Y “nun?
‘ , .5. , ‘ . . ., rsw
a g. , :3 g _ . , _ , s
. A p.21, . ‘ ,, . 371.“ . ».._ .1! I... ‘ ‘ . .
.wmwwhw? . , ‘ £me a. a mu)... _
farmt . . . . mkw.§wwaflm.s
. . 7 Nat“? .62?

, A?
mafia“... . h
any 4% . . . , . . .. . .v .. . ‘ . . , www.isWm‘.
I :2. . . . . . . . .
. A),
a:

;. Gran;

.
4>~

gamma. , n, g

‘ :3.“ .m. Sufi ‘ . , ‘

9 2., Bar ‘ . (5.4:?
g .

any .i , ‘ . ‘ 25...“.

‘4 . . . .

.1 s
11:53..
.. . .1 .
{.13 .

ii...
3.4.x...)
E

‘ $1.:

1......»

 

twat:- .

: $5.).

1: .1. 3:. (l
mfi. “Aymra .
"flit Riki???
.r. 34:. a?
“‘¢1l.~‘- ,
£23131?
4 sun? fin.“
J... .- 01:4: ‘.
1 . ‘ {Adapt

.l:

 

 

,2;

‘3,

 

v.Lx.....

. 1x. b.7522: ‘ . V . wmstrfix .

. :32: 3.1.?

 

WES]:

pm

This is to certify that the
thesis entitled
An Evaluation of the Michigan Department of Natural
Resources' White-tailed Deer (Odocoileus
virginianus) Field Survey Methodologies
presented by

Sarah Laggner Cook

has been accepted towards fulfillment
of the requirements for

M.S. degreeinFisheries & Wildlife

 

 

gatnom

Major professor

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

 

 

 

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

 

Earl; gugm3| DATE DUE

4117051

0511383 02.
“ 233’
3—1

 

MAY 01,98‘ zansl

 

he 3'0““20052 (I;

 

 

 

 

 

 

 

 

 

 

 

 

am a/CWIS

AN EVALUATION OF THE MICHIGAN DEPARTMENT OF NATURAL
RESOURCES’ WHITE—TAILED DEER (ODOCOILEUS VIRGINIANUS) FIELD
SURVEY METHODOLOGIES
By

Sarah Laggner Cook

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

2001

ABSTRACT
AN EVALUATION OF THE MICHIGAN DEPARTMENT OF NATURAL
RESOURCES WHITE-TAILED DEER (ODOCOILEUS VIRGINIANUS) FIELD
SURVEY METHODOLOGIES
By

Sarah Laggner Cook

The Michigan Department of Natural Resources (MDNR) needs reliable data for
public accountability and sound scientific management of the state’s white-tailed deer
(Odocoz'leus virginianus) herd. Three white-tailed deer field surveys were examined to
evaluate the quality of the data collected. The biophysical data (biodata) collected from
harvested deer brought to voluntary check stations provides the largest available source
of data on Michigan’s deer herd, but biases, errors, and insufficient data limit the use of
the biodata. However, the biodata can be used in the sex-age-kill estimates of population
size, or to develop indices of herd health, track herd or harvest composition, and compare
the herd composition of different years or geographic areas. The lactation data do not
provide an estimate of annual recruitment, but they can be used to develop an index or
minimum estimate of reproductive success. The inconsistent manner in which the winter
severity index (WSI) data are collected prevents the index from being a useful tool with
which to predict winter mortality or yearling beam diameters. Standardizing the
collection process or developing an alternate WSI could make such predictions possible.
By improving the quality of the data collected in the field surveys (e. g. by increasing the
number of check stations or other methods), the MDNR will be better able to manage

Michigan’s deer herd and will increase public confidence in their management decisions.

To my parents,
David and Kathi Cook

Who give their children unending love and support

iii

ACKNOWLEDGMENTS

Financial support was provided by the Michigan Department of Natural
Resources (MDNR).

Several MDNR Wildlife Division personnel provided valuable comments and
assistance. Dr. Harry Hill, Dr. William Moritz, and John Urbain reviewed drafts of each
chapter and suggested new questions or new interpretations of the data. Brian F rawley
provided the mail survey harvest data, and Marshall Strong provided much needed
ArcView support.

I would especially like to thank Dr. Scott Winterstein, my major professor. He
has provided unending patience, advice, and support throughout the project, and I have
greatly enjoyed working with him. The other members of my committee, Drs. Henry R.
Campa, Harry Hill, and Robert Tempelman, also deserve thanks and acknowledgement
for their guidance.

Several students in the Department of Fisheries and Wildlife provided me with
support and assistance. Lori Corteville was especially helpful in introducing me to the
biodata and helping me learn SPSS. Brandi Hughey also provided much appreciated
assistance in literature searches, teaching me how to use ArcView, organizing the
biodata, developing the B-WSI, and in many other ways. Without her help, this project
would have taken much longer. I would also like to thank the many other graduate

students who answered my questions or provided me with support.

TABLE OF CONTENTS

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

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

CHAPTER 1

INTRODUCTION .................................................................................... 1

CHAPTER 2

THE BIODATA EVALUATION
Introduction ................................................................................... 4
Question 1 ................................................................................... 10
Question 2 ................................................................................... 28
Question 3 ................................................................................... 35
Question 4 ................................................................................... 41
Question 5 ................................................................................... 49
Question 6 ................................................................................... 58
Question 7 ................................................................................... 67
Question 8 ................................................................................... 77
Question 9 ................................................................................... 84
Question 10 ................................................................................ 104
Conclusion ................................................................................. 1 10

CHAPTER 3

THE LACTATION SURVEY EVALUATION
Introduction ................................................................................ 1 13
Methods .................................................................................... 114
Results ................................ 115
Discussion ................................................................................. 127

CHAPTER 4

THE WINTER SEVERITY INDEX EVALUATION
Introduction ................................................................................ 1 33
Methods .................................................................................... 134
Results ...................................................................................... 139
Discussion ................................................................................. 158

CHAPTER 5

CONCLUSION .................................................................................... 163

APPENDICES ..................................................................................... 167

LITERATURE CITED ........................................................................... 176

LIST OF TABLES

Table 1. Increase in the number of antlerless deer checked when antlerless hunting was
permitted on at least two-thirds more of the county in Year 2 than in Year 1.

Table 2. Percent of deer aged incorrectly by field personnel according to the ages
determined at Rose Lake in 1999. Bold lines indicate deer of several ages were grouped
in a single category.

Table 3. Locations of the check stations from 1996 through 1999.

Table 4. Change in the number of deer checked and harvested following the addition of a
check station to a county.

Table 5. Results of chi-squared tests comparing the 1999 distribution of the biodata to
the distribution of the harvest data within the counties of each management unit.

Table 6. The number (and proportion) of deer of each harvest-type category checked
from each region and season. The chi-squared tests compared the harvest composition of
the firearm and archery season in the biodata. All tests have 1 degree of freedom.

Table 7. The number (and proportion) of antlered deer of each age category checked
from each region and season. The chi-squared tests compared the antlered deer age
composition of the firearm and archery seasons in the biodata. All tests have 1 degree of
freedom.

Table 8. The number (and proportion) of antlerless deer of each age category checked
from each region and season. The chi-squared tests compared the antlerless composition
of the firearm and archery seasons in the biodata. All tests have 2 degrees of freedom.

Table 9. The number (and percent) of deer harvested and checked from private and
public lands in 1999.

Table 10. The 1999 numbers (and proportions) of antlered and antlerless deer harvested
on public land according to the mail survey and the number (and proportion) checked
from public land in the biodata. The chi-squared tests compare the public land harvest
composition as reported in the mail survey to that reported in the biodata.

Table 11. The 1999 numbers (and proportions) of antlered and antlerless deer harvested
on private land according to the mail survey and the number (and proportion) checked
from private land in the biodata. The chi-squared tests compare the private land harvest
composition as reported in the mail survey to that reported in the biodata.

Table 12. The 1999 numbers (and proportions) of antlered and antlerless deer of each

age category checked from private and public land. The chi-squared tests compare the
public land composition to that of private land.

vi

Table 13. The 1999 numbers (and proportions) of antlered and antlerless deer checked
from private and public lands. The chi-squared tests compare the private land harvest
composition to that of public lands.

Table 14. The number (and proportion) of deer checked from public and private lands at
highway and field stations. The chi-squared tests compare the land-type composition of
deer checked at the highway stations to that of those checked at field check stations
during the firearm season. All tests have 1 degree of freedom.

Table 15. The number (and proportion) of antlered and antlerless deer checked at
highway and field stations. The chi-squared tests compare the harvest-type composition
of deer checked at the highway stations to that of those checked at field check stations
during the firearm season. All tests have 1 degree of freedom.

Table 16. The number of yearling and adult antlered deer checked at highway and field
stations. (Proportions are in parentheses.) The chi-squared tests compare the age
composition of antlered deer checked at the highway stations to those checked at field
check stations during the firearm season. All tests have 1 degree of freedom.

Table 17. The number (and proportion) of fawn, yearling, and adult antlerless deer
checked at highway and field stations. The chi-squared tests compare the age
composition of antlerless deer checked at the highway stations to that of antlerless deer
checked at field check stations during the firearm season. All tests have 2 degrees of
freedom.

Table 18. Number and percent of checked deer aged as ‘A’ or ‘AA’ each year.

Table 19. Examples of some regional sample sizes from 1999 biodata.

Table 20. The number of counties missing data necessary to calculate the listed ratios,
and the number of counties, management units, and regions whose sample sizes are

sufficient to calculate the listed ratios with a margin of absolute error (E) of 5%, 10%, or
20%. All numbers are based on 1999 data.

Table 21. Examples of some management unit sample sizes from 1999 biodata.
Table 22. Examples of some county samples sizes from 1999 biodata.

Table 23. Crosstabulation of records between the ‘antler’ variable and the ‘killtype’
variable.

Table 24. Results of a chi-squared test comparing the proportion of lactating does in the
first week of firearm season to that of the second week of firearm season.

vii

Table 25. Average annual number of lactation records and the average annual margin of
error associated with the estimates of the percent of does lactating in each region and
management unit. The margins of error are based on absolute percentages. All values
are averaged from 1993-1999.

Table 26. Percent of does lactating from 1993 to 1999 in each region and associated
sample sizes. Data are from October.

Table 27. Percent of does lactating from 1993 to 1999 in each management unit and
associated sample sizes. Data are from Firearm Week 1.

Table 28. The mean corrected WSI values of the UP, averaged over the stations in the
Western UP and Eastern UP management units, and the coefficient of variation.

Table 29. The mean corrected WSI values of the Northern LP, averaged over the stations
in the Northwestern LP, the Northeastern LP, and the Saginaw Bay management units,
and the coefficient of variation.

Table 30. The mean corrected WSI values of the Southern LP, averaged over the stations
in the Southwestern LP, the South Central LP, and the Southeastern LP management
units, and the coefficient of variation.

Table 31. Results of the regression analyses of total annual WSI values and the first and
fourth month annual WSI values against several dependent variables.

Table 32. Results of regression analyses of monthly and annual B-WSI values against

yearling beam diameter. Monthly and annual B-WSI values were divided by 10 and 25,
respectively, to place them on a scale similar to that of the MDNR WSI.

viii

LIST OF FIGURES
Figure 1. Regional (dashed lines) and wildlife management unit (solid lines) boundaries.
Figure 2. Locations of the 1999 check stations. Highway check stations are circled.

Figure 3. Percent of harvested deer checked by county during 1999. The 1999 TB
surveillance counties are marked with a “*”.

