHIGAN

ABSTRACT .......................................................................................................... 60
INTRODUCTION ................................................................................................ 6 1
OBJECTIVES ....................................................................................................... 64
METHODS ........................................................................................................... 65
Egg Survival .............................................................................................. 65
Vegetation Cover ...................................................................................... 66
Vegetation Cover and Survival ................................................................. 67
RESULTS ............................................................................................................. 68
Vegetation Comparisons ........................................................................... 68

High vs. Low Water Sites ............................................................. 68

Vegetation Cover and Survival Estimates ................................................ 68
DISCUSSION AND MANAGEMENT IMPLICATIONS .................................. 72
LITERATURE CITED ......................................................................................... 75
APPENDIX ........................................................................................................... 77

CHAPTER 3: ANALYZING POPULATION VIABILITY OF COMMON TERNS IN
THE ST. MARY’S RIVER, MICHIGAN USING VORTEX

ABSTRACT .......................................................................................................... 78

INTRODUCTION ................................................................................................ 79

vi

OBJECTIVES ....................................................................................................... 82

METHODS ........................................................................................................... 83
Modeling Method ...................................................................................... 83
Demographic Input Parameters ................................................................. 83
VORTEX Simulations .............................................................................. 85
Initial Population Projections .................................................................... 85

Density Dependent Breeding ........................................................ 86
Inbreeding Depression .................................................................. 88
Catastrophes .................................................................................. 88
Assumptions .............................................................................................. 89
Sensitivity Analyses .................................................................................. 90
Simulation Experiments ............................................................................ 92
Single-parameter Manipulations ................................................... 92
Two-parameter Manipulations ...................................................... 92

RESULTS ............................................................................................................. 94
Initial Population Projections .................................................................... 94
Sensitivity Analyses .................................................................................. 97
Simulation Experiments ............................................................................ 99

Single-parameter Manipulations ................................................... 99
11% Probability of Flooding ............................................. 99
52% Probability of Flooding ............................................. 99
Two-parameter Manipulations ...................................................... 99
11% Probability of Flooding ............................................. 99
52% Probability of Flooding ........................................... 101
Stable Population ........................................................................ 101
1 1% Probability of Flooding ........................................... 101
52% Probability of Flooding ........................................... 101

DISCUSSION AND MANAGEMENT IMPLICATIONS ................................ 105

LITERATURE CITED ....................................................................................... 111

APPENDIX 1 ...................................................................................................... 115

APPENDIX 2 ...................................................................................................... 118

vii

LIST OF TABLES

CHAPTER 1: IMPACTS OF VEGETATION COVER ON COMMON TERN
REPRODUCTIVE SUCCESS IN THE ST. MARY’S RIVER, MICHIGAN

Table 1. Number of breeding pairs (greatest number of active nests on a given day),
total number of chicks banded, number of chicks hatched per pair, mean
number of chicks hatched per pair, number of chicks fledged per pair, and
mean number of chicks fledged per pair for common terns nesting in the St.
Mary’s River, Michigan, summers 1997-1998 and 2000-2001 ............................ 26

Table 2. Mean (SE) vegetation characteristics at common tern nests and random
points on Lime Island, Michigan (summers 1997 and 1998). .............................. 27

Table 3. Mean (SE) vegetation characteristics at common tern nests and random
points on Andrews Reef, Michigan (summers 2000 and 2001) ............................ 29

Table 4. Mean (SE) vegetation characteristics at common tern nests and random
points on Harbor Island Reef, Michigan (summers 2000 and 2001). ................... 30

Table 5. Mean (SE) vegetation characteristics at successful (those that hatched at
least one young) and unsuccessful (those that did not hatch any young)
common tern nests on Lime Island, Michigan (summers 1997 and 1998). .......... 31

Table 6. Mean (SE) vegetation characteristics at successful (those that hatched at
least one young) and unsuccessful (those that did not hatch any young)
common tern nests on Andrews Reef, Michigan (summers 2000 and 2001). ...... 32

Table 7. Mean (SE) vegetation characteristics at successful (those that hatched at
least one young) and unsuccessful (those that did not hatch any young)
common tern nests on Harbor Island Reef, Michigan (summers 2000 and
2001) .................................................................................................................... 34

Table 8. Mean (SE) vegetation characteristics around common terns nests on all
nesting sites in the St. Mary’s River, Michigan, summers 1997-1998, 2000-
2001 ....................................................................................................................... 35

Table 9. One-variable models for predicting nesting success from vegetation
variables collected at common tern nests in the St. Mary’s River, Michigan,
1997-1998, and 2000-2001. .................................................................................. 37

Table 10. Daily (Sd) and period (Sp) survival rates (SD; Mayfreld 1961) of
common tern eggs on three nesting Sites in the St. Mary’s River, Michigan,
summers 1997-1998 (Lime Island) and 2000-2001 (Andrews Reef and
Harbor Island Reef) ............................................................................................... 38

viii

Table 11. Daily (Sd) and period (Sp) survival rates (SD; Mayfield 1961) of
common tern chicks on three nesting Sites in the St. Mary’s River,
Michigan, summers 1997-1998 (Lime Island) and 2000-2001 (Andrews
Reef and Harbor Island Reef). .............................................................................. 39

Appendix 1. Table 1. Common vegetation Species found growing on the Lime
Island coal dock, Michigan in summers 1997 and 1998 ....................................... 55

Appendix 1. Table 2. Common vegetation species found growing on Andrews
Reef, Michigan in summers 2000 and 2001 ......................................................... 56

Appendix 1. Table 3. Common vegetation species found growing on Harbor Island
Reef, Michigan in summers 2000 and 2001. ........................................................ 58

Appendix 2. Number of adults trapped, number of adults trapped with bands
already present (band returns), percent band returns, and number of band returns
from 1997 (birds banded as juveniles on Lime Island in 1997) in the St. Mary’s
River, summers 2000-2001. .................................................................................. 59

CHAPTER 2: IMPACTS OF VEGETATION COVER ON COMMON TERN EGG
SURVIVAL ON HIGH AND LOW WATER NESTING SITES IN THE ST. MARY’S
RIVER AND SAGINAW BAY, MICHIGAN

Table 1. Mean (SE) vegetation characteristics at common tern nests on high
water sites (CDF, 1996-1997; Lime Island 1997-1998) and low water Sites
(Andrews Reef and Harbor Island Reef, 2000-2001) in Michigan. ...................... 69

Table 2. Daily (Sd) and period (Sp) survival rates (Mayfield 1961) of
common tern eggs on high water nesting sites (CDF, 1996- 1997; Lime
Island, 1997-1998) and low water nesting sites (Andrews and
Harbor Island Reef, 2000-2001) in Michigan. ...................................................... 70

CHAPTER 3: ANALYZING POPULATION VIABILITY OF COMMON TERNS IN
THE ST. MARY’S RIVER, MICHIGAN USING VORTEX

Table 1. Initial VORTEX population simulations for common terns in the St.
Mary’s River. K = carrying capacity. Severity (S) = severity with respect to

survival .................................................................................................................. 87

Table 2. Juvenile and sub-adult mortality rate combinations used in the 49 two-
parameter VORTEX simulations for common terns in the St. Mary’s River ....... 93

ix

Table 3. Growth rate (r) (SD), probability of extinction (SD), and mean years to
extinction (SD) for initial VORTEX population simulations for common
terns in the St. Mary’s River. Results are based on 1,000 replicates for each
lOO-year simulation .............................................................................................. 96

Table 4. Sensitivity indices for increases and decreases in baseline values for
juvenile mortality, sub-adult mortality, carrying capacity, and flooding rates
in VORTEX simulations for common terns in the St. Mary’s River.
Separate sets of simulations were run with probability of flooding = 11%
and probability of flooding = 52% ........................................................................ 98

Appendix 1. Summary of initial VORTEX parameter values and data sources for
the common tern population in the St. Mary's River ............................................ 115

Appendix 2. Assumed age distribution of common terns in the St. Mary's River in
1997 (N = 1,298). Values are based on percentages reported by Austin and
Austin (1956). ....................................................................................................... l 18

LIST OF FIGURES

STUDY AREA

Fig. 1. Location of Andrews Reef, Harbor Island Reef, and Lime Island in the St.
Mary’s River, Michigan. ....................................................................................... 3

Fig. 2. Aerial view of Andrews Reef, Michigan. Common terns nest on the entire
area of Andrews Reef, except within the dotted line. Ring-billed gulls
nested within the dotted line. The Shaded regions represent distinct areas of
vegetation cover. ................................................................................................... 4

Fig. 3. Aerial view of Harbor Island Reef, Michigan. In 2000, common terns
nested north of the upper dotted line and south of the lower dotted line.
Ring-billed gulls were located between the dotted lines. In 2001, common
terns nested north of the dashed line and on the rocky area north of Harbor
Island Reef. Ring-billed gulls nested south of the dashed line. ........................... 5

Fig. 4. Aerial view of the west side of Lime Island, Michigan. Common terns (C)
were located within the fence line on the coal dock and on the rock pile in
1997 and 1998. Ring-billed gulls (R) were also located on the west side of
the fence line in 1997 and 1998. .......................................................................... 7

CHAPTER 1: IMPACTS OF VEGETATION COVER ON COMMON TERN
REPRODUCTIVE SUCCESS IN THE ST. MARY’S RIVER, MICHIGAN

Fig. 1. Egg and chick survival rates and mean percent total cover, bare ground, and
litter cover on the three nesting Sites used by common terns in the St.
Mary’s River, Michigan, 1997-1998 and 2000-2001. .......................................... 41

CHAPTER 2: IMPACTS OF VEGETATION COVER ON COMMON TERN EGG
SURVIVAL ON HIGH AND LOW WATER NESTING SITES IN THE ST. MARY’S
RIVER AND SAGINAW BAY, MICHIGAN

Fig. 1. Comparisons of percent total cover, bare ground, and litter cover with
common tern egg survival rates on high water nesting sites and low water
nesting Sites in the St. Mary’s River and Saginaw Bay, Michigan. ...................... 71

Appendix 1. Location of the Confined Disposal Facility (CDF) in Saginaw Bay,

Michigan. The CDF was used as a common tern nesting site during summers
1996-1997 ............................................................................................................. 77

xi

CHAPTER 3: ANALYZING POPULATION VIABILITY OF COMMON TERNS IN
THE ST. MARY’S RIVER, MICHIGAN USING VORTEX

Fig. 1.

Fig. 2.

Fig. 3.

Fig. 4.

Initial population projections for common terns in the St. Mary’s River.

Density dependence (1): max. % breeding at 0 = 100%, % breeding at K =

50%, exponential steepness = 2, Allee parameter = 1. Density dependence

(2): max. % breeding at 0 = 100%, % breeding at K = 80%, exponential
steepness = 16, Allee parameter = 0. Inbreeding: 3.14 lethal alleles, 50% of
genetic load due to lethal alleles. Flood (1): probability = 52%, severity

with respect to survival = 95%. Flood (2): probability = 11%, severity with
respect to survival = 95%. Flood (3): probability = 52%, severity with

respect to survival = 99%. Flood (4): probability = 11%, severity with

respect to survival = 99% ...................................................................................... 95

Probabilities of extinction with increasing juvenile and sub-adult mortality

rates for common terns in the St. Mary’s River, M1 for simulations with an

11% probability of flooding (a) and a 52% probability of flooding (b). The

only input value changed in each simulation was either juvenile or sub-adult
mortality rate. ........................................................................................................ 100

Targeting 0% extinction and a stable population of common terns in the St.
Mary’s River. Varying combinations of juvenile and sub-adult mortality

rates and resulting probabilities of extinction are shown for simulations with

an 11% probability of flooding (a) and a 52% probability of flooding (b) ........... 102

Stable common tern population in the St. Mary’s River for simulations with

an 11% probability of flooding (a) and a 52% probability of flooding (b).

Juvenile mortality rate = 57% and sub-adult mortality rate = 37.7% (a).

Juvenile mortality rate = 40%, sub-adult mortality rate = 35% (b). ..................... 103

xii

THESIS STRUCTURE

This thesis is divided into three chapters:

Chapter 1: Impacts of Vegetation Cover on Common Tern Reproductive Success
in the St. Mary’s River, Michigan.

Chapter 2: Impacts of Vegetation Cover on Common Tern Egg Survival on High
and Low Water Nesting Sites in the St. Mary’s River and Saginaw Bay, Michigan.

Chapter 3: Analyzing Population Viability of Common Tems in the St. Mary’s

River, Michigan using VORTEX.

In Chapter 1, I investigated the relationship between vegetation cover and
common tern reproductive success among three nesting sites used in the St. Mary’s
River, Michigan (1997-2001). The focus of Chapter 2 is a comparison of vegetation
cover and egg survival between nesting sites used during high water years (1996-1998)
and low water years (2000—2001) in the St. Mary’s River (1997-2001) and in Saginaw
Bay (1996-1997), Michigan. In Chapter 3, I have developed a population viability
analysis (PVA) for common terns in the St. Mary’s River using VORTEX. Using
VORTEX, I have predicted the long-term survival potential of common terns in the St.
Mary’s River.

Each chapter contains Abstract, Introduction, Objectives, Methods, Results,
Discussion and Management Implications, and Literature Cited sections as well as

Appendices. A study area description precedes Chapter 1.

STUDY AREA

In the St. Mary’s River, common terns (Sterna hirundo) nested on Lime Island
from 1997-1998 and on Andrews Reef and Harbor Island Reef from 2000-2001.

Andrews Reef is located approximately 6 km from the eastern shore of
Michigan’s Upper Peninsula in Chippewa County (46°4’N 83°53’W; Figure 1). In 2000
and 2001, the entire area of Andrews Reef was used as a nesting site by common terns
(Figure 2). One pair and 25 pairs of ring-billed gulls (Lams delawarensis) nested on the
southwestern end of Andrews Reef in 2000 and 2001, respectively. During both years,
double-crested cormorants (Phalacrocorax auritus) were seen roosting along the western
shore of the reef. There are no mammalian residents on Andrews Reef. Andrews Reef is
approximately 0.13 ha in size. The highest point on Andrews Reef is approximately 1 m
above the water level, and the substrate ranges from cobble to boulders. Distinct areas of
vegetation occurring on Andrews Reef include spotted knapweed (Centaurea maculosa)
and lady’s thumb (Polygonum persicaria; Figure 2).

Harbor Island Reef is located approximately 10.4 km from the eastern shore of
Michigan’s Upper Peninsula in Chippewa County (46°3’N 83°47’W; Figure 1). In 2000,
common terns nested on the north and south ends of Harbor Island Reef, with the center
of the island used as a nesting site by a colony of ring-billed gulls (approximately 400
breeding pairs; Figure 3). In 2001, common terns nested within the first 11.5 m on the
north end of Harbor Island Reef and 700 — 800 ring-billed gull pairs nested on the
remainder of the reef south of the tern nesting area. Common terns also nested on a small
rocky area 5 m north of Harbor Island Reef. This area was 36 m long, 1 m wide at the

narrowest point, and 9.3 m at the widest point. During both years, double-crested

 

\
i CANADA
|

  
  

-' Chippewa
County ' "

   

 

   

if Mackinac
" County

  
 
  

 

 

Drummond
Island

  

 

UNITED STATES

Fig. 1. Location of Andrews Reef, Harbor Island Reef, and Lime Island in the St. Mary’s
River, Michigan.

 

Scale: 2.54 cm = 10 m
E] Spotted Knapweed
E] Lady’s Thumb

Ring-billed gull
nesting area

 

 

 

 

Fig. 2. Aerial view of Andrews Reef, Michigan. In 2000 and 200], common terns nested
on the entire area of Andrews Reef, except within the dotted line. Ring-billed gulls
nested within the dotted line. The shaded regions represent distinct areas of vegetation

cover.

 

 

 

Scale: 2.54 cm = 20 m

    

.‘.o" I D Pond
D Gravel

“*5 Lady’s Thumb

.IIIIIID‘

___>z

 

[:j Stinging Nettle

I Unidentified grass

 

 

 

Fig. 3. Aerial view of Harbor Island Reef, Michigan. In 2000, common terns nested
north of the upper dotted line and south of the lower dotted line. Ring—billed gulls were
located between the dotted lines. In 2001, common terns nested north of the dashed line
and on the rocky area north of Harbor Island Reef. Ring-billed gulls nested south of the
dashed line.

cormorants were seen roosting along the southwestern end of Harbor Island Reef. No
mammalian residents are found on Harbor Island Reef, but garter snakes (Thamnophis
spp.) are common. Harbor Island Reef is approximately 0.43 ha is size. The highest
point on Harbor Island Reef is approximately 1.5 m above the water level, and the
substrate includes sand, gravel, cobble and boulders. There is a pond approximately 20
m long by 5 m wide located near the center of the island (Figure 3). Distinct areas of
vegetation occurring on Harbor Island include lady’s thumb, stinging nettle (Urtica
dioica), and an unidentified grass species (Figure 3).

Lime Island is located approximately 4.0 km from the eastern shore of Michigan’s
Upper Peninsula in Chippewa County (46°05’N 84°01 ’W; Figure 1). Historically, two
areas on Lime Island were used as nesting sites by common terns: 1) an old coal dock
(along the west side of Lime Island; Figure 4) and 2) a rock pile adjacent to the coal dock
(Figure 4). The coal dock, connected to the north end of Lime Island via a bridge (5.5 m
long by 1.8 m wide), is 0.8 ha in size and is covered with mowed grass. In 1997 and
1998, common terns nested on the southern end of the coal dock and shared the area with
a small colony of ring-billed gulls (Cook 1999; Figure 4). The vegetation in the colony
area of the coal dock is mowed twice a year: once in early May and once in late August.
The adjacent rock pile rises 1.9 m out of the water, is approximately 0.01 ha in size, and
the substrate ranges from cobble to boulders. The vegetation on the rock pile consists
primarily of forb species (e. g., catnip [Nepeta cataria], pale touch-me—not [Impatiens
pallida], and goldenrod [Solidago spp.]).

Permanent mammalian residents found on Lime Island include weasels (Mustela

spp.), mink (Mustela vision), raccoon (Procyon lotor), black bear (Ursus americanus),

.—.’z

BRIDGE

 

   

LIME
ISLAND

COAL
DOCK

FENCE

 

no
00000
00000

CCC
CCC
R CCC
R CCC
CCCC
CCCC
CCCC
CCCC
CCCC
CCC
CCC

 

Scale: 2.54 cm = 45 m

 

 

 

 

 

 

 

 

Fig. 4. Aerial view of the west side of Lime Island, Michigan. Common terns (C) were
located within the fence line on the coal dock and on the rock pile in 1997 and 1998.
Ring-billed gulls (R) were also located on the west side of the fence line in 1997 and
1998.

coyote (Cam's latrans), red fox (Vulpes vulpes), white-tailed deer (Odocoileus
virginianus), beavers (Castor canadensis), and river otter (Lutra canadensis). Avian
residents include Canada geese (Branta canadensis), ring-billed gulls, herring gulls
(Larus argentatus), bald eagles (Halia leucocephalus), osprey (Pandion haliaetus),
killdeer (Charadrius vociferus), spotted sandpipers (Actitis macularia), common

mergansers (Mergus merganser) and many songbirds.

