Illliliflljlllljlflllllllllll "

 

LIBRARY

Michigan State
University

 

 

 

This is to certify that the

thesis entitled

Monitoring and Prediction of the Fall Flight of
Wood Ducks (Aix sponsa) in Michigan: An
Investigation of Survey Methods

 

presented by

Linda A. Briggs

has been accepted towards fulfillment
of the requirements for

Master of Science degree in Fish. & Wildl.

 

Major professor

Date lei/2‘! (/47

0.7639 MS U is an Affirmative Action/Equal Opportunity Institution

 

 

 

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
l:3‘*J112010
‘ 24 1 '

 

‘lfi

 

 

 

 

 

 

 

 

 

 

 

 

 

 

5/08 KlProleocauProleIRCIDatoDuojndd

MONITORING AND PREDICTION OF THE FALL FLIGHT
OF WOOD DUCKS (fl amuse) IN MICHIGAN:
AN INVESTIGATION OF SURVEY METHODS
By

Linda A. Briggs

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

1997

ABSTRACT
MONITORING AND PREDICTION OF THE FALL FLIGHT OF WOOD DUCKS
(AIx SPONSA) 1N MICHIGAN:
AN INVESTIGATION OF SURVEY METHODS
By

Linda A. Briggs

Direct estimates of wood duck (Aix spam) numbers and production are dificult
to derive and improved methods are needed to estimate fall flight. The goal of this project
is to evaluate survey methods that might provide a finer scale of monitoring wood duck
numbers and production and that correlate with estimates derived from the traditional
analyses of harvest surveys. In 1994, night surveys for wood duck broods were
conducted on 16 to 32 km sections of three rivers in southwest Michigan. Nine road
surveys were established in southern Michigan in 1994 and expanded to 17 routes
conducted in 1995 and 1996. The road surveys were conducted in early and late time
periods. Over 700 nest boxes were checked at 11 sites in 1995 and 1996.

In 1994, only 1 brood was observed over 105 miles of stream surveyed,
eliminating this as a potential survey method. The road surveys were not correlated with
the Michigan Breeding Waterfowl Surveys (aerial surveys) conducted at corresponding
sites. The road surveys and nest box surveys were not correlated at the site level while
significant correlations were detected at the unit level (r, = 0.94, P = 0.0048). The
USFW S harvest estimates were not correlated with the road surveys and the nest box
surveys. Of the survey methods conducted in this project, the road surveys may have the

best potential and should be evaluated further.

ACKNOWLEDGMENTS

I would first like to thank my major professor, Dr. Harold H. Prince for his
wisdom, patience, support, and encouragement during my graduate work. Drs. Scott
Winterstein and Don Beaver also have my gratitude for their input, advice, and willingness
to serve on my graduate committee. -

Gratitude is extended to the Michigan Department of Natural Resources for
funding this project through the Wildlife Conservation and Restoration Act, Pittman-
Robertson Project Number W-127-R I would like to thank Jerry Martz and Bill Moritz
from the MDNR for their help in gathering preliminary data and conducting road surveys.
Thanks are also extended to Al Stewart and Barry Loper from the MDNR for their many
contributions to this project, including their input on methods to conduct nest box surveys
and use of their facilities.

I acknowledge and thank all the MDNR personnel from the nest box survey sites
for their help in locating and surveying the boxes. This includes Jetr Greene and Ron
Hoflinan from Somerset S.G.A., Amie Carr and Brian Vogel fiom Shiawassee S.G.A.,
Rodney Clute from Crane Pond S.G.A., John Lerg fiom Allegan S.G.A, Dave Smith from
Flat River S.G.A, Bill Fuchs fiom Muskegon S.G.A., and Bill Stark fi'om Bald Mountain
Recreation Area.

I would like to thank additional wildlife professional for their contributions to this

iii

project. This includes Joe Johnson fi'om Kellogg Biological Station for his help in locating
nest boxes and with the radio telemetry design and Paul Padding fiom the USFWS for his
help and advice with the wood duck harvest surveys. I thank Tom Oliver for his work
initiating this project and employees, Barb Lercel, Daria Hyde, Tom Bryant, and Scott
Graves for their help and hard work in conducting surveys. I also extend my gratitude to
my interns, Adam Bump, Heather Catlin, and Cal Steinorth for their fiiendship and hours
of hard labor given to this project. I also thank Cathy Flegel, John Nrewoonder, and Mike
Wasilco for their help and support with my graduate work.

I extend my heartfelt thanks to my family, especially my parents, Stan and Ruth
Briggs, and my fi'iends for their encouragement and support throughout my time in
graduate school. I extend special thanks to Stacie Force, Cindy VanNeste, Allison
Gormley, and Julie Weeder for their fiiendship and emotional support.

Finally, I thank God for the opportunity to complete my Master’s Degree and for

seeing me through this challenging experience.

iv

TABLE OF CONTENTS

LIST OF TABLES .................................................................................

LIST OF FIGURES .................................................................................

INTRODUCTION .................................................................................

METHODS .................................................................................

Brood Surveys .................................................................................
Road Surveys .................................................................................
Nest Box Surveys ............................................................................
Data Analysis .................................................................................

RESULTS .................... ‘ .............................................................

Brood Survey .................................................................................
Road Survey .................................................................................
Comparison of Road Surveys to Aerial Surveys ...............................
Nest Box Surveys ............................................................................
Fall Harvest .................................................................................

DISCUSSION .................................................................................
CONCLUSIONS AND RECOMMENDATIONS ........................................

APPENDIX A:

A radio telemetry pilot study of female wood ducks in Maple
River and Gratiot Saginaw State Game Areas during the 1996
brooding season ..............................................................................

APPENDIX B:

Predictions of the fall flight of wood ducks with mechanistic
model equations ...............................................................................

LITERATURE CITED ...............................................................................

36

40

46

LIST OF TABLES

Table l. Broods observed and estimation of effort for night float counts fiom
20 June, 1994 to 27 July, 1994 ..................................................................

Table 2. Mean number of wood ducks Observed and/or heard in Michigan
along 17 difi‘erent roadside surveys conducted between 15 April and 14
May (Early) and between 15 May and 6 June (Late). Nine of the surveys
were conducted on the same routes for 3 consecutive years (1994 - 1996).
Eight other routes were conducted for 2 years (1995 and 1996). In 1994,
routes were conducted 8 times. In 1995 and 1996, routes were conducted
4 times .......................................................................................................

Table 3. Analyses of Variance for the 9 road surveys conducted from 1994 -
1996. Due to the significant interaction of site and time in the first
AN OVA, the analysis was repeated twice, first separating the data by
years and then by time periods ...................................................................

Table 4. Analysis of Variance for the road surveys conducted fiom 1995 -
1996 ..........................................................................................................

Table 5. Indicated pairs calculated fi'om road survey data for 1994 - 1996.
Indicated pair‘s = [Pairs + [Lone Males + [Lone Females + [Lone
Birds of unknown sex ................................................................................

Table 6. Numbers of indicated breeding wood ducks from MBWS helicopter
and fixed wing transects (Data fiom MDNR) and mean indicated breeding
wood ducks fi'om the road surveys .............................................................

Table 7. Number of nest boxes and use by wood ducks and hooded mergan-
sers by site, based on surveys conducted fi'om May through September in
1995 and 1996 ...........................................................................................

Table 8. Estimated wood duck harvest by year for 3 units in lower
Michigan including sample sizes of total number of birds sampled and
hunters participating. (Data from USFWS) ................................................

11

13

14

15

16

17

19

21

Table 9. Comparison of region means between the number of wood ducks per
sampling unit fiom the road surveys and the occupancy rates from the nest
box surveys. The road surveys and nest box surveys were significantly

correlated (r, = 0.943, P = 0.0048) at the unit level in 1995 and 1996 ......... 22
Table 10. Fall flight estimates based on varied numbers of indicated pairs
fi'om 1995 and 1996 road surveys .............................................................. 33

Table A1. Summary of individual records for wood duck females trapped and
marked in the radio telemetry pilot Study in 1996. Twenty female wood
ducks were trapped, radio collared, and monitored from 10 June to 16
August. The number of locations and the top 3 habitat types birds
occupied are listed in the order of use. Number of ducklings observed and
the date of the last radio contact with the female are listed ......................... 38

Table A2. Sizes (in ha) of home ranges for hens monitored in radio telemetry
study. Home ranges were calculated using harmonic home range formulas

in Telem88. 50 %, 80 %, and 95 % contours of use were calculated ......... 39
Table Bl. Parameters and equations used as the basis for fall flight model
(Walters et al. 1974) .................................................................................. 41

Table 32. Estimates of parameters for fall flight model of wood ducks, using
indicated birds fiom 1995 and 1996 road surveys to replace inputs fiom
aerial surveys for adult spring populations calculations .............................. 34

vii

LIST OF FIGURES

Figure 1. Location of night stream brood surveys conducted in 1994 ............. 5

Figure 2. Locations of Michigan road survey routes for 1994 to 1996 and
nest box surveys conducted in 1995 and 1996 .......................................... 6

Figure 3. Locations of units A, B, and C in lower Michigan. Unit A =
Allegan, Barry, Berrien, Branch, Calhoun, Cass, Eaton, Kalamazoo, St.
Joseph, and Van Buren counties. Unit B = I-Iillsdale, Ingham, Jackson,
Lenawee, Oakland, and Washtenaw counties. Unit C = Clinton,
Genesee, Gratiot, Ionia, Kent, Montcalm, Newaygo, Oceana, Ottawa,
Saginaw, and Shiawassee counties ............................................................ 23

Figure 4. Results of cluster analysis on USFWS harvest estimates (harvest/
1011’) means for 36 southern Michigan counties. Means were calculated
for 13 years, fiom 1983 to 1995 ............................................................... 25

Figure 5. Comparison of the frequency distribution of harvest estimates for
36 counties of Michigan for 1983 - 1996 and the indicated pairs fi'om the
road surveys for 1994 - 1996. The indicated pairs are marked above the
frequency bar containing the harvest estimate for that year ....................... 32

viii

INTRODUCTION

Wood ducks (A ix sponsa) have been a challenge to wildlife managers for the past
century. In the early 1900's, wood duck numbers were believed to be so low that many
feared they were near extinction. Hunting of wood ducks was stopped in 1918. The
recovery of the population is well documented (Bellrose and Holm 1994) and hunting
resumed about 60 years ago.