Figure 4. Percent of all harvested deer checked from each management unit.

Figure 5. Total number of deer checked (solid line) and total number of deer harvested
(broken line) in each region.

Figure 6. Total number of antlered and antlerless deer checked and harvested in each
region.

Figure 7. Histograms of the number of counties whose difference between the harvest
estimates and biodata estimates of the antlerless to antlered ratio falls within the listed
categories.

Figure 8. Number of deer given younger ages at the check stations than at Rose Lake
(under-aged) or given older ages at the check stations than at Rose Lake (over-aged) in
1999.

Figure 9. Percent of does lactating in Michigan, averaged from 1993-1999 (using
October data). Error bars are :I: 1 standard error.

Figure 10. Average beam diameter of all antlered deer checked in Michigan, averaged
from 1992-1999. Error bars are i 1 standard error.

Figure 11. Percent of yearling does lactating in each region, averaged from 1993-1999
(data from first week of firearm season). Error bars are + 1 standard error.

Figure 12. Average yearling beam diameter in each region, averaged from 1987—1999.
Error bars are + 1 standard error.

Figure 13. The percent that each county contributed to the total number of deer checked
(solid) and harvested (striped) in each management unit in 1999.

Figure 14. Percent of deer checked during firearm and archery season and percent of
deer harvested during archery and firearm season in Michigan.

Figure 15. The percent of the firearm (solid line) and total (dashed line) biodata collected
at the highway check stations: Alma, Big Rapids, Birch Run, and the Mackinac Bridge
(added in 1999).

Figure 16. Histogram of the number of counties in the archery season, firearm season,
and all seasons combined whose difference between the harvest estimates and biodata
estimates of the antlerless to antlered ratio falls within the listed categories in 1999.

Figure 17. Distribution of public lands (shaded areas) in Michigan

Figure 18. The percent of the firearm (solid line) and total (dashed line) biodata collected
at the highway check stations: Alma, Birch Run, Big Rapids, and the Mackinac Bridge
(added in 1999).

Figure 19. Harvest locations of the deer checked at the four highway check stations in
1999. Approximate station locations are marked as white (Alma, Birch Run, Big Rapids)
or black (Mackinac Bridge) circles.

Figure 20. Age structure of the female and male deer classified as ‘A’ and aged at Rose
Lake (in 1999), excluding fawns.

Figure 21. Age structure of all female and male deer checked during 1999, excluding
fawns.

Figure 22. Percent of checked deer aged as (a) ‘A’ or (b) ‘AA’ during all seasons at
highway check stations (solid line) and field stations (broken line).

Figure 23. Number of deer checked from archery (solid line) and firearm (broken line)
seasons from 1987 through 1999.

Figure 24. Estimates of the number of deer harvested by county from the 1999 mail
survey.

Figure 25. Total number of deer checked from each management unit between 1987 and
1999.

Figure 26. Number of bucks checked in 1999 by township and range coordinates.

Figure 27. Percent of adult females lactating in Michigan from October to December,
averaged from 1993-1999. Error bars are :I: 1 standard error.

Figure 28. Percent of females in Michigan lactating in each age class (averaged from
1993-1999 using data from all seasons combined). Error bars are :t 1 standard error.

Figure 29. Percent of females lactating in each age class and in each region (averaged
from 1993 -1999 using October data). Error bars are + 1 standard error.

Figure 30. Percent of females lactating from 1993 to 1999 in each age class and in each
region (using October data). (Exact values and sample sizes can be found in Table 26)

Figure 31. Correlations between the annual average beam diameter and proportion of
lactating does (from October data) among yearlings. (UP: r2=0.3668, p=0.1495; NLP:
r2=0.1929, p=0.3241; SLP: r2=0.1044, p=0.4796).

Figure 32. Correlations between the annual average beam diameter and proportion of
lactating does (from October data) among adults (2.5+ years old). (UP: r2=0.8561,
p=0.0028; NLP: r2=0.0971, p=0.4964; SLP: r2=0.3909, p=0.1332).

Figure 33. The WSI station locations.

Figure 34. The availability of data from each WSI station from 1969 to 1998. The
station codes correspond to the codes from Figure 33. A station with usable data
indicates data were collected and used in the corrected WSI value. A station with
unusable data indicates data were collected but could not be used in the corrected WSI
value. A blank space indicates data were not collected that year.

Figure 35. The time period WSI data were collected in each region from 1969 to 1999.
The bold lines mark days 42 (December 13) and 168 (April 18).

Figure 36. The annual corrected WSI values recorded at Allegan Farm (solid line) and
Allegan Forest (broken line).

Figure 37. The (a) cumulative and (b) weekly WSI values from three selected years with
similar final WSI values in the Upper Peninsula.

Figure 38. Corrected annual WSI values from 1977 to 1999 in the Upper Peninsula,
Northern Lower Peninsula, and Southern Lower Peninsula.

Figure 39. Corrected annual WSI values of the Western UP and the Eastern UP.

Figure 40. Corrected annual WSI vaers of the management units of the Northwestern
LP, the Northeastern LP, and Saginaw Bay.

Figure 41. Corrected annual WSI values of the management units of the Southwestern
LP, the South Central LP, and the Southeastern LP.

xi

CHAPTER 1
INTRODUCTION

In 1996, the Michigan public passed Proposal G, a ballot initiative calling for the
scientific management of Michigan’s wildlife resources. White-tailed deer (Odocoileus
virginianus) are one of Michigan’s largest wildlife resources, and the Michigan
Department of Natural Resources (MDNR) management of the deer impacts virtually
every Michigan resident. The MDNR has found that Michigan residents have different
opinions of and goals for the deer population (W. Moritz, MDNR, personnal
communication). Many hunters prefer a large population of healthy animals with a large
proportion of older, larger bucks. Farmers want low deer densities to reduce crop
damage. The general public values deer viewing opportunities, but people do not want
the deer eating their ornamental shrubs or causing car-deer collisions. With so many
different opinions on the desired size and condition of rhe deer herd, the MDNR needs
sound scientific data to develop effective management strategies and to justify its
management techniques to the public.

The MDNR collects data on the Michigan deer herd through several different
field survey techniques. The annual field surveys of the white-tailed deer population of
Michigan allow the MDNR to estimate population size, predict changes in population
size due to harvests and winter mortality, and to gather information on the composition
and health of the Michigan deer herd. Field surveys are techniques used to collect data
fi'om the deer herd to estimate population parameters and to develop population indices
(relative measures of population parameters) of wildlife species based on statistical

principles. Field surveys allow wildlife managers to approximate the true population

parameters based on results of sampling a fraction of the population. The field survey
data, however, cannot be used confidently unless they are collected in a consistent and
statistically valid manner. I evaluated the procedures used in data collection and analysis
of 3 field surveys (the biodata, the lactation survey, and the winter severity index) to
determine if they provide the most accurate data possible. I also evaluated the statistical
validity of the data at the relevant scales.

The results of the 3 different evaluations are presented in the following 3 chapters.
The second chapter of this thesis discusses the evaluation of the check station data. The
MDNR runs the voluntary check stations during the deer hunting seasons to collect
biophysical data (known as the biodata) on the harvested deer. Chapter 2 is divided into
several sections. The first section describes the biodata and the purpose of the evaluation.
The following sections present several questions which the evaluation addressed, the
methods used to answer each question, and the results of the evaluation. The final
section draws conclusions about the quality of the biodata and their potential uses and
suggests potential improvements in data use and collection.

The third chapter discusses the evaluation of the lactation survey. The lactation
data are collected as part of the biodata, but these data were evaluated separately.
Chapter 3 is comprised of 4 sections, organized as a traditional research report. An
Introduction describes the lactation data and the purpose of the evaluation. The Methods
section describes the methods used to evaluate the lactation survey, and the Results and
Discussion sections present the findings of the evaluation and suggestions for the

improvement and use of the lactation data.

The fourth chapter presents the evaluation of the winter severity index (WSI).
The MDNR collects weather data throughout the winter to try to predict the impact of the
winter severity on the deer population. The Introduction of Chapter 4 describes the WSI
and the purpose for its evaluation. The Methods section describes the methods used to
collect the WSI data and the methods used to evaluate the data. The Results and
Discussion sections present the findings of the evaluation, the potential uses of the WSI
data, and suggestions for the improvement of the WSI.
Each chapter addresses specific objectives for the individual evaluation, but for
each survey, my objectives were to:
1. evaluate the procedures used in data collection and analysis to determine if they
provide the most accurate data possible;
2. evaluate the statistical validity of the data at the relevant scales;
3. determine if the survey provides the data necessary to accurately satisfy its
present uses; and
4. make recommendations on the improvement of the current data collection and
analysis procedures or on the development of new procedures.
By meeting the above objectives, each evaluation will insure the quality of the data and
the analyses in which the data are used. The recommendations will also suggest more
accurate and efficient survey methodologies, resulting in a more effective use of MDNR

I'CSOUI'CCS.

CHAPTER 2

THE BIODATA EVALUATION

Introduction

Michigan’s current white-tailed deer population estimate greatly exceeds the
MDNR’S deer population goal of 1.3 million deer (W. Moritz, MDNR, personal
communication). The current estimated population of approximately 2 million deer is
split fairly evenly among Michigan’s three regions (MDNR 2000a; see Figure 1 for
regional boundaries). The primary means of managing the deer herd to achieve the
desired population goal lies in regulating the annual deer harvest, during which Michigan
hunters have recently harvested more than 500,000 deer in a single year (F rawley 2000).
For the 2000 deer hunting season, the MDNR created harvest regulations to try to meet a
target harvest ratio of 3 antlerless deer for every 2 antlered to reduce population growth
(MDNR 2000b). The annual harvest is divided among several different seasons, which
always include the split archery season from October 1 through November 14 and
December 1 through early January, the firearm season from November 15-30, and the
muzzleloader season in early December in the Upper Peninsula (UP) and mid December
in the Lower Peninsula (LP) (MDNR 2000b). Recently the MDNR has also instituted
some special Early and Late Firearm seasons for antlerless deer only to try to further
reduce the deer population in areas with especially high deer densities.

The MDNR needs a source of biological information to create a profile of the
annual deer harvest, draw inferences about the state’s deer herd, develop population
estimates, and examine the effects of current management practices. The largest source

of such data is the biophysical data, known as the biodata, collected from voluntary deer

 

 

 

    

Northern Lower
Peninsula

 

 

 

 

 

 

 

8; new Bay

 

m
m

m .... .. out -
Southern Lower .. .. 5mm“ __°a.; ....

Peninsula

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Figure 1. Regional (dashed lines) and wildlife management unit (solid lines)
boundaries.

check stations. The goal of the voluntary check system is not to develop a harvest
estimate, but rather to collect information on the harvest composition and the biological
characteristics of the deer herd. Hunters in Michigan are not required to register their
deer. Instead, they are encouraged to voluntarily bring their deer to a MDNR check
station where the biodata are collected. To encourage participation in the voluntary deer
checking system, the MDNR provides a patch for every deer a hunter brings to a check
station. The check stations are distributed across the state. In 1999, the MDNR
maintained 4 highway stations and 75 field stations at field offices and state parks, game
areas, and recreation areas (Figure 2). During several days of the firearm season, the four
highway check stations (circled in Figure 2) are located along three major southbound
arteries and at the Mackinac Bridge. The other check stations are scattered throughout
the state. The exact number and location of check stations varies slightly from year to
year and has generally been increasing.