LITERATURE CITED

Cook, B. S. 1999. Impacts of vegetation manipulations on the reproductive success of
common terns (Sterna hirundo) on Lime Island in the St. Mary’s River, Michigan.
MS. Thesis, Mich. State Univ. 101 pp.

CHAPTER 1
IMPACTS OF VEGETATION COVER ON COMMON TERN REPRODUCTIVE
SUCCESS IN THE ST. MARY’S RIVER, MICHIGAN

ABSTRACT

Common terns are believed to prefer 10-30% total vegetation cover around their
nests. However, research on Lime Island (1997-1998) in the St. Mary’s River, Michigan
has shown that common tern nest success is better in areas with moderate amounts (40-
50%) of total vegetation cover. To determine in which amount of vegetation cover
reproductive success is the greatest, I investigated the relationship between reproductive
success and vegetation cover on three nesting sites in the St. Mary’s River: Lime Island
(1997-1998), Andrews Reef (2000-2001) and Harbor Island Reef (2000-2001). Percent
total vegetation cover was 21.2% on Andrews Reef, 36.3% on Harbor Island Reef, and
70.3% on Lime Island. Percent bare ground was 61.9% on Andrews Reef, 45.5% on
Harbor Island Reef, and 5.9% on Lime Island. Egg and chick survival rates were the
highest on Andrews Reef (69.5% and 88.7%, respectively), followed by Harbor Island
Reef (66.3% and 86%, respectively) and Lime Island (25.3% and 43.6%, respectively).
Percent total cover was negatively correlated with egg survival (rs = -0.94, P = 0.005) and
percent bare ground was positively correlated with egg (rS = 0.89, P = 0.02) and chick (rs
= 0.83, P = 0.04) survivals. Analyses suggest that common terns in the St. Mary’s River
have the highest survival in areas supporting 20-40% vegetation cover and > 40% bare
ground. On naturally occurring islands (e.g., Andrews and Harbor Island Reefs),
common terns in the St. Mary’s River have better nesting success in low to moderate
amounts (20-40%) of vegetation cover but on human-made sites (i.e., Lime Island), they

have better nesting success in moderate amounts (40-50%) of total vegetation cover.

10

More research is needed to support or refute these conclusions. Management should
include vegetation manipulations to provide the amount of vegetation cover under which

reproductive success is the greatest.

ll

INTRODUCTION

Common terns prefer nest sites in early stages of plant succession with 10-30%
cover (Soots and Parnell 1975). Although this amount of cover is minimal, it offers a
number of advantages. First, this sparse amount of cover offers open spaces that adult
common terns use in courtship displays prior to copulation (Burger and Gochfeld 1991).
Second, these open spaces allow for movement within vegetation without the risk of
entanglement, thus making common terns less susceptible to predation (Burger and
Shisler 1978). Third, this amount of cover provides protection of chicks against avian
and mammalian predators (by providing hiding cover) and protection of adults and chicks
from harsh weather conditions (e. g., sun, wind, rain; Blokpoel et a1. 1978). Finally, the
vegetation creates a barrier that decreases the amount of visual contact between nests,
thereby reducing territorial conflict between neighboring terns (Burger and Shisler 1978).

As vegetation grows tall (> 0.60 m) and dense (> 30% total cover) with
succeeding years, common terns may abandon nest sites (Courtney and Blokpoel 1983,
Shields and Townsend 1985, Burger and Gochfeld 1991). Dense vegetation creates a
number of disadvantages for nesting common terns: disappearance of required landing
sites (Palmer 1941), reduced visibility resulting in a reduction of social stimulation
among colony members (Palmer 1941), increased risk of entanglement in thick
vegetation making adults and chicks more susceptible to predation (Burger and Shisler
1978), and an inability to detect predators through loss of visual contact. These factors
can decrease common tern reproductive success by either reducing mating (Palmer 1941)

or causing adult, chick, and egg mortality (Burger and Shisler 1978).

12

Previous research has investigated the impacts of vegetation on common tern
reproductive success and nest site selection on natural and human-made sites. In a newly
established colony at the Eastern Headland, Toronto, Ontario, the relationship between
nest sites and vegetation was investigated (Blokpoel et a1. 1978). In a 130 x 215 m study
plot on the headland (an elevated breakwater), vegetation cover within the plot was
compared to vegetation cover around nests within the plot. Mean vegetation cover within
the study plot (11 = 495) was 16%. Greater (> 31%, f = 44%) vegetation cover occurred
more frequently (x2 = 89.58, P < 0.001) around nests (n = 49) Within the plot than areas
where no nests where located. In the study plot, the majority of terns nested on or near
broad bands of low and clumped vegetation parallel to bare beaches, indicating terns
avoided nesting in areas with little or no vegetation (Blokpoel et a1. 1978).

To determine how vegetation encroachment affected common tern nesting habits,
researchers cleared vegetation from two islands in the Eastern Headland of Lake Ontario
in April and May of 1982 (Morris et a1. 1992). Common tern numbers rose from 13
breeding pairs on each island in 1981 to 218 and 562 breeding pairs in 1982. In
succeeding years, no vegetation control was undertaken, and common terns abandoned
the islands as vegetation height and density increased (Morris et a1. 1992).

Other studies have examined the use of different substrate types and vegetation in
determining preferred nesting sites of common terns. In one of these studies, researchers
provided a choice of three substrate types on a breakwall in Lake Erie to test nesting
substrate preferences of late-nesting common terns (Richards and Morris 1984). Tems
preferred “super-enhanced” areas (103 nests) covered with small rocks and clumps (15-

40 cm diameter) of mossy stonecrop (Sedum acre) and driftwood randomly distributed

l3

over the area to the enhanced (small rocks; 48 nests) and control (bare concrete; 15 nests)
areas (Richards and Morris 1984). In addition, mean clutch size and fledging success
were all higher on the super-enhanced substrate, suggesting that enhancement procedures
are valuable management tools for improving reproductive success of late-nesting
common terns (Richards and Morris 1984).

In a similar study of habitat enhancement in Lake Erie near Port Colboume,
Ontario, large and small rocks were spread over a breakwater area, and then small logs,
driftwood, and debris were placed on the rocky substrate (Morris et a1. 1992).
Additionally, mossy stonecrop was planted in the substrate. Following these
manipulations, common tern numbers rose from 906 breeding pairs in 1988 to 1,052
breeding pairs in 1989 (Morris et a1. 1992).

In another experimental study on Long Island, Oneida Lake, New York, 5 nesting
substrate types with 2 replicates each were set up in parallel strips to investigate nest site
selection by common terns (Severinghaus 1982). The typical nesting substrate at this site
was dried grass interspersed with stones of different sizes (control). The experimental

substrates added were: dried grass, large stones (2 2 times the size of a tern egg), medium

stones (1-2 times egg size), and small stones (3 1 egg size). Tems preferred (x2 =
25.1251, df = 4, P < 0.05) dried grass (21 nests) as a nesting substrate followed by the
control (14 nests; Severinghaus 1982). Relatively fewer nests were found on the stone
substrates (13 nests in 3 substrates; Severinghaus 1982). Nest site selection was not
linked to nesting success in this study.

Vegetation manipulations were performed on Lime Island, located in the St.

Mary’s River, Michigan, to determine if reduced vegetation cover increased reproductive

14

success (Cook 1999, Cook-Haley and Millenbah 2002). Common terns used an old coal
dock connected to Lime Island as a nesting site in 1997 and 1998. The vegetation cover
on Lime Island (81.5% in 1997) exceeded the 10 - 30% cover normally preferred by this
species (Soots and Parnell 1975) and was believed to be negatively impacting common
tern reproductive success by causing possible entanglement or becoming a barrier to
predator detection (Cook 1999, Cook—Haley and Millenbah 2002).

In cooperation with the MDNR, vegetation manipulations were performed in
1998 to determine if vegetation cover was affecting common tern reproductive success.
Two treatment types, complete herbiciding (spraying of all vegetation within the plot)
and partial herbiciding (spraying of dense spots of vegetation within the plot), and a
control were randomly applied to 5 x 5 m plots (n = 16 or 17 for each treatment).
Common tern nest success on Lime Island in 1998 was affected by treatment type (x2 =
24.5, df = 9, P < 0.005; Cook-Haley and Millenbah 2002). Partial herbicide areas (41.3%
total cover) had a lower than expected number of nests that failed to hatch any young
(partial 12 = 4.5; P s 0.10) and greater than expected number of nests that successfully
hatched 2 young (partial x2 = 42.7; P s 0.10; Cook-Haley and Millenbah 2002).
However, in the complete herbicide areas (24.4% total cover) a greater than expected

number of nests failed to hatch young (partiaI x2 = 4.5; P s 0.10) and a lower than

expected number of nests hatched 3 young (partial x2 = 3.2; P s 0.10; Cook-Haley and
Millenbah 2002). This study suggested that common terns have better nest success in
areas that supported moderate amounts (40-50%) of total standing vegetation cover

(partial herbicide), indicated by the tendency to hatch more young per nest (Cook-Haley

15

and Millenbah 2002). All of these studies suggest that common terns prefer nesting in
sparsely vegetated areas to areas with no vegetation or heavily vegetated areas.

The findings on Lime Island suggest that nesting common terns may tolerate
greater amounts of cover in the northern Great Lakes than predicted by the literature (10-
30%; Soots and Parnell 1975; Burger and Gochfeld 1991). Soots and Parnell (1975) and
Burger and Gochfeld (1991) researched common terns inhabiting the east coast of the
U. S. that may be susceptible to different weather conditions (i.e., hurricanes) and have
different available nesting substrate than in the Great Lakes. The difference in these
factors may explain why common terns seem to tolerate greater than predicted vegetation
cover in the northern Great Lakes region.

Originally, one of the purposes of this study was to continue investigating the
impacts of vegetation manipulations on common tern reproductive success on Lime
Island by repeating the 1998 study and determining if the conclusions of the 1998 study
could be either supported or refuted. However, in 1999, following flooding and predation
by a long-tailed weasel (Mustelafrenata) and ring-billed gulls, common terns abandoned
Lime Island as a nesting site in early June (Millenbah and Lamp 1999). In 1999,
researchers were unable to follow common terns afier they abandoned Lime Island;
therefore reproductive success in 1999 is unknown. In 2000 and 2001, common terns did
not return to Lime Island to nest but were found nesting on two natural sites in the St.
Mary’s River: Andrews Reef and Harbor Island Reef.

The purpose of this research was to assess common tern reproductive success in
the St. Mary’s River (1997-1998 and 2000-2001) and determine how vegetation cover

affects reproductive success. The common tern is listed as a threatened species in

16

Michigan (Scharf 1991), with the St. Mary’s River currently supporting the largest
number of common terns in the state of Michigan (> 50% of breeding pairs; Cuthbert et
a1. 1997). Therefore, efforts to assess common tern reproductive success and factors
affecting success in the St. Mary’s River are necessary. By assessing vegetation cover
and its effects on common tern reproductive success, management recommendations can
be made to continue the existence of this species in Michigan. These management
recommendations may also be applicable to other common terns colonies in the northern

Great Lakes region.

17

OBJECTIVES

Specific objectives of this study were to:
1) determine the reproductive success of common terns in the St. Mary’s River;
2) quantify the vegetation cover of different common tern nesting sites in the St.
Mary’s River;
3) determine the effects of vegetation cover on common tern reproductive
success among different nesting sites in the St. Mary’s River; and
4) make management recommendations for common terns in the St. Mary’s

River.

18

METHODS

The Lime Island common tern colony was monitored and observed from 5 June to
8 August 1997 and 21 May to 1 August 1998 (Cook 1999). Andrews Reef and Harbor
Island Reef common tern colonies were monitored and observed from 12 June to 3
August 2000 and 8 May to 21 July 2001 to obtain data on reproductive success and

vegetation cover.

Nest Checks and Survival Estimates

Nest checks, completed every 3-6 days, were used to estimate reproductive
success. Nesting variables recorded at nests included date of egg laying, number of eggs,
date of completed clutch, number of eggs lost, and reason(s) for loss, hatching date,
number of chicks hatched, number of chicks Iost, and reason(s) for loss, fledging date,
and number of chicks fledged. Each nest was given a unique number using a permanent
marker on a rock or wooden dowel to allow nests to be followed until hatching, fledging,
or until the nest was destroyed or abandoned. Attempts were made to follow all chicks
until fledging. To facilitate following chicks to fledging, chicks were hand captured and
banded using a USFWS #2 steel band and one stripe celluloid color band. Chicks were
banded only after they were completely dry. Appropriate state and federal banding,
capturing, and handling permits were obtained from the MDNR and USFWS. Chicks
were aged according to criteria presented by Nisbet and Drury (1972). Re-nesting

attempts were also followed. All methods for capturing and banding adults and chicks

19

were approved by Michigan State University's All-University Committee on Animal Use
and Care (AUF # 02/99-017-00).

Daily and period survival estimates for eggs and chicks were calculated using the
Mayfield method (Mayfield 1961). Daily survival is the survival from the beginning of
one day to the next, and period survival is the survival from initiation of incubation to
hatching (eggs) and hatching to fledging (chicks). The Mayfield method calculates
survival based on deaths and exposure days. Survival estimates were made for two time
periods: 1) from initiation of incubation to hatching and 2) from hatching to fledging.
Eggs were used as the experimental unit from initiation of incubation to hatching and
chicks were used as the experimental unit from hatching to fledging. Nests were
considered active if at least one egg or chick was present in the nest cup. Abandoned
nests were incorporated into the estimate using a mean of 12 exposure days. Egg and
chick period survivals were calculated using 24 days (L; Mayfield 1961) as the mean
number of incubation days and 28 days (L) as the mean number of days from hatching to
fledging (Burger and Gochfeld 1991). Chicks were considered fledged if they were last
recorded at 218 days of age. Chicks younger than 18 days of age on last capture and not
found dead were considered censored (fate unknown).

Exposure days for censored eggs and chicks were incorporated into the Mayfield
estimates using the midpoint method that terminates exposure days at the midpoint
between the last day eggs or chicks were known to be active and the first day eggs or
chicks were known to be inactive (Manolis et al. 2000). Censored individuals were
included in Mayfield estimation because their exclusion results in an underestimate

(Manolis et al. 2000)

20

Vegetation Cover

A 50 X 50 cm modified Daubenmire frame (Daubenmire 1959) was centered at
each nest and vegetation measurements were taken to determine vegetation cover around
nests and to compare differences in vegetation cover 1) between successful and
unsuccessful nests within a nesting site, 2) between nests and random points within a
nesting site, and 3) at nests among nesting sites. Vegetation variables included percent
live, dead, forb, grass, and woody cover as well as estimates of percent bare ground,
percent litter cover, and vegetation height. Vegetation variables were measured after a
completed clutch of eggs was present in the nest cup. Vegetation variables were
measured at 79 nests and paired random points and 69 nests and paired random points on
Lime Island in 1997 and 1998, respectively (Cook 1999). In 2000 and 2001, vegetation
variables were measured at 50 nests and paired random points on Andrews Reef and 52
and 50, respectively, nests and paired random points on Harbor Island Reef. Vegetation
measurements were taken at paired random points to determine if particular vegetation
characteristics are associated with nest placement. Paired points were chosen randomly
in places where no nests were located by walking three paces in a random direction from

each nest location.

Data Analysis

Vegetation Comparisons within Nesting Sites

Vegetation variables between successful and unsuccessful nests at each nesting
site were compared (P S 0.05; Mann-Whitney U (MWU); Mann and Whitney 1947) to

determine if specific vegetation variables were associated with successful nesters. A nest

21

was considered successful if at least one young hatched and unsuccessful if no young
hatched. Vegetation variables were also compared (P S 0.05; MWU) between nests and
random points on each nesting site. This comparison was used to determine if nest
placement was associated with specific vegetation variables. Multi-year data were
pooled within nesting sites. Although there were significant differences (P S 0.05)
between years for some vegetation characteristics at each nesting site, the vegetation
variables identified as significantly different between successful and unsuccessful nests
and nests and random points were similar when multi-year data were pooled for each
nesting site. Therefore, multi-year data were pooled to evaluate each nesting site as a

whole.

Vegetation Comparisons among Nesting Sites

A Kruskal-Wallis (KW) one-way analysis of variance (P S 0.05; Siegel 1956) was
used to determine if there was a significant difference in vegetation cover among nesting
sites used by common terns in the St. Mary’s River from 1997-2001 (Lime Island,
Andrews Island Reef, and Harbor Island Reef). Multi-year data were pooled within each
nesting site for reasons similar to those outlined in the previous section. If significant
differences were found using a KW, a KW mutiple-comparison z-test (z > 1.96) was used
to identify specific differences. Since no vegetation manipulations were performed on
Andrews Island Reef and Harbor Island Reef in 2000 or 2001, only vegetation data
collected in the unmanipulated (Cook 1999) area on the Lime Island coal dock in 1997

and 1998 were used in the statistical comparisons mentioned above.

22

Vegetation Cover and Nest Success

Logistic regressions were performed to determine if vegetation cover could be
used to predict nest success. Logistic regression was chosen because nest success is a
binary response variable (i.e., successful or unsuccessful). One and 2—variable models
were run for common tern nests at each nesting site (multi-year data pooled for reasons
mentioned previously). Covariates in the l-variable models were either percent total
cover, bare ground, litter cover, or vegetation height. Percent total cover and vegetation
height, percent bare ground and vegetation height, and percent litter cover and vegetation
height were the covariates in the 2-variable models. Since percent total cover, bare
ground, and litter cover are not independent variables (i.e., % total cover + % bare
ground + % litter cover = 100%), these were not incorporated in the same logistic
regressions. Percent live, dead, grass, forb, and woody cover were not included in the
regression analysis because they are not independent of percent total cover. Height was
included in the 2-variable models because vegetation height is independent of the other
vegetation variables measured. All Iogistic regressions were conducted with a 0.20
model selection cutoff criteria (Hosmer and Lemeshow 2000) using a forward variable

selection. Variables were considered significant at P S 0.10.

Vegetation Cover and Survival
A Spearman rank correlation (P S 0.05) was used to determine correlations
between vegetation characteristics and common tern egg and chick survivals. Using the

same rationale outlined in the logistic regression section, percent total and litter cover,

23

percent bare ground, and vegetation height were the only vegetation characteristics used

in the correlation analysis.

24

RESULTS

Nest Checks and Banding

Common tern nests and chicks were first observed between 17 May and 17 June
and 5 June and 24 June, respectively, from 1997-2001 (Cook 1999). Common terns
departed the St. Mary’s River area between mid-July and mid-September from 1997-
2001 (Cook 1999).