Wood ducks are currently a significant portion of Michigan’s annual harvest of
waterfowl, second only to the mallard (Anas platyhynchos) (Brakhage 1990). The
number of wood ducks harvested in Michigan has increased in the last few decades, from a
low of 9,900 in 1961 to a high of 60,800 in 1985, under a stable bag limit of 2 birds per
day (Bartlet and Trost 1990). The estimated harvest has plateaued this past decade.

There is some evidence that breeding wood ducks have saturated the prime
lowland forest habitats, forcing expansion in lower quality areas (Bellrose and Holm
1994). This is supported by breeding birds being detected in areas with secondary habitat
such as northern coniferous forests.

The increases in the numbers of breeding and harvested wood ducks should be
evaluated cautiously. Gamble (1990) believes that increases in the length of the hunting
season in several areas could be responsible for increases in wood duck harvest. Similarly,

Bellrose and Holm (1994) suggested that the increased importance of wood ducks to the

2

annual harvest could be the result of simply harvesting a greater proportion of the wood
duck population. They rejected this as a possibility since an expected increase in the rate
of band returns did not occur consistently over the areas experiencing an increase of wood
duck harvest.

Shifts in bag limit from 1 to 2 birds were easily accommodated in the 1970's and a
daily bag of 3 birds in response to population increases could be possible in future seasons
(Kelley 1994). However, an increased loss and degradation of primary wood duck habitat
(Tiner 1984) makes it imperative that an accurate means of assessing fall flight be
developed before any filrther liberalization of the wood duck harvest occurs.

Direct estimates of wood duck numbers and production are dificult to derive
(Geis 1966, Sauer and Droege 1990). A habitat preference of flooded woods and swamps
and secretive behavior makes this species difl'lcult to survey (Brakhage 1990).
Additionally, the wood ducks’ unique breeding range within the continental United States
(Bellrose and Holm 1994) is well beyond most of the range of aerial surveys used for
counts of breeding waterfowl in the more northern latitudes of the U. S. and Canada (Geis
1966). Currently, only indirect means, such as band returns and harvest data, can be used
to monitor status. Although these methods are suitable for the detection of general trends
over large areas, finer scaled population estimates are needed for monitoring the
libralization of hunting regulations (Mississippi Flyway Council 1994).

Cleveland (1994) estimated the wood duck breeding population based on aerial
surveys and input this data in a model designed to calculate the fall flight for Michigan.
The estimates of fall flight appeared to overestimate production by 100 to 300 %. The

aerial survey’s visibility correction factors (VCF) are high for wood ducks compared to

3

other species meaning that wood ducks are more difficult to detect. Wood duck VCFs
ranged from 9.0 to 28.0, while the VCFs for mallards range from 2.3 to 5.5 during the
1992 through 1996 surveys (J. Martz, pers. comm.) In addition to poor aerial survey
results, available data on clutch sizes and survival rates of wood ducks in Michigan are
limited.

A few alternate survey methods for wood ducks have been suggested as
independent indices. Road surveys, like those conducted in the Breeding Bird Survey
(Sauer and Droege 1990), annual nest box surveys (Zicus and Hermes 1987), and brood
surveys on stream (Cottrell and Prince 1990) have been suggested as potential population
indices for wood ducks.

The goal of this project is to evaluate survey methods that might provide a finer
scale of monitoring of wood duck numbers and production. The degree to which survey
methods studied correlate with the estimates derived from the traditional analyses of

harvest surveys will also be assessed.

METHODS

Brood Surveys

Night surveys on streams for wood duck broods were conducted from 01 June to
31 July 1994. Surveys were conducted bi-weekly along 16 to 32 km (10-20 mile) sections
of three rivers located in south and western Michigan (Figure 1). The surveys began at
the upstream end of the study areas at least 30 minutes after sunset. Using a 14 foot flat
bottomed boat powered by a 6 hp engine, 2 observers motored downstream, visually
scanning the shoreline for wood duck broods. Each observer used a 450,000 candlepower
hand held spotlight powered by a 12 volt battery, to monitor the stream channel and
banks. The numbers of adults, broods, ducklings per brood, and age classes of the
ducklings were recorded.
Road Surveys

Nine road survey routes were established in the southern half of the lower
peninsula of Michigan in 1994 and expanded to 17 routes in 1995 (Figure 2). In 1994,
four of the routes, Allegan, Maple, Mendon, and Muskegon l, were placed in areas
adjacent to forested livers used for the stream brood surveys and the five remaining survey
routes, Albion, Ann Arbor, Cambria, Fine Lake, and Hastings, were established in sites

that corresponded with the flight paths used in the Michigan Breeding Waterfowl Survey

Muskegon

n

= Maple

: Kalamazoo Grand

L l

_—f

 

D Location of River Survey Routes

Figure 1. Location of night stream brood aurveya conducted in 1994.

 

* Road Survey Route
Nest Box Survey Site

A = Muskegon l G = Hastings
E= Muskegon 2 H = Fine Lake
= Flat River IRose Lake

D = Maple River = Bald Mm.

Gratiot-Saginaw I“ Gull Lake

F = Allegan L = Albion

M = Ann Arbor
IE= Crane Pond

O = Mendon
Somerset

Q = Cambria

R = Shiawassee

Figure 2. Locations of Michigan road survey routes for 1994 to 1996 and

nest box surveys conducted in 1995 and 1996.

D- road survey routes conducted only In 1995 and 1996.

7

(MBWS). In 1995, eight additional routes were established and conducted. The new
routes, Bald Mountain, Crane Pond, Flat River, Gull Lake, Gratiot-Saginaw, Muskegon 2,
Rose Lake, and Somerset, traversed areas with active nest box programs. The Maple
River route is the only survey that met all of the road survey criteria since it is adjacent to
the Maple River stream survey route, corresponds with flight path number 60 from the
MBWS, and is in Maple River State Game Area which has an active nest box program.
All 17 routes were repeated in 1996.

Each 10 mile route was started 30 minutes before sunrise. Surveys were not
conducted on days with any precipitation or wind speeds greater than 15 mph. The
temperature, time of sunrise, percent cloud cover, time of first observation, and initial
odometer readings were recorded at the beginning of each survey. The observer stopped
at every halfmile, exited the car, and counted the numbers of wood ducks seen and heard
over a three minute period. The number and sex (when possible) of the wood ducks
observed were recorded. The final time and odometer reading were recorded upon
completion of the survey.

In 1994, each route was conducted 8 times on different mornings. The surveys
were conducted on 4 days between 15 April and 14 May and repeated on 4 days between
15 May and 5 June. The early period was established because it overlaps the timing of the
peak nest initiation for wood ducks in Illinois (17 April to 2 May), Minnesota (15 April to
25 April) (Bellrose 1980), and Michigan (Bellrose and Holm 1994, Prince 1991). The late
period was selected by the USFWS Office of Migratory Bird Management as the desig-
nated time period for nationwide wood duck road surveys to be conducted (Kelley 1994).

To allow for increases in the number of sites surveyed, the number of counts was reduced

8
fi'om 4 to 2 in the early time frame (15 April and 14 May) and the late period (15 May and
5 June) for all 17 survey routes in 1995 and 1996.
Nest Box Surveys
Wood duck nest boxes at 11 sites with active nest box programs (>20 boxes)
throughout southern Michigan were checked in 1995 and 1996 (Figure 2). To facilitate
comparisons between survey methods, nest box surveys sites were conducted in areas with
road surveys. The boxes were checked at least once between 9 May and 30 September.
Boxes checked during the breeding season (before 30 June) and boxes with active nests
were rechecked after 1 August when all nesting had been completed.
The presence, species, and outcome of nests were recorded for each nest box.
The size, shape, and color of the eggs were used to identify species. Nest boxes
containing at least 1 wood duck egg were counted as being used. Nests with no down and
few or many eggs (as in the case of dump nests) were designated as being in the laying
stage. Nests lined with down and at least 8 eggs were recorded as being in the incubating
stage. Hatching stage included successfully hatched nests and nests in the process of
hatching at the time the nest was surveyed. The success of a nest was determined by the
presence of at least 1 shell membrane. Nests with cold whole eggs were recorded as being
abandoned. Nests with broken eggs were considered to be unsuccessful. Since nests were
only checked once, cases in which nests were abandoned before they were predated could
not be detected.
Data Analysis
A General Linear Model (GLM) Analysis of Variance (AN OVA) (SAS Institute

Inc. 1990) was used to analyze numbers of wood ducks fiom the road surveys and rates of

9
nest box use. ANOVAS were completed for the 9 sites surveyed fi'om 1994 to 1996 and
the 17 sites run in 1995 and 1996. A three factor AN OVA with interactions using the
parameters year, site, and the 2 time periods (early: 15 April to 14 May, late: 15 May to
June 5) were tested for road surveys. The assumptions of this analysis are that the effects
of the treatment and the environment are additive and that the experimental errors are
random, independently and normally distributed about a mean of 0 (Steel et al. 1997).
When significant interactions were detected, violating the assumption of independence, the
components of the interactions were separated and tested again. Year and site without
interactions were the parameters tested for the rate of nest box use. The percent use rates
of the nest boxes were adjusted with the inverse sine (arcsin Y‘°") transformation (Steel et
al. 1997). Tukey’s test (honestly significant difference) (Steel et al. 1997) was used on
years and sites for the road surveys and on the sites for the nest box surveys to detect
which years or sites may be significantly difi‘erent (a = 0.05).