The check stations are run by MDNR employees and other volunteers who have
participated in annual training sessions. The check station workers are trained to collect a
variety of data on each deer brought into the check station. Data are collected from each
hunter and deer on the location and season of harvest, as well as the sex, age, antler size,
lactation status (since 1993), and bovine tuberculosis (TB) status based on chest cavity
inspection (since 1996). An example of a check station data sheet is in Appendix 1.
Once all of the season’s data are collected, they are transcribed into an SPSS data file.
Descriptions of each variable in the SPSS data file and the data contained within them
can be found in Appendix 2. These data provide a wealth of information about the

Michigan deer herd and annual harvest.

    

             
   

      

 

   

      

 

 

 

 

 

 

    

 

 

 

                 

 

 

 

 

 

 

 

 

 

 

 

  

 

 

 

 

 

 

 

-
u
.r
u
.n. ...... .. ...---1 .
M. ”.0 iii. "
4- _ 4.-- _ ..- ..... .-
1 u m on." 3
nnnnnn no 0 _ 0mm 4.111” c m" i.
w. o m o ._ ..lm.nu.uJ-I--.MJ o "m «a ...... LU ..... 0
1.1.1: t m o m n u s 1 .u lit-
.I u. r w v m" n
n m" a u 0 m V] rill” m
. O u. m n. u w 71 _ In
_ :- o . o . . KK . M _. -lI
. T it i. a a p n 0 III . IIIIIL
3 ,. .1 w w r
a... . _ u . m .
t. . L W will” . l
.......... l A A o u . .rlial.lil
. _
. o W m .n
n n O IIWIIIIIo m
I u . O. . m
up w w 0 n m a
a nnnnnnn O" n 1 O u C
u . r». u“ .
_ ‘
n u n
n . in:
r
H
J
u
—
_.
~

aw.

Figure 2. Locations of the 1999 check stations. Highway check stations are

circled.

Harvest data are also collected through the annual mail survey of Michigan deer
hunters. The mail survey is sent to thousands of randomly selected people who
purchased Michigan deer hunting licenses. In 1999, the survey was sent to 5.7% of
license buyers and had a 76% response rate (F rawley 2000). The survey asks for
information on where the respondent hunted and what the respondent harvested. The
statistical design of the random sampling method of the mail survey ensures that it
provides a more accurate representation of the deer harvest than do the voluntary check
stations, which are not a statistically designed, random representation. The mail survey
does not provide as much data on each deer as the biodata, however. For example, the
mail survey data does not include information on the age of the deer harvested.
Therefore, although the MDNR also collects information on the deer herd using summer
herd observations, winter dead deer surveys, deer-vehicle accident reports, and spring
pellet surveys, the biodata represent the most comprehensive source of data on the herd.

This chapter describes the results of an evaluation of the MDNR check station
data collected between 1987 and 1999. The quality of these records is first dependent on
the procedures used to collect the data. The procedures should insure that the data are
random and accurately represent the true state of the parameter they are used to estimate.
Biases in the data could result in inaccurate estimates. The distribution of the biodata is
entirely dependent on the distribution of harvested and checked deer, but hunters do not
tend to harvest a strictly random sample of the deer population. For example, hunters are
more likely to harvest male fawns than female fawns (Coe et al. 1980). Hunters are also
more likely to check older, antlered deer than younger does (Bull and Peyton 2000).

Sampling regimes that favor a particular geographic area, age class, or sex will provide

biased data. I evaluated the methodologies used to collect the biodata to identify biases

or sources of error in the biodata.

Question 1: Do the biodata represent the composition of the true harvest?

Significance
The biodata would not reflect the true composition of the harvest if they were

susceptible to selection bias or measurement error. The biodata are collected from deer
checked at voluntary check stations. The checked deer are therefore not likely a random
representation of the true population of harvested deer. Due to hunter attitudes, certain
sex and age classifications may be more likely to be checked than others. A pilot study
(Bull and Peyton 2000) of Michigan deer hunters suggested that larger and older deer are
'more likely to be checked than smaller, younger deer. Hunters are also more likely to
check antlered deer than antlerless deer (Bull and Peyton 2000). The biodata are also
susceptible to measurement errors. For example, Ryel et a1. (1961) found that Michigan
check station agers tend to under-age older deer (4.5 years old and older) and over-age
younger deer. Although the mail survey is probably a more accurate representation of the
harvest, it does not contain as much information as the biodata. The biodata are the only
data available that provide detailed information about the composition and biological
condition of the deer herd and harvest. It is necessary, therefore, to identify possible

sources of bias and error within the check station survey process.

Methods

Although there is no census of harvested deer, the MDNR conducts an annual
mail survey to provide an independent and more reliable and random representation of
the annual deer harvest than the check station survey. I compared the biodata to the data

collected in the mail survey (hereafter called harvest data) to compare the harvest

10

composition as represented by both surveys. I had to first assume that the mail survey
provides an unbiased, random representation of the deer harvest. The random sampling
procedure used to distribute the mail survey (F rawley 2000) should ensure that the
harvest data meets these assumptions. I was able to calculate the percent of deer checked
from each county by dividing the number of deer checked from a particular county by the
mail survey’s county harvest estimate. Using the biodata and the harvest data, I
calculated antlerless to antleredl ratios (the only designations available in the harvest
data) to compare the composition of the checked deer to the composition of the harvested
deer. By comparing these two surveys, I was able to identify hunter biases toward
checking certain deer and geographic biases in the collection process.

The second possible source of error is in measurement methods. I had little
opportunity, however, to check the accuracy of the aging, sexing, and measuring involved
in the data collection process. The sole opportunity to check the aging process was in
comparing the ages determined in the field to the ages determined at the Rose Lake
Wildlife Research Station when deer were tested for TB in 1999. Deer taken for TB
testing are first sent to Rose Lake where they are aged by highly experienced agers. At
both the check stations and Rose Lake, the deer are aged using the tooth wear and
replacement patterns described by Severinghaus (1949). The Rose Lake agers have more

experience than the field agers, however, and I assumed the age determined at Rose Lake

 

‘ Unless otherwise noted, all ‘bucks’ and ‘antlered’ deer are deer considered antlered according to a
definition set by MDNR harvest regulations (used in the mail survey), rather than the biological definition.
The differences between these definitions primarily affect yearling (1.5 year old) males whose antlers may
be either short spikes or true antlers. The harvest definition defines a buck as any antlered deer with at least
one antler greater than 3 inch spikes, placing some yearling males in the antlerless category. The biological
definition describes a buck as any male deer that is not a fawn, placing all yearling males in the buck
category.

11

was the more accurate age. Comparing the two ages allowed me to estimate the aging

error of the field data within several different age categories.

Results and Discussion

The counties of the UP and Northern LP checked the greatest percent of harvested
deer, especially those counties with intensive TB surveillance (Figure 3). The percent of
deer checked peaked in 1994 (Figure 4) when the MDNR celebrated 100 years of
licensed deer hunting in Michigan (MDNR 1994). Since dropping again in 1995, the
percent of harvested deer checked has increased dramatically in the Northeast LP (Figure
4). The counties with the lowest percentages of deer checked in the Southern LP were
also generally those counties without check stations (mostly along the border of the
Southwestern management unit) (Figure 3 compared to Figure 2). The recent emphasis
on checking deer from the Northeastern LP for signs of TB (since the TB testing began in
1996) has caused the higher checking rates in the Northern LP. The MDNR sent
mailings to hunters in the TB surveillance region asking them to check their deer (W.
Moritz, MDNR, personal communication). The TB outbreak was probably also at least
partly responsible for the 1999 increase in the percent of deer checked from all
management units (Figure 4) as hunters wanted to assure themselves that their deer were
not diseased

The greater checking rates in the northern regions of the state and the severe
under-representation of the deer from the Southern LP in the biodata will cause statewide
data to be biased in favor of the northern areas. The under-representation of the Southern

LP also suggests that hunters from the southern counties were less likely to check their

 

 

% checked
[:1 < 5%

5 - 10%
10 - 20%
20 - 30%
> 30%

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Figure 3. Percent of harvested deer checked by county during 1999. The
1999 TB surveillance counties are marked with a “*”.

l3

-—X—West UP - i- East UP - - O - -Northwest LP

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

— + . Northeast Lp — o- - Saginaw Bay - ‘A - Southwest LP
- - O - - South Central LP —I— Southeast LP
20
18 [:1
I/’
16 71
l/’
14 h
'3 f\\ ;I- - — - 4
g 12 I V. /’ /
. \
a \ . ' /
U 10 ‘ - . X, ,/ .1 I"
~ ‘ \ . / \ ¥ /
fl \ /
E s ’\ * - u .. 4 \\
u . v \ ’ I ’ ,
m ..... ~ ‘oyu/A. ‘ ’x— -’-' *. 0/0
6 . . I h u ‘ - - ‘~" ‘ ‘ a '
’0 ‘-‘K ". ...... ‘b‘~L-:‘;’"‘i
4 . - . - t " . . l ‘ ‘..
' . ~.' """ O ............
2
o r 1 7 1' r fi' T I'
1990 1991 1992 1993 1994 1995 1996 1997 1998 1999
Year

Figure 4. Percent of all harvested deer checked from each management unit.

deer, therefore not providing enough data to analyze much of the Southern LP on the
county level (see Question 9). Many of the southern counties could provide adequate
sample sizes for county-level analyses, if a greater proportion of hunters checked their
deer from these counties. The tendency of southern Michigan hunters to check their deer
may increase, however, if the number and convenience of check stations increases (see
Question 3). The disproportionate number of deer checked from the TB counties will
cause the data for the Northern LP region and the Northeastern LP management unit to be
biased toward the TB counties. If the deer of the TB positive counties are not similar to
the deer of the entire Northern LP, regional averages will reflect the status of the
Northeastern LP, rather than the entire region. The biodata’s bias toward the TB counties
could therefore be of some concern when examining the data at the management unit or
regional scales.

I next compared the annual trends in the number of deer checked within each
region to the estimated harvest numbers. Within the UP, the trends deviated from one
another during the early 1990s, but recently have closely tracked one another (Figure 5a).
In the Northern LP, the harvest and check numbers follow the same general trend, but
their relative positions reversed in 1999 (Figure 5b). In the Southern LP, the number of
deer harvested increased dramatically beginning in 1995, but the number of deer checked
remained steady (Figure 5c). The lack of sufficient check stations in the Southern LP
may explain the absence in a corresponding increase in the number of deer checked as the
harvest numbers increased (Figure 2). The number of deer checked in the Southern LP is
also significantly less than 10% of deer harvested in every year, although in the UP and

Northern LP the number of deer checked is only slightly less than or even greater than

(a) Upper Peninsula

 

 

 

 

 

 

Number Checked

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

12000 120000
10000 100000
8000 80000 E
E
e-
(b
"I
6000 60000 a
w
2
G
v:
4000 40000 8
D.
2000 20000
O 1 l l -' 1 i i 1 O
1990 1991 1992 1993 1994 1995 1996 1997 1998 1999
Year
(b) Northern LP
25000 250000
20000 200000
2
E t - 5
35000 , ’ 150000g
5 a
I... E
Q 9
@0000 100000?b
z a.
5000 50000
0 l 1 1 1 i. 1 i E 0
1990 1991 1992 1993 1994 1995 1996 1997 1998 1999
Year

Figure 5. Total number of deer checked (solid line) and total number of deer
harvested (broken line) in each region.

16

Number Checked

(0) Southern LP

 

 

 

 

 

 

 

 

 

 

 

35000 350000
30000 7‘ , 300000
I
25000 to 250000
I. '
20000 , 200000
0
15000 " "e , - 9 " 4 150000
0
10000 100000
5000 50000
0 l l 4- 1 i l + i 0
1990 1991 1992 1993 1994 1995 1996 1997 1998 1999
Year

Figure 5 (cont'd).