The total number of chicks banded and the average number of chicks hatched per
pair were the greatest on Lime Island in 1997 (1,475 and 2.27, respectively; Cook 1999;
Table 1). When common terns were not nesting on Lime Island, the greatest number of
chicks was banded on Andrews Reef in 2001 (n = 254; Table 1) and the greatest number
of chicks hatched per pair occurred on Andrews Reef in 2000 (1.92; Table 1). More
chicks fledged per pair on Andrews Reef in 2000 (0.70) than any other year. The poorest
fledging success occurred on Lime Island in 1998 (0.21 chicks fledged/pair; Table 1).

The mean number of chicks hatched per pair ranged from 1.78 on Harbor Island
Reef to 1.88 on Andrews Reef (Table l). The mean number of chicks fledged per pair

was the greatest on Harbor Island Reef (0.69) and the lowest on Lime Island (0.42; Table

1).

Vegetation Comparisons within Nesting Sites
Nests vs. Random Points
On Lime Island, percent bare ground was greater (P = 0.005) around nests than

random points (Table 2). Conversely, percent litter cover was greater (P < 0.001) around

25

.38: :88 as: 32885::
8: 98893988 899 8:208 9993 .98: .30 98:3 885 89: mo :8: 8988 88 :0 98853 :8: 809 8: 98:38 9898:...

 

 

mod mm; RN :2 .839 .79 8983 Sow
mod wk;
mod 21.9 39 mo 88% .99 898E ooom
:3 mo.— me wwm .839 8898»: Bow
$6 $9
end $9 909 cm: .839 mace: ooom
Ed $9 om: Rm :9 883 wmfl
$6 2:
3: ANN mi} 39 .29 885 309
9:3 8% 9:3 83 9:3 83 9:3 83 898m 89:3 8% 8: 88>
9383 $9990 :3on 89990 9898:m 89990 898:3 8.9990 88990.90 898on
no 898:2 :82 .9: 898:2 mo 898:2 8:82 we 898:2 898:2 gob .90
898:2

 

.Soméoom 8: woo—-33 8888:: 8:89:92 89% 30:2 :m 89: :9 3:8: 88: :0888 8.9 93 83 wowwoc
88990 .90 898:: :88 :8: .93 83 8389.9 @9990 .8 898:: .93 83 9898:: 88990 .90 898:: 8:88 .93 83 898:9 8.8990
30 8898:: 988:9 89990 .90 898:: 9:89 63: 83w : :o 88: 898: me 898:: 88:83 893 89889 mo 898:2 .9 29:3.

26

Table 2. Mean (SE) vegetation characteristics at common tern nests and random points on

Lime Island, Michigan (summers 1997 and 1998).

 

 

Nests Random Points
Characteristic (n = 148) (n = 148)
% Total Cover 70.3 (2.2) 65.6 (2.3)
% Live Cover 59.7 (2.1) 54.6 (2.2)
% Grass Cover 57.8 (2.1) 52.0 (2.2)
% Forb Cover 2.2 (0.8) 2.6 (0.9)
% Woody Cover 0.0 (0.0) 0.0 (0.0)
%Dead Cover 10.6 (1.0) 11.0 (1.1)
% Bare Ground* 5.9 (1.1) 4.7 (1.2)
% Litter Cover* 23.8 (1.9) 29.7 (2.1)

 

*significantly different (Mann-Whitney U (MWU) test, P S 0.05)

27

random points than around nests (Table 2). Nests and random points on Lime Island were
not significantly different for any other vegetation variables. Since no measurements of
vegetation height were taken in 1997, vegetation height was not compared between nests
and random points on Lime Island.

There was a greater (P = 0.03 and P < 0.0001, respectively) percentage of dead
and litter cover around common tern nests than random points and a greater (P < 0.001)
amount of bare ground around random points than nests on Andrews Reef (Table 3). No
other vegetation variables were significantly different between nests and random points on
Andrews Reef.

On Harbor Island Reef, percent total and live cover were greater (P = 0.04 for
both variables) around random points than around nests (Table 4). Percent litter cover was
greater (P < 0.0001) around common tern nests than around random points (Table 4). No
significant differences were found between nests and random points for any other
vegetation variables (Table 4).

Common vegetation species found growing on the Lime Island coal dock in 1997

and 1998 and on Andrews Reef and Harbor Island Reef in 2000 and 2001 are listed in

Appendix 1.

Successful vs. Unsuccessful Nests

There were no significant differences between successful and unsuccessful nests
for any vegetation variables measured on Lime Island (Table 5) or Andrews Reef (Table
6).

On Harbor Island Reef, successful nests had a greater (P = 0.04, P = 0.03,

28

Table 3. Mean (SE) vegetation characteristics at common tern nests and random points on

Andrews Reef, Michigan (summers 2000 and 2001).

 

 

Nests Random Points
Characteristic (n = 100) (n = 100)
% Total Cover 21.2 (2.6) 23.2 (3.0)
% Live Cover 21.1 (2.6) 23.2 (3.0)
% Grass Cover 14.9 (2.3) 14.4 (2.6)
% Forb Cover 5.7 (1.7) 8.6 (2.2)
% Woody Cover 0.1 (0.1) 0.1 (0.1)
% Dead Cover* 0.6 (0.3) 0.2 (0.2)
% Bare Ground* 61.9 (3.0) 73.4 (3.1)
% Litter Cover* 17.0 (1.2) 3.4 (0.4)
Height (cm) 12.2 (1.4) 12.4 (1.5)

 

*significantly different (Mann-Whitney U (MWU) test, P S 0.05)

29

Table 4. Mean (SE) vegetation characteristics at common tern nests and random points on
Harbor Island Reef, Michigan (summers 2000 and 2001).

 

 

Nests Random Points
Characteristic (n = 102) (n = 102)
% Total Cover* 36.3 (2.2) 46.6 (3.4)
% Live Cover* 36.0 (2.2) 46.3 (3.4)
% Grass Cover 20.1 (2.4) 28.3 (3.5)
% Forb Cover 13.0 (1.9) 16.1 (2.8)
% Woody Cover 0.5 (0.4) 0.0 (0.0)
% Dead Cover 2.6 (0.5) 2.1 (0.7)
% Bare Ground 45.5 (2.4) 46.8 (3.5)
°/o Litter Cover* 18.3 (1.1) 6.6 (1.1)
Height (cm) 26.3 (1.6) 24.8 (1.8)

 

*significantly different (Mann-Whitney U (MWU) test, P S 0.05)

30

Table 5. Mean (SE) vegetation characteristics at successful (those that hatched at least
one young) and unsuccessful (those that did not hatch any young) common tern nests on

Lime Island, Michigan (summers 1997 and 1998).

 

 

Successful Unsuccessful

Characteristic (n = 93) (n = 40)

% Total Cover 66.9 (2.9) 74.3 (3.4)
% Live Cover 58.4 (2.7) 62.1 (3.5)

% Grass Cover 56.5 (2.7) 61.1 (3.5)

% Forb Cover 1.9 (0.9) 1.1 (0.6)

% Woody Cover 0.0 (0.0) 0.0 (0.0)

% Dead Cover 8.5 (1.5) 12.1 (1.9)

% Bare Ground 5.6 (1.4) 5.8 (1.5)
% Litter Cover 27.4 (2.6) 20.0 (3.2)

 

31

Table 6. Mean (SE) vegetation characteristics at successful (those that hatched at least
one young) and unsuccessful (those that did not hatch any young) common tern nests on
Andrews Reef, Michigan (summers 2000 and 2001).

 

 

Successful Unsuccessful

Characteristic (n = 20) (n = 38)

°/o Total Cover 28.8 (6.1) 15.4 (3.7)
% Live Cover 28.8 (6.1) 15.4 (3.7)

% Grass Cover 13.8 (4.5) 10.4 (3.2)

% Forb Cover 15.0 (6.1) 4.3 (2.1)

% Woody Cover 0.0 (0.0) 0.0 (0.0)

% Dead Cover 0.0 (0.0) 0.7 (0.5)

% Bare Ground 50.8 (6.6) 67.8 (4.6)
% Litter Cover 20.5 (3.0) 16.8 (2.0)
Height (cm) 15.3 (2.8) 8.4 (1.7)

 

32

P < 0.0001, and P = 0.004, respectively) percentage of total, live, grass, and dead cover
around them than did unsuccessful nests (Table 7). Percent forb cover and bare ground
were greater (P < 0.0001 and P = 0.006, respectively) around unsuccessful nests than

around successful nests (Table 7).

Vegetation Comparisons among Nesting Sites

Percent total, live, grass, forb, and dead cover and percent bare ground were
different (P < 0.0001 for all 6 variables) among all nesting sites (Table 8). Percent total
cover was greater around nests on Lime Island than on either Andrews Reef or Harbor
Island Reef (2 = 11.52 and z = 7.95, respectively; Table 8). Percent total cover was
greater (2 = 3.33) around Harbor Island Reef nests than around Andrews Reef nests
(Table 8). Percent bare ground was lower around Lime Island nests than around either
Andrews Reef or Harbor Island Reef nests (z = 13.66 and z = 11.17, respectively; Table
8). Andrews Reef nests had a greater (2 = 2.35) percent bare ground around them than
did Harbor Island Reef nests (Table 8). Vegetation was taller (P < 0.0001) around
Harbor Island Reef nests than around Andrews Reef nests (Table 8). Neither percent

woody cover nor percent litter cover was different among nesting sites (Table 8).

Vegetation Cover and Nest Success

On Lime Island, the 1-variab1e model with percent litter cover was the only model
that predicted nest success (Table 9). Percent litter cover was negatively associated (B =
-0.014, P = 0.10) with nest success. On Andrews Reef, the 1- variable models with

percent total cover, percent bare ground, and vegetation height correctly predicted nest

33

Table 7. Mean (SE) vegetation characteristics at successful (those that hatched at least
one young) and unsuccessful (those that did not hatch any young) common tern nests on
Harbor Island Reef, Michigan (summers 2000 and 2001).

 

 

Successful Unsuccessful
Characteristic (n = 33) (n = 30)
% Total Cover* 42.0 (4.0) 28.7 (4.0)
% Live Cover* 41.8 (4.0) 28.0 (4.0)
% Grass Cover* 35.5 (3.9) 12.5 (3.9)
% Forb Cover* 1.4 (1.0) 14.9 (3.5)
% Woody Cover 0.0 (0.0) 0.0 (0.0)
% Dead Cover* 5.1 (1.2) 1.3 (0.8)
% Bare Ground* 37.3 (3.9) 54.8 (4.6)
% Litter Cover 21.1 (2.0) 16.5 (1.8)
Height (cm) 27.4 (2.2) 21.1 (2.5)

 

*significantly different (Mann-Whitney U (MWU) test, P S 0.05)

34

.306 w 3 .3236 89m 398: 88389 888.99: 38:99:39....
Ga; v N .88-: 80983800 0:99:8v 8083.9:

38.89988 90: 8: 8:0— 08:: 09: 83:... 8:08 .38 : 8993 A36 w m .Bv: v.09: w::8: 308: 888.99: 38:09:88...

.33 :9 908808 8: 83 8309 89:88.»—

 

 

..G: :8 <3: :2 :83 9:6:
3.: 2: a: 6.: 6.: :2 .260 6:: .\.
bad 6.? mad 6.6 <3: mm .6580 66m .x.
63.8 no 63.8 :0 <88 6.2 .360 66: .x.
3.8 no 88 No 88 80 6.60 68>» Ox.
68.: :8 ma. 3 6.8 <6. 3 N: .1260 :6: .x.
m9: 2 68: o: (3.: 3m .1660 66:0 .x.
09.6 oem 66.3 :8 <86 5% .660 65 .x.
66:0 :3 66.8 N: <26 :8 .1660 66.: .x.
8: u 5 82 n 5 a: n 5 883880
88-88 58-88 82-32
66: 8:2 6:6: .66: $222. 8:: 68:

 

.Soméoom £3ng2 8088::

8399092 83¢ 30:2 :m 09H :9 89m 398: =: :0 $8: :88 808800 8:8: 8989889890 8988?; Ammv :82 .w 29:3.

35

success 63.8%, 62.1%, and 70.7% of the time, respectively (Table 9). Percent total cover
and height were negatively associated (B = -2.14, P = 0.06 and B = -5.03, P = 0.04,
respectively) with nest success whereas percent bare ground was positively associated

(B: 2.00, P = 0.04) with nest success. On Harbor Island Reef, all of the 1-variable models
were significant (Table 9). Percent total cover, bare ground, and litter cover and
vegetation height correctly predicted nest success 61 .9%, 60.3%, 63.55, and 60.3% of the
time, respectively (Table 9). Percent total cover, litter cover, and height were negatively
associated (B = -2.65, P = 0.03; B = -4.11, P = 0.10; and B = -3.67, P = 0.07, respectively)
with common tern nest success. Conversely, percent bare ground was positively
associated (B = 3.10, P = 0.01) with nest success. None of the Z-variable models for any

nesting site were significant.

Survival Estimates

Daily survival estimates for common tern eggs in the St. Mary’s River from
initiation of incubation to hatching ranged from 94.4 % to 98.5% (Table 10). Period
survival rates (L = 24) from egg-laying to hatching ranged from 25.3% on Lime Island to
69.5% on Andrews Reef (Table 10). Censoring of nests was greater than 10% in 1998,
2000, and 2001, violating the assumption of allowable censoring (< 10%; Mayfield
1961). Therefore, care should be taken in interpreting these results.

Daily survival estimates for common tern chicks in the St. Mary’s River from
hatching to Hedging ranged from 97.1% to 99.6% (Table 11). Period survival (L = 28) of
chicks from hatching to fledging ranged from 43.6% on Lime Island to 88.7% on

Andrews Reef (Table 11). Censoring of chicks exceeded 10% in all years; therefore care

36

Table 9. One-variable models for predicting nesting success from vegetation variables
collected at common tern nests in the St. Mary’s River, Michigan, 1997-1998, and 2000-
2001.

Model X2 P-value Concordancea

 

Lime Island - 1997 and 1998 (n = 133)

 

% Total Cover 2.29 0.13 69.92
% Bare Groundb 0.00 1.00 69.92
% Litter Cover 2.89 0.09 69.92
Heightb 0.00 1.00 81.16

 

Andrews Reef- 2000 and 2001 (n = 58)

 

% Total Cover 3.71 0.05 63.79
% Bare Ground 4.40 0.04 62.07
% Litter Coverb 0.00 1.00 65.52
Height 4.43 0.04 70.69

 

Harbor Island Reef - 2000 and 2001 (n = 63)

 

% Total Cover 5.39 0.02 61.90
% Bare Ground 8.17 0.004 60.32
% Litter Cover 2.86 0.09 63.49
Height 3.52 0.06 60.32

 

' percent likelihood of correctly classifying nests as successful or unsuccessful based on the model.
b using forward variable selection, these variables were not included in the model; therefore, a x2 of 0.00
was obtained.

37

Table 10. Daily (Sd) and period (Sp) survival rates (SD; Mayfield 1961) of common tern
eggs on three nesting sites in the St. Mary’s River, Michigan, summers 1997-1998 (Lime
Island) and 2000-2001 (Andrews Reef and Harbor Island Reef).

 

 

Nesting Site Sd Sp
Lime Island“ 0.9444 0.2530
Andrews Reef 0.9849 (0.0010) 0.6949 (0.0167)
Harbor Island Reef 0.9830 (0.0017) 0.6627 (0.0281)

 

* Cook 1999; data needed to calculate SD not available for Lime Island

38

Table 11. Daily (Sd) and period (Sp) survival rates (SD; Mayfleld 1961) of common tern
chicks on three nesting sites in the St. Mary’s River, Michigan, summers 1997-1998
(Lime Island) and 2000-2001 (Andrews Reef and Harbor Island Reef).

 

 

Nesting Site Sd Sp
Lime Island* 0.9708 0.4361
Andrews Reef 0.9957 (0.0012) 0.8866 (0.0285)
Harbor Island Reef 0.9946 (0.0015) 0.8597 (0.0332)

 

* Cook 1999; data needed to calculate SD not available for Lime Island

39

should be taken when interpreting these results.
Although adults were trapped for band returns, no survival estimates could be
calculated for sub-adult or adult breeders (due to lack of adequate return data). A

summary of adult trapping results and band returns is shown in Appendix 2.

Vegetation Cover and Survival

Percent total cover was negatively correlated with egg survival (IS = -0.94, P =
0.005). Percent bare ground was positively correlated with egg (rs = 0.89, P = 0.02) and
chick survivals (rS = 0.83, P = 0.04). Percent total (rs = -0.66, P = 0.16) and litter cover (rS
= —0.77, P = 0.07) were negatively correlated with chick survival, but these differences
were not significant. There were no significant correlations between percent litter cover
and egg survival or between vegetation height and either egg or chick survival. Figure 1
shows a comparison between egg and chick survival rates and vegetation cover on Lime

Island and Andrews and Harbor Island Reefs.

4o

1 00 -.-,---------.- -- -- -- - ----- - WM- ----.--.......---_-, _ -- . ---.--- W-w - -m-_--- ..--,

90 4» . __..
80 IL. A 7- .2. .7-

...,._......_.._ --..—...-.._ M-_._- -

Percentage

 

 

 

 

 

 

 

 

Lime Island Andrews Reef Harbor Island Reef

Nesting Site

Total Cover iBare Ground Cl Litter co}? T Egg Survival 0 Chic‘kSurvival

 

Fig. 1. Egg and chick survival rates and mean percent total cover, bare ground, and litter
cover on the three nesting sites used by common terns in the St. Mary’s River, Michigan,
1997-1998 and 2000-2001.

41

DISCUSSION AND MANAGEMENT IMPLICATIONS

Nest Checks and Banding

It is likely that common terns began nesting on Lime Island before 5 June 1997
and 21 May 1998, but due to weather, researchers were unable to reach the island before
those dates (Cook 1999). In 1999, researchers arrived on Lime Island on 26 May, and
common terns had already established nests. Following a flood on 3 June, which left 2
inches of standing water on the coal dock, ring-billed gulls and a long-tailed weasel were
seen eating common tern eggs (W. Mieras, pers. com.) I observed the long-tailed
weasel entering and leaving the colony on two occasions and captured the weasel
crossing the bridge to the coal dock on film with a Trail-Master trail monitor (Goodson
and Associates, Inc., Lenexa, Kansas) connected to a 35-mm camera. Cached and
crushed eggs were further evidence of weasel predation within the colony (pers. observ.).
Weasels have been reported to destroy colonies in Massachusetts (Austin 1929), and
ring-billed gulls can be significant egg predators of common terns (Cuthbert et al. 1984).