Spearman’s Rank correlations (Steel et al. 1997) were used to compare the results
of the difi‘erent survey techniques using geographical locations and the year as the basis for
pairing the survey methods. For each year, Spearman’s Rank correlations were on the
road survey and nest box survey data pairing the surveys first by Sites and then by units
which are groupings of the counties based on geographical proximity. Two external
surveys were also compared to the nest box and road survey data using Spearman’s Rank
correlations. USFW S harvest estimates were compared to both road surveys and nest
box surveys by regions. The indicated breeding birds calculated from road survey data
were compared to indicated breeding birds calculated from Michigan Breeding Waterfowl

Surveys conducted at 6 corresponding sites, Albion, Ann Arbor, Cambria, Fine Lake,

10

Hastings, and Maple River from 1994 to 1996. The data were grouped for the 3 years for
the comparison (11 = 17).

Cluster Analysis on USFW S county harvest estimates were conducted using
SYSTAT to visually evaluate the level of harvest across lower Michigan. The cluster
analysis used means of 13 years of USFW S harvest estimates for each county from 1983

to 1995. The Euclidian distance and Wards linkage were used for the cluster analysis.

RESULTS

Brood Survey

In 1994, 8 night brood surveys were done between 20 June and 27 July in southern
Michigan (Table 1). One brood of wood ducks was observed on the Kalamazoo River in
Allegan State Game Area on 25 July. No other broods were contacted. This survey

method was not repeated in 1995 or 1996.

Table l. Broods observed and estimation of effort for night float counts from 20 June, 1994 to 27

 

 

July, 1994.'

Stream n" Survey Time Miles Broods Broods
dates started surveyed observed per mile

Kalamazoo 3 6/22 2205 16 0 0.00
7/1 1 2200 16 0 0.00
7/25 2225 16 l 0.06

Maple 2 6/‘20 2150 13 0 0.00
7/27 2143 14 0 0.00

Muskegon 3 6/2 1 2203 10 0 0.00
7/5 2212 10 0 0.00
7/26 2202 10 0 0.00

Total 8 105 1 0.01

 

' Four additional surveys were attempted, l on the Muskegon River (7/18) and 3 on the Maple
River (7/12, 7/14, and 7/20). These were not successful due to weather and equipment failures.
" n = # of surveys conducted

11

12

Road Survey

The number of wood ducks detected on road surveys varied between sites and
years (Table 2). On 9 routes conducted 8 times in 1994, 355 wood duck contacts were
made on 1428 three minute listening sessions. An average of 7.6 wood ducks were seen
and/or heard per route in the early time period and 3 .2 wood ducks were counted per
route in the late time period. Although the number of routes was nearly doubled in 1995
and 1996, the reduced sampling intensity per route still amounted to 1407 three minute
listening stations in 1995 and 1428 in 1996. The expanded surveys resulted in 154 and
193 wood ducks contacts in 1995 and 1996, with an efl‘ort of about 310 person-hours
each year. The combined average number of birds per count declined from 5.2 in 1994 to
2.3 and 2.9 birds per route in 1995 and 1996, respectively.

The year (P = 0.0001), site (P = 0.0001), the time period (P = 0.0011) and the
interaction between the year and the time period (P = 0.03 69) of the 9 surveys conducted
from 1994 to 1996 were significant (Table 3). Due to the significant interaction,
ANOVAS were run with the data separated first by year and then by time period. Site (P
= 0.0001), time period (P = 0.0001), and the site and time period (P = 0.0011) interaction
were significant sources of variation in 1994. In 1995, the site (P = 0.0687) and time
period (P = 0.0900) seemed to play a role in the variation, while the site and time
interaction (P = 0.7707) was not significant. Site was the only significant main efi‘ect in
1996 (P = 0.0086). For the original 9 routes, site (P = 0.0687) and time (P = 0.0900)
appeared to play a role in the variation in 1995, while the site and time interaction was not
significant (P = 0.7707). Site was the only significant main efl‘ect for the 9 routes in 1996

(P = 0.0086).

13

Table 2. Mean number of wood ducks observed and/or heard in Michigan along 17 roadside survey sites
conducted between 15 April and 14 May (Early) and between 15 May and 6 June (Late). Nine ofthe surveys
were conducted on the same routes for 3 consecutive years (1994 - 1996). Eight other routes were conducted
for 2 years (1995 and 1996). In 1994, routes were conducted 8 times. In 1995 and 1996, routes were
conducted 4 times.

 

 

 

1994A 19958 1996B
Survey
Route Early Late Combinedc Eariy Late Combined Early Late Combined
Albion 7.5 5.3 6.4 1.5 0.5 1.0 1.5 3.5 2.5
Allegan 3.0 1.0 2.0 4.0 1.5 2.8 3.5 4.5 4.0
Ann Arbor 2.3 0.3 1.3 1.0D 0.5 0.7 4.0 1.0 2.5
Bald Mtn. 3.0 1.5 2.3 4.0 2.0 3.0
Cambria 1.3 1.3 1.0 0.5 0.8 1.0 0.5 0.8
Crane Pond 3.0 0.5 1.8 0 0.5 0.3
Fine Lake 8.0 3.0 5.5 1.5 1.5 1.5 1.0 3.0 2.0
Flat River 3.0 4.5 3.8 3.5 3.0 3.3
Gratiot- 0 1.0 0.5 1.0 1.5 1.3
Saginaw
Gull Lake 2.0 3.5 2.8 0 2.0 1.0
Hastings 5.8 5.8 5.8 4.5 0.5 2.5 6.0 2.5 4.3
Maple River 22.0 6.8 14.4 12.5 4.5 8.5 8.5 11.5 10.0
Mendon 8.0 4.0 6.0 2.0 0 1.0 2.5 2.0 2.3
Muskegon 1 3.8 1.3 2.5 0 0 0 2.5 0.5 1.5
Muskegon 2 4.5 1.5 3.0 3.5 3.0 3.3
Rose Lake 4.0 2.0 3.0 4.0 1.0 2.5
Somerset 2.0 4.0 3.0 7.0 2.5 4.8
Total Birds 355 154 8 196
Mean I route 7.6 3.2 5.2 2.9 1.6 2.3 3.2 2.4 2.9

 

A: n=4 foreach surveyperiod

B: n=2foreach surveyperiod

C: Combined was derived as the mean of the early and late time periods
Dzn=1forthissurveypcriod

14

Table 3. Analyses of Variance for the 9 road surveys conducted from 1994 -1996. Due to the significant
interaction of site and time in the first ANOVA, the analysis was repeated twice, first separating the data by

 

 

years and then by time periods.
ANOVA Parameters Source of Variation DF Mean Square P Significance

All Years and Times Year 2 103.9 0.0002 "*
Site 8 123.2 0.0001 “"‘
Time 1 122.9 0.0011 "*

Year "' Site 16 12.5 0.3145
Year " Time 2 36.9 0.0369 “’t

Site " Time 8 13.1 0.2997

Year " Site ‘ Time 15 15.7 0.1378

Separated by Year

1994 Site 8 129.0 0.0001 ""
Time 1 240.3 0.0001 ”"'
Site ‘ Time 7 48.2 0.0011 "*

1 995 Site 8 25.9 0.0687

Time 1 36.0 0.0900

Site " Time 8 6.6 0.7707
1996 Site 8 30.0 0.0086 *"

Time 1 0.3 08605

Site ‘ Time 8 5.4 0.6943

Separated by Time Period

Time 1 Year 2 109.1 0.0033 *"
Site 8 102.4 0.0001 "*

Year "' Site 15 22.0 0.2306
Time 2 Year 2 30.6 0.0066 *“
Site 8 33.3 0.0001 "*

Year ‘ Site 16 6.7 0.2801

 

15

The 2 time periods were separated grouping data for all 3 years. In the early time
period, the year (P = 0.0033) and site (P = 0.0001) were both significant with Maple River
as the only site significantly higher than all other sites (P < 0.05). The 1994 surveys were
significantly higher (P < 0.05) than the surveys fi'om 1995 and 1996. Year (P = 0.0066)
and site (P = 0.0001) were also significant sources of variation for the late time period. In
this case, the Maple River counts were significantly higher (P < 0.05) than all sites except
Hastings and Albion. Hastings and Albion were not significantly different fi'om any of the
Sites. For the late time period, counts were lowest in 1995 (P < 0.05) and not different
between 1994 and 1996.

For the 17 read surveys conducted in 1995 and 1996, sites (P = 0.0001) and time
periods (P = 0.0444) were significant sources of variation while the year (P = 0.1731) was
not (Table 4). Maple River continued to be the only site with significantly higher counts

compared to the other sites (P < 0.05).

Table 4. Analysis of Variance for the 17 road surveys conducted fiom 1995 - 1996.

 

 

Source of variation DF Mean Square P Significance

Year 1 12.2 0.1731

Site 16 32.1 0.0001 “"

Time 1 27.0 0.0444 ”*
Year l"Site 16 2.5 0.9815
Year " Time 1 4.5 0.4041
Site " Time 16 4.1 0.8382
Year " Site " Time 16 6.6 0.4421

 

The total number of indicated pairs of breeding birds was calculated by adding the

16

number of indicated pairs from all road surveys for each year (Table 5). The number of
indicated pairs for each road survey was calculated using the formula:

Indicated pairs - [Pairs + [Lone Males +21»... Females +210» Birds with unlmorvn sex.
This formula is Similar to Cleveland’s (1994) indicated bird formula with modifications
allowing the more audible lone females to be counted as an indicated pair. The number of

indicated pairs ranged from 80 to 125.