DOJSQMBH .IaqurnN

10% of the deer harvested (Figures 5a-c). The relationship between the total number of
deer checked to the total number of deer harvested is very similar to the relationship
between the number of antlered or antlerless deer checked and the number of antlered or
antlerless deer harvested (Figures 6a—c). Any discrepancies between the trends of the
number of deer checked and harvested are therefore due to differences in both the
antlered and antlerless deer numbers.

The fairly consistent relationship between the number of deer checked and
harvested in the northern areas of the state suggests that the biodata from these regions
have represented a similar subset of the harvest over the past ten years. Annual data are
therefore comparable to one another. The parallel patterns also indicate that the biodata
were not dependent on the same group of hunters each year. As more hunters were
successful, more hunters checked their deer. The dramatic increase in the harvest data,
accompanied by only a slight increase in the check data from the Southern LP indicates
that recently a large portion of the harvest has gone unchecked. As this was not the case
during the first half of the decade, the earlier data may not be comparable to the most
recent data.

Although the hunters may have checked an adequate percentage of the harvested
deer to theoretically represent the true harvest, the distribution of the checked deer did
not appear to be random and the composition of the deer harvested may not be equivalent
to the composition of the deer checked. To address this issue, I compared the harvest
data’s and biodata’s antlerless to antlered ratios. Although the MDNR uses only the data
from the yearling population to calculate the adult buck and doe composition, the total

antlerless to antlered ratio was the only ratio I was able to calculate using the harvest

l8

+Antlered Checked +Antlerless Checked
- .- Antlered Harvested ' 'X - Antlerless Harvested

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

(a) Upper Peninsula
8000
7000
!L
6000 , - ~ 2
g I . E
f, 5000 ’ t 5
0 er
I: 0
U "I
5 4000 E
.9 x _ 2
E 3000 ,- ar o
:1 I \ '3
Z ’ \ CD
a.
2000 -
1000
0 1 1 1 1 1 1 1 1 O
1990 1991 1992 1993 1994 1995 1996 1997 1998 1999
Year
(b) Northern LP
14000 140000
12000 120000
.51 1000002
.93 :1
o E
,2 80000 g
D
33 E
.9 60000 3
E o
g a
40000 a
20000
O 1 1 1 1 1 1 1 1 0
1990 1991 1992 1993 1994 1995 1996 1997 1998 1999
Year

Figure 6. Total number of antlered and antlerless deer checked and harvested in
each region.

(c) Sonthern LP

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

18000 75 180000
16000 . 160000
AK
14000 w 140000
I I . ~

3 12000 1 I -. I 1 f 120000

:1 I I ‘ I

8 , " I

.1: 10000 , .I 100000

0 j ,3"

1... . I

3 8000!. ‘f- _., 80000

a 1 ~ —*

z 6000 60000
4000 40000
2000 20000

0 1 1 1 1 1 1 1 1 0
1990 1991 1992 1993 1994 1995 1996 1997 1998 1999
Year
Figure 6 (cont'd).

20

parse-urn .raqrnnN

data. If the biodata and the harvest data represented the same population of harvested
deer, the shape of the distributions in Figure 7 would be normal; there should be some
variation, but the distribution should peak at 0 (the ratios are the same) and fall off
symmetrically on either side. Such a pattern would indicate that the ratio of antlerless to
antlered deer was similar between each county’s biodata and harvest data. The
distribution of the difference between the harvest ratio and the checked ratio tended to be
skewed to the right, however, indicating that counties more frequently had a higher
harvest antlerless to antlered ratio than checked ratio (Figure 7). Hunters, therefore,
checked a greater proportion of antlered deer than antlerless ones. The biodata do not
appear to represent the same population of deer as the complete harvest, which may
provide an inaccurate view of the state of the deer herd. Management decisions based on
information contained within the biodata could be affected, leading to undesired effects
on the composition of the deer herd.

In one instance I was able to check the biodata against itself. I looked at the trend
of the number of antlerless deer checked as the number of antlerless permits increased
(Table 1). When an entire county issued antlerless permits following a year in which no
antlerless permits were issued (except block permits) the number of antlerless deer
checked increased an average of 622%. When the area of a county that issued antlerless
permits increased by at least two-thirds of the county’s area, the number of antlerless deer
checked increased an average of 261%. These results suggest that the biodata do not just
include records from the same successful hunters every year. When more antlerless
permits were available, more hunters were successful, and the number of checked

antlerless deer increased.

21

(1)1990 (11) 1991

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

   

35 35
3 30
E
:1 25
O
U 20
S
g 15
a 10
5
z 5 -
O .1
< -0.4 -0.2 0.0 0.2 > 0.4 < .04 -0,2 0,0 02 > 04
Harvest Ratio - Checked Ratio Harvest Ratio - Checked Ratio
(iii) 1992 (iv) 1993
35 35
330 a 30
1: a
I:25 5 25
520 5 20
“a “a
I. 15 1... 15
.8 .8
a 10 a 10
= 5
z 5 Z 5
0 0
< -0.4 -0.2 0.0 0.2 > 0.4 < -0.4 -0.2 0.0 0.2 > 0.4
Harvest Ratio - Checked Ratio Harvest Ratio - Checked Ratio
(v) 1994 (vi) 1995
35 35
g 30 g 30
g 25 E 25
U 20 U 20
"5 “a
I- 15 1.. 15
0 0
'2 10 '5 10
:2 5 2 5
O O
< -0.4 -0.2 0.0 0.2 > 0.4 < .04 -02 0,0 02 > 04
Harvest Ratio - Checked Ratio Harvest Ratio - Checked Ratio

Figure 7. Histograms of the number of counties whose difference between the
harvest estimates and biodata estimates of the anterless to antlered ratio falls
within the listed categories.

22

(vii) 1996
35

30
g
5 25
9
o 20
h-

 

 

 

 

 

 

 

 

< -0.4 -0.2 0.0 0.2 > 0.4
Harvest Ratio - Checked Ratio

(ix) 1998
35

830
"2
=.25
O
020
‘5'
I-.15
0

£10

 

 

 

 

 

 

 

 

 

< -0.4 -O.2 0.0 0.2 > 0.4
Harvest Ratio - Checked Ratio

Figure 7 (cont'd).

23

(viii) 1997
35

830
'g
=25
c
020
"e"
11.15
0

510

b

Nu

 

0

< -0.4 -0.2 0.0 0.2 > 0.4
Harvest Ratio - Checked Ratio

(x) 1999
35

830
E
=25
o
020
"5
1.15
B
510
:1
Z 5—
0-

< -0.4 -0.2 0.0 0.2 > 0.4
Harvest Ratio - Checked Ratio

 

 

 

 

 

 

 

 

Table 1. Increase in the number of antlerless deer checked when antlerless hunting
was permitted on at least two-thirds more of the county in Year 2 than in Year 1.

 

 

 

 

 

 

County Year 1 Year 2 Area Yr 11 Area Yr 22 % increase3
Mason 1994 1995 0 1 206.25
Mecosta 1994 1995 0 1 796.55
Newaygo 1994 1995 O 1 285.96
Emmett 1994 1995 0 1 1200.00
average 622.19
Gogebic 1998 1999 0 2/3 185.71
Iron 1998 1999 0 2/3 436.59
Kalkaska 1994 1995 0 2/3 100.00
Lake 1994 1995 0 2/3 72.22
Mackinac 1998 1999 0 2/3 1171.43
Missaukee 1994 1995 0 2/3 192.50
Ontonogon 1998 1999 0 2/3 217.65
Houghton 1998 1999 0 2/3 500.00
Muskegon 1987 1988 1/3 1 0.00
Osceola 1994 1995 1/3 1 128.89
Montmorency 1995 1996 1/3 1 405.45
Wayne 1997 1998 1/3 1 0.00
Otsego 1998 1999 1/3 1 175.73
Alpena 1994 1995 1/3 1 109.52
Benzie 1994 1995 1/3 1 537.50
Dickinson 1994 1995 1/3 1 174.26
Iron 1994 1995 1/3 1 411.50
Cheyboygan 1995 1996 1/3 1 156.25
Manistee 1995 1996 1/3 - 1 250.00
Benzie 1997 1998 1/3 1 37.50
Cheyboygan 1997 1998 1/3 1 43 1 .82
Grand Traverse 1997 1998 1/3 1 45.00
average 260.89

 

l. The area of the county for which antlerless permits were issued in Year 1.
2. The area of the county for which antlerless permits were issued in Year 2.
3. The percent increase in the number of antlerless deer checked from Year 1 to Year 2.

24

The final check of the biodata against the true harvest was the check of the
accuracy of the aging techniques. When compared to the age determined at Rose Lake
(assumed to be the true age), the age determined in the field decreased in accuracy with
increasing age (Table 2). Field personnel always aged fawns correctly and aged yearlings
correctly 94.85% of the time (male ages may be more accurate than female ages due to
field agers use of antlers as additional evidence). When separated into their individual
age categories, deer 2.5 years old or older were aged incorrectly more than 25% of the
time in almost all categories. Deer 2.5 years old or younger tended to be overaged
(assigned higher than true ages) at the check stations, while older deer tended to be
underaged (assigned lower than true ages) at the check stations (Figure 8).

The field—determined ages could be used to lump the older deer into a single 2.5+
category with 93.35% accuracy (males ages were less accurate than female ages in this
category). Although reducing the number of age categories to three (fawns, yearlings,
and 2.5+) would potentially increase the accuracy of the aging process, knowing the
complete age structure of population is still necessary to address landscape level issues.
Determining the true age of the deer may also be important for public relations issues,
because many hunters are interested in knowing the exact age of their deer (H. Hill,

MDNR, personal communication).

25

Table 2. Percent of deer aged incorrectly by field personnel according to the
ages determined by Rose Lake personnel in 1999. Bold lines indicate deer
of several ages were grouped in a single category.

 

Age Male Female Total
(in field): N % Incorrect N % Incorrect N % Incorrect

 

 

fawn 31 0.00 24 0.00 55 0.00
1 547 2.93 171 12.28 718 5.15
2 120 26.67 140 30.00 260 28.50

2+ 170 15.88 386 2.59 556 6.65
3 40 40.00 96 29.17 136 28.46
3+ 50 28.00 246 7.72 296 l 1.1 5

4 7 71.43 60 43.33 67 46.27
5 2 50.00 33 36.36 35 37.14
6+ 1 100.00 57 17.54 58 18.97

 

 

 

 

 

26

 

iUnder—aged IS Over-aged

 

 

50

45

 

4o

 

 

 

 

35

 

 

30

25

 

20

 

Number of Deer

 

15

10

 

 

 

 

//////////////

 

N

 

 

1.5 Yrs

2.5 Yrs

 

 

 

 

 

 

3.5 Yrs
‘ Rose Lake Age

 

 

4.5 Yrs

 

 

 

 

5.5 Yrs

Figure 8. Number of deer given younger ages at the check stations than at Rose
Lake (under-aged) or given older ages at the check stations than at Rose Lake

(over-aged) in 1999.

27

 

Question 2: Do the biodata represent the composition of the true population?

Significance

Data collected from harvested deer, whether at the voluntary check stations or the
more structured mail survey, are not necessarily representative of the entire Michigan
deer herd. Hunting regulations strictly control the number of antlerless deer that can be
harvested. Hunter attitudes also affect harvest composition. Many hunters would rather
take a “big buck” than a fawn or a doe (Bull and Peyton 2000). Some deer are also more
susceptible to harvest than others. Males may be more susceptible to harvest due to
larger home ranges and greater movements (Roseberry and Klimstra 1974). Lone fawns
may also be highly susceptible to harvest (Coe et al. 1980). The biodata collected from
the harvest may, therefore, not be useful for creating estimates of the true population

parameters, but the data may be useful to indicate general trends in the herd.