Following the flooding and predation events in 1999, common terns abandoned
Lime Island as a nesting site in early June (pers. observ.) Because I was unable to follow
common terns to other nesting sites in 1999, survival rates are unknown in the St. Mary’s
River for 1999. Some of the terns from Lime Island re-nested on Little Cass Island in the
St. Mary’s River (T. Ludwig, pers. comm.) In 2000, common terns did not return to
Lime Island to nest, and I was unable to locate the new common tern nesting sites until
mid-June of that year. In 2001, common terns began nesting sometime between 8 or 9
May and May 17. When I visited Andrews Reef and Harbor Island Reef on 8 May and 9

May, respectively, common terns were observed displaying courtship rituals and

42

copulating, but no nests had been laid. When I returned on 17 May, common tern nests
were observed on both islands.

The number of common tern breeding pairs in the St. Mary’s River fluctuated
from 649 in 1997 to 727 in 1998, 249 in 2000, and 425 in 2001. Since I was unable to
reach common tern nesting sites until mid-June in 2000, the estimate of number of
breeding pairs in that year is an underestimate. By mid-June, many chicks had already
hatched and were moving out of their nests, making it difficult to associate chicks with
nests and estimate the total number of nests. The number of breeding pairs in 2000 is
also an underestimate for the St. Mary’s River as a whole. There were 4 small colonies
(< 50 breeding pairs on each) in the northern part of the St. Mary’s River (T. Ludwig
pers. comm.) that were too far away for researchers to monitor intensively. The total
number of breeding pairs in the St. Mary’s River during 2000 was therefore at least 400 —
500 pairs. The only islands used by common terns in the St. Mary’s River in 2001 were
Andrews Reef and Harbor Island Reef (T. Ludwig, pers. comm), therefore 425 breeding
pairs is an accurate estimate of the population size in 2001. Overall, the number of
breeding pairs, and thus the population size, of common terns using the St. Mary’s River
area appears to have decreased since 1997. It is possible that some individuals that
nested on Lime Island in 1997 and 1998 may have nested on the Canadian side of the St.
Mary’s River or other areas in Michigan in 2000 and 2001, resulting in the apparent
decrease in breeding pairs over time.

It is believed that common tern colonies require 1.1 young fledged/pair to
maintain a stable breeding population (DiConstanzo 1980). The weighted mean number

of chicks fledged/pair on Lime Island, Andrews Reef, and Harbor Island Reef were

43

61.8%, 52.7%, and 37.3% less than the required chicks/pair, respectively (Table 1). The
weighted mean number of chicks fledged per pair on all nesting sites was less than the
required 1.1 chicks fledged pair, indicating a declining population in the St. Mary’s
River. Long-term population persistence is discussed thoroughly in Chapter 3 of this
document.

As mentioned previously, the number of breeding pairs on both Andrews and
Harbor Island Reefs in 2000 was an underestimate. Since the numbers of chicks hatched
per pair and fledged per pair is based on the number of breeding pairs, these values for
Andrews and Harbor Island Reefs in 2000 (Table 1) are overestimates. It is thus likely
that the number of chicks hatched and fledged per pair were less than 1.83 and 0.70,

respectively, on Andrews Reef and 1.77 and 0.68, respectively, on Harbor Island Reef.

Vegetation Comparisons within Nesting Sites
Nest vs. Random Points

Across all nesting sites the only consistently significant difference in vegetation
cover between areas where common terns nested and areas they did not nest (i.e., random
points) was mean percent litter cover. On Lime Island, there was significantly less litter
cover around nests (23.8%) than around random points (29.7%; Table 2). On Andrews
Reef and Harbor Island Reef 2000, there was significantly more litter cover around nests
(17.0% and 18.3%, respectively) than around random points (3.4% and 6.6%,
respectively; Table 3 and Table 4). These results indicate that common terns nesting in
the St. Mary’s River are preferentially building their nests in areas where there is

approximately 15-25% litter cover. Excess litter cover can increase the chance of chick

44

entanglement and susceptibility to predation (Burger and Shisler 1978); therefore,

common terns may be avoiding litter cover greater than 25% on these nesting sites.

Successful vs. Unsuccessful Nests

There were no consistent differences in vegetation characteristics between
successful and unsuccessful nests among nesting sites. On Harbor Island Reef,
successful nests had greater percent total, live, and grass cover and less percent bare
ground than did unsuccessful nests (Table 7). However, there were no significant
differences between successful and unsuccessful nests on either Lime Island (Table 5) or
Andrews Reef (Table 6). Therefore, no vegetation characteristics seemed to have a
consistently strong impact on the success of common tern nests.

It is possible that nest success was influenced by more minute differences in
vegetation that were not measured in this study or the study done in 1997 and 1998 (Cook
1999). Additionally, there are other factors such as predation (Hatch 1970, Morris and
Hunter 1976, Scharf 1991, Ludwig 1991, Bumess and Morris 1992), competition with
ring-billed gulls (Morris and Hunter 1976, Scharf 1981, Scharf and Shugart 1985,
Ludwig 1991), food availability (Courtney and Blokpoel 1980; Safina et al. 1988)
parental incubation attentiveness (Courtney 1976), and weather conditions (e. g., wind,

heavy rain, and waves; Blokpoel et a1. 1978) influencing nest success.
Vegetation Comparisons among Nesting Sites

Among nesting sites, Andrews Reef supports the lowest percent total cover

(21.2%) and highest percent bare ground (61 .9%), followed by Harbor Island Reef nests

45

with intermediate percentages of total cover (36.3%) and bare ground (45.5%) and Lime
Island nests with the highest percent total cover (70.3%) and lowest percent bare ground
(5.9%; Table 8). Percent litter cover does not differ among nesting sites. It is interesting
to note that 82.2% of the total cover on Lime Island is comprised of grass cover, while on
Andrews Reef and Harbor Island Reef, forb cover accounts for 94.3% and 90.1% of the
total cover on each site. This difference is not surprising and is due to the nature of the
three nesting sites. The Lime Island coal dock is a man-made site that was seeded with
grasses by the MDNR (Cook 1999) whereas Andrews Reef and Harbor Island Reef are
natural, rocky reefs in the early stages of succession with vegetation composed primarily
of forb species.

During 2001, vegetation on Andrews Reef and Harbor Island Reef was denser and
taller than 2000 (pers. observ.). By mid-July 2001, common terns had vacated Andrews
Reef and Harbor Island Reef. Since common terns are known to abandon sites when the
vegetation grows taller than 0.60 m (Morris et al. 1980, Courtney and Blokpoel 1983,
Shields and Townsend 1985, Burger and Gochfeld 1991, Morris et al. 1992) and the
vegetation had grown to a height of approximately 1 meter on Andrews Reef and 1.5 m
on Harbor Island Reef by mid-July 2001, it is likely that common terns vacated both sites

early because of tall, dense vegetation.

Vegetation Cover and Nest Success
Logistic regression analyses indicate that none of the vegetation variables
measured (e. g., percent total cover, percent bare ground, percent litter cover, and

vegetation height) consistently predicts nest success among the nesting sites used in the

46

St. Mary’s River. Percent total cover, percent bare ground, and vegetation height were
predictors of nest success on Andrews and Harbor Island Reefs but not on Lime Island
(Table 9). Percent litter cover predicted nest success on Lime Island and Harbor Island
Reef, but not on Andrews Reef (Table 9).

Predictions from logistic regressions run for Lime Island, Andrews Reef, and
Harbor Island Reef vegetation variables support the pattern seen in comparisons between
successful and unsuccessful nests (i.e., quantitative differences in mean vegetation
characteristics between successful and unsuccessful nests) for each nesting site.
However, due to the dissimilarity of logistic regressions and MWU tests, statistical
differences were not always found for the same vegetation variable in both analyses. For
instance, there was no significant difference between successful and unsuccessful nests in
percent total cover, bare ground, or vegetation height on Andrews Reef (MWU; Table 6)
but the same vegetation variables were predictors of nest success in logistic regression

analyses for Andrews Reef (Table 9).

Survival Estimates

Common tern egg survival was highest on Andrews Reef (69.5%) followed by
Harbor Island Reef (66.3%) and Lime Island (25.3%; Table 10). Following the same
pattern, common tern chick survival was highest on Andrews Reef (88.7%) followed by
Harbor Island Reef (86.0%), and Lime Island (43.6%; Table 11). On all nesting sites,
chick censoring exceeded the allowable level of censoring (10%; Mayfield 1961).
Therefore, chick survival rates should be interpreted with caution.

Lower egg and chick survivals on Lime Island were due in part to heavy weasel

47

predation on the site in 1998 (Cook 1999). Because excess vegetation cover can cause
chicks to become entangled and increase susceptibility to predation (Burger and Shisler
197 8) and decrease open spaces used by adults to detect predators (Palmer 1941), effects
of weasel predation were likely exacerbated by excess total standing vegetation cover on
Lime Island.

Common terns nesting on Andrews Reef and Harbor Island Reef were not
affected by mammalian predation because the reefs are not connected to a mainland area
(a potential source of mammalian predators). Therefore, unlike common terns nesting on
Lime Island, those nesting on Andrews and Harbor Island Reef did not experience

decreased survival due to weasel or other mammalian predation.

Vegetation Cover and Survival

There were distinct differences in vegetation cover and common tern egg and
chick survival rates between Lime Island and Andrews Reef and Harbor Island Reef (Fig.
1). Total vegetation cover was greater and the amount of bare ground was less on Lime
Island than on Harbor Island Reef; and total vegetation cover was greater and the amount
of bare ground was less on Harbor Island Reef than on Andrews Reef. The opposite
trend was seen in survival rates. Overall survival was the highest on Andrews Reef
followed by Harbor Island Reef and Lime Island (Fig. 1). Correlation analyses provide
evidence egg and chick survival rates may be linked to vegetation cover: as percent total
cover increases, egg survival decreases and as percent bare ground increases, egg and
chick survivals increase.

Since survival rates were the highest on Andrews Reef and Harbor Island Reef,

48

and vegetation cover was lowest on these two sites (2.12% on Andrews Reef and 36.3%
on Harbor Island Reef), these results suggest that common terns in the St. Mary’s River
have the highest survival in areas with 20-40% cover and > 40% bare ground.

Results from the study conducted on Lime Island in 1997 and 1998, which
investigated the impacts of vegetation manipulations on common tern reproductive
success, suggest that terns have better nest success in moderate amounts (40-50%) of
total standing vegetation cover and lower amounts of litter cover (< 53%; Cook-Haley
and Millenbah 2002). In another study of common terns in the Great Lakes, researchers
found that vegetation cover around common tern nests on a human-made peninsula in
Ontario ranged from 30-50% (Blokpoel et al. 1978). However, survival was not
estimated or related to vegetation cover around nests.

I speculate that common terns in the St. Mary’s River and the northern Great
Lakes region may experience better reproductive success in different amounts of total
vegetation cover depending on whether they nest on a human-made or natural site. When
nesting on naturally occurring islands like Andrews Reef or Harbor Island Reef, common
terns appear to have better reproductive success in low to moderate amounts (20-40%) of
total vegetation cover, whereas when nesting on human-made sites like Lime Island terns
experience better reproductive success in moderate amounts (40-50%) of total vegetation
cover (Cook-Haley and Millenbah 2002). The coal dock common terns utilized as a
nesting site on Lime Island is connected to the main island by a bridge, providing
predators with easy access to the nesting colony (Cook 1999, pers. observ.). Forty to 50
percent vegetation cover may provide the needed hiding cover for chicks yet allow for

enough open space for visual detection and signaling of predators by adults. On natural

49

sites like Andrews and Harbor Island Reefs, which have no direct source of mammalian
predators, common terns may not require as much vegetation cover for hiding from
predators. Future research on sources of predation and predation rates on both types of
nesting sites is needed to address this hypothesis.

It is also possible that common terns have simply adapted to a greater amount of
vegetation cover on human-made sites like Lime Island, which supports greater
vegetation cover than natural sites. Natural sites are periodically submerged, depending
on seasonal and annual water level fluctuations, and generally tend to support less
vegetation cover.

Another year of research on Lime Island, in which managers replicate the
vegetation manipulations performed in 1998 (Cook 1999), is needed to either support or
refute the results of that study. If managers conclude common terns experience better
reproductive success in 40-50% total vegetation cover, the Lime Island coal dock should
be manipulated to reduce the vegetation cover on the entire nesting area to 40-50%. If
common terns experience better reproductive success in amounts of vegetation cover
different from 40-50%, vegetation should be manipulated to provide that particular
amount of cover.

Another year of research on natural sites like Andrews Reef and Harbor Island
Reef, is needed to support or refute the results of this study. If common tern reproductive
success if better in areas with 20-40% vegetation cover, managers should maintain
vegetation cover on natural sites within that range. Otherwise, vegetation should be
manipulated to provide the amount of cover that supports the best reproductive success.

Managers should reduce vegetation cover either through herbiciding, removing

50

vegetation, or adding substrate such as sand, rocks, or gravel (Richards and Morris 1984,
Morris et a1. 1992, Cook 1999, Cook—Haley and Millenbah 2002).

It is important to note that vegetation cover is not the only factor affecting
common tern egg and chick survival rates. Other factors such as predation, competition
with ring-billed gulls, food availability, weather conditions, and loss of habitat due to
human development influence nesting success. Therefore, future studies assessing the
effects of these factors on survival are warranted.

To determine long-term patterns in survival and changes in vegetation cover, at
least 5-10 more years of intensive monitoring is needed for the common tern population
in the St. Mary’s River. Through more intensive monitoring and further study of the
relationship between reproductive success and vegetation cover, managers can better

manage common terns in the St. Mary’s River for increased reproductive success.

51

LITERATURE CITED

Austin, 0. L., Jr. 1929. Contributions to the knowledge of the Cape Cod Steminae.
Bull. NE Bird Band. Assoc. 5:123-140.

Blokpoel, H., P. M. Catling, and G. T. Haymes. 1978. Relationship between nest sites of
common terns and vegetation on the eastern headland, Toronto outer harbor. Can.
J. Zool. 56: 2061-2067.

Burger, J ., and M. Gochfeld. 1991. The common tern. Columbia Univ. Press, New
York. 419 pp.

Burger, J ., and J. Shisler. 1978. Nest site selection and competitive interactions of
herring and laughing gulls in New Jersey. Auk. 95: 252-266.

Bumess, G. P., and R. D. Morris. 1992. Shelters decrease gull predation on chicks at a
common tern colony. J. Field Omithol. 63(2): 186-189.

Cook, B. S. 1999. Impacts of vegetation manipulations on the reproductive success of
common terns (Sterna hirundo) on Lime Island in the St. Mary’s River, Michigan.
MS. Thesis, Mich. State Univ. 101 pp.

Cook-Haley, B. S., and K. F. Millenbah. 2002. Impacts of vegetative manipulations
on common tern nest success on Lime Island, Michigan. J. Field Omithol.
73(2): 174-179.

Courtney, P. A. 1976. The relationship between incubation attention and reproductive
success in the common tern. Proc. of fish-eating birds of the Great Lakes and
env. contaminants symposium. Quebec, Canada.

Courtney, P. A., and H. Blokpoel. 1980. Food and indicators of food availability for
common terns on the lower Great Lakes. Can. J. Zool. 58: 1318-1323.

Courtney, P. A., and H. Blokpoel. 1983. Distribution and numbers of common terns on
the lower Great Lakes during 1900-1980: a review. Colonial Waterbirds. 6: 107-
120.

Cuthbert, F. J ., J McKeaman, and T. E. Davis. 1984. Status of common terns nesting at
the Duluth port terminal 1982-83. The Loon. 56: 20-24.

Cuthbert, F. J ., J. McKeaman, and L. Wemmer. 1997. US. Great Lakes Tern and
Cormorant Survey: 1997 Progress Report. 16 pp.

Daubenmire, R. F. 1959. A canopy coverage method of vegetational analysis.
Northwest Sci. 33: 43-64.

52

DiCostanzo, J. 1980. Population dynamics of a common tern colony. J. Field Omithol.
51(3): 229-243.

Hatch, J. J. 1970. Predation and piracy by gulls at a temery in Maine. Auk 87: 244-254.

Hosmer, D. W., and S. Lemeshow. 2000. Applied logistic regression. 2nd ed. Eds. N.
A. C. Cressie et al. John Wiley and Sons, Inc., New York. 375 pp.

Ludwig, J. P. 1991. Ring-billed gull. Pages 216-217 in R. Brewer, G. A. McPeek, and
R. J. Adams, eds. The atlas of breeding birds of Michigan. Mich. State Univ.
Press, East Lansing.

Mann, H. B., and D. R. Whitney. 1947. On a test of whether one or two random
variables is stochastically larger than the other. Ann. Math. Statist. 18: 50—60.

Manolis, J. C., D. E. Andersen, and F. J. Cuthbert. 2000. Uncertain nest fates in
songbird studies and variation in Mayfield estimation. The Auk. 117 (3): 615-
626.

Mayfield, H. M. 1961. Nesting success calculated from exposure. Wilson Bull. 73:
255-261.

Millenbah, K. F., and N. E. Lamp. 1999. Evaluating the reproductive success and
long-term survival potential of common terns (Sterna hirundo) on Lime Island,
Michigan. Annual Report to the Michigan Department of Natural Resources.

12 pp.

Morris, R. D., and R. A. Hunter. 1976. Factors influencing desertion of colony sites by
common terns (Sterna hirundo). Can. Field Nat. 90: 137-143.

Morris, R. D., I. R. Kirkham, and J. W. Chardine. 1980. Management of a declining
common tern colony. J. Wildl. Manage. 44(1): 241-245.

Morris, R. D., H. Blokpoel, and G. D. Tessier. 1992. Management efforts for the
conservation of common tern Sterna hirundo colonies in the Great Lakes: Two

case histories. Biol. Conserv. 60: 7-14.

Nisbet, I. C. T., and W. H. Drury. 1972. Measuring breeding success in common and
roseate terns. Bird-Banding. 43: 97-106.

Palmer, R. S. 1941. A behaviour study of the common tern (Sterna hirundo hirundo L.).
Proc. Boston Soc. Nat. Hist. 42: 1-119.

Richards, M. H., and R. D. Morris. 1984. An experimental study of nest site selection in
common terns. J. Field Omithol. 55: 457-466.

53

Safina, C., J. Burger, M. Gochfeld, and R. H. Wagner. 1988. Evidence for prey
limitation of common and roseate tern reproduction. The Condor. 90: 852-859.

Scharf, W. C. 1981. The significance of deteriorating man-made island habitats to
common terns and ring-billed gulls in St. Mary’s River, Michigan. Colonial
Waterbirds. 4: 155-159.

Scharf, W. C. 1991. Common tern. Pages 221-222 in R. Brewer, G. A. McPeek, and R.
J. Adams, eds. The atlas of breeding birds of Michigan. Mich. State Univ. Press,
East Lansing.