Table 5. Indicated pairs calculated from road survey data for 1994 - 1996. Indicated pairs = [Pairs + ZLone
Males +£Lone Females +£Lone Birds with unknown sex

 

Year Number Total Pairs Lone Lone Lone Sex Flocks Total Indicated

 

of . Birds Males Females unknown (>2 and Pairs per
surveys Counted unknowns) Season
1994 68 355 34 24 31 36 197 125
1995 67 154 23 21 15 21 51 80
1996 68 196 46 13 16 37 38 112

 

Comparison of Road Surveys to Aerial Surveys

A total of 146 to 158 wood ducks were observed during helicopter surveys while 4
to 12 % of those totals Were counted fi'om a fixed wing aircraft flying the same route
(Table 6). The total indicated breeding wood ducks fiom the road surveys conducted in
the vicinity of the aerial routes ranged fi'om 58 to 88.

The indicated birds from helicopter and fixed-wing airplane surveys for the
Michigan breeding Waterfowl Survey were compared to the total number of indicated
wood ducks detected at each of the 6 road survey routes corresponding to the flight paths

using Spearman Rank correlation analysis. Since the road surveys from 1994 had 8

17

surveys conducted per route, the total indicated birds from this year were not directly
comparable to 1995 and 1996 when 4 surveys were conducted per year. To reduce the
number of surveys for 1994, 2 surveys were randomly selected from each time period for
each site, resulting in 4 surveys used per site in 1994. The 1994 indicated breeding birds
were then calculated with the subset of the 1994 surveys. The years were combined for
the correlation analysis (it = 17). The road surveys were not significantly correlated with
the helicopter surveys (r, = 0.144, P = 0.583) or the fixed-wing surveys (r, = 0.075, P =

0.776) on the 6 areas over the 3 year period.

Table 6. Numbers of indicated brwding wood ducks from MBWS helicopter and fixed wing transects (Data
from MDNR) and the mean indicated breeding wood ducks fiom the road surveys.

 

 

 

1994 1995 1996
Tranaea Site Helicopter Fixed Road Helicopter Fixed Road Helieorxa’ Fined Road
4 cm 14 2 2 36 0 6 l4 0 2
19 AA; 16 4 4 44 0 4 26 0 10
21 “5°“ 21 0 28 14 0 6 20 2 6
F“
27 kt 58 0 10 42 2 8 32 6 12
41 my 30 10 20 22 4 8 38 2 16
M
60 R3: 10 2 16 - - 26 16 O 42
Total 149 18 80 158 6 58 146 10 88

 

While the helicopter surveys appear to be more eficient than the road surveys at
detecting wood ducks, differences in the methods and the size of area surveyed should be
noted. The helicopters zig-zag along 18 mile by 1/4 mile transect while the road surveys

are conducted in sampling plots with a 1/4 mile radius. An area of 4.50 mi2 are surveyed

18

for each helicopter survey while 1.03 mi2 are surveyed for each road survey route and a
total of4. 12 nri2 are surveyed for each road survey site. Adjusting the total indicated pairs
for the area surveyed results in 5.5, 5.9, and 5.4 wood ducks/mi2 for the helicopter surveys
and 3.2, 2.3, and 3.6 wood ducks/mi2 which are fairly similar even though the road
surveys are a more passive survey.

Nest Box Surveys

Eleven sites with over 700 nest boxes were checked in 1995 and 1996 (Table 7)
with an efl‘ort of 1160 person-hours. Wood ducks nested in 40 % of the nest boxes while
10 % contained clutches with wood duck and hooded merganser (Mergus cucullatus)
eggs. Hooded merganser nests were found in an additional 11 % of the nest boxes.
Almost 78% of the merganser nests (8% of total number of nests) were located in boxes
along the Consumer’s Power line in Muskegon on the west side of the state. The
remaining 39 % of nest boxes were not used by nesting waterfowl.

All of the Muskegon Power line data were deleted from further analysis since the
condominium style nest boxes used at this site with 4 boxes within a 2 meter space at each
of the outer support beams for the power line were different fiom the other sites. The
remaining boxes from Muskegon State Game Area were similar in placement to the boxes
fiom the other sites and were included in the analyses.

The percentage of nest boxes used at each site was transformed using an inverse
sine transformation before the GLM Anova was applied to the data. Use of nest boxes by
wood ducks varied significantly among sites (P = 0.0001), ranging fiom 21 % to 76 %
(0.21 to 0.87 upon transformation). Total use rates of 51 and 50 % of the nest boxes was

similar between years (P = 0.8195).

19

.8382 :68 8 .8ng Boa—q 8x3 v ac 8:8» 55 .98... 8:383:00 a S 203 3?.
£5 5 8x3 2:. dons—$9.8 x3 5 80:20.9“. 3 one 8553 mafia—=2 05 Set 38350 83 37. 2.: been coma—32 2F e

82 Aug 0.233 can..— 83. 5 86:0 new E was v5 5:80 Sue 80555 05 a 5&3
.623 v8.— 33 omox 25. £023 55 .8: 98 £255 38 e833 885958 8882a 3 8:358 an: 82a 25. 65m 0893
e5 .3300 mg June muons 3.8 8.8 men—8.0 one 8.3. £888.: 0.32:5 8:3 83. 555 8.633 8x3 3580 g 8.3 89m e

.92: a e .86 as see: use... 5.: assume Susanne .8 as see. sea a: 3 832.8 882 .

 

 

 

 

em 2 z. 2. an an n: N: .58.
an 8 an R a a 2 a: mu .2: .25.. senses—2
.— an e e e _ m e z a Sea 2:5
am ea _ e n o e e R 2 83x82
"ma e... e e e _ e. e Z a seesaw
"mac 2 n e e e 2 a: 3 S as?
mun—om an e o m e on an 8. e: .38 2%:
38¢ a m e e 2 w e ”a em 25:82 28
DR? an e N e e no G «2 2: sansméeemv
one. 3 e m m n en mu 9. mm .25. an
m< z _ ~ N e z. 8 an 2. .33 seem
< 2. n n : 2 2 R 8 S Beaafim
ea: 32 e8— 82 82 32 ea: 32
.325 e83 3 legal decline 882. a is... e83 3 new
has 8.. :82 38.. 3 as... a case .85 Baa. a 8...: a axes sea:
.83 e8 32 E Seesaw

49.25 .32 863:8 manta no 8mg .86 .3 £8592: v38: was 8.96 395 .3 on: 93 8x3 .8: me 3952 H «Eek

20

The mean of the total number of wood ducks observed and the means for the early
and late time periods of the road surveys were examined for correlation with the use rates
from nest box surveys. The road survey and nest box data were paired by site and
analyzed for each year, giving each site a data point for both road surveys and nest box
surveys for each year. No correlation using Spearman’s Rank correlation analysis was
found between the nest box surveys and the total observed on the road survey (r, = 0.059,
P = 0.816). The same is true between the percent use rate of nest boxes and totals

observed in early (r, = 0.148, P = 0.558) and late (r, = 0.123, P = 0.627) time periods.

Fall Harvest

The USFWS estimates the number of wood ducks harvested each year by county
for every state in the United States by surveying hunters through sampling post ofices
that sell hunting licenses (Crissey 1984). Harvest estimates for 36 counties of southern
Michigan were summarized for 13 years from 1983 to 1996 (Table 8). During this period,
between 60 and 122 hunters were surveyed, sending in between 129 and 342 wood duck
wings annually. The highest year was 1984 when 41,950 birds were harvested. The
estimates for 1986 to 1993 ranged from 9490 to 19,490 wood ducks and were lower than
1994 and 1995 when 27,690 and 26,180 birds were harvested.

The harvest data, used as the baseline index for monitoring wood duck populations
trends in Michigan, were compared to both road surveys and nest box surveys. The
USFWS calculates harvest estimates at both the county and state levels. Ifin one year no
surveys were sent in from a county, then no estimates were calculated for that county.

Due to missing data, the county level is not an appropriate comparison.

21

Table 8. Estimated wood duck harvest by year for 3 units in lower Michigan including sample sizes of total
number of birds sampled and hunters participating. (Data from USFWS)

 

 

 

 

Birds/1cm2 Sample Size

Year Unit A“ Unit 3" Unit C ‘ Birds Hunters Total Harvest
1983 0.7 0.5 0.3 314 122 28780
1984 1.0 0.6 0.6 342 122 41950
1985 0.9 0.3 0.3 262 102 27510
1986 0.4 0.2 0.3 129 60 17770
1987 0.3 0.2 0.3 138 67 15990
1988 0.2 0.1 0.2 134 70 9490
1989 0.3 0.2 0.3 194 96 15810
1990 0.4 0.2 0.1 132 72 13200
1991 0.4 0.4 0.2 154 72 19490
1992 0.4 0.1 0.3 227 87 15450
1993 0.4 0.3 0.2 197 85 16490
1994 0.7 0.3 0.4 228 84 27690
1995 0.8 0.4 0.2 132 61 26180
1996 0.6 0.3 0.2 196 N.A. 21120

Means 0.5 0.3 0.3 199 85 21215

 

‘ Unit A = Allegan, Barry, Berrien, Branch, Calhoun, Cass, Eaton, Kalamazoo, St Joseph, and Van Brn'en
counties. Unit B = I-Iillsdale, Ingham, Jackson, Lenawee, Livingston, Macomb, Monroe, Oakland, St Claire,
and Washtenaw counties. Unit C = Clinton, Genesee, Gratiot, Huron, Ionia, Kent, Lapeer, Montcalm,
Muskegon, Newaygo, Oceana, Ottawa, Saginaw, Sanilac, Shiawassee and Tuscola counties.