Methods

If the biodata represent even a moderately accurate picture of the state’s herd,
they should present general trends in antler size and the percent of does lactating
(hereafter refered to as the lactation rate for simplicity) that the literature suggests is true
of deer biology. I examined the lactation trends of the checked deer by age class and
geographic region to compare them to what the literature suggests the true trends should
be. I similarly examined beam diameters of checked deer. If the patterns presented in the
literature are reflected in the biodata, then the biodata could possibly provide an indicator

of general trends among the population as a whole. Eberhardt (1960) also developed a

28

method to check for the accurate representation of fawns in the harvest by comparing the
female fawn to 1.5+ doe ratio of one year to the 1.5 doe to~2.5+ doe ratio of the following
year. I was unable to perform Eberhardt’s (1960) analysis, however, because necessary
assumptions were not satisfied. First, Bull and Peyton (2000) found that hunters are less
likely to check fawns than older deer so the biodata would not represent the proportion of
fawns in the true harvest. The analysis also assumes that mortality rates are consistent
from 6 months to 18 months and from 18 months to 30 months. F awns are more
susceptible to predation and harsh winter conditions than older deer, however. I was
therefore unable to check how well the biodata may represent the age structure of the

deer population.

Results and Discussion

The percent of does lactating (Figure 9) and mean beam diameter (Figure 10) both
increased with increasing age, according to the biodata. Although there is some error
associated with the ages of the older deer (Question 1), the general trends of lactation and
antler size are still apparent. Scanlon and Urbston (1978) found that higher lactation rates
among adults than among yearlings in South Carolina. Ozoga et al. (1994) also found
that, in Michigan, older deer are more likely to breed than younger deer, especially under
stressful conditions. Increases in beam diameter with increasing age were also expected
based on results in the literature (Severinghaus et al. 1950; McCullough 1982; Ozoga et
al. 1994).

The geographic variations in the lactation rates and antler size are most apparent

among the yearlings. The percent of lactating does decreased with increasing latitude

29

Percent Lactating

9O

 

 

80
70

60 fl
.. /

4O /

30$,

20

 

 

 

 

 

 

 

10

 

 

 

O T I I
1.5 Years 2.5 Years 3.5 Years 4.5 Years 5.5+ Years

 

Age

Figure 9. Percent of does lactating in Michigan, averaged from 1993-1999
(using October data). Error bars are d: 1 standard error.

30

35

33

31

29

27

25

23

21

Mean Beam Diameter (mm)

l9

17

15
1.5

 

 

 

 

 

 

/

 

/

 

/

 

 

/

 

 

 

 

Years 2.5 Years

3.5 Years
Age

4.5 Years

5.5+ Years

Figure 10. Average beam diameter of all antlered deer checked in Michigan,
averaged from 1992-1999. Error bars are :t 1 standard error.

31

(Figure l 1); fewer deer were lactating in the northern than in the southern regions.

Ozoga et al. (1994) describe the same pattern and suggest it is due to the greater stress in
the more northern regions. Northern Michigan has harsher winters, less productive soils,
and less farmland, on which deer could find abundant food, than southern Michigan. The
pattern is similar for yearling beam diameters. The deer of the Southern LP had larger
average beam diameters than the deer of the Northern LP and UP (Figure 12). As with
the lactation trends, the patterns of beam diameters may be due to the quality of the deer
habitat or population density across the state. Severinghaus et al. (1950) found that antler
development was related to the availability of forage due to habitat quality and deer
density in New York. As Ozoga et al. (1994) report, the Northern LP and UP have higher
deer population densities than the Southern LP, but poorer deer habitat.

The trends of lactation rates and beam diameter in the biodata reflect the trends
expected from the literature. The data also support the concern that the recent heavy bias
of the data collection in favor of the Northeastern LP, which is not representative of the
entire state, could bias the data, leading to erroneous conclusions about the condition of
the herd. These results support the use of the biodata to provide an index of the true
condition of the state’s entire herd, rather than just the harvested population. If the
checked animals were grossly misrepresenting the population as a whole, the trends in

lactation and antler size would not follow expected patterns.

32

 

35

30

 

N
Ll.

 

 

N
C

 

Percent Yearlings Lactating

 

 

 

 

 

 

   

 

 

15
10
5
0 _
Upper Peninsula Northern Lower Peninsula Southern Lower Peninsula
Region

Figure 11. Percent of yearling does lactating in each region averaged from 1993-
1999 (data from first week of firearm season). Error bars are + 1 standard error.

33

 

 

25

 

N
O

 

 

_
M
1

 

 

H
O
I

 

 

Average Yearling Beam Diameter (mm)
LII

 

 

 

 

 

0..

Upper Peninsula Northern Lower Peninsula Southern Lower Peninsula

Region

Figure 12. Average yearling beam diameter in each region averaged from 1987-
1999. Error bars are + 1 standard error.

34

Question 3: How do the locations of the check stations affect the spatial distribution

of the biodata?

Significance

Many of the problems associated with the biodata are due to under-representation
of certain geographic areas or to generally low sample sizes. The only way to improve
the situation is to collect additional data from those areas that are under-represented. One
way to increase data collection may be to add additional check stations, a tactic that will
work only if the location of the check stations affects the tendency of hunters to check
their deer. If hunters will check more deer if check stations are more convenient, then the
distribution of the check stations across the state may play a large role in the spatial
distribution of the biodata. A pilot study of Michigan check stations (Bull and Peyton
2000) found that hunters who check their deer are more likely to live within 16 miles of a
check station than do hunters who do not check their deer. Hunters also expressed the
importance of the convenience of checking a deer, both in terms of location and the
amount of time it takes, in their decision to check a deer (Bull and Peyton 2000). The
results of the pilot study suggest that if increasing the number of check stations increases
the accessibility and decreases the time it takes to check a deer, hunters will be more

likely to check their deer.

Methods

Accurate lists of check station locations were available for only as far back as

1996 (Table 3), so I first identified any counties in which check stations were added since

35

Table 3. Locations of the check stations from 1996 through 1999.

 

Station Name 1996 1997 1998 1999

 

 

Algonac State Park
Allegan State Game Area
Alma HighwayStation
Alpena Field Office
Atlanta Field Office
Bald Mtn. Recreation Area
Baldwin Field Office
Baraga Field Office
Barry State Game Area
Bay City Office
Bellaire Field Office
Big Rapids Highway Station
Birch Run Highway Station
Brighton Recreation Area
Cadillac Field Office
Cass City Field Office
Cheboygan Field Office
Crane Pond State Game Area
Crystal Falls Field Office
Curran Check Station
Escanaba Field Office

Evart Field Office x
Fish Point Wildlife Area
Flat River State Game Area
Fort Custer Recreation Area
Fort Wilkins State Park
Gaylord Field Office
Gladwin Field Office
Grand Rapids Field Office
Grayling Field Office
Gwinn Field Office
Harrison Field Office
Harsen's Island
Holly Wildlife Area
Houghton Lake Field Office
Indian River Field Office
Ishpeming Field Office
Island Lake Recreation Area
Kalkaska Field Office
Lake Hudson Recreation Area
Lapeer State Game Area
Lincoln Field Office
Livonia Office
Mackinac Bridge Highway Station
Manton Field Office
Marquette Field Office x x x

X

XXXXXXXXXX
XXXXXXXXXX
XXXXXXXXX

XXXXXXXX
XXXXXXXX

XXXXXXXX XX X
XXXXXXXXXXX XX
XXXXXXXXXXXXXXXXXXXXXXX

XXX XXXXXXXXXXXXXXXXXXXXXXXX

XXXX
XXXX
XXXXXXXXXX

X
X

XXXXXXXX
X
X

X
X
X

 

XXXXXX

 

36

 

Table 3 (cont'd).

 

 

Maybury State Park
McLain State Park
Mio Field Office
Mitchell State Park
Morrice Field Office
Mt. Clemens Field Office
Muskegon State Game Area
Naubinway Field Office
Nayanquing Point Wildlife Area
Newberry Field Office
Norway Field Office
Onaway Field Office
Paris Field Office
Pellston Field Office
Plainwell Office
Platte River Field Office
Porcupine Mtn. State Park
Port Huron Field Office
Posen Check Station
Pte. Mouillee State Game Area
Rifle River Recreation Area
Roscommon Field Office
Rose Lake Field Office
Sand Lakes Corners
Sault Ste. Marie Field Office
Shingleton Field Office
St. Charles Field Office
Standish Field Office
Stephenson Field Office
Tawas Point State Park
Traverse City Field Office
Van Buren State Park
W J Hayes State Park
Warren Dunes State Park
Waterford Field Office
Waterloo State Game Area
West Branch Field Office
West Walker Sportsman Club
Wolf Lake Fish Hatchery

 

XXXX XX XXXXXXXX

XXXX

XXXXXXXXXXXX

XXXX X XXXXXXXX

XXXX

XXXXXXXXXXXX

XX

XXXXXXXXX XXXXXXXX

XXXXXXXXXX

X

XX

XXXXXXXX

XXXXXXXXXXXXXXXXXXXXX

XXXX

 

37

 

1996. I then compared the number of deer checked from a particular county for the year
the station was added to the number of deer checked in the previous year. I also
compared the number of deer harvested (based on the mail survey estimates) from the
same counties in the same years to determine if any trends found in the number of deer

checked was due to similar changes in the number of deer harvested.

Results and Discussion

The number of deer checked in a county increased by an average of 88.21%
following the addition of a check station to the county (Table 4). The increase in the
number of checked deer was accompanied by only a 17.19% increase in the number of
deer harvested in the corresponding counties and years (Table 4). The number of deer
checked decreased in both Keweenaw and Bay counties following the addition of a check
station. The Keweenaw decrease was probably due to the larger decrease in the number
of deer harvested. The decrease in the number of checked deer in Bay County is not due
to a decrease in harvest, but the total number of deer checked between the two years
differs only by seven deer, suggesting the decrease is not as great as it may appear.
Adding a check station, therefore, tended to increase the percent of deer checked from the
county in which the station was added. Increasing the number of check stations may
decrease the pressure on any individual station, reducing the amount of time necessary to
check a deer. Perhaps more importantly, adding check stations increases the convenience
of checking a deer for the hunters by reducing the amount of time they must drive to

reach a check station. The distribution and placement of check stations could be altered

38

Table 4. Change in the number of deer checked and harvested following the
addition of a check station to a county.

 

 

 

% Change
County Station(s) Added Year added Checked Harvested
Cheboygan Cheboygan Field Office 1999 53.79 -4.15
Mackinac Mackinac Bridge 1999 290.52 107.23
Emmet Pellston Field Office 1999 228.09 4.93
Antrim Bellaire Field Office 1999 130.77 24.38
Iosco Sand Lake Corners 1999 50.81 -12.39
Roscommon Houghton Lake Field Office 1999 80.42 6.84
Presque Isle Posen Check Station and 1998 89.18 29.64
Onaway Field Office
Alcona Curren Check Station 1998 127.11 71.06
Ogemaw West Branch Field Office 1998 24.89 33.24
Bay Nayanquing Point Wildlife Area 1998 -8.14 9.21
Tuscola Fish Point Wildlife Area 1998 8.98 18.50
Keweenaw Fort Wilkins State Park 1997 -17.95 -82.20
Averafi 88.21 17.19

 

 

39

 

to increase the sample size of biodata in different counties or to balance the unequal

emphasis on certain counties within the biodata.