Scharf, W. C., and G. W. Shugart. 1985. Population sizes and status recommendation
for double-crested cormorants, black-crowned night herons, caspian terns,
common terns, and F orster’s terns in the Michigan Great Lakes in 1985. Report
to the Michigan Department of Natural Resources.

Severinghaus, L. L. 1982. Nest site selection by the common tern Sterna hirundo on
Oneida Lake, New York. Colonial Waterbirds. 5: 11-18.

Shields, M. A., and T. W. Townsend. 1985. Nesting success of Ohio’s endangered
common terns. Ohio J. Sci. 85: 45-49.

Siegel, S. 1956. Nonparametric statistics for the behavioral sciences. McGraw-Hill
Book Co., New York. 312 pp.

Soots, R. F., and J. F. Parnell. 1975. Ecological succession of breeding birds in relation
to plant success on dredge islands in North Carolina. Univ. North Carolina Sea
Grant Program publ. UNC-SG-75-27. 91 pp.

54

Appendix 1. Table 1. Common vegetation species found growing on the Lime Island
coal dock, Michigan in summers 1997 and 1998.

Common Name

Scientific Name

 

Black-eyed Susan
Cinquefoil spp.

Common evening primrose
Common milkweed
Common mullen

Field peppergrass

F ireweed (or Great willow herb)
Oxeye daisy

Poison ivy

Red clover

Red-osier dogwood
Shrubby St. J ohnswort
Sumac spp.

Tansy

White sweet clover

Wild red raspberry

Wild strawberry

Yellow goatsbeard

Yellow sweet clover

Rudbeckia serotina
Potentilla spp.
Oenothera biennis
Asclepias syriaca
Verbascum thapsus
Lepidium campestra
Epilobium angustifolium
Chrysanthemum Ieucanthemum
T oxicodendron radicans
T rlfolium pratense

C ornus stolonifera
Hypericum spathulatum
Rhus spp.

T anacetum vulgare
Melilotus alba

Rubus idaeus

F ragaria virginiana

T ragopogon pratensis

Melilotus oflicinalis

 

55

Appendix 1. Table 2. Common vegetation species found growing on Andrews Reef,
Michigan in summers 2000 and 2001.

Common Name Scientific Name

 

Blue vervain

Bull thistle

Canada thistle
Common dandelion
Common milkweed
Common reed
Cress spp.

Elm spp.
Goldenrod spp.
Grass spp.

Hairy willow herb
Knotweed spp.
Lady’s thumb

Pale touch-me-not
Path rush

Queen Anne’s lace
Redtop grass

Sedge spp.
Silverweed
Spiny-leaved sow thistle

Spotted knapweed

Verbena halei
Cirsium vulgare

C irsium arvense

T araxacum officinale
Asclepias syriaca
Phragmites australis
Brassicaceae spp.
Ulmus spp.

Solidago spp.
Poaceae spp.
Epilobz'um hirsutum
Polygonum spp.
Polygonum persicaria
Impatiens pallida

F uncus tenuis
Daucus carota

A grostis gigantea

C arex stipata
Potentilla anserina
Sanchus asper

C entaurea maculosa

 

56

Appendix 1. Table 2 (cont’d).

 

Common Name Scientific Name
White clover T rifolr'um repens
Wintercress or Yellow rocket Barbarea vulgaris

 

57

Appendix 1. Table 3. Common vegetation species found growing on Harbor Island
Reef, Michigan in summers 2000 and 2001.

Common Name Scientific Name

 

American hazelnut

Blue vervain

Common evening primrose
Flat-topped aster
Goldenrod spp.

Hairy willow herb
Knotweed spp.

Lady’s thumb

Orchard grass

Pale touch-me-not
Queen Anne’s lace

Reed canarygrass
Rough-fruited Cinquefoil
Rush spp.

Rush, Path

Sedge spp.

Sedge spp.

Stinging nettle

White clover

Wintercress or Yellow rocket

Corylus americana
Verbena halei
Oenothera biennis
Aster umbellatus
Solidago spp.
Epilobium hirsutum
Polygonum spp.
Polygonum persicaria
Dactylis glomerata
Impatiens pallida
Daucus carota
Phalaris arundinacea
Potentilla recta
Juncus spp.

F uncus tenuis

C arex stipata

C arex spp.

Urtica droica

T rifolium repens

Barbarea vulgaris

 

58

Appendix 2. Number of adults trapped, number of adults trapped with bands already
present (band returns), percent band returns, and number of band returns from 1997 (birds
banded as juveniles on Lime Island in 1997) in the St. Mary’s River, summers 2000-2001.

 

 

Number of Number of Band Percent Number of Returns from
Adults Trapped Returns Band Lime Island, 1997
Year Returns
2000 39 9 23.1 5
2001 49 8 16.3 3

 

59

CHAPTER 2
IMPACTS OF VEGETATION COVER ON COMMON TERN EGG SURVIVAL
ON HIGH AND LOW WATER NESTING SITES IN THE ST. MARY’S RIVER
AND SAGINAW BAY, MICHIGAN

ABSTRACT

Among the causes of the population decline of common terns in Michigan and the
Great Lakes region is the loss of suitable breeding habitat due to high water levels in the
Great Lakes. During years of high water levels (> 175.8 m, International Great Lakes
Datum (IGLD) 1985; 1996 — 1998) in the Great Lakes, common terns nested on human-
made sites in the St. Mary’s River (Lime Island) and Saginaw Bay (Confined Disposal
Facility; CDF), Michigan. During years of low water levels (< 174.8 m, IGLD 1985;
2000 — 2001), common terns nested on naturally occurring islands in the St. Mary’s River
(Andrews Reef and Harbor Island Reef). The purpose of this research was to assess and
compare common tern egg survival and vegetation cover on nesting sites used during
high and low water years. Nesting sites used by common terns during years of low water
levels supported less vegetation cover (28.8%) and had greater egg survival (67.9%) than
nesting sites used during years of high water levels (66.1% total vegetation cover, 35.9%
egg survival). Depending on Great Lakes water levels, managers should be prepared to
employ vegetation management strategies specific to nesting sites used during high and

low water years.

60

INTRODUCTION

Among the causes of the population decline of common terns in Michigan and the
Great Lakes region is the loss of suitable breeding habitat due to high water levels that is
exacerbated by loss of habitat due to human development in the Great Lakes (Scharf
1991). During years of high water levels [> 175.8 m, International Great Lakes Datum
(IGLD) 1985, Lakes Michigan-Huron; A. Fox pers. comm.]1, natural island nesting
habitat that typically supports more preferred vegetation cover is either partially or
completely submerged under water (Morris and Hunter 1976, Scharf 1991), forcing
common terns to select human-made sites that generally tend to support greater
vegetation cover (Chapter 1 of this document).

Great Lakes water levels are influenced by a combination of precipitation,
upstream inflows, groundwater, surface water runoff, evaporation, diversions into and out
of the system and climatic conditions (Hanna 1964, Office of the Great Lakes: Michigan
Department of Environmental Quality 2000). Great Lakes water levels fluctuate
seasonally, with higher water levels in summer than in winter and average seasonal
variation less than 0.2 - 0.6 meters (Tovell 1979, Office of the Great Lakes: Michigan
Department of Environmental Quality 2000). Long-term water level fluctuations occur at
intervals varying from 10 to 30 years between high and low water levels (Tovell 1979).

Extremely low water levels (< 174.8 m, IGLD 1985, Lakes Michigan-Huron; A. Fox,

 

1 Minimum value for high water year is based on the water level shown in a hydrograph for Lakes
Michigan-Huron during 1998, which was reported as a high water year and was a year when common terns
used Lime Island as a nesting site.

61

pers. comm.)2 were recorded in the mid-19208, mid—19303 and early 19605 while
exceptionally high water levels occurred in 1929-1930, 1952, 1973-1974, 1985-1986, and
1997-1998 (Office of the Great Lakes: Michigan Department of Environmental Quality
2000)

From 1997-1998 (Office of the Great Lakes: Michigan Department of
Environmental Quality 2000), a period coinciding with high Great Lakes water levels, the
only nesting site used by common terns in the St. Mary’s River was Lime Island, a
human-made site (Cook 1999). In 1999, a year of falling water levels (Office of the
Great Lakes: Michigan Department of Environmental Quality 2000), common terns
abandoned Lime Island (following flooding caused by a storm and predation by a long-
tailed weasel and ring-billed gulls; Millenbah and Lamp 1999) and nested on a number of
small, rocky reefs in the St. Mary’s River area which were exposed during that year (T.
Ludwig, pers. comm.) In 2000 and 2001, two years of low water levels (A. Fox pers.
comm), common terns did not nest on Lime Island but did nest on Andrews Reef and
Harbor Island Reef (Chapter 1 of this document), rocky reefs that were exposed due to
low water levels.

During another period of high water levels (1996-1997; Office of the Great Lakes:
Michigan Department of Environmental Quality 2000), common terns nested on the
Saginaw Bay Confined Disposal Facility (CDF) in Saginaw Bay, Michigan (Millenbah
1997; see Appendix 1 for map). The Army Corps of Engineers built the CDF in 1978 to

house contaminated dredge material from the Saginaw River and Saginaw Bay shipping

 

2 Maximum value for low water year is based on the water level shown in a hydrograph for Lakes
Michigan-Huron during 2000, which was reported as a low water year and was a year when common terns
used Andrews Reef and Harbor Island Reef as nesting sites.

62

channel (Millenbah 1997). The CDF is a kidney shaped island (118.5 ha) near the mouth
of the Saginaw River. The outer retaining wall of the CDF is approximately 3 — 5 m
above water level with the top of the interior dike wall 0.5 m below the outer retaining
wall. The interior cell, where dredge material is deposited, is 2 — 2.5 m below the interior
dike wall. From 1996-1997, common terns nested on the east-central section of the CDF,
in an area approximately 75 m long by 2 m wide (the width of the interior dike;
Millenbah 1997).

From 1996-2001, common tern nest site selection followed a similar pattern in the
St. Mary’s River and Saginaw Bay, Michigan. During years of high water levels (1996-
1998), common terns nested on human-made sites in both the St. Mary’s River (Lime
Island) and Saginaw Bay (CDF). During years of low water levels (2000-2001), common
terns nested on naturally occurring islands in the St. Mary’s River (Andrews Reef and
Harbor Island Reef; common terns in Saginaw Bay were not studied during the same
period of low water levels). Vegetation cover on nesting sites used during low water
years appeared different from nesting sites used during high water years. Therefore, the
purpose of this research was to assess and compare common tern egg survival and
vegetation cover high water (1996-1998) and low water (2000-2001) nesting sites in the

St. Mary’s River and Saginaw Bay, Michigan.

63

OBJECTIVES

Specific objectives of this study were to:
1) compare the effects of vegetation cover on common tern egg survival between
high water and low water nesting sites;
2) make recommendations for managing common terns in the St. Mary’s River on

high water and low water nesting sites.

64

METHODS

Common terns nesting on the CDF were observed and monitored from 22 May to
3 August 1996 and 28 May to 1 August 1997 (Millenbah 1997). The Lime Island
common tern colony was monitored and observed from 5 June to 8 August 1997 and 21
May to 1 August 1998 (Cook 1999). Andrews Reef and Harbor Island Reef common tern
colonies were monitored and observed from 12 June to 3 August 2000 and 8 May to 21

July 2001 to obtain data on egg survival and vegetation cover.

Egg Survival

Nest checks, completed every 3-6 days, were used to estimate egg survival.
Nesting variables recorded at nests included date of egg laying, number of eggs, date of
completed clutch, number of eggs lost, and reason(s) for loss. Chick censoring exceeded
the allowable level of censoring (10%; Mayfield 1961). Therefore, no chick data was
analyzed in this chapter. Each nest was given a unique number using a permanent marker
on a rock or a wooden dowel to allow nests to be followed until hatching, fledging, or
until the nest was destroyed or abandoned. Appropriate state and federal banding,
capturing, and handling permits were obtained from the MDNR and USFWS. Re-nesting
attempts were also followed. All methods for capturing and banding adults and chicks
were approved by Michigan State University's All-University Committee on Animal Use
and Care (AUF # 02/99-017-00).

Daily and period survival estimates for eggs were calculated using the Mayfield

method (Mayfield 1961). Daily survival is the survival from the beginning of one day to

65

the next, and period survival is the survival from initiation of incubation to hatching.
Eggs were used as the experimental unit from initiation of incubation to hatching. Nests
were considered active if at least one egg was present in the nest cup. Abandoned nests
were incorporated into the estimate using a mean of 12 exposure days. Egg period
survival was calculated using 24 days (L; Mayfield 1961) as the mean number of
incubation days.

Exposure days for censored eggs were incorporated into the Mayfield estimates
using the midpoint method that terminates exposure days at the midpoint between the last
day eggs were known to be active and the first day eggs or chicks were known to be
inactive (Manolis et al. 2000). Censored individuals were included in Mayfield

estimation because their exclusion results in a downward bias (Manolis et al. 2000)

Vegetation Cover

A 50 X 50 cm modified Daubenmire frame (Daubenmire 1959) was centered at
each nest and vegetation measurements were taken to compare differences in vegetation
cover between nesting sites used during high water years and low water years in the St.
Mary’s River and Saginaw Bay. Vegetation variables included percent live, dead, forb,
grass, and woody cover as well as estimates of percent bare ground, percent litter cover,
and vegetation height. Vegetation variables were measured after a completed clutch of
eggs was present in the nest cup. In 1996 and 1997, vegetation characteristics were
measured at 50 and 95, respectively, nests on the CDF (Millenbah 1997). Vegetation
variables were measured at 79 and 69 nests on Lime Island in 1997 and 1998,

respectively (Cook 1999). In 2000 and 2001, vegetation variables were measured at 50

66

nests (each year) on Andrews Reef and 52 and 50, respectively, nests on Harbor Island

Reef.

Vegetation Cover and Survival

Vegetation variables between nests at high water nesting sites (CDF, 1996-1997;
Lime Island, 1997-1998) and low water nesting sites (Andrews Reef and Harbor Island
Reef, 2000-2001) were compared (P S 0.05; Mann-Whitney U (MWU); Mann and
Whitney 1947) to determine if different vegetation characteristics were associated with
high and low water nesting sites. Successful (nests in which at least one young hatched)
and unsuccessful (nests in which no young hatched) nests were pooled based on Chapter
1 findings that there were no consistently significant differences between successful and
unsuccessful nests. Although there were significant differences (P S 0.05) between years
at a nesting site and nesting sites for some vegetation characteristics, the trends were
similar when multi-year and multi-site data were pooled to compare high water and low
water nesting sites. Therefore, multi-year data were pooled to evaluate vegetation

differences between high water and low water nesting sites.

67

RESULTS

Vegetation Comparisons
High vs. Low Water Sites

There were differences (P S 0.05) between high and low water nesting sites for all
vegetation characteristics measured (Table 1). Percent total, live, grass, and dead cover
were greater (P < 0.0001 for all 4 variables) around common tern nests on high water
sites than on low water sites (Table 1). Conversely, on low Water sites, percent forb,
woody, and litter cover as well as percent bare ground were greater (P = 0.001, P = 0.02,
P < 0.0001, and P < 0.0001, respectively) around nests than on high water sites (Table 1).
Vegetation was taller (P < 0.001) around nests on low water sites than on high water sites

(Table l).

Vegetation Cover and Survival Estimates

Daily survival for common tern eggs was 95.3% on high water sites and 98.4% on
low water sites. Period survival for common tern eggs was 35.9% and 67.9% on high
and low water sites, respectively (Table 3). Figure 1 shows a comparison between egg

survival and vegetation cover on high and low water nesting sites.

68

Table 1. Mean (SE) vegetation characteristics at common tern nests on high water sites

(CDF, 1996-1997; Lime Island 1997-1998) and low water sites (Andrews Reef and
Harbor Island Reef, 2000-2001) in Michigan.

 

 

High Water Sites Low Water Sites

Characteristic (n = 2931) (n = 202)

% Total Cover* 66.1 (1.3) 28.8 (1.8)
% Live Cover* 59.8 (1.3) 28.6 (1.8)

% Grass Cover* 40.9 (1 .7) 1.9 (0.3)

% Forb Cover* 19.5 (1.4) 26.4 (1.7)

% Woody Cover* 0.0 (0.0) 0.3 (0.2)

% Dead Cover* 5.9 (0.6) 0.2 (0.1)

% Bare Ground* 18.2 (1.1) 53.6 (2.0)
% Litter Cover* 15.6 (1.2) 17.7 (0.8)
Height (cm)* 11.9 (0.5) 19.3 (1 .2)

 

1 height measurements were only made on Lime Island in 1998, therefore n = 69 for

height on high water sites.

*significantly different (Mann-Whitney U (MWU) test, P S 0.05)

69

Table 2. Daily (Sd) and period (Sp) survival rates (Mayfield 1961) of common tern eggs
on high water nesting sites (CDF, 1996-1997; Lime Island, 1997-1998) and low water
nesting sites (Andrews and Harbor Island Reef, 2000-2001) in Michigan.

 

 

Nesting Category Sd Sp
High Water Sites 0.9528 0.3589
Low Water Sites 0.9840 0.6790

 

7O

Percentage

 

 

 

High Water Sites Low Water Sites

 

I'%WrbTaiWch¢r mafia-crud burner cave-{rig survival

 

Fig. 1. Comparisons of percent total cover, bare ground, and litter cover with common
tern egg survival rates on high water nesting sites and low water nesting sites used for
nesting in the St. Mary’s River and Saginaw Bay, Michigan.

71

DISCUSSION AND MANAGEMENT IMPLICATIONS

Nesting sites used by common terns during years of low water levels supported
less vegetation cover (28.8%; Table 1) and had greater egg survival (67.9%; Table 3)
than did nesting sites used during years of high water levels (66.1% total vegetation
cover, 35 .9% egg survival; Tables 1 and 3).

Whether or not common terns nest on human-made or natural sites in the St.
Mary’s River and Saginaw Bay is largely dependent on Great Lakes water levels,
especially when nesting locations are limited. When water levels are high in the Great
Lakes, natural islands like Andrews Reef and Harbor Island Reef become either
completely or mostly submerged under water, and human-made sites like Lime Island (>
1.5 m high during high water years; Cook 1999) or the CDF (3 — 5 m high during high
water years; Millenbah 1997) are relatively unaffected by changes in water levels.
During high water years, common terns are forced to select from fewer available nest
sites. As human development in the Great Lakes has increased, loss of suitable, natural
nesting habitat has also increased. Therefore, during years of high water levels, human-
made sites like Lime Island may be the only available nesting sites for common terns.

When the Great Lakes water levels rise once again and submerge more suitable
nesting sites like Andrews and Harbor Island Reefs, Lime Island will likely become an
important (and possibly exclusive) nesting site for common terns in the St. Mary’s River.
Therefore, during years of high water levels, it will become imperative to provide the
amount of vegetation cover in which reproductive success is the greatest on Lime Island.