22

To facilitate comparisons using the harvest estimates, the road surveys, and nest
box surveys, the counties of southern Michigan were grouped into 3 units (Figure 3).
Each unit contained 10, 10, or 16 counties. The harvest data standardized for area fi'om
1983 to 1995 show some difi’erences between units, with Unit A having the highest rate of
harvest (Table 8). Units B and C had similar rates of harvest estimates. Data for road and
nest box surveys were pooled for each unit.

Spearman Rank correlation analysis was used on combinations of road survey,
rates of nest box use, and harvest surveys at the unit level. Harvest data were not
positively correlated with either road surveys ( n = 9, rs = -0.270, P = 0.482) or rates of
nest box use (n = 6, r8 = -0.754, P = 0.084). However, the mean number of wood ducks
observed on the road surveys were correlated with the rate of nest box use (it = 6, r8 =

0.943, P = 0.0048) (Table 9).

Table 9. Comparison of rmit means between the numbers of wood ducks per sampling point Sun the road
surveys and the occupancy rates from the nest box surveys. The road surveys and nest box surveys were
significantly correlated (rs = 0.943, P = 0.0048) at the unit level in 1995 and 1996.

 

 

 

 

Road survey Nest box survey
Year Unit n' Mean # wood ducks n" occupancy rate
(ducks/sampling point) (% of boxes used)
1995 A 576 0.09 79 46.8
B 383 0.08 84 42.9
C 448 0.15 404 62.3
1996 A 576 0.1 1 84 44.0
B 404 0.15 76 48.7
C 448 0.16 408 62.2

 

‘ n = number of 3 minute sampling sites in each unit
" n = number of nest boxes surveyed in each unit

23

 

 

 

 

 

 

 

 

 

 

 

 

 

Figure 3. Locations of proposed units A, B, and C in lower Michigan Unit A = Allegan,
Barry, Berrien, Branch, Calhoun, Cass, Eaton, Kalamazoo, St. Joseph, and Van Buren counties.
Unit B = Hillsdale, Ingham, Jackson, Lenawee, Livingston, Macomb, Monroe, Oakland, St.
Clair, and Washtenaw counties. Unit C = Clinton, Genesee, Gratiot, Huron, Ionia, Kent,
Lapeer, Montcalm, Muskegon, Newaygo, Oceana, Ottawa, Saginaw, Sanilac, Shiawassee, and
Tuscola counties.

24

Counties with similar harvest rates were grouped using cluster analysis on the
mean harvest rate fi'om 1983 to 1995. This analysis allowed the counties to be grouped by
similarities in harvest estimates, showing regions of high and low harvest. The analysis
resulted in 3 clusters with similar harvest rates (Figure 4). Muskegon and Allegan with
the highest harvest rates, 1.30 and 1.06 birds/1cm2 respectively, are both on the west side
of the state. The counties with low harvest rates, between 0.30 and 0.03 birds/kmz, are in
the thumb area and the middle of the state. Overall, Unit A had more counties in the high

and medium categories of harvest than Units B and C.

25

Harvest rates

I High - 1.30-r .06
E Medium -= 0.754135

Low = 0.30-0.03

 

Figure 4. Results of cluster analysis on USFWS harvest estimates
(harvest/kmz) means for 36 southern Michigan counties. Means
were calculated for 13 years, from 1983 to 1995.

DISCUSSION

Improved methods for monitoring wood ducks to estimate fall flight have been
sought by wildlife researchers and managers for the past 30 years. With several states
looking to liberalize hunting regulations for wood ducks, the need for better survey
methods has never been greater.

Several methods aiming to monitor the number of wood ducks have been
attempted. Researchers have examined aerial surveys (Cleveland 1994, Hein 1966) and
Kelley (1996) evaluated line-transects that traversed wood duck habitat in Missouri.
Other methods including stream brood surveys (Cottrell and Prince 1990), surveys of use
of nest boxes (Zicus and Hennes 1987), and road surveys (Sauerand Droege 1990) are
designed to provide indices that monitor trends in numbers (Martin et al. 1979). Banding
data and harvest surveys are used to estimate survival and annual harvest. Models
predicting fall flight of wood ducks in Michigan based on recruitment and survival
estimates have also been attempted (Cleveland 1994). These methods have limitations that
should be addressed.

Aerial surveys for waterfowl initiated in 1947 and expanded in 1955 to cover most
of the breeding habitat of waterfowl are one of the longest running waterfowl survey
methods (Martin et al. 1979). These systematic surveys seem to work well for most

waterfowl species nesting in prairie potholes and marshes, including mallards and Canada

26

27

geese (Branta canadensis). However, the low visibility in the densely vegetated swamps
and lowland forests used as wood duck breeding habitat prohibits this method fi'om
working cost effectively for wood ducks (Cleveland 1994).

Line-transects may have more potential than aerial surveys for estimating wood
duck densities because the habitat is traversed at the ground level. Kelley (1996) found
that in areas with medium wood duck densities similar to the Mingo National Wildlife
Refuge in Missouri, 30 km of transects are needed to achieve Coemcients of Variation
(CV) of 20 % while 480 km are needed for CV’s of 5 %. In areas with low visibility such
as dense scrub-shrub habitat, this sampling method would be ineflicient. Areas with low
densities of wood ducks would also require a higher sampling intensity. Another problem
with this method is that it requires areas large enough to run 30 km of transects in sites
with uniform habitat. Application of this method across the landscape would be dificult
where difl'erent habitat and wood duck densities would confound estimates.

Stream brood surveys, which are less arduous than line-transects, can be treated as
an index to wood duck populations or, with assumptions made about the area sampled, as
density estimates. Stream brood surveys seemed to have potential for monitoring wood
duck populations on the Holston River in Tennessee, with 20 % of marked broods
observed through biweekly night surveys (Cottrell and Prince 1990). In their study,
ducklings were easily observed roosting along the shorelines of the river (Cottrell and
Prince 1990). However, in Michigan, only one brood was observed for 105 miles of
stream floated in 1994. Obviously, stream brood surveys can not be used to monitor
wood duck populations in Michigan.

Major differences between brood survey results between Tennessee and Michigan

28

seem to reflect difi‘erences in habitat. Tennessee has several large rivers with the upper
Holston River boasting the highest wood duck populations of any streams in the nation
(Bellrose and Holm 1994). These river systems are the major brooding habitat available in
Tennessee and brooding hens may be more visible on the less densely vegetated and more
navigable river systems. In contrast, while Michigan has several smaller rivers, it also has
more than 2 million ha of wetlands (Dahl 1990), over 7 times the area in Tennessee. In
Tennessee, Cottrell et al. (1990) observed consistent use of the Holston River shoreline
habitats by wood duck broods from hatching to fledging. In 1996, a pilot telemetry study
at Maple River and Gratiot Saginaw State Game Areas in Michigan (Appendix A) showed
many of the brooding hens to be using forested wetlands and lowland hardwoods. Twice
a week, visual locations were attempted on radioed hens during daylight hours. None of
the hens were found to be using the main river channel of the Maple River.

Nest box surveys, which require 3.5 times the effort of road surveys, have been
proposed as a population index. Although the rates of nest box use did correlate with
road surveys at the region level, variation in the use of nest boxes may result in less
dependable estimates than road surveys. Since most wood ducks nest in natural cavities
(Bellrose and Holm 1994), use of nest boxes may not be representative of use in natural
cavities (Ryan, et al. Unpublished, Semel et al. 1990). Some areas may have a lower
number of natural cavities causing the boxes to increase the numbers of females using the
areas (Soulliere 1990). In other areas, low nest box use rates may be caused by poor
maintenance of the nest boxes.

The positioning and density of the nest boxes can afl‘ect use rates. Nest boxes are

usually placed directly over water causing them to be the most obvious nest sites

29

(Soulliere 1990). In areas with both natural cavities and nest boxes present, the more
visible nest boxes probably result in higher use rates (Soulliere 1990). Nest boxes placed
in high densities can influence behavior. Females observing other females entering and
leaving nest boxes may be attracted to the same nest boxes causing dump nesting even
when other nest sites are available (Semel and Sherman 1986).

Since the use rates of the nest boxes may depend on the surrounding habitat,
natural cavity availability, maintenance, position, and density of nest boxes, this index may
not be reliable for monitoring the population at the site level. However, liens and Hennes
(1987) found use of nest boxes correlated with the number of hens and juveniles harvested
each year over 14 difi‘erent counties over a 5 year period. As with the road survey data, it
is possible that the site by site variation masks any trends that could be detected over time
in larger areas.

Harvest surveys, derived from the combination of hunter surveys on the numbers
of birds harvested with species and sex composition from Parts Collection Surveys
(Martin et al. 1979) are used to monitor wood duck harvest. Since the harvest estimates
are derived from a small sample of hunters, especially when broken down to the county
level, the data should be used cautiously (P. Padding pers. commun.) Because the
USFWS accumulates and analyzes data from the entire United States, they currently do
not have the capacity to increase sample sizes for the county level. Even at large scale
levels, the USFW S harvest estimates are problematic (Wright 1978, Tautin et al. 1989,
and Pendleton 1992).

Harvest estimates were not correlated with nest box use rates or road surveys at

the unit level. Numbers of breeding wood ducks based on the road surveys and nest box

30
use seemed to be highest for Unit C while Unit A had the highest harvest estimates.

Road surveys, similar to call counts and Breeding Bird Surveys (BBS), have been
used to detect trends of wood duck populations over groups of years (Sauer and Droege
1990). Route regression analysis of wood duck numbers from the BBS can detect
population trends (Sauer and Droege 1990). Kelley (1994) suggested that shortening the
survey from 40 km to 16 km and focussing efl‘orts in areas with prime wood duck habitat
might improve the efiiciency of the surveys for wood ducks. The limitations of route -
regression analysis include the requirement of at least 5 years of data to detect trends and
that changes in one year or changes at the local level, such as a management area can not
be detected (Kelley 1994).