40

Question 4: Does the spatial distribution bias the biodata?

Significance

Due to the voluntary nature of deer checking, the location of check stations, and
the distribution of lands available for deer hunting, not all counties are represented
equally in the biodata. Each county does not contribute equally to the deer harvest,
however, so each county need not check an equivalent number of deer. Ideally the
distribution of the biodata would be equivalent to the distribution of the harvest data. For
several reasons, however, certain counties may account for more or less of the biodata

than they may of the harvest data, potentially biasing the biodata.

Methods

I calculated the percent of total check station records that each county contributed
to the biodata within a particular wildlife management unit in 1999. I then did the same
for the harvest data and compared the distribution of the check station data within a
management unit to the distribution of the harvest data. Chi-squared tests, comparing the
observed biodata distribution to the expected distribution based on the harvest data,

determined the statistical significance of the differences between the distributions.

Results and Discussion
Chi-square tests revealed a statistically significant difference between the
distribution of the biodata and the distribution of the harvest data among the counties

within all eight management units (Table 5). In the Western UP, the largest differences

41

Table 5. Results of chi-squared tests comparing the 1999 distribution of the
biodata to the distribution of the harvest data within the counties of each
management unit.

 

2

 

 

Management Unit d.f. x p-value
Western UP 10 135.541 <0.001
Eastern UP 3 38.418 <0.001

Northwestern LP 12 1865.851 <0.001
Northeastern LP 13 2468.658 <0.001

Saginaw Bay 9 3980.185 <0.001
Southwestern LP 11 2389.501 <0.001
Southcentral LP 1 1 531.97 <0.001
Southeastern LP 6 419.457 <0.001

 

 

 

42

were in Dickinson and Gogebic counties where the percent harvested exceeds the percent
checked (Figure 13a). Both could be due to inaccessibility of check stations; Gogebic did
not have a check station in 1999, and Dickinson’s check station was located on the
extreme southern edge of the county (Figure 2). In the Eastern UP, the greatest
differences were in Chippewa and Mackinac counties (Figure 13b). Mandatory deer
checking on Drummond Island probably caused the greater percent of deer checked in
Chippewa. The difference in Mackinac is difficult to explain. The placement of the new
Mackinac Bridge check station should allow a greater percentage of the harvested deer to
be checked.

Many of the counties in the Northwestern LP exhibited large differences between
the percent deer checked and harvested (Figure 13c). Mecosta, Newaygo, Mason, and
Oceana counties were all under-represented in the biodata, probably due to the lack of
check stations in the area (Figure 2). Osceola, Lake, Missaukee, and Kalkaska counties
may have been over-represented in the biodata because they contained several check
stations among them, and they are located near the TB surveillance area. The TB
positive counties in the Northeastern LP were all over-represented in the biodata due to
hunters checking their deer for TB test results (Figure 13d). The over-representation in
these counties caused under-representation in the remaining counties of this management
unit. In the Saginaw Bay management unit, the counties without check stations (Figure
2) tended to be underrepresented in the biodata (Figure 136). Clare was over-represented
due to a quality deer management project which required participating organizations to
check a minimum number of deer (J. Urbain, MDNR, personal communication), while

Saginaw’s two check stations may have contributed to its dominance in the biodata.

43

(a) Western UP

1,)
O

 

N
M

 

N
o
1

 

 

 

 

Percent Checked or Harvested

 

 

 

 

 

 

 

 

(b) Eastern UP

 

 

40

 

 

 

354

301 /

25-

 

 

 

 

 

 

20‘

15+

 

10 -~—— ——

Z: 1A A A

Chippewa Mackinac Schoolcraft Luce

 

Percent Checked or Harvested

    

 

 

 

 

 

 

 

 

Figure 13. The percent that each county contributed to the total number of
deer checked (solid) and harvested (striped) in each management unit in 1999.

44

Percent Checked or Harvested

Percent Checked or Harvested

(c) Northwestern LP
25

 

20

 

15+

 

 

\\\\\ X
m

10-

 

 

 

 

 

 

 

0-

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

89"" .319 $9 049 gas as"? 4? @040 be? 63° ”3°
18 (d) Northeastern LP
16 -
14 -
12 4
10 E
r
I
8 - f
“ I
41 . 'f
l | 5.5
2 - I I
I ‘I‘
o z a
\0 4‘ 9 o b . § 0'). '6'
\069 do? 630496"0 {5 955$ 665$) 060% (5130‘ x049 V9660 663% 496° \5’0
V4990 0&6??? 0 09° 0 C80 C365>

Figure 13 (cont'd).

45

30

Percent Checked or Harvested

N
M

Percent Checked or Harvested

 

O
1

M
1

O
1

(e) Saginaw Bay

 

 

 

 

 

 

 

 

i\\\\\\\\\\\\\V

 

 

 

 

 

 

 

Clare Saginaw Tuscola Arenac Sanilac Midland Isabella Gladwin Huron

(1) Southwestern LP

Bay

 

 

 

 

 

 

 

 

 

 

Figure 13 (cont'd).

46

 

 

 

(g) South Central LP

 

 

 

 

 

 

Percent Checked or Harvested

 

 

 

10 a
8 _
6 — V
5
4 « /
5
2 . é _
ol 4
e e o «5
\féfigp 3&0 $15? 06“? 061$ @965 xoéy
v .51" 31° a“?

60

   
 

(h) Southeastern LP

\ \\\\ N

 

 

 

 

 

 

 

 

 

 

 

 

M
C
1

 

&
o
1

 

 

N
C
i

 

Percent Checkgd or Harvested
O

 

 

—
O

 

 

 

 

Genesee St. Clair

Figure 13 (cont'd).

Lapeer

 

47

 

 

 

 

M

Oakland Macomb Monroe Wayne

Areas with low check-station density are under-represented in the biodata (i.e. Sanilac,
Isabella, and Huron counties).

In the Southwestern LP, Allegan and Van Buren counties both had two check
stations (Figure 2) and were over-represented in the biodata (Figure 13f), while
Kalamazoo had no check stations and yet was also over-represented. The border counties
of the Southwestern management unit (Branch, Calhoun, and Kent) lacked sufficient
check stations and so tended to be under-represented in the biodata. The South Central
management unit had the same problem with the western border counties (Hillsdale,
Eaton, Ionia, Montcalm) (Figure 13 g). The distribution of the biodata and harvest data in
the Southeastern LP (Figure 13h) did not appear to be related to the distribution of check
stations; Genesee county had no check stations and yet was over-represented in the
biodata, while Lapeer did have a check station and was under-represented.

Generally, the management units provide the most practical and statistically valid
division for combining and averaging the biodata. Ideally the distribution of the data
within each management unit should reflect the harvest distribution as closely as
possible. The over— and under- representation of particular counties within the biodata is
usually due to the distribution of check stations. Opening new check stations in areas that
are under-represented (such as many counties in the Southern LP) could balance the

distribution of the biodata throughout the management units.

48

Question 5: Does the seasonal distribution bias the biodata?

Significance

The majority of the biodata are collected from deer harvested during the firearm
season, both because the majority of the deer are harvested during the firearm season and
because the highway check stations are only open during the firearm season. If the data
harvested during the firearm season do not represent the same population as the deer
harvested during the other seasons, especially the archery season, then the biodata could

be biased in favor of the population harvested during the firearm season.

Methods

Although the majority of the biodata come from deer harvested during the firearm
season, the second largest source of data is the archery season. I therefore compared the
composition of the firearm season and archery season to explore any differences between
the two seasons in the age structure of the antlered and antlerless harvest separately and
the harvest composition (antlered vs. antlerless). All comparisons were made using
Pearson’s chi-squared test for each region and year separately. I also compared the
biodata ratio of antlerless to antlered deer in the archery season, firearm season, and all

seasons combined to the corresponding harvest ratio of the same seasons.

Results and Discussion

Generally, over 80% of the biodata were collected from deer harvested during the

firearm season, although usually no more than 70% of the harvest came from the firearm

49

season (Figure 14). The difference in the percent of data from the firearm season and the
percent of deer harvested during the firearm season is probably due to the highway check
stations. The highway check stations provide a convenient location for many hunters to
check their deer. Over 15% of the firearm biodata have been collected at highway check
stations over the past thirteen years (Figure 15). The majority of the non-firearm season
biodata are collected from deer harvested during the archery season.

In most cases, the composition of the firearm harvest was significantly different
from the composition of the archery harvest. For the past decade, the antlered to
antlerless ratio of the archery season has differed significantly from that of the firearm
season every year (Table 6). The firearm season generally had a higher proportion of
antlered deer than the archery season, except for a few year in the Southern LP. The age
structure of the buck harvest was significantly different between the two seasons in all
cases except recently in the UP (Table 7). Generally, the archery harvest consisted of a
higher proportion of yearlings than the firearm harvest. The two seasons differed less in
the composition of the antlerless harvest. Although the Northern LP always presented
Significant differences in the antlerless harvest (except in 1999), the significance of the
differences in the UP and the Southern LP tended to fluctuate (Table 8). Mattson Hansen
(1998) reported similar results in her comparison of Michigan’s archery and firearm data
using smaller geographical units. As Mattson Hansen (1998) reported, the causes of
these differences are unknown but could be related to biological factors, equipment
biases, or hunter selection biases that differ between archery and firearm season.

The significant differences in the composition between the archery and the

firearm season indicate that when the data of the two seasons are combined in a single

50

+Archery-Checked
' ‘I " Archery-Harvest

+ Firearm-Checked
' i ' Firearm-Harvest

 

100

 

 

 

 

 

 

 

 

 

 

Percent Deer Checked or Harvested

 

 

 

0 T I I I I I I
1990 1991 1992 1993 1994 1995 1996 1997

Year

I

1998

 

1999

Figure 14. Percent of deer checked during firearm and archery season and percent

of deer harvested during archery and firearm season in Michigan.

51

Percent of Checked Deer

 

30

 

 

 

 

 

 

,. /\/\ .
.-/ V\/\

 

 

1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999

Year

Figure 15. Percent of the checked deer harvested during firearm season that are
checked at highway check stations: Alma, Big Rapids, Birch Run, and the Mackinac
Bridge (added in 1999).