Managers should be prepared to reduce vegetation cover on Lime Island by either

72

applying herbicide to the vegetation (Cook 1999) or adding substrate such as sand, rocks,
or gravel (Richards and Morris 1984, Morris et a1. 1992). An additional year of study on
Lime Island, in which the vegetation manipulations performed in 1998 are replicated
(Cook 1999), is needed support or refute the conclusion than common terns have better
nest success in areas with 40-50% vegetation cover (Cook-Haley and Millenbah 2002,
Chapter 1 of this document). If it is determined that reproductive success is better in
areas supporting 40-50% vegetation cover, managers should reduce the vegetation cover
on Lime Island to 40-50%. Otherwise, vegetation on Lime Island should be managed for
the percent cover in which reproductive success is determined to be the greatest.

When common terns return to nest on Lime Island in future years, a proactive
approach to predator management should be taken (based on the predation effects seen
during 1997-1999). Conibear trapping was shown to be effective for mammalian
predator control of a population of California least terns (Butchko and Small 1992).
Therefore, conibear traps targeted at weasels should be used on Lime Island as soon as
possible after common terns arrive on the nesting site in early May. No predation was
observed in the common tern colonies nesting on Andrews Reef and Harbor Island Reef
in 2000 or 2001, but managers should be prepared to employ predator management (if
necessary) on any islands common terns use as nesting sites in the future.

During years of low water levels, it is likely that common terns will nest on
Andrews Reef and Harbor Island Reef or other similar natural sites in the St. Mary’s
River. If common terns nest on these sites during a number of consecutive years, it is
likely the vegetation will succeed and become taller and denser. Managers should

monitor the vegetation cover on these sites to determine if cover has exceeded the range

73

within which nest success is better. Another year of research on low water sites is needed
to support or refute the conclusion that common tern reproductive success is better in
areas with 20-40% vegetation cover (Chapter 1 of this document). If vegetation cover
exceeds the amount in which reproductive success is the greatest, managers should
reduce cover either through applying herbicide, removing vegetation, or adding substrate
(Richards and Morris 1984, Morris et a1. 1992, Cook 1999, Cook-Haley and Millenbah
2002)

Because common terns move to different nesting sites based on Great Lakes water
levels, it is essential to manage both human-made and naturally occurring islands for

increased reproductive success during years in which they are used by common terns.

74

LITERATURE CITED

Butchko, P. H., and M. A. Small. 1992. Developing a strategy of predator control for the
protection of the California least tern: A case history. Proc. 15th vertebrate pest
conf. Univ. of California, Davis. 29-31.

Cook, B. S. 1999. Impacts of vegetation manipulations on the reproductive success of
common terns (Sterna hirundo) on Lime Island in the St. Mary’s River, Michigan.
MS. Thesis, Mich. State Univ. 101 pp.

Cook-Haley, B. S., and K. F. Millenbah. 2002. Impacts of vegetative manipulations
on common tern nest success on Lime Island, Michigan. J. Field Omithol.
73(2): 174-179.

Daubenmire, R. F. 1959. A canopy coverage method of vegetational analysis.
Northwest Sci. 33: 43-64.

Hanna, J. E. 1964. Great Lakes water level fluctuations. Shore and Beach. October: 12-
15.

Mann, H. B., and D. R. Whitney. 1947. On a test of whether one or two random
variables is stochastically larger than the other. Ann. Math. Statist. 18: 50-60.

Manolis, J. C., D. E. Andersen, and F. J. Cuthbert. 2000. Uncertain nest fates in
songbird studies and variation in Mayfield estimation. The Auk. 117 (3): 615-
626.

Mayfield, H. M. 1961. Nesting success calculated from exposure. Wilson Bull. 73:
255-261.

Millenbah, K. F. 1997. Assessment of common tern (Sterna hirundo) reproductive
success in Saginaw Bay of eastern Michigan. Ph.D. Dissertation, Mich. State
Univ. 111 pp.

Millenbah, K. F., and N. E. Lamp. 1999. Evaluating the reproductive success and
long-term survival potential of common terns (Sterna hirundo) on Lime Island,
Michigan. Annual Report to the Michigan Department of Natural Resources.

12 pp.

Morris, R. D., and R. A. Hunter. 1976. Factors influencing desertion of colony sites by
common terns (Sterna hirundo). Can. Field Nat. 90: 137-143.

Mom's, R. D., H. Blokpoel, and G. D. Tessier. 1992. Management efforts for the

conservation of common tern Sterna hirundo colonies in the Great Lakes: Two
case histories. Biol. Conserv. 60: 7-14.

75

Richards, M. H., and R. D. Morris. 1984. An experimental study of nest site selection in
common terns. J. Field Omithol. 55: 457-466.

Office of the Great Lakes: Michigan Department of Environmental Quality. 2000. Great
Lakes Trends: Into the New Millenium. 41 pp.

Scharf, W. C. 1991. Common tern. Pages 221-222 in R. Brewer, G. A. McPeek, and
R. J. Adams, eds. The atlas of breeding birds in Michigan. Mich. State Univ.

Press, East Lansing.

Tovell, W. M. 1979. The Great Lakes: geology at work. Joum. of Soil and Water
Conservation. March-April: 65-67.

76

 

 
 
  
 

CD F Outer Dike

Wall

  
 
 

Center Dike

  

Interior Cell

]75m

Common
Tern
Nesting
Area

 

 

 

 

 

Appendix. Location of the Confined Disposal Facility (CDF) in Saginaw Bay, Michigan.
The CDF was used as a common tern nesting site during summers 1996-1997.

CHAPTER 3
ANALYZING POPULATION VIABILITY OF COMMON TERNS IN THE ST.
MARY’S RIVER, MICHIGAN USING VORTEX

ABSTRACT

Common terns (Sterna hirundo) in Michigan were listed as threatened in 1978
and have undergone an assessment in the Great Lakes for listing as federally endangered.
Although researchers have evaluated the minimum population parameters necessary to
maintain stable populations using deterministic models, no studies have assessed the
viability of common tern populations using models that simulate stochastic events
affecting population dynamics. Therefore, I developed a population viability analysis
(PVA) for common terns in the St. Mary’s River, Michigan using VORTEX, a stochastic
simulation of the extinction process. Data for the model were assembled from survival
information gathered in the St. Mary’s River (1997 - 2001) and from existing literature.
Analyses indicate that the population is declining at a rate ranging from 6.3% (11%
flooding probability each year) to 10% (52% flooding probability each year) per year
with a probability of extinction over 100 years ranging from 70.4% (11% flooding
probability) to 99.9% (52% flooding probability). Sensitivity analyses suggest that the
most significant factors affecting the population’s long-term persistence are juvenile
(egg-laying to fledging) and sub-adult (fledging to age 3) survival. Therefore,
management strategies should be directed at increasing juvenile and sub-adult survivals.
This PVA is limited by the lack of species—specific data regarding inbreeding depression,
carrying capacity, and density dependence. Thus, PVA is an important tool for managers

but has limitations and should be used as only one step in the management process.

78

INTRODUCTION

Declining numbers of common terns have caused concern over the future of this
species in the Great Lakes region. In Michigan, estimates for common terns have
declined from 6,000 breeding pairs in the 19605 (Ludwig 1962) to between 1,200 and
1,400 breeding pairs in the 19905 (Evers 1992, Cuthbert et a1. 1997). As a result of this
decline, in 1978 the common tern was listed as a state threatened species in Michigan and
has undergone a status assessment in the Great Lakes for possible listing as federally
endangered (Scharf 1991 ). Several factors, including loss of habitat due to vegetation
encroachment (Scharf 1981), shortages of breeding areas due to high water levels (Scharf
1991), competition with gulls for nesting space (Morris and Hunter 1976, Scharf 1981,
Scharf and Shugart 1985, Ludwig 1991), predation (Hatch 1970, Morris and Hunter
1976, Scharf 1981, Ludwig 1991, Bumess and Morris 1992), and effects of contaminants
(Weseloh et al. 1989), have contributed to the decline of this species.

The decline of common terns throughout the Great Lakes and in the eastern
United States has incited researchers to evaluate the minimum population parameters
necessary to maintain stable populations of common terns (Austin 1942, Austin and
Austin 1956, Nisbet 1978, DiConstanzo 1980, Penning 1993, Millenbah 1997).
However, no research has evaluated the viability of common tern populations using a
model that simulates both the effects of deterministic forces and stochastic or random
events. Deterministic models (life-table analyses) generate long-term projections of
population grth or decline but do not reveal fluctuations in population size caused by

stochastic processes (Clark et al. 1991, Lacy and Clark 1993, Lindenmayer et al. 1993,

79

Kendall 1998, Forys and Humphrey 1999). A model that incorporates stochasticity is
important because a number of interacting demographic, environmental, genetic, and
catastrophic processes determine the vulnerability of small populations to extinction
(Lacy 1993, Lacy and Clark 1993).

The St. Mary’s River, located along the eastern Upper Peninsula of Michigan,
currently supports the largest number of common terns in the state (> 50% of breeding
pairs; Cuthbert et al. 1997), making it one of Michigan's important nesting areas. Since
this area is one of Michigan’s strongholds for common terns, it is necessary to understand
the current population status, impacts to survival, and long-term survival potential of the
common tern colonies in the St. Mary’s River. Previous research has shown that
colonies in the St. Mary’s River are vulnerable to stochastic processes (i.e., predation and
flooding; Millenbah and Lamp 1999). Therefore, to adequately evaluate the potential for
long-term persistence of the common tern colonies in the St. Mary’s River, a population
viability analysis (PVA) that models stochastic processes was needed.

PVA is a process for modeling and analyzing the vulnerability of small
populations to extinction and comparing alternative management options (Shaffer 1981,
Clark et al. 1991, Boyce 1992, Lacy and Clark 1993, Brook and Kikkawa 1998).
Extinction probabilities are estimated using computer simulation models that incorporate
interacting demographic, environmental, catastrophic, and genetic processes, all of which
occur in real populations and determine the vulnerability of a population to extinction
(Shaffer 1981, Gilpin and Soule 1986, Boyce 1992, Lacy and Clark 1993). These four
extinction processes can be simulated in PVA models, and the effects of deterministic

and stochastic forces can be investigated (Clark et al. 1991, Lacy 1993). PVA is used to

80

rank the outcomes of different management options, such as reducing mortality or
increasing carrying capacity, and is thus an important planning tool for managing
threatened and endangered species (Clark et a1. 1990, Brook et al. 1997b).

VORTEX (Miller and Lacy 1999), a computer simulation model for PVA, is one
model that incorporates the effects of deterministic forces as well as stochastic events on
the dynamics of wildlife populations. The purpose of this project was to use VORTEX to
predict the long-term survival potential of common terns in the St. Mary’s River and

determine which factors have the most significant effects on long-term survival.

81

1)

2)

3)

OBJECTIVES

predict the long-term survival potential of common terns in the St. Mary’s
River using PVA;

determine which factors most affect the long-term survival potential of
common terns in the St. Mary’s River; and

make recommendations for managing common terns in the St. Mary’s River.

82

METHODS

Modeling Method

VORTEX 8.21 (Miller and Lacy 1999) was used to simulate the deterministic and
stochastic factors affecting common tern population dynamics. VORTEX, a Monte Carlo
simulation written in C programming language, models population dynamics as discrete,
sequential events that occur according to random variables based on user-defined
distributions. VORTEX simulates a population by following each individual through a
series of events that describe the annual cycle of a sexually reproducing diploid organism
(Miller and Lacy 1999). I chose VORTEX because it has been widely used for PVA
analysis (Lacy 1993, Lacy and Clark 1993) and because it incorporates a wide range of
stochastic factors such as environmental variability and catastrophes (Brook and Kikkawa
1998). VORTEX has also been shown to model avian population dynamics realistically

(Brook et al. 1997a).

Demographic Input Parameters

Parameter values (see Appendix 1 for a list of VORTEX input values) used in the
simulation model were taken either from five years (1997-2001) of breeding season field
data collected in the St. Mary's River (Cook 1999, Millenbah and Lamp 1999, Chapter 1
of this document) or from data in published literature regarding common tern population
dynamics (Austin 1942, Austin and Austin 1956, Nisbet 1978, DiConstanzo 1980,
Penning 1993, Millenbah 1997). Since I did not know the specific age distribution of the

St. Mary's River common tern population, I estimated the number of individuals of each

83

age (Appendix 2) by multiplying the initial population size (1997) by the percentage of
individuals of each age in a common tern population on the east coast (Austin and Austin
1956). Under the assumption of a 1:1 sex ratio, the number of individuals of each age
was divided evenly between males and females.

Based on previous research, I assumed that males and females initially breed at 3
years of age (Austin and Austin 1956, Nisbet 1978, DiConstanzo 1980). No birds were
projected to live and breed past 21 years of age. Although birds > 10 years of age
comprise only a small portion of the total breeding population in the eastern U. S. (Austin
and Austin 1956), those age classes were included in the simulation because the age
distribution was based on a population comprised of ages 1-21 (Austin and Austin 1956).

Maximum clutch size (6 eggs) and the percent of clutches with 1, 2, 3, 4, 5, and 6
eggs were based on data collected in the St. Mary's River from 1997-2001. The
percentages of clutches with 1, 2, 3, 4, 5, and 6 eggs were calculated for each of the five
years of field data and averaged.

Juvenile mortality (0.796) was calculated as 1 minus the average of overall
breeding season survival (egg survival*chick survival) for each year. Overall juvenile
survival was 0.21, 0.13, 0.70, and 0.51 during 1997 (Cook 1999), 1998 (Cook 1999),
2000 (this study), and 2001(this study), respectively. Overall survival during 1999 is
unknown because common terns abandoned known nesting sites (after 266 nests had
been laid) due to flooding and predation (Millenbah and Lamp 1999). Since researchers
were unable to follow common terns to re-nesting locations, I could not determine
survival rates during 1999. The environmental variation (EV) of juvenile mortality was

calculated as the standard deviation of the mean of 1997, 1998, 2000, and 2001 overall

84

survivals. EV is defined as the fluctuation in the probabilities of births and death that
results from fluctuations in the environment (Miller and Lacy 1999).

It has been estimated that 14.3% of common tern fledglings survive to breeding _
age (DiConstanzo 1980, Penning 1993). Assuming that survival is constant during the
three years until breeding, an annual survival rate of 0.523 is calculated as the cubed root
of 0.143. Subtracting 0.523 from 1 gives a mortality rate of 0.477 for age classes 0-1
(fledging to age 1), 1-2, and 2-3 (sub-adults). The EV of mortality (1.38) of each of these
age-classes is calculated as the cube root of the standard deviation of survival (2.6;
DiConstanzo 1980, Penning 1993).

It has also been estimated that 92% of common tern adults survive annually
(DiConstanzo 1980, Penning 1993), giving an adult mortality rate of 8%. The standard
deviation of adult survival (1.4; DiConstanzo 1980, Penning 1993) was used as the EV of

adult mortality.

Vortex Simulations

Population simulations were projected for 100 years. Each simulation was
replicated 1,000 times to minimize standard errors around mean results (Lacy 1993).
Reported probabilities of extinction are the average of 1,000, 100-year simulations. The
initial population size for all simulations was 1,298 individuals, the population size of

common terns in the St. Mary’s River in 1997 (Cook 1999).

Initial Population Projections

To determine whether or not density dependent breeding, inbreeding, and

85

flooding were affecting long-term survival and probability of extinction, Iran a series of
simulations in which each factor was modeled with the demographic parameters while
the other two factors were omitted (Table 1). For example, I ran a simulation with the
demographic parameters and only inbreeding. Density dependent breeding and flooding

were not incorporated in this simulation.

Density Dependent Breeding
VORTEX models density dependent breeding with an equation that specifies the
proportion of adult females breeding as a function of the total population size (Miller and
Lacy 1999). Since there was no information regarding density dependent breeding
specific to common terns in the literature or from the field data, common tern population
dynamics were modeled with two different sets of VORTEX default values (Table 1):
(1) The percent of adult females breeding near N = 0 (N = population size) and at
K (carrying capacity) was 100% and 50%, respectively. The exponential
steepness (B) was 2, indicating that the percent of females breeding is a
quadratic fimction of N. The Allee effect (A), the decrease in the proportion
of females breeding at low densities, was modeled with a magnitude of 1.
(2) The percent of adult females breeding near N = 0 and at K was 100% and
80%, respectively. The exponential steepness (B) was 16, indicating that the
percent of females breeding does not begin to decrease dramatically until N is
large. The Allee effect (A) was not incorporated in this scenario, which
implies that there is no decrease in the proportion of females breeding at low

population densities.

86

Table 1. Initial VORTEX population simulations for common terns in the St. Mary’s
River. K = carrying capacity. Severity (S) = severity with respect to survival.

Scenario

Values

 

Demographics only

See Appendix 1

 

Demographics + density dependent
breeding (1)A

Demographics + density dependent

breeding (2)B

Max % breeding near 0 = 100%
% breeding at K = 50%
exponential steepness (B) = 2
Allee parameter (A) = 1

Max % breeding near 0 = 100%
% breeding at K = 80%
exponential steepness (B) = 16

Allee parameter (A) = 0

 

Demographics + inbreedingA

3.14 lethal alleles

50% of genetic load due to lethal alleles

 

Demographics + flood (1)
Demographics + flood (2)
Demographics + flood (3)

Demographics + flood (4)

Probability = 52%, Severity (S) = 95%

Probability = l 1%, Severity (S) = 95%

Probability = 52%, Severity (S) = 99%

Probability = 11%, Severity (S) = 99%

 

Adefault values

Bvalues from a density dependent breeding curve chosen within VORTEX

87

Inbreeding Depression

Since there was no information regarding inbreeding depression specific to
common terns in the literature or from the field data, common tern population dynamics
were modeled with the VORTEX default values (Table 1). The default value of 3.14
lethal alleles per diploid genome is a median value based on a study of 40 captive
mammalian populations (Ralls et al. 1988). Studies of Drosophila flies and scant data for
other species indicate that 50% of the inbreeding effects is due to lethal alleles (Miller

and Lacy 1999).