The surveys focussing only on wood ducks were found to have higher variances
associated with them than the BBS (Kelley 1997). To detect a population trend of 2 %
change/year (a= 0.10, power = 0.80) in the Lake States (MI, OH, and IN) over a 3 year
sampling period, 2675 wood duck road survey routes with at least 1 wood duck observed
are required while only 299 BBS routes are required. Similarly, to detect a 4%
change/year in the same area over a 3 year period, 669 wood duck routes are required
while only 75 BBS are needed. Kelley (1997) concluded that establishing wood duck road
surveys to replace BBS would be inefl'rcient. However, their analysis was based on a
limited number of wood duck routes (only 24 for eastern North America).

Road surveys may have value as an index to wood duck populations in Michigan.
Although road surveys were not consistently correlated with nest box use at the site level,
unit correlations were significant. The lack of a correlation found between the road

surveys and the harvest estimates at the county or regional level could imply difi‘erences in

31

the distributions of the density of breeding birds and the fall harvest.

The number of indicated birds from the road surveys are higher for 1994 and 1996
compared to 1995 while similar numbers were harvested each year (Figure 5). The
discrepancies could be due to biases in the harvest estimates. The hunters who are more
likely to respond to the harvest surveys may be hunting their traditional sites until they are
successfirl instead of going to sites where the birds are at higher densities. Hunters putting
in higher levels of efi‘ort lead to overestimates of the harvest for that area.

One strength the wood duck road surveys have over the BBS is the sampling of
wood ducks throughout the pre-breeding stage. While the BBS is limited to 1 sampling
efl‘ort in early June (Martin et al. 1979, Kelley 1994), each of the 17 wood duck routes
were conducted in early and late time periods between April 15 to June 5 resulting in
approximately 68 total routes conducted. The timing of the road surveys did prove to be
significant in 1995 and 1996 and more wood ducks were consistently detected in the early
time period all 3 years. The variation in annual nesting chronology is an important
consideration in the enumeration of wood ducks. In brood surveys conducted in
Tennessee, Cottrell and Prince (1990) found variations in the timing of hatching fi'om year
to year. They concluded that using specific dates for comparing annual surveys would
result in unreliable estimates of production. The same may be true for surveys of the
spring breeding population of wood ducks. Surveying wood ducks once annually in June
may result in surveys of birds at different points in their breeding chronology from year to
year. However, spreading the sampling period over the entire early breeding stage and
summing the results for the total indicated birds may result in a more accurate estimate of

the number of wood ducks and intensity of the breeding efl‘ort along the sample route.

32

do? - 32 no.“ gage. use 82m when @3863 me £3852 n .1
.30» 35 com 82258 emote: 05 wee—5:8 an 5532.“ 05 gene wee—BE

as sea Bases 2: .eea .. :2 see sees... eee 82.. mane Bases 2: es. 3% - 32 Lee
fiwfiomz me 82:50 on com 88:33 emote: .«o momentum—e 5:262.“ 05 mo acme—3:50 .m oSwE

82: xv seesaw

om

ow

ow

o2

when

on _
833

o:
oo—

03

 

 

39%

mm 93”

mm 98

 

 

mea—

veg

among

 

k.

N92

 

83 .1

w— 3 E

 

29¢

 

ro—

rm—

tom

rmN

rem

rmm

row

.9.

 

.96—
aets. See
a 692 9

82 as:
Booze e\e

33

Fall flight estimates of wood ducks in southern Michigan were calculated using a
model based on survival and recruitment with the number of indicated birds fiom the road
surveys input as the adult spring breeding population (Appendix B). A fall flight of
81,504 and 112, 428 wood ducks in 1995 and 1996, respectively was estimated (Table
10). The harvest estimates computed by the USFWS are 30 % and 17 % of the fall flight
estimates using the unmodified indicated pairs for 1995 and 1996. The harvest estimates
from 1983 to 1996 ranged from roughly '/2 to 2 times the current estimates fiom harvest.
To simulate a similar range of road survey results, the numbers of indicated birds were
halved and doubled and the model equations were recalculated with these numbers.

Estimated fall flight ranged fi'om 40,752 to 224,856 wood ducks.

Table 10. Fall Flight estimates based on varied numbers of indicated pairs fi'om 1995 and
1996 road surveys.

 

Factor of indicated pairs # of indicated pairs Year Fall flight

 

Indicated pairs 112 1995 112,428
80 1996 81,504

‘/2 * Indicated pairs 56 1995 56,214
40 1996 40,752
2 * Indicated pairs 224 1995 224,856

160 1996 163,009

 

CONCLUSIONS AND RECOMMENDATIONS

Of the survey methods conducted in this project, road surveys appear to have the
best potential and should be evaluated further. The road survey methodology could easily
be expanded into a point sampling survey by refining the estimated sample area which
would facilitate the calculation of density. Density could be used to refine estimates of fall
flight. Road surveys are relatively low effort compared to nest box surveys or line-
transects. Although line-transects can be used to estimate densities of wood ducks
(Kelley 1996), the sampling effort required by this method would be extremely dificult to
expand to state-wide estimates.

Road surveys also fit well with techniques traditionally used to monitor other game
Species. Call counts have been evaluated for pheasants (Phasianus colchicus), and turkey
(Meleagris gallopavo meniami) (Scott and Boeker1972). Route regression of road
surveys is currently used to monitor woodcock (Scolopax minor) (Straw 1992) and
mourning dove (Zenaida aurita) (Dalton 1993) populations. The methodology is simple
and sampling efforts throughout the breeding season may provide useful counts of the
breeding wood ducks even with 5 year constraints on the data.

Currently, the predictions from the modified Walters et al. (1974) model with road
surveys input as spring breeding populations have high variance estimates (Appendix B).

Expanding the results of the road surveys to represent the 31 counties is a large source of

34

35

this variation. The model may be improved by stratifying the sample points by habitat
types. The amount of each habitat type present in the 31 counties could be estimated and
the wood duck densities could be expanded separately for each habitat type.

Kelley (1997) found the BBS to be more eflicient than routes focussing on wood
ducks. However, since the analysis was based on a limited number of wood duck routes
(only 24 for eastern North America), further evaluation of the two techniques is
warranted. Sampling throughout the breeding season instead of one time may provide a
better representation of the breeding wood ducks. Therefore, the wood duck road surveys

should continue and undergo further evaluation as a method of monitoring and predicting

the fall flight of wood ducks.

APPENDICES

APPENDIX A

A radio telemetry pilot study of female wood ducks in Maple River and Gratiot-Saginaw
State Game Areas during the 1996 brooding season.

APPENDIX A

A radio telemetry pilot study of female wood ducks in Maple River and Gratiot-
Saginaw State Game Areas during the 1996 brooding season.

To determine where brood survey efi‘orts should be directed, radio telemetry was
used in the summer of 1996 to monitor the movements of a small sample of brooding
hens. Hens captured through nest box surveys at Maple River State Game Area and
Gratiot-Saginaw State Game Area were fitted with necklace style telemeters (Lotek
Engineering, Inc., Newmarket, Ontario, Canada) weighing 15 grams equipped with a 15
inch antenna and a 9 hour mortality sensor.

Hens were located with directional antennas using triangulation from 2 locations.
Compass directions of each signal were marked on laminated Michigan Inventory
Resource Information System (MIRIS) maps of the area. The intersection of the 2 lines
from the 2 locations was assumed to be the hen’s approximate location. The mapped
location was recorded to the nearest 0.01 mi2 using the X,Y coordinates of a grid overlay.
The habitat, as described by the MIRIS map, was recorded for each location. Locations
were attempted 7 times for each hen each week. Four of the locations were during
daylight hours and 3 of the locations were taken at least 30 minutes after sunset.

Visual locations were attempted for each hen at least once every other week. The
radio telemetry equipment, including the receivers, directional antennae, and headphones
was used to walk or canoe into the location of the hen. At least 2 observers would use the
equipment to walk into the hen from difi‘erent angles. When hens were visually located,

the site, brood size, habitat type, and a qualitative description of the visibility of the area

36

37

were recorded. Other broods and hens seen in the area of the hen were also recorded.

Twenty female wood ducks were radio-collared in 1996 (Table A1). Twelve of
the hens had successful nests, 4 nests were unsuccessful with signs of predation, 3 were
abandoned, and 1 did not successfully complete hatching. By the end of the summer, only
4 hens were alive and not censored. Twelve hens had malfunctioning radios or moved far
out of the study region. Radios were collected for 4 hens. Upon collection of the radios,
3 of the sites had feathers or showed signs of that the hen died while only the radio was
found for the last hen.

Forested wetlands, represented by 611 (Table A1) were the habitat type recorded
as the most frequently used by the radio collared wood ducks. Thirteen hens were found
to be using forested wetlands as the first of their 3 primary habitats. Lowland hardwoods
(414) and herbaceous open fields (31) were also found to be used frequently.

The home range size was calculated for each hen using Telem88 (Coleman and
Jones 1988) (Table A2). The harmonic center and harmonic home range calculations

were selected in this program.