52

Table 6. The number (and proportion) of deer of each harvest-type category
checked from each region and season. The chi-squared tests compared the

composition of the firearm and archery seasons in the biodata. All tests have
1 degree of freedom.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Firearm Archery
Year Region Antlered Antlerless Antlered Antlerless I: p—value
1987 UP 3900 (0.6) 2566 (0.40) 137 (0.49) 141 (0.51) 13.508 <0.001
NLP 8325 (0.70) 3605 (0.30) 731 (0.43) 979 (0.57) 489.864 <0.001
SLP 3165 (0.63) 1849 (0.37) 554 (0.61) 350 (0.39) 1.111 0.292
1988 UP 4344 (0.83) 904 (0.17) 143 (0.51) 137 (0.49) 174.788 <0.001
NLP 8739 (0.64) 4983 (0.36) 727 (0.44) 933 (0.56) 247.543 <0.001
SLP 3484 @58) 2475 (0.42) 616 (0.63) 362 (0.37) 7.099 0.008
1989 UP 4430 (0.78) 1217 (0.22) 200 (0.52) 183 (0.48) 138.421 <0.001
NLP 7053 (0.60) 4656 (0.40) 753 (0.41) 1076 (0.59) 235.536 <0.001
SLP 3810 (0.56) 2958 (0.44) 743 (0.59) 519 (0.41) 2.885 0.089
1990 UP 4175 (0.79) 1141 (0.21) 182 (0.51) 176 (0.49) 144.377 <0.001
NLP 8358 (0.61) 5327 (0.39) 871 (0.47) 993 (0.53) 139.967 <0.001
SLP 3215 (0.57) 2449 (0.43) 748 (0.64) 426 (0.36) 19.286 <0.001
1991 UP 4339 (0.78) 1232 (0.22) 194 (0.39) 304 (0.61) 366.491 <0.001
NLP 7658 (0.71) 3183 (0.29) 836 (0.37) 1394 (0.63) 893.241 <0.001
SLP 3288 (0.60) 2237 (0.40) 727 (0.55) 593 (0.45) 9.644 0.003
1992 UP 3962 (0.74) 1396 (0.26) 197 (0.44) 248 (0.56) 178.209 <0.001
NLP 5353 (0.67) 2622 (0.33) 719 (0.39) 1122 (0.61) 499.408 <0.001
SLP 2907 (0.62) 1782 (0.38) 830 (0.57) 623 (0.43) 11.056 0.001
1993 UP 4424 (0.90) 514 (0.10) 239 (0.42) 331 (0.58) 893.77 <0.001
NLP 5567 (0.78) 1542 (0.22) 616 (0.44) 783 (0.56) 691.601 <0.001
SLP 2922 (0.63) 1691 (0.37) 863 (0.59) 595 (0.41 8.136 0.004
1994 UP 6967 (0.87) 1052 (0.13) 427 (0.41) 620 (0.59) 1308.37 <0.001
NLP 9180 (0.92) 812 (0.08) 1076 (0.49) 1137 (0.51) 2525.739 <0.001
SLP 4856 (0.67) 2435 (0.33) 1276 (0.62) 794 (0.38) 17.555 <0.001
1995 UP 5297 (0.65) 2807 (0.35) 426 (0.47) 482 (0.53) 119.88 <0.001
NLP 8479 (0.80) 2101 (0.20) 1111 (0.49) 1136 (0.51) 925.71 <0.001
SLP 4641 (0.64) 2587 (0.36) 1232 (0.58) 876 (0.42) 23.242 <0.001
1996 UP 3325 (0.69) 1471 (0.31) 227 (0.44) 294 (0.56) 140.608 <0.001
NLP 7514 (0.65) 4100 (0.35) 915 (0.48) 975 (0.52) 183.77 <0.001
SLP 4174 (0.55L 3355 (0.45) 1407 (0.62) 848 (0.38) 34.262 <0.001
1997 UP 2545 (0.79) 664 (0.21) 146 (0.39) 224 (0.61) 282.381 <0.001
NLP 7419 (0.66) 3881 (0.34) 941 (0.44) 1206 (0.56) 365.449 <0.001
SLP 3742 (0.47) 4222 (0.53) 1307 (0.56) 1039 (0.44) 55.21 <0.001
1998 UP 3319 (0.88) 441 (0.12) 222 (0.44) 284 (0.56) 623.186 <0.001
NLP 7816 (0.48) 8490 (0.52) 1154 (0.43) 1554 (0.57) 26.363 <0.001
SLP 3964 (0.49) 4143 (0.51) 1306 (0.59) 902 (0.41) 72.999 <0.001
1999 UP 5287 (0.77) 1577 (0.23) 412 (0.50) 419 (0.50) 290.665 <0.001
NLP 9825 (0.51) 9516 (0.49) 1217 (0.49) 1287 (0.51) 4.279 0.039
SLP 3845 (0.48) 4133 (0.52) 1510 (0.62) 938 (0.38) 136.415 <0.001

 

53

 

Table 7. The number (and proportion) of antlered deer of each age category
checked from each region and season. The chi-squared tests compared the

antlered deer age composition of the firearm and archery seasons in the biodata.

All tests have 1 degree of freedom.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Firearm Archery
Year Region 1.5 2.5+ 1.5 2.5+ x2 p-value
1987 UP 2110(0.58) 1542(0.42) 100 (0.76) 32 (0.24) 16.955 <0.001
NLP 5797 (0.72) 2290 (0.28) 521(0.76) 164(0.24) 6.000 0.014
SLP 2111(0.69) 964 (0.31) 405 (0.78) 111(022) 20.389 <0.001
1988 UP 2252 (0.52) 1882 (0.48) 97 (0.70) 41 (0.30) 13.494 <0.001
NLP 5791 (0.68) 2679 (0.32) 512(077) 149 (0.23) 23.685 <0.001
SLP 2396 (0.71) 998 (0.29) 464 (0.80) 117(020) 21.109 <0.001
1989 UP 1630(0.43) 2137 (0.57) ll6(0.63) 69 (0.37) 27.001 <0.001
NLP 3404 (0.39) 2152 (0.61) 506 (0.78) 140 (0.22) 72.302 <0.001
SLP 2472 (0.69) 1122(031) 544 (0.78) 154 (0.22) 23.453 <0.001
1990 UP 1714 (0.43) 2291 (0.57) ll3(0.65) 61 (0.35) 33.238 <0.001
NLP 5159 (0.64) 2933 (0.36) 584 (0.74) 206 (0.26) 32.575 <0.001
SLP 2109 (0.68) 1011(032) 550 (0.78) 154 (0.22) 30.058 <0.001
1991 UP 1899 (0.47) 2165 (0.53) 103 (0.57) 79 (0.43) 6.805 0.009
NLP 4882 (0.66) 2517(034) 598 (0.78) 173 (0.22) 42.395 <0.001
SLP 2095 (0.66) 1074 (0.34) 510(075) 171 (0.25) 19.752 <0.001
1992 UP 1389 (0.38) 2249 (0.62) ll8(0.63) 70 (0.37) 45.260 <0.001
NLP 3308 (0.65) 1799 (0.35) 499 (0.77) 147 (0.33) 39.843 <0.001
SLP 1859 (0.66) 959 (0.34) 592 (0.77) 181(0.33) 31.551 <0.001
1993 UP 1765(0.42) 2435 (0.58) 132 (0.59) 91(041) 25.484 <0.001
NLP 3206 (0.59) 2201 (0.41) 396 (0.73) 149 (0.27) 37.023 <0.001
SLP 1758 (0.62) 1059 (0.38) 575 (0.72) 220 (0.28) 26.678 <0.001
1994 UP 3798 (0.57) 2832 (0.43) 286 (0.71) 119(029) 27.860 <0.001
NLP 5739 (0.66) 3003 (0.34) 718(0.74) 253 (0.26) 26.989 <0.001
SLP 3003 (0.65) 1592 00.35) 873 (0.76) 275 (0.24) 47.855 <0.001
1995 UP 2600 (0.53) 2276 (0.47) 272 (0.72) 106 (0.28) 49.158 <0.001
NLP 5888 (0.73) 2185 (0.27) 807 (0.82) 172 (0.18) 40.887 <0.001
SLP 2913(0.67) 1410(033) 825 (0.76) 267 (0.24) 27.191 <0.001
1996 UP 897 (0.29) 2168 (0.71) 92 (0.46) 109 (0.54) 24.340 <0.001
NLP 4599 (0.63) 2667(0.37) 639 (0.76) 206 (0.24) 50.281 <0.001
SLP 2827 (0.70) 1190 (0.30) 1004(0.78) 288 (0.22) 26.168 <0.001
1997 UP 678 (0.28) 1760 (0.72) 42 (0.35) 77 (0.65) 3.142 0.076
NLP 4562 (0.64) 2594 (0.36) 622 (0.74) 216(0.26) 36.100 <0.001
SLP 2435 (0.67) 1200 (0.33) 887 (0.75) 300 (0.25) 25.004 <0.001
1998 UP 1991 (0.62) 1230 (0.38) 127 (0.62) 79 (0.38) 0.002 0.963
NLP 5001 (0.66) 2592 (0.34) 773 (0.74) 270 (0.26) 28.168 <0.001
SLP 2542 (0.66) 1290 (0.34) 929 (0.77) 281 (0.23) 46.735 <0.001
1999 UP 3007 (0.58) 2178 (0.42) 225 (0.60) 147(040) 0.884 0.347
NLP 5905 (0.62) 3608 (0.38) 818(0.73) 309 (0.27) 47.841 <0.001
SLP 2540 (0.69) 1163 (0.31) 1101 (0.79) 295 (0.21) 52.423 <0.001

 

54

 

Table 8. The number (and proportion) of antlerless deer of each age category
checked from each region and season. The chi-squared tests compared the antlerless
composition of the firearm and archery seasons in the biodata. All tests have 2
degrees of freedom.

 

Year

Reg.

0.5

Firearm
1.5

2.5+

0.5

Archery
1.5

2.5+

213

p-value

 

1987

UP
NLP
SLP

1086 (0.43)
1330 (0.38)
801 (0.44)

368 (0.15)
737 (0.21)
443 (0.24)

1051 (0.42)
1461 (0.41)
576 (0.32)

43 (0.33)
420 (0.45)
156 (0.46)

21 (0.16)
206 (0.22)
90 (0.27)

67 (0.51 )
313 (0.33)
91 (0.27)

5.852
21.928
2.961

0.054
<0.001
0.227

 

1988

UP
NLP
SLP

284 (0.34)
1779 (0.37)
1049 (0.43)

168 (0.20)

387 (0.46)

1095 (0.23) 1970 (0.41)

631 (0.26)

763 (0.31)

33 (0.26)
351 (0.40)
174 (0.50)

34 (0.27)
258 (0.29)
82 (0.23)

59 (0.47)
269 (0.31)
94 (0.27)

4.503
35.673
5.81

0.105
<0.001
0.055

 

1989

UP
NLP
SLP

392 (0.41)
1639 (0.42)
1217 (0.43)

144 (0.15)
707 (0.18)
732 (0.26)

431 (0.45)
1546 (0.40)
895 (0.31)

45 (0.26)
431 (0.44)
250 (0.50)

46 (0.27)
223 (0.23)
114 (0.23)

79 (0.46)
318 (0.33)
135 (0.27)

20.369
20.154
9.257

<0.001
<0.001
0.010

 

1990

UP
NLP
SLP

344 (0.32)
1915 (0.37)
987 (0.41)

235 (0.22)

484 (0.46)

1095 (0.21) 2199 (0.42)

604 (0.25)

795 (0.33)

50 (0.29)
422 (0.46)
199 (0.48)

43 (0.25)
232 (0.25)
105 (0.25)

78 (0.46)
271 (0.29)
108 (0.26)

1.054
54.717
9.487

0.590
<0.001
0.009

 

1991

UP
NLP
SLP

405 (0.36)
1245 (0.40)
941 (0.43)

175 (0.16)
553 (0.18)
481 (0.22)

543 (0.48)
1303 (0.42)
775 (0.35)

93 (0.33)
618 (0.47)
245 (0.44)

59 (0.21)
288 (0.22)
141 (0.25)

130 (0.46)
422 (0.32)
171 (0.31)

4.702
41.333
5.147

0.095
<0.001
0.076

 

1992

UP
NLP
SLP

425 (0.34)
930 (0.37)
765 (0.44)

238 (0.19)
548 (0.22)
428 (0.24)

584 (0.47)
1032 (0.41)
555 (0.32)

64 (0.29)
406 (0.39)
288 (0.49)

62 (0.28)
264 (0.25)
137 (0.23)

94 (0.43)
375 (0.36)
168 (0.28)

9.645
9.537
4.298

0.008
0.008
0.1 17

 

1993

UP
NLP
SLP

208 (0.42)
568 (0.38)
707 (0.43)

93 (0.19)
295 (0.20)
396 (0.24)