Catastrophes

Since the common tern population in the St. Mary's River experienced severe egg
mortality (83%) from flooding during 1999 (Millenbah and Lamp 1999), this event was
modeled as a catastrophe. On June 1, 1999, the day of the flooding event, the amount of
precipitation recorded at the weather station closest (12 km) to Lime Island (DeTour
Village, COOP ID 202094) was 2.44 cm (National Climatic Data Center, National
Oceanic and Atmospheric Administration 2001). It is more likely that that amount of
rainfall would be catastrophic if it fell in a short period of time (1-2 hours), but there was
not enough data available based on hourly precipitation levels to conduct an analysis at
that level of detail. Using 2.44 cm as a minimum amount of rainfall causing a flooding
event, I calculated the number of breeding seasons during which the daily precipitation
exceeded 2.44 cm at least once. This occurred during 28 of 54 years of records (1948-
2001), giving a flooding probability within a breeding season of 52%. Although a 52%

flooding probability is based on the amount of rainfall that caused the catastrophic

88

flooding event, this percentage may be too high. There may have been other factors such
as wind and impermeability of the soil that contributed to the catastrophe. Therefore to
examine the effects of a range of flooding probabilities, I calculated the number of
breeding seasons during which the daily precipitation exceeded 5.08 cm at least once.
This occurred during 6 of 54 years, giving a flooding probability of 11%. Since 83% of
the eggs laid in 1999 were lost (Millenbah and Lamp 1999), the severity of flooding with
respect to reproduction was 17% (read by VORTEX as an 83% reduction in
reproduction). The exact ntunber of adults experiencing mortality directly from the
flooding is unknown, so I assumed a low reduction (either 1% or 5%, depending on the
simulation) in survival resulting from flooding. Since they can fly, adults can easily
leave sites affected by flooding, and only a small percentage would experience direct
mortality due to flooding.

I ran a series of four flooding simulations to evaluate the effects of a 52%
probability of catastrophe versus an 11% probability of catastrophe and to determine
whether or not the long-term survival was different between a 1% reduction (severity =

99%) in survival and a 5% reduction (severity = 95%) in survival (Table 1).

Assumptions
Some assumptions were necessary to assign values to or address a number of
input parameters:
(1) Since chicks were not sexed after hatching, I assumed a 1:1 sex ratio at birth.
(2) 91.1% of adult males and females in the population were breeders. This was

calculated by determining the percentage of the initial population size

89

comprised of individuals from ages 3-10, the reproductive cohort of the
population (Austin and Austin 1956).

(3) No correlation of environmental variation between reproduction and survival
was modeled because the reproduction rate (juvenile survival) is more
sensitive to weather and other outside disturbances than adult survival
probabilities, which are more stable (Li and Li 1998).

(4) No emigration from or immigration into the population is occurring.
Although it is likely immigration and emigration are occurring in this
population (J. Spendelow, pers. comm), there was not enough information for
other nearby common tern populations to create a metapopulation model.

(5) The carrying capacity of the St. Mary's River is unknown. Lime Island was
the only nesting site used in the St. Mary’s River during 1997 and 1998, and
nearly 1500 individuals nested on half the available nesting habitat on that
site. Therefore, I multiplied 1500 by 2 to obtain a carrying capacity of 3000

for common terns in the St. Mary’s River.

Sensitivity Analyses

Sensitivity analysis is used to determine the effects of inaccuracies of parameter

estimates on model predictions as well as to target management efforts that would have

the greatest impact on increasing population viability (Reed et al. 1998). Sensitivity

analyses were performed on four parameters: juvenile mortality, sub-adult mortality,

frequency of flooding, and carrying capacity. Juvenile and sub-adult mortalities were

chosen for sensitivity analysis because these mortalities are high in comparison to adult

90

mortality and should therefore have a greater effect on population size. Additionally,
these parameters were chosen because they were shown to be the most important factors
affecting the long-term persistence of another common tern population in Saginaw Bay,
Michigan (Millenbah 1997). Since field data and observations from 1999 have shown
that catastrophic flooding has a considerable impact on common tern reproduction,
sensitivity analyses were conducted on flooding. Carrying capacity was chosen for
sensitivity analysis because it is one of the most uncertain yet critical components in the
simulation.

To measure the sensitivity of the model to the aforementioned parameters, I used

the following sensitivity index (Jorgensen 1986):

s. = (dX/X) / (dP/P) (1)

where P is the parameter value under normal conditions and X is the dependent variable
of interest (population size in this model). dX is the change in the dependent variable
caused by the change in parameter value (dP). A larger absolute value of S)( indicates a
higher sensitivity of X to a change in P.

To determine the sensitivity of extinction probability in response to changes in
juvenile mortality (baseline = 61%), sub-adult mortality (baseline = 47.7%), and
frequency of flooding (baseline = 52% or 11%), I increased and decreased the baseline
values of the three parameters by 10% and evaluated the model responses. Carrying
capacity (baseline = 3,000) was increased and decreased by 1,000 to evaluate the model

response to change in carrying capacity. The extinction probability under the increased

91

and decreased values for each of the four parameters was determined. Sx values were

calculated using equation 1.

Simulation Experiments
Single-Parameter Manipulations

Since juvenile and sub-adult mortalities were determined to be the most sensitive
parameters in the VORTEX analyses (based on sensitivity analyses), these two
parameters were manipulated singly to determine how probability of extinction is
affected by varying mortality rates. Two sets of simulations were run: one with juvenile
mortality rate manipulated by increments of 10% above and below the current rate of
61% and one with sub-adult mortality rate manipulated by increments of 10% above and
below the current rate of 47.7%. Including simulations with 0% and 100% mortality,

there were a total of 12 simulations for each parameter (juvenile and sub-adult mortality).

T wo-parameter manipulations

To determine how extinction probability is affected by manipulating the two most
sensitive parameters together, 49 different scenarios involving the simultaneous
manipulation of juvenile and sub-adult mortality were run (Table 2). Results from the 49
simulations were plotted as surface graphs in Microsoft Excel to interpolate extinction

probabilities between selected rate combinations.

92

Table 2. Juvenile and sub-adult mortality rate combinations used in the 49 two-parameter
VORTEX simulations for common terns in the St. Mary’s River.

 

 

 

Sub-adult Mortality Rate
7.7% 17.7% 27.7% 47.7% 57.7% 67.7% 87.7%
11% 1* 2 3 4 5 6 7
31% 8 9 10 11 12 13 14
41% 15 16 17 18 19 20 21
Juvenile 51% 22 23 24 25 26 27 28
Mortality
Rate 61% 29 30 31 32 33 34 35
81% 36 37 38 39 40 41 42
91% 43 44 45 46 47 48 49

 

 

 

 

 

* scenario number for rate combinations

93

RESULTS
Initial Population Projections

Long-term population projections with density dependent breeding (1), density
dependent breeding (2), and inbreeding depression were similar to the projection with
only demographic parameters modeled (Fig. 1). These four projections had similar
stochastic growth rates (r)3, ranging from —0.050 to — 0.064 (Table 3), as well as similar
extinction probabilities (0.173 - 0.348; Table 3) and mean times to extinction (93.6 - 94.7
years; Table 3).

Projections with a 52% chance of flooding within a breeding season [Flood (2)
and Flood (4)] had a greater negative impact on long-term survival than projections with
an 11% chance of flooding within a breeding season [Flood (1) and Flood (3)].
Projections with a 52% probability of flooding had lower growth rates, higher extinction
probabilities, and shorter mean times to extinction than projections with an 11%
probability of flooding (Table 3).

Projections with the same probability of flooding but different severities with
respect to survival [Flood (1) vs. Flood (3) and Flood (2) vs. Flood 4)] were not different.
Growth rates, extinction probabilities, and mean times to extinction were similar between
projections with 52% flooding probability and either 95% or 99% severity with respect to
survival as well as between projections with 11% flooding probability and either 95% or
99% severity with respect to survival (Table 3).

I chose to model common tern population dynamics with only demographics

parameters and Flood (3) or Flood (4) for all subsequent simulations. Density dependent

 

3 When r < 0, the population is declining; when r = 0, the population is stable; and when r > 0, the
population is increasing.

94

 

 

 

 

 

1600
1500
1400 .
1300 --—Demograplncs
fig +DensityDependence(l)
@ 1000 +DensityDependence(2)
g 700 +Flood(1)
a 600 +Flood(2)
:5 :28 +Flood(3)
300 -—Flood(4)
200
100
0
:- O B IO M v- 0, N ID 10 v- Q h
.- N n 1- v n e rs co no a»
Year

Fig. 1. Initial population projections for common terns in the St. Mary’s River. Density
dependence (1): max. % breeding at 0 = 100%, % breeding at K = 50%, exponential
steepness = 2, Allee parameter = 1. Density dependence (2): max. % breeding at 0 =
100%, % breeding at K = 80%, exponential steepness = 16, Allee parameter = 0.
Inbreeding: 3.14 lethal alleles, 50% of genetic load due to lethal alleles. Flood (1):
probability= 52%, severity with respect to survival= 95%. Flood (2): probability— —
11%, severity with respect to survival= 95%. Flood (3): probability= 52%, severity
with respect to survival= 99%. Flood (4): probability— — 11%, severity with respect to
surviva1=99°/o.1mges in th'g thgis g; pmntgg In cglor.

95

Table 3. Growth rate (r) (SD), probability of extinction (SD), and mean years to
extinction (SD) for initial VORTEX population simulations for common terns in the St.
Mary’s River. Results are based on 1,000 replicates for each 100-year simulation.

 

 

 

 

 

Scenario r Probability of Mean Years to
Extinction Extinction
Demographics only -0.064 (0.092) 0.289 (0.44) 93.6 (4.95)
Demographics + density -0.051 (0.087) 0.196 (0.41) 93.6 (5.20)
dependent breeding (1)
Demographics + density -0.050 (0.084) 0.173 (0.38) 94.7 (4.87)
dependent breeding (2)
Demographics + inbreeding -0.056 (0.094) 0.348 (0.47) 94.1 (4.83)
Demographics + flood (1) -0.118 (0.168) 1.000 (0.00) 51.5 (7.52)
Demographics + flood (2) -0.068 (0.132) 0.842 (0.38) 85.0 (8.76)
Demographics + flood (3) -0.100 (0.151) 0.999 (0.03) 60.5 (8.08)
Demographics + flood (4) -0.063 (0.122) 0.704 (0.46) 88.1 (7.48)

 

96

breeding and inbreeding were not included because the simulations with these factors
were similar to the simulation with only demographic parameters included. Flooding was
modeled because this is an event known to affect the St. Mary’s River population
(Millenbah and Lamp 1999). I did not run any further simulations using a 95% severity
with respect to survival because it is likely that the effect on adult survival was very low
and the severity is probably closer to 99%. Since they can fly, adults can easily leave
sites affected by flooding, and only a small percentage would experience mortality due to
flooding.

The projection with a 52% probability of flooding and a 99% severity with
respect to survival predicts a 99.9% probability of extinction over 100 years
[demographics and Flood (3); Table 3]. Under this scenario, mean time to extinction is
60.5 years (SD = 8.08), with the population declining at a rate of 10% each year (Table
3). With a probability of flooding of 11% and a 99% severity with respect to survival,
VORTEX predicts a 70.4% probability of extinction within 100 years (Table 3). Mean
time to extinction is 88.1 years (SD = 7.48), with the population declining at a rate of

6.3% each year (Table 3).

Sensitivity Analyses

Common tern population size is particularly sensitive to increases in juvenile and
sub-adult mortalities and flooding when the probability is 52% (Table 4). Since the
absolute values of sensitivity indices for juvenile and sub-adult mortalities were the
highest, I focused on juvenile and sub-adult mortalities as the two most sensitive

parameters affecting long-term common tern population dynamics.

97

Table 4. Sensitivity indices for increases and decreases in baseline values for juvenile
mortality, sub-adult mortality, carrying capacity, and flooding rates in VORTEX
simulations for common terns in the St. Mary’s River. Separate sets of simulations were
run with probability of flooding = 11% and probability of flooding = 52%.

 

 

 

 

 

 

 

 

Value . Sensitivity Index
Parameter 1:100de
(baseline value) Increase Decrease Probability Increase Decrease
0 _ _
Juvenile mortality 0 o 11 /° 1'72 3'29
61°/ 71 /o 51 /o
( °) 52% -097 -1.20
Sub-adult mortality 0 , 1 1% '1 '43 "2°76
47 70/ 57.7% 37.7/o
( : °) 52% -1.12 .2.84
C 'n ca acit 11% 0.01 0.004
W3 %OOP y 4,000 2,000
( ’ ) 52% -001 0.03
Flooding 21% 1% -0.15 -0.18
(11%)
Flooding 62% 42% -0.61 -0.57
(52%)

 

98

Simulation Experiments
Single—Parameter Manipulations

11% Probability of Flooding

At juvenile and sub-adult mortality rates of 2 71% and 2 57.7%, respectively, the
long-term probability of extinction is 1 (Fig. 2a). If juvenile mortality is decreased to
41%, with all other initial input values remaining the same, probability of extinction is 0
(Fig. 2a). However, r = —0.018. The population does not stabilize over 100 years until
juvenile mortality is decreased to at least 30% (r = 0.001). At sub-adult mortality rates
S 37.7%, probability of extinction is 0 (Fig. 2a). The population declines at a sub-adult
mortality rate of 37.7% (r = -0.006), but at rates S 27.7% r is > 0.

52% ProbabilityI of Flooding

 

At juvenile and sub-adult mortality rates of 2 71% and 2 57.7%, respectively, the
long-term probability of extinction is 1 (Fig. 2b). Probability of extinction does not reach
0 until juvenile mortality is decreased to 0% (Fig. 2b). Even with a 0% juvenile mortality
rate the population declines over 100 years (r = -0.008). At sub-adult mortality rates S
27.7%, probability of extinction is 0 (Fig. 2b). The population declines at 27.7% sub-
adult mortality (r = -0.005), but at sub-adult mortality rates S 17.7% the population is

growing over 100 years (r > 0).

T wo-Parameter Manipulations

11% Probability of Floodigg

When juvenile and sub-adult mortality rates are approximately S 61% and

99

 

p—s
N

 

1

 

 

 

\\
\

P (Extinction)

CC

N
\
\

 

 

 

 

 

Mortality Rate (%)
+ Juvenile Mortality
—I——- Sub-adult Mortality

 

 

 

b.)

 

—
N

 

 

 

P (Extinction)
O
O‘

 

GO

N-b
\
\

 

 

 

 

0 20 40 60 80 100
Mortality Rate (%)

Fig. 2. Probabilities of extinction with increasing juvenile and sub-adult mortality rates
for common terns in the St. Mary’s River, M] for simulations with an 11% probability of
flooding (a) and a 52% probability of flooding (b). The only input value changed in each
simulation was either juvenile or sub-adult mortality rate.

100

S 37.7%, respectively, probability of extinction is 0 (Fig. 3a). At juvenile mortality rates
2 51% and sub-adult mortality rates 2 67.7%, extinction probability is 1 (Fig. 3a).
Between areas where probability of extinction is 0 and 1, extinction probability rises

steeply for small changes in mortality rates (Fig. 3a).

52% Probability of Flooding

 

Probability of extinction is 0 when juvenile and sub-adult mortality rates are
approximately S 61% and S 35%, respectively (Fig. 3b). Combinations of juvenile
mortality rates 2 41% and sub-adult mortality rates 2 57.7% result in a probability of
extinction of 1 (Fig. 3b). Between areas where probability of extinction equals 0 and 1,

the probability of extinction rises steeply for small changes in mortality rates (Fig. 3b).

Stable Population

11% Probability of Flooding

With a juvenile mortality rate of 57% and a sub-adult mortality rate of 37.7%, the
population grows during the first 15 years, decreases slightly, begins to increase in year
20, and stabilizes during the last 25 years (r = 0.004; Fig. 4a). In this scenario, the final
population size in year 100 is 1,981 individuals. Modeled with an 11% probability of
flooding, the population is either stable or growing over 100 years when mortality rates
are S 57% and S 37.7% for juveniles and sub—adults, respectively.

52% Probability of Flooding

With a juvenile mortality rate of 40% and a sub-adult mortality rate of 35%, the

population grows during the first 15 years, then decreases slightly and begins to stabilize

101

a.)

b.)

 

Sub-adult Mortality

Fig. 3. Targeting 0% extinction and a stable population of common terns in the St.
Mary’s River. Varying combinations of juvenile and sub-adult mortality rates and
resulting probabilities of extinction are shown for simulations with an 11% probability of
flooding (a) and a 52% probability of flooding (b).

102

 

 

 

 

 

2500 -~-~-- —-—--

r = 0.004

2000

1500

 

 

1000 - - g- - - -

Population Size (N)

500 , - ~ ~ -

 

 

Year

b)

 

2500 --7
2000

1500

 

 

1000 ,, ~ ~~ ~ ~ ~ ~ -

Population Size (N)

500

 

 

O fia—‘rlrrl ill-IillTl“-‘IT1T1T1111 .4 7- T
0 10 20 30 40 50 60 70 80 90 100

Year

Fig. 4. Stable common tern population in the St. Mary’s River for simulations with an
1 1% probability of flooding (a) and a 52% probability of flooding (b). Juvenile mortality
rate = 57% and sub-adult mortality rate = 37.7% (a). Juvenile mortality rate = 40%, sub-

adult mortality rate = 35% (b).

103

near the end of the 100-year simulation (r = 0.006; Fig. 4b). The final population size in
year 100 is 1,783 individuals. Under a scenario with a 52% probability of flooding, the
population is either stable or growing over 100 years when mortality rates are S40% and

S 35% for juveniles and sub-adults, respectively.

104

DISCUSSION AND MANAGEMENT IMPLICATIONS

Under existing environmental conditions, the long-term survival potential of
common terns in the St. Mary's River appears to be in jeopardy. According to the results
of the VORTEX simulations, this population is decreasing at a rate between 6.3% (11%
probability of flooding) and 10% (52% probability of flooding) per year. The upper limit
of this annual rate of decrease in the St. Mary’s River is slightly less than the annual rate
of decrease for a different common tern population in Saginaw Bay, Michigan (17%
decrease per year; Millenbah 1997).

Sensitivity analyses indicated that the factors with the most significant impacts on
population size and persistence in the St. Mary’s River are juvenile mortality and sub-
adult mortality. To obtain a stable population with a flooding probability of 11%,
juvenile mortality needs to be decreased to 30% or sub-adult mortality needs to be
decreased to 27.7%. When flooding probability is 52%, a stable population cannot be
achieved even when juvenile mortality is reduced to 0%. However, when sub-adult
mortality is decreased to at least 17.7%, the population stabilizes over 100 years. It
would be difficult to reduce juvenile mortality from the current rate of 61% down to 30%
(11% flooding probability). If flooding probability is closer to 52% in the St. Mary’s
River, a stable population cannot be achieved even if managers could reduce juvenile
mortality to 0%. Sub-adult mortality should be decreased from the current rate of 47.7%
down to between 27.7% and 17.7% (depending on flooding probability), but it would also

be difficult for managers to reduce mortality to these levels.

105

The difficulty in reducing sub-adult mortality rates lies in the delayed breeding of
this cohort. Little can be done in Michigan to impact the survival of sub-adults (ages 1-
3), as these birds do not generally return to their breeding grounds until they reach
breeding age (2 3 years old; Austin and Austin 1956, Nisbet 1978, DiConstanzo 1980).
Therefore, through the cooperation of local, national, and international agencies,
management must be directed at increasing the survival of sub-adults along migration
routes and on over-wintering grounds (Haymes and Blokpoel 197 8, Blokpoel et al. 1987).