38

Table A1. Summary of individual records for wood duck females trapped and marked in the radio telemetry

pilot study in 1996. Twenty female wood ducks were trapped, radio collared, and monitored from 10 June to
16 August. The number of locations and the top 3 habitat types birds occupied are listed in the order of use.
Number of ducklings observed and the date of the last radio contact with the female are listed

 

 

Site Radio NoningAaanpt #of #of Primary Ming- Fate
Freq. Membranes locations habitats@ observed
Maple 148.496 Successful 13 45 611, 31, 622 2 Alive 8/16
148.834 Successful 6 48 611,31,21 5 Dead8/6
148.914 Successful 6 36 611, 31,622 0 Dead7/25
149.115 Successful 6 32 611,21,31 0 Lost7/l9
149.676 Successful 12 31 31, 611, 6 Radio
622/414 failed 701
149.706 Successful 5 40 611, 31,414 8 Alive 8/16
150.076 Unsuccessful - 53 611, 31,414 0 Alive 8/16
Shot&
collected
150.085 Successful 3 8 611, 31 0 Lost6f22
150.355 Unsuccessful - 17 622, 611, 0 Lost 7/5
3101
150.476 Unsuccessful - 51 622, 61 1, 31 0 Alive 8/16
Gratiot- 150.126 Successful 10 20 611, 414, 21 0 Collected
Saginaw radio 7/9
150.145 Abandoned - 13 414,611 0 Dead7/l
150.165 Abandoned - 24 611, 414, 0 Lost7/l9
612/21
150.175 Unsuccessful - 7 414, 611 0 Lost6/24
150.185 Successful 4 33 611,32,4l4 3 Lost7/22
150.204 Abandoned - 38 414, 611, 0 Lost 7/26
3281
150.274 Unsuccessful" 1 22 '611/21, 414 0 Lost 7/21
150.306 Successful 6 36 611,414,31 5 Lost7/26
150.515 Successful 10 41 31, 414, 2+ Lost8/4
611/622
150.546 Successful 3 11 611, 622, 414 0 Lost 6/27

 

‘1mernhranewuf0tmdinthisnestalongwithldeadehick. hismnnedthatnoehickswerefledgedfiumthisneat.
@Habitarclasaificatiom:21=agriar1tureaopland.3l=herbaceousopenfield,32=s1uubopenfield4l4=lowlandhardwood,61l=
foreateduetland,612-s1uubwetland,622=unergerawetlmd.

39

Table A2. Sizes (in ha) of home ranges for hens monitored in radio telemetry study. Home ranges were
calculated using harmonic home range formulas in Telem88. 50%, 80% and 95% contours of use were

 

 

calculated
Site Frequency Nesting Size of 50 % Size of 80% Size of 95 %
(hen id.) attempt contour (ha) contour (ha) contour (ha)
Maple River 148.496 Successful 82.90 165.69 327.41

148.834 Successful 78.43“ 135.25" 266.33
148.914 Successful 16.96 40.88 84.94
149.115 Successful 38.25 144.93 461.39
149.676 Successful 20.50 143.56 447.35
149.706 Successful 91.51 183.69 515.09
150.076 Unsuccessful 31.58 84.37 183.95
150.085 Successful 23.18 49.64 95.92“
150.355 Unsuccessful 31.22 89.97 189.97
150.476 Unsuccessful 36.47 82.48 185.52

Gratiot-Saginaw 150. 126 Successful 98. 12 255.50“ 347.20“
150.145 Unsuccessful 33.82 85.10 187.03"
150.165 Unsuccessful 1218.91 218725” 260109“
150.175 Unsuccessful l 1.95 47.09" 62.19‘
150.185 Successful 39.84 432.81 666.28“
150.204 Unsuccessful 51.61 " , 170.64 285.50
150.274 Unsuccessful 69.87 986.17 368904"
150.306 Successful 94.46 1660.90 4093.61
150.515 Successful 682.95" 2381.31“ 416927"
150.546 Successful 25.68 65.65 787.37‘I

 

 

:I'hese contours were made up of more than 1 polygon

APPENDIX B

Predictions of the fall flight of wood ducks with mechanistic model equations.

APPENDIX B

Predictions of the fall flight of wood ducks with mechanistic model equations

To predict the fall flight of wood ducks in Michigan, models will need to be
developed to expand results from the selected surveys across the area in question. As an
exploratory exercise, the results of the indicated pair calculations (Table 5) from the road
surveys were input into a model that estimates fall flight of wood ducks (Cleveland 1994).
The adult spring population size using road survey data was based on an assumption that
the size of the area sampled at each survey point was a 0.403 km (1/4 mile) and 0.201 km
(1/8 mile) radius circular plot. Plot sizes of 0.508 km2 and 0.127 km2 were multiplied by
the number of plots sampled in each survey and expanded across all sites, representing 723
km2 and 181 km2 over the 50,383 km2 survey zone. The adult spring population was

calculated using the following formula:

Adult spring pop. = (Total indicated birds) * (Total area of counties/Area surveyed).

Fall Flight was based on the variables identified in Table B 1.

The fall flight estimate for the 1/4 mile radius plot sample was much smaller than
the 1/8 mile radius estimate (Table 82). For 1995, the 1/4 mile radius estimate was
20,376 wood ducks in the fall flight, while the 1/8 mile radius was 81,504. The 1995
USFWS harvest estimate for the counties is 24,192 wood ducks. Obviously, the 1/4 mile
estimate for the fall flight must be low, with a deficit of 3816 wood ducks after harvest.

The USFW S harvest estimates are thought to overestimate the harvest (Pendleton 1992)

40

41

and some wood duck harvested may have been birds migrating through the area.
However, the 1/8 mile radius estimate may be more accurate with an estimate of 30 % of

the population harvested.

Table B1. Parameters and equations used as the basis for fall flight model (Walters et al.
1974)

 

 

Parameter Variables p

Production Rate (Eggs produced/Adult) ‘ (Surv. rate eggs to hatching) ‘ (Surv. rate chicks to
fledging) "' (Surv. rate through early flight period)

Production (New adults in spring ‘ Production rate for lst year) + (Old adults in spring ‘
Adult production rate)

Fall Adult Population (New and old spring adults) " (Adult summer surv. rate)
Fall Juvenile Population (Production) "' (Juvenile summer survival rate)
Fall Flight Fall adult population + Fall juvenile population

 

The calculations of variance were similar to those from Cleveland’s (1994) model,
with adjustments for the input of the road survey data. The variance of the fall flight was
the total of the variance from the adult fall population and the juvenile fall population. The

formula for the variance'of the Adult Fall Population was:

Var(Ad. fall pop.) - (Ad. spring pop.)’ ‘Var(Survival rate) + (Survival rate)’ 1"Var (Ad. spring pop.)

Since the survival rate was assumed to be 0.8 for both males and females, no variance was

42

Table B2. Estimates of parameters for fall flight model of wood ducks, using indicated
birds from 1995 and 1996 road surveys to replace inputs fi'om aerial surveys for adult
spring population calculations.

 

 

 

 

0.25 mi2 radius 0.125 mi2 radius
Model parameters Year Estimate Variance Estimate Variance
Adult spring population 1995 11,290 4.19 ‘ 10’ 45,150 6.71 "' 10‘
1996 15,570 4.67 " 105 62,280 7.47 " 10‘
Production’ 1995 12,610 2.33 ‘ 107 50,430 3.73 "' 108
1996 17,390 4.39 "' 107 69,570 7.02 " 10s
Adult fall population 1995 9,030 2.68 * 105 36,120 4.29 "‘ 106
1996 12,460 2.99 "‘ 10’ 49,820 4.78 * 10°
Juvenile fall population 1995 11,350 2.44 " 107 45,390 1.59 "' 108
1996 15,650 4.60 "' 107 62,610 3.00 * 108
Fall flight 1995 20,380 3.85 "' 107 81,500 1.63 " 10’
1996 28,110 7.24 "‘ 107 112,430 3.04 " 10I

 

I'The same production rate estimate fi'om Cleveland (1994) was used. Since the
production rate was in units of clutch size/hen, it was multiplied by only the number of
hens (half of the adult spring population).

43

estimated for this term, simplifying the formula to:

Var(Ad. fall pop.) - (Survival rate)’ *Var (Ad. spring pop.).

The variance of the adult spring population was calculated as:

Var(Ad. Spring pop.) 8 (Area of counties/area surveyed)’ " Varflndicated birds).

To calculate the variance of the indicated birds, the road surveys results were treated as a
binomial distribution with birds either observed or not observed at a sample point. The

formula for the variance of the adult spring population was:

Var(Ad. Spring pop.) =- (Area of counties/area surveyed)‘ "' (n)(p)(q).

where n = the number of sampling points, p = (sample points with birds observed/n), and q
= (sample points without birds observed/n).

The variance of the juvenile fall population was calculated as:

Var(Juv. fall pop.) - l/4(Prod.)’* Var(Surv. juv. female) + (Surv. juv. female)’ * Var(Production) +

1I4(Prod.)’* Var(Surv. juv. male) + (Surv. juv. male)’ * Var(Production).

The variance of the survival rates for juvenile males (0.00526) and females (0.00731) were

derived from the study on survival of post-fledgling wood ducks in Minnesota (Kirby

44

1990) used in Cleveland’s (1994) model. The variance of production was calculated as:

Var(Prod.) - ll4[(Ad. spring popJZ)z * Var(Prod. rate) +( Prod. rate)’ " ll4 Var(Ad. spring pop.)]

where the variance of the production rate is 0.714675 fi'om Cleveland’s (1994) model.

The variance of the fall flight was high for both density estimates for both years
(Table 10). Although the 1/8 mile radius may be a closer estimate to the population size,
it had a much larger variance than the 1/4 mile radius, due to the larger expansion factor
from the area surveyed to the total area represented in the survey. In both years, the 1/8
mile variance was about 16 times as large as the 1/4 mile estimate.