Since it unlikely managers can reduce either juvenile or sub-adult mortalities
singly to levels that ensure long-term survival, both types of mortality should be
decreased simultaneously. To obtain a stable population under a scenario with 11%
flooding probability, juvenile mortality must be decreased from the current rate of 61% to
at least 57.7%; sub-adult mortality must be decreased from the current rate of 47.7% to at
least 37.7%. With a 52% flooding probability affecting survival and reproduction,
juvenile mortality must be decreased to 40% and sub-adult mortality must be decreased
to 35%. Thus, researchers must focus on increasing the survival of the population’s
juveniles and sub-adults to increase the number of breeding birds.

To decrease juvenile mortality, the overall survival of eggs and chicks must
increase. Excess vegetation cover was shown to negatively impact egg and chick
survival on Lime Island in 1997 and 1998 (Cook 1999, Cook-Haley and Millenbah
2002). Vegetation encroachment leads to a decrease in visual contact (used to signal the
presence of intruders) and makes the movement of young difficult because they become
entangled in vegetation. These two factors make young more susceptible to predation

(Hatch 1970). After decreasing vegetation cover (through herbiciding) in an

106

experimental area on Lime Island in 1998, it was shown that common tern egg survival
was greatest in areas with 40-50% total standing vegetation cover (Cook-Haley and
Millenbah 2002). Therefore, it may be possible to reduce overall egg and chick mortality
from 61% to between 40% and 57.7% (depending on flooding probability) by decreasing
vegetation cover on nesting sites. Herbicide or the addition of substrate such as rocks,
sand, or gravel can be used to reduce vegetation cover (Richards and Morris 1984, Morris
et al. 1992, Cook 1999). It is important to note that the amount of vegetation
manipulations required may depend on the type of nesting site (human-made site vs.
naturally occurring island) used, which can be dependent on Great Lakes water levels
(Chapter 1 and 2 of this document).

' Although predation was not modeled explicitly in VORTEX (deaths due to
predation were accounted for in the juvenile mortality rate), predation may potentially
have a significant effect on common tern population dynamics. Predation by a single
long-tailed weasel in 1997, 1998, and 1999 had a negative impact on common tern
survival and reproductive success (Cook 1999, Millenbah and Lamp 1999). If effective
predator management is employed it may be possible to reduce or remove this effect on
the common tern population in years in which predation is a problem. If common terns
return to nest on Lime Island in future years, a proactive approach to predator
management should be taken. Conibear trapping was shown to be effective for
mammalian predator control of a population of California least terns (Butchko and Small
1992). Therefore, conibear traps should be used on Lime Island as soon as possible after
common terns arrive on the nesting site in early May. Although no predation was

observed in the common tern colonies on Andrews Reef and Harbor Island Reef in 2000

107

or 2001, managers should be prepared to employ predator management (if necessary) on
nesting sites used in the future.

To increase the power and predictability of the VORTEX simulation model, more
field data is required on the St. Mary’s River common tern population. One of the
limitations of this model is the limited amount of field data available for this particular
population. More than four years of data are needed to show long-term fluctuations in
demographics and catastrophes. Field research should continue on a yearly basis for at
least 10-20 years to show any long-term trends in population dynamics. A PVA
conducted with 5 years of field data on Capricorn silvereyes (Zosterops lateralis
chlorocephala) produced different predictions of extinction risk compared with a PVA
based on 15-25 years of data (Brook and Kikkawa 1998). Thus, more comprehensive
biological data will provide more accurate parameter estimates and thus greater
confidence in VORTEX predictions (Lindenmayer et al. 1993).

Future field research should also be directed at determining inbreeding
depression, density dependence, carrying capacity, and immigration and emigration rates.
Inbreeding depression could be estimated through genetic analysis conducted on the
population. Although population size in the St. Mary’s River is currently several times
greater than the threshold where inbreeding depression would likely have a considerable
effect (i.e., S 50 individuals; Franklin 1980), it would be wise to conduct genetic analyses
on the population. Carrying capacity could be better estimated by conducting research on
resource (i.e., food and habitat) availability and how common terns used these resources.

To better estimate the age distribution, sub-adult survival rate, and immigration

and emigration rates of the St. Mary’s River population, adults should be trapped for

108

band recoveries. By determining the year in which a bird was originally banded, a
researcher will know the age of a bird and can more reliably estimate age distribution.
Since current estimates of common tern sub-adult survival rates are uncertain, more
information is required to generate a more accurate estimate. Since sub-adults typically
return to their natal sites to breed at age 3 and researchers began banding chicks in the St.
Mary’s River in 1997, the timing is conducive for trapping adults for band recoveries. It
is likely that common terns in the St. Mary’s River have permanent emigrants from and
permanent immigrants into the population from nearby nesting sites either on the
Canadian side of the St. Mary’s River or Lake Huron. If immigration and emigration
were incorporated into the model, long-term survival potential would be higher than
predicted in this study (J. Spendelow, pers. comm.) Movement between neighboring
population centers has been shown to have a mitigating effect on extinction probability
(Brook and Kikkawa 1998). To determine immigration and emigration rates affecting the
common tern population in the St. Mary’s River, a long-term, multisite band recovery
study should be undertaken. Directing future research at the aforementioned parameters
will allow a more empirically based model to be built, thereby better substantiating a
number of assumptions and increasing the predictability of the simulation.

As previously mentioned, excess vegetation cover can have a significant negative
impact on common tern survival and reproduction (Cook 1999). Therefore, effects of
vegetation cover should be modeled with VORTEX. Effects of habitat quality on
populations cannot be directly modeled in VORTEX, but they can be modeled indirectly.
To indirectly model habitat effects in VORTEX, separate simulations with varying

mortality rates related to different percentages of vegetation cover could be run and

109

compared. A model that incorporates habitat quality will allow managers to better
predict what level of vegetation cover and vegetation management is required for the
greatest common tern reproductive success.

It is important to note that PVA is a prediction tool and not a definitive answer.
PVA provides managers with a powerful tool to determine where management should be
targeted (such as reducing juvenile and sub-adult mortality rates in this study) and to
assess the potential outcome of various management options (Clark et al. 1990,
Lindenmayer et a1. 1993, Hamilton and Moller 1995). PVA is most effective if applied
within an adaptive management framework (Lacy and Clark 1993, Lindenmayer et al.
1993). As better data and better models become available, PVA modeling for common
terns in the St. Mary’s River should be repeated and reexamined. Used in this manner,
VORTEX can be used to better understand and manage common terns for continued
existence in the St. Mary’s River. It is hoped that these models will allow managers to
determine the management scenario that will ensure the continued existence and stability
of common terns in the St. Mary’s River, Michigan. The models may also prove valuable

in protection efforts at other common terns colonies in Michigan and the U. S.

110

LITERATURE CITED

Austin, 0. L. 1942. The life span of the common tern. Bird-Banding. 11: 155-159.

Austin, 0. L., and O. L. Austin, Jr. 1956. Some demographic aspects of the Cape Cod
population of common tern (Sterna hirundo). Bird-Banding. 27: 55-66.

Blokpoel, H., G. D. Tessier, and A. Harfenist. 1987. Distribution during post-breeding
dispersal, migration, and overwintering of common terns color-marked on the
lower Great Lakes. J. Field Omithol. 58(2): 206-217.

Boyce, M. S. 1992. Population viability analysis. Annu. Rev. Ecol. Syst. 23: 481-506.

Brook, B. W., and J. Kikkawa. 1998. Examining threats faced by island birds: a
population viability analysis on the Capricorn silvereye using long-term data.
Journal of Applied Ecology. 35: 491-503.

Brook, B. W., Lim, L, Harden, R., and R. Frankham. 1997a. Does population viability
analysis predict the behaviour of real populations? A retrospective study on the
Lord Howe Island woodhen T richolimnas sylvestris (Sclater). Biological
Conservation. 82: 119-128.

Brook, B. W., Lim, L, Harden, R., and R. Frankham. 1997b. How secure is the Lord
Howe Island woodhen? A population viability analysis using VORTEX. Pacific
Conservation Biology. 3: 125-133.

Burger, J ., and M. Gochfeld. 1991. The common tern: its breeding biology and social
behavior. Columbia Univ. Press, New York. 413 pp.

Bumess, G. P., and R. D. Mon'is. 1992. Shelters decrease gull predation on chicks at a
common tern colony. J. Field Omithol. 63(2): 186-189.

Butchko, P. H., and M. A. Small. 1992. Developing a strategy of predator control for the
protection of the California least tern: A case history. Proc. 15th Vertebrate Pest
Conf. Univ. of California, Davis.

Clark, T. W., Backhouse, G. N., and R. C. Lacy. 1990. The population viability
assessment workshop: a tool for threatened species management. Endangered
Species Update. 8(2): 1-5.

Clark, T. W., Backhouse, G. N., and R. C. Lacy. 1991. Report of a workshop on

population viability assessment as a tool for threatened species management and
conservation. Australian Zoologist. 27(1 and 2): 28-35.

111

Cook, B. S. 1999. Impacts of vegetation manipulations on the reproductive success of
common terns (Sterna hirundo) on Lime Island in the St. Mary’s River, Michigan.
MS. Thesis, Mich. State Univ. 101 pp.

Cook-Haley, B. S., and K. F. Millenbah. 2002. Impacts of vegetative manipulations
on common tern nest success on Lime Island, Michigan. J. Field Omithol.
73(2): 174-179.

Cuthbert, F. J ., J. McKeaman, and L. Wemmer. 1997. US. Great Lakes Tern and
Cormorant Survey: 1997 Progress Report. 16 pp.

DiConstanzo, J. 1980. Population dynamics of a common tern colony. J. Field Omithol.
51 : 229-243.

Evers, D. C. 1992. A guide to Michigan’s endangered wildlife. Univ. Michigan Press,
Ann Arbor. 103 pp.

Forys, E. A., and S. R. Humphrey. 1999. Use of population viability analysis to evaluate
management options for the endangered lower keys marsh rabbit. Joum. of
Wildl. Manage. 63(1): 251-260.

Franklin, I. A. 1980. Evolutionary change in small populations. Pages 135-150 in O. H.
Frankel and M. E. Soule, eds. Conservation biology: an evolutionary perspective.
Sinauer Associates: Sunderland, Massachusetts.

Gilpin, M. E., and M. E. Soule. 1986. Minimum viable populations: processes of
extinction. In Conservation biology: the science of scarcity and diversity. Ed. M.
E. Soule. Sinauer Associates: Sunderland, MA. pp. 19-34.

Hamilton, S. and Moller, H. 1995. Can PVA models using computer packages offer
useful conservation advice? Sooty shearwaters Puffinus griseus in New Zealand
as a case study. Biological Conservation. 73: 107-117.

Hatch, J. J. 1970. Predation and piracy by gulls at a temery in Maine. Auk 87: 244-254.

Haymes, G. T., and H. Blokpoel. 1978. Seasonal distribution and site tenacity of the
Great Lakes common tern. Bird-Banding. 49(2): 142-151.

J orgensen, S. E. 1986. Fundamentals of ecological modelling. Elsevier, Amsterdam.

Kendall, B. E. 1998. Estimating the magnitude of environmental stochasticity in
survivorship data. Ecological Applications. 8(1): 184-193.

Lacy, R. C. 1993. VORTEX: A computer simulation model for population viability
analysis. Wildlife Research. 20: 45-65.

112

Lacy, R. C., and T. W. Clark. 1993. Simulation modeling of American marten (Martes
americana) populations: vulnerability to extinction. Great Basin Naturalist.
53(3): 282-292.

Li, X., and D. Li. 1998. Current state and the future of the crested ibis (Nippom'a
Nippon): A case study by population viability analysis. Ecological Research. 13:
323-333.

Lindenmayer, D. B., Clark, T. W., Lacy, RC, and V. C. Thomas. 1993. Population
viability analysis as a tool in wildlife conservation policy: with reference to
Australia. Environmental Management. 17(6): 745-758.

Ludwig, J. P. 1962. A survey of the gull and tem populations of Lakes Huron, Michigan,
and Superior. Jack-pine Warbler. 40(4): 104-119.

Ludwig, J. P. 1991. Ring-billed gull. Pages 216-217 in R. Brewer, G. A. McPeek, and
R. J. Adams, eds. The atlas of breeding birds of Michigan. Mich State Univ.
Press, East Lansing.

Millenbah, K. F. 1997. Assessment of common tern (Sterna hirundo) reproductive
success in Saginaw Bay of eastern Michigan. Ph.D. Dissertation, Mich. State
Univ. 111 pp.

Millenbah, K. F., and N. E. Lamp. 1999. Evaluating the reproductive success and
long-term survival potential of common terns (Sterna hirundo) on Lime Island,
Michigan. Annual Report to the Michigan Department of Natural Resources.

12 pp.

Miller, P. S., and R. C. Lacy. 1999. VORTEX: A stochastic simulation of the extinction
process. Version 8 User’s Manual. Apple Valley, MN: Conservation Breeding
Specialist Group (SSC/IUCN).

Morris, R. D., and R. A. Hunter. 1976. Factors influencing desertion of colony sites by
common terns (Sterna hirundo). Can. Field Nat. 90: 137-143.

Morris, R. D., H. Blokpoel, and G. D. Tessier. 1992. Management efforts for the
conservation of common tern Sterna hirundo colonies in the Great Lakes: Two

case histories. Biol. Conserv. 60: 7-14.

National Climatic Data Center, National Oceanic and Atmospheric Administration.
2001. ftp://ftp.ncdc.noaa.gov/pub/upload/727417199416.inv.txt.

Nisbet, I. C. T. 1978. Dependence of fledging success on egg size, parental performance
and egg composition among common and roseate terns. Ibis. 120: 207-215.

113

Penning, W. L. 1993. Common terns in western Lake Superior: history, management,
and population modeling. M. S. Thesis, Univ. Minnesota, Minneapolis. 61 pp.

Ralls, K., Ballou, J. D., and A. Templeton. 1988. Estimates of lethal equivalents and the
cost of inbreeding in mammals. Conservation Biology. 2: 185-193.

Reed, J. M., Elphick, C. S., and L. W. Oring. 1998. Life-history and viability analysis of
the endangered Hawaiian stilt. Biological Conservation. 84: 35-45.

Richards, M. H., and R. D. Morris. 1984. An experimental study of nest site selection in
common terns. J. Field Omithol. 55: 457-466.

Scharf, W. C. 1981. The significance of deteriorating man-made island habitats to
common terns and ring-billed gulls in St. Mary’s River, Michigan. Colonial
Waterbirds. 4: 155-159.

Scharf, W. C. 1991. Common tern. Pages 221-222 in R. Brewer, G. A. McPeek, and
R. J. Adams, eds. The atlas of breeding birds in Michigan. Mich State Univ.
Press, East Lansing.

Scharf, W. C., and G. W. Shugart. 1985 . Population sizes and status recommendation
for double-crested cormorants, black-crowned night herons, caspian terns,
common terns, and F orster’s terns in the Michigan Great Lakes in 1985. Report
to the Michigan Department of Natural Resources.

Shaffer, M. L. 1981. Minimum population sizes for species conservation. BioScience.
31(2): 131-134.

Weseloh, D. V., T. W. Custer, and B. M. Braune. 1989. Organochlorine contaminants in
eggs of common terns from the Canadian Great Lakes. Env. Poll. 58: 141-160.

114

Appendix 1. Summary of initial VORTEX parameter values and data sources for the
common tern population in the St. Mary's River.

Parameter

Value

Data Source

 

Time span for simulation
Replications

Population number
Inbreeding depression?

EVA (reproduction) correlated
with EV (survival)

Number of catastrophes
Mating system

Female initial breeding age

Male initial breeding age

Maximum breeding age
Sex ratio at birth

Density dependent breeding
Maximum clutch size
Percent clutch size 1
Percent clutch size 2
Percent clutch size 3
Percent clutch size 4
Percent clutch size 5

Percent clutch size 6

100 years
1,000
1
No

No

1
Monogamous

3

21
1:1
No
6
15.5%
23.94%
59.29%
1.13%
0.11%

0.03%

Author specification
Lacy 1993
Author specification
Assumption

Li and Li 1998

Field data 1997-2001
Burger and Gochfeld 1991

Austin and Austin 1956;
Nisbet 1978; Burger and
Gochfeld 1991

Austin and Austin 1956;
Nisbet 1978; Burger and
Gochfeld 1991

Austin and Austin 1956
Assumption
Assumption

Field data 1997-2001
Field data 1997-2001
Field data 1997-2001
Field data 1997-2001
Field data 1997-2001
Field data 1997-2001

Field data 1997-2001

 

115

Appendix 1. (cont’d.)

 

Parameter Value Data Source
Percent females breeding 91.1% Assumption based on
Austin and Austin 1956
Mortality age 0-1* 79.6% Field data 1997-2001;
DiConstanzo 1980; Penning

1993

EV of mortality" 0.5999% Field data 1997-2001;

DiConstanzo 1980; Penning

1993

Mortality age 1-2 47.7% DiConstanzo 1980; Penning
1993

EV of mortality 1.38% DiConstanzo 1980; Penning
1993

Mortality age 2-3 47.7% DiConstanzo 1980; Penning
1993

EV of mortality 1.38% DiConstanzo 1980; Penning
1993

Adult mortality 8% DiConstanzo 1980; Penning
1993

EV of mortality 1.4% DiConstanzo 1980; Penning

Probability of catastrophe
(flooding)

Severity to reproduction
Severity to survival

All males breeders?

Percent of males present in
breeding pool

Start at stable age
distribution?

52% or11%

0.17
0.99
No

91.1%

No

1993

Field data 1997-2001;
National Climatic Data
Center 2001

Field data 1997-2001
Assumption
Austin and Austin 1956

Assumption based on
Austin and Austin 1956

Austin and Austin 1956

 

116

Appendix 1. (cont’d.)

 

 

Parameter Values Data Source

Age distribution Appendix 2 Austin and Austin 1956
Carrying capacity (K) 3,000 Field data 1997-2001
EV of K 0 Assumption
Trend in K? No Assumption
Harvest? No Assumption
Supplement? No Assumption

 

AEV = environmental variation

*calculated as 0.61 + (0.39 * 0.477); 0.61 = juvenile mortality (egg-laying to fledging),

0.477 = sub-adult mortality (fledging to age 1).

** property of variances used to calculate value of juvenile mortality SD (0.265) and sub-

adult mortality SD (1.38) together

117

Appendix 2. Assumed age distribution of common terns in the St. Mary's River in 1997
(N = 1,298). Values are based on percentages reported by Austin and Austin (1956).

 

 

Age Number of Individuals
(1:1 sex ratio)
1 10
2 34
3 230
4 260
5 196
6 158
7 118
8 84
9 58
10 38
11 30
12 26
13 18
14 14
15 12
16 6
17 4
18 2
19 0
20 0

21 0

 

118