A number of issues must be clarified before these estimates can be verified.
Habitat quality and wood duck population density are confounded which makes the
expansion of the surveys to the total area unrealistic. Since there was significant variation
within the sites of the 1995 and 1996 road surveys, with Maple River consistently having
the highest number of wood ducks seen, the assumption of constant density is violated
immediately. Also, most of the road surveys are placed in areas with higher quality wood
duck habitat, along rivers (Maple, Mendon, Allegan, Muskegon l), or in state game areas
actively managing for wood ducks, at least the in the form of nest boxes (Maple River,
Allegan, Muskegon 2, Bald Mountain, Flat River, Gull Lake, Crane Pond, Somerset, Rose
Lake). The density of wood ducks in these areas is likely to be higher than in other areas.
In defense of the assumption, the remaining sites (Ann Arbor, Albion, Cambria, Hastings,
Fine Lake), which were set up to correspond with the flight surveys and had no actively

managed areas nearby, were not significantly difl'erent fiom the other sites. In fact,

45

Hastings ranked in the top 3 sites for the numbers of wood ducks observed. However, it
is highly unlikely that the density of wood ducks on the surveys is comparable to more
urban areas which are included in the calculations of area. A model that adjusts for the
variation in habitat quality and quantity would be more accurate.

Another problem with this model is the assumptions of size of the area surveyed.
No adjustment for visibility was made in the calculations of the adult spring population
from the indicated bird estimate from the road surveys. It is unlikely that the same sized
area would be surveyed at each site. Some areas are more forested with trees lining the
road, while agricultural areas have more open areas allowing for wood ducks to be seen at

much greater distances.

LITERATURE CITED

Bartlet, G. A. and R. E. Trost. 1990. Populations, harvest, and survival of wood ducks
in Wisconsin, Minnesota, and Michigan, 1961-87. Pages 341-348 in L. H.
Fredrickson, G. V. Burger, S. P. Havera, D. A. Graber, R E. Kirby, and T. S.
Taylor, eds. Proc. 1988 N. Am. Wood Duck Symp., St. Louis, MO.

Brakhage, D. H. 1990. Techniques currently used for monitoring wood duck
populations. Pages 201-203 in L.H.Fredrickson, G. V. Burger, S. P. Havera, D.
A. Graber, R. E. Kirby, and T. S. Taylor, eds. Proc. 1988 N. Am. Wood Duck
Symp., St. Louis, MO.

Bellrose, F. C. 1980. Ducks, geese, and swans of North America. Third ed. Stackpole
books, Harrisburg, PA 544pp.

Bellrose, F. C. and D. J. Holm. 1994. Ecology and management of the wood duck.
Stackpole Books, Mechanicsburg, PA. 588pp.

Cleveland, K. T. 1994. Predictive Models for the flight of selected waterfowl species in
Michigan. MS. Thesis, Michigan State University, East Lansing, Michigan. 91

PP.

Coleman, J. S. and Jones, A. B. 111. 1988. User‘s guide to TELEM88: Computer analysis
system for radio-telemetry data. Dept. of Fisheries and Wildlife, Virginia
Polytechnic Institute and State University, Blacksburg, VA. Research Series No.1 .
49pp.

Cottrell, S. D. and H. H. Prince. 1990. Comparison of wood duck survey techniques on
the Holston River in East Tennessee. Pages 201-203 in L. H. Fredrickson, G. V.
Burger, S. P. Havera, D. A. Graber, R. E. Kirby, and T. S. Taylor, eds. Proc. 1988
North Amer. Wood Duck Symp., St. Louis, MO.

46

47

Cottrell, S. D., H. H. Prince, and P. I. Padding. 1990. Nest success, duckling survival,
and brood habitat selection of ducks in a Tennessee Riverine System. Pages 191-
197 in L. H. Fredrickson, G. V. Burger, S. P. Havera, D. A. Graber, R. E. Kirby,
and T. S. Taylor, eds. Proc. 1988 North Amer. Wood Duck Symp., St. Louis,
MO.

Crissey, W. F. 1984. Calculators and Ouija Boards. Pages 259-271 in A. S. Hawkins, R.
C. Hanson, H. K. Nelson, H. M. Reeves, eds. Flyways: Pioneering Waterfowl
Management in North America. US. Dept. of Interior, F .W.S., Washington DC

Dahl, T. E. 1990. Wetland Losses in the United States, 1780's to 1980's, US. Dept. of
Interior, F .W.S., Washington, DC, 21pp.

Dolton, D. D. 1983. Mourning dove breeding population status, 1983. US. Dept.
Interior, Fish Wildl. Serv. 15 pp.

Gamble, K. E. 1990. Status of wood ducks in the Mississippi Flyway. Pages 115-119 in
L. H. Fredrickson, G. V. Burger, S. P. Havera, D. A. Graber, R. E. Kirby, and T.
S. Taylor, eds. Proc. 1988 North Amer. Wood Duck Symp., St. Louis, MO.

Geis, A. D. 1966. Establishing the Status of wood duck populations - successes and
problems (Information needs and procedures for establishing status of the wood
duck population. Pages 183-192 in J .B. Trefethen RD., ed., Wood Duck
management and Research: a symposium. Wildl. Manage. Inst, Washington DC.

Hein, D. 1966. Float counts vs flight counts as indices to abundance of nesting wood
ducks. J. Wildl. Manage. 30(1):13-16.

Kelley, J. R. Jr. 1994. Study proposal: Wood duck population monitoring initiative.
US. Fish and Wildl. Serv. Oflice of Migratory Bird Management. Unpublished.

50 pp.

Kelley, J. R Jr. 1996. Line-transect sampling for estimating breeding wood duck density
in forested wetlands. Wildl. Soc. Bull. 24(1):32-36.

48

Kelley, J. R. Jr. 1997. Wood duck population monitoring initiative: Final report. US.
Fish and Wildl. Serv. Oflice of Migratory Bird Management. Unpublished drafi.
207 pp.

Kirby, R. E. 1990. Survival of postfledging wood ducks in northcentral Minnesota.
Pages 185-189 in L.H.Fredrickson, G. V. Burger, S. P. Havera, D. A Graber, R
E. Kirby, and T. S. Taylor, eds. Proc. 1988 N. Am. Wood Duck Symp., St. Louis,
MO.

Martin, F. W., R S. Pospahala, and J. D. Nichols. 1979. Assessment and population
management of North American migratory birds. Pages 187-239 in J. Cairns Jr.,
G. P. Patil, and W. E. Waters, eds. Environmental Biomonitoring, Assessment,
Prediction, and Management-Certain Case Studies and Related Quantitative
Issues. International Co-operative Publishing House, F airland, MD.

Mississippi F lyway Council. 1994. A wood duck harvest strategy for the Mississippi
Flyway. Mississippi Flyway Technical Section. Unpublished. 16pp.

Pendleton, G. W. 1992. Nonresponse patterns in the Federal Waterfowl Hunter
Questionnaire Survey. J. Wildl. Manage. 56(2):344-348.

Prince, H. H. 1991. Species account: Wood duck (Aix sponsa) Pages 122-123 in R. C.
Brewer, G. A. McPeek, R. J. Adams, Jr., eds. The Altlas of Breeding Birds of
Michigan. Michigan State University Press, E. Lansing, MI.

Ryan, D. C., R. J. Kawula, and R. J. Gates. Unpublished report. Breeding biology of
wood ducks in natural cavities in southern Illinois. 31 pp.

SAS Institute Inc. 1989. SAS/STAT User’s Guide, Version 6, 4th edition, Volume 2.
SAS Institute Inc., Cary, NC. 1686 pp.

Sauer, J. R, and S. Droege. 1990. Wood duck population trends fi'om the North
American Breeding Bird Survey. Pages 162-168 in L. H. Fredrickson, G. V.
Burger, S. P.‘ Havera, D. A. Graber, R. E. Kirby, and T. S. Taylor, eds. Proc. 1988
North American Wood Duck Symp., St. Louis, MO.

49

Scott, V. E. And E. L. Becker. 1972. An evaluation of wild turkey call counts in
Arizona. J. Wildl. Manage. 36(2):628-630.

Semel, B., and P. W. Sherman. 1986. Dynamics of nest parasitism in wood ducks. Auk
103:813-816.

Semel, B., P. W. Sherman, and S. M. Byers. 1990. Nest boxes and brood parasitism in
wood duck: A management dilemma. Pages 163-170 in L. H. F redrickson, G. V.
Burger, S. P. Havera, D. A. Graber, R E. Kirby, and T. S. Taylor, eds. Proc. 1988
North American Wood duck Symp., St. Louis, MO.

Soulliere, G. J. 1990. Regional and site-specific trends in wood duck use of nest boxes.
Pages 23 5-244 in L. H. Fredrickson, G. V. Burger, S. P. Havera, D. A Graber, R
E. Kirby, and T. S. Taylor, eds. Proc. 1988 North American Wood duck Symp.,
St. Louis, MO.

Steel, R G. D., J. H. Torrie, and D. A Dickey. 1997. Principles and procedures of
statistics, a biometrical approach. The McGraw-Hill Companies, Inc., New York,
NY. 666 pp.

Straw, J. A. Jr. 1992. American woodcock harvest and breeding population status, 1993.
US. Dept. Interior, Fish Wildl. Serv. 15 pp.

Tautin, J., S. M. Carney, and J. B. Bortner. 1989. A national migratory gamebird harvest
survey: A continuing need. Trans. N. Amer. Wildl. Natur. Resour. Conf. 54:545-
551.

Tiner, R W., Jr. 1984. Wetlands of the Unites States: Current status and recent trends.
National wetlands inventory. US. Fish and Wildl. Serv., Washington DC. 59 pp.

Walters C. J., R Hilborn, E. Oguss, R M. Peterman, J. M. Stander, 1974. Development
of a simulation model of mallard duck populations. Canadian Wildlife Service
Occasional Paper 20. 34pp.

Wright, V. L. 1978. Causes and efl‘ects of biases on waterfowl harvest estimates. J.
Wildl. Manage. 42(2):251-262.

50

Zicus, M. C., and S. K. Hermes. 1987. Use of nest boxes to monitor cavity nesting
waterfowl populations. Wildl. Soc. Bull. 15:525-532.

kg..- -

MICHIGAN star: UNIV. LIBRARIES
llHlllllllllllllllllllllllllllllllllllllllllllllllllllllll
31293017821541