194 (0.39)
637 (0.42)
558 (0.34)

91 (0.32)
352 (0.48)
235 (0.42)

69 (0.24)
161 (0.22)
154 (0.27)

127 (0.44)
216 (0.30)
173 (0.31)

8.607
35.43
3.217

0.014
<0.001
0.200

 

1994

UP
N LP
SLP

389 (0.38)
296 (0.38)
988 (0.42)

212 (0.21)
155 (0.20)
564 (0.24)

415 (0.41)
333 (0.42)
814 (0.34)

194 (0.33)
447 (0.42)
345 (0.46)

131 (0.22)
242 (0.23)
185 (0.25)

264 (0.45)
385 (0.36)
216 (0.29)

4.661
8.462
7.974

0.097
0.015
0.019

 

1995

UP
NLP
SLP

940 (0.37)
627 (0.31)
956 (0.38)

503 (0.20)
487 (0.24)
606 (0.24)

1079 (0.43)
881 (0.44)
928 01.37)

104 (0.24)
410 (0.39)
364 (0.44)

135 (0.31)
296 (0.28)

195 (0.45)
347 (0.33)

213 (0.26L257 (0.31)

40.339
36.576
12.004

<0.001
<0.001
0.002

 

1996

UP
NLP
SLP

390 (0.29)
1 1 19 (0.28)
1341 (0.41)

255 (0.19)
871 (0.22)
729 (0.22)

714 (0.53)
1965 (0.50)
1226 (0.37)

l

66 (0.24)
293 (0.33)
339 (0.42)

55 (0.20)
255 (0.28)
186 (0.23)

156 (0.56)
351 (0.39)
278 (0.35)

2.721
34.735
1.854

0.256
<0.001
0.396

 

1997

UP
NLP
SLP

237 (0.37)
1132 (0.30)
1613 (0.39)

112 (0.18)
721 (0.19)
969 (0.23)

285 (0.45)
1918 (0.51)
1580 (0%)

46 (0.24)
327 (0.29)
336 (0.34)

32 (0.16)
257 (0.23)
262 (0.26)

117 (0.60)
537 (0.48)
397 (0.40)

15.408
7.978
9.198

<0.001
0.019
0.010

 

1998

UP
NLP
SLP

151 (0.35)
2668 (0.32)

89 (0.21)

188 (0.44)

1635 (0.20) 3914 (0.48)
1635(0.41) 898 (0.22) 1480(0.37)

68 (0.25)
447 (0.31)
355 (0.4;)

69 (0.25)
360 (0.25)
201 (0.23)

138 (0.50)
639 (0.44)
305 (0.35

8.774
18.896
0.746

0.012
<0.001
0.689

 

 

1999

NLP
SLP

 

423 (0.28) 278 (0.18) 808 (0.54)
2671 (0.29) 1967 (0.21) 4578 (0.50)
1613 (0.40) 897 (0.22) 1519 (0.38)

 

79 (0.20)
348 (0.29)
330 (0.37)

99 (0.25)
274 (0.22)
238 (0.27)

217 (0.55)
597 (0.49)
325 (0.36)

 

14.744
0.821
8.195

0.001
0.663
0.017

 

55

 

analysis, the results of the analysis will be heavily biased by the effect of the firearm
season data. For example, Mattson Hansen (1998) found that SAK model estimates
differ when the firearm and archery seasons’ data are combined or when the firearm
season’s data are considered separately. I found that the firearm and archery data
considered separately provide a more accurate view of the true harvest of their respective
seasons, than the combined data provide of the entire harvest (Figure 16). The antlerless
to antlered ratios from the firearm and archery seasons’ biodata generally match more
closely to the same ratio calculated using the harvest data than does the combined season
data (Figure 16). The biodata do appear to be biased in favor of the firearm data. The
composition of the biodata is not equal to that of the true harvest, and combining the data
of both the archery and firearm seasons may provide a less accurate picture of the true

harvest than considering them separately.

56

All Seasons I Archery El Firearm

 

N
0'5

 

N
&

 

N
N
\K\

 

 

N
o

 

u—s
oo

 

i—n
Oi

g

 

i— I—I
oo o N «h
1

 

 

 

 

 

H

 

 

Number of Counties

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

< -0.4 -0.3 -0.2 -0.1 0 0.1 0.2 0.3 >0.4
Harvest Ratio - Checked Ratio

Figure 16. Histogram of the number of counties in the archery season, firearm
season, and all seasons combined whose difference between the harvest estimates
and biodata estimates of the antlerless to antlered ratio falls within the listed
catgories in 1999.

57

 

Question 6: Does the composition of checked deer differ between private and public

lands?

Significance

Many hunters believe that the public land deer harvest has a different age and sex
composition from that of the private land harvest. The tendencies of hunters to check
their deer are also thought to differ between those hunting on private and public land (H.
Hill, MDNR, personal communication). If the composition of the harvest differed
between public and private lands, and the checking tendencies differed between public
and private land hunters, then the biodata could be biased. Since 1998, therefore, the
MDNR has collected information on whether checked deer were taken on private or
public land (data collected since 1997 in the harvest data) so it is now possible to identify
differences in the checking tendencies and composition between private land and public

land harvests.

Methods

Although Michigan has separate quotas for public and private land harvests, the
MDNR has collected data on the private or public land harvest classification only since
1998 in the biodata and since 1997 in the harvest data. The 1997 and 1998 harvest data
were not available, so I restricted the evaluation to the 1999 data. I compared the private
land harvest composition to the public land harvest composition as reported in the
biodata, and I compared the harvests reported in the mail survey and biodata on public

and private lands. All comparisons were made using Pearson’s chi-squared tests. One

58

potential error in the data is the identification of Commercial Forest Reserve (CF R) lands
as public lands. These private lands get a tax exemption if they are open to hunting, and
many hunters may report CF R lands to be public lands since they are usually open to
hunting (W. Moritz, MDNR, personal communication). The errors should be consistent
between the biodata and the harvest data, however, and should have little effect on the

relative relationships between the two datasets.

Results and Discussion

Statewide, approximately 17% of the deer harvest was taken on public lands in
1999 according to the mail survey. The distribution of the harvest across private and
public lands varied among the regions as the availability of public lands varies. More
than one third of the UP’s harvest came from public lands (Table 9) where public land is
abundant (Figure 17). In the Southern LP, however, there is’very little public land
(Figure 17) and less than 10% of the harvest came from public lands (Table 9). The
percent of deer checked from public lands greatly exceeded the percent of deer harvested
from public lands in both regions of the LP, implying that hunters were more likely to
check their deer taken from public lands than from private lands in these regions (Table
9). Hunters hunting on public land have to leave their property and are more likely to
pass a check station while transporting their deer than hunters hunting on private land.
Check stations also tend to be located on public land hunting areas, which could increase
the tendency of public land hunters to check their deer. In the UP, however, the percent
of deer checked from public lands was only slightly higher than the percent of deer

harvested from public lands (Table 9). In all regions, more deer were checked and

59

Table 9. The number (and percent) of deer harvested and checked from
private and public lands in 1999.

 

Region

Public

Harvested Checked

Private
Harvested Checked

 

 

Upper
Peninsula
Northern

Lower Peninsula
Southern
Lower Peninsula

 

23657 (31.7) 2577 (32.9)
46314 (23.5) 7111 (30.9)

23394 (8.8) 1524 (13.6)

 

51001 (68.3) 5247 (67.1)
150626 (76.5) 15873 (69.1)

242579 (91.2) 9705 (86.4)

 

60

 

 

Figure 17. Distribution of public lands (shaded areas) in Michigan.

61

harvested from private lands than from public lands. On both public (Table 10) and
private (Table 11) lands, the harvest composition differed significantly between the mail
survey results and the biodata results. The biodata, therefore, do not represent the true
harvest composition on either private or public land.

The biodata also suggest there are significant differences in the age composition
of the deer harvested on private and public land. The age composition of the antlered and
antlerless deer differed significantly between private and public land in all regions except
the antlerless harvest in the Southern LP (Table 12). The harvest composition also
differed significantly between private and public lands in all regions except the Southern
LP (Table 13). These differences between the private and public lands indicate that,
because the majority of deer are harvested and checked from private lands, the biodata
reflect the condition of the deer harvested on private lands, which is different from that of
the deer harvested on public lands. The situation may not be as serious as these results
imply. In much of the state, hunters are more likely to check deer from public lands than
private lands. The motivation for checking deer may be different for private and public
land hunters. Private land hunters may check their deer to get accurate sex, age, and
beam measurements so they can keep track of the deer herd on their lands. Public land
hunters may check their deer to cooperate with the MDNR. Different motivations may
skew the picture of the herd composition the biodata provides. The biodata do not
represent the true harvest, so the differences seen between the composition of the deer
checked from private and public lands may only be an artifact of checking habits and not

a reflection of true differences in the harvest.

62

 

Table 10. The 1999 numbers (and proportions) of antlered and antlerless
deer harvested on public land according to the mail survey and the number
(and proportion) checked from public land in the biodata. The chi-squared

tests compare the public land harvest composition as reported in the mail

survey to that reported in the biodata.

 

 

2

 

 

 

 

 

 

Region Harvest Mail Biodata X. d.f. p-value
Upper Antlered 19006 (0.80) 2154 (0.86) 49.149 1 <0.001
Peninsula Antlerless 4632 (0.20) 345 (0.14)
Northern Antlered 26720 (0.58) 3538 (0.53) 59.019 1 <0.001
LP Antlerless 19606 (0.42) 3174 (0.47)
Southern Antlered 10366 (0.44) 724 (0.50) 19.039 1 <0.001
LP Antlerless 13056 (0.56) 720 (0.50)

 

63

 

 

Table 11. The 1999 numbers (and proportions) of antlered and antlerless deer
harvested on private land according to the mail survey and the number (and
(proportion) checked from private land in the biodata. The chi-squared tests
compare the private land harvest composition as reported in the mail survey to that
reported in the biodata.

 

Region Harvest Mail Biodata X. d.f. p-value
Upper Antlered 30310 (0.59) 3448 (0.67) 117.62 1 <0.001
Peninsula Antlerless 20676 (0.41) 1681 (0.33)
Northern Antlered 70946 (0.47) 7410 (0.49) 22.422 1 <0.001

LP Antlerless 79684 (0.53) 7676 (0.51)
Southern Antlered 108536 (0.45) 4652 (0.51) 125.25 1 <0.001
LP Antlerless 134055 (0.55) 4531 (0.49)

 

 

 

 

 

 

 

 

 

64

 

 

 

 

 

N

 

 

 

82 N Sou A383“ A285 @538 @8882 6868 828 a: 82.2.5. .5
Bod _ 63 A88: €3va A385 63.8 a: 8.22.... 52.2.8
Sod N 83 A888; $6886 @888 A884:- am8m5 5888 82.2.5 .3
25¢ _ E3 A883 93.838 . A8862 33.885. 8.22.... 5222
Bed N 8? $3. 2 5.83 5.88 38% p233 @6843. 8222.... 252.3
mm; 2 :3. 898% 23.882 A3822 A8888 622......- .25:
+3 3 m... +3 3 m... 25.
2:9»-.. a... x 33:.— SSEQ «metal nemwom

 

 

65w— 3§ta mo 35 8 8360888 wea— ozna 05 8888 memo“ Begum-Eu 2E. 65: 333 was
033.:— Eat woo—cone bowoumo owe :23 no .808 macro—:3 Ea 822:8 mo Amcomtoaoa 2:8 838:: 32 23. .N— 0.3—ah.

65

Table 13. The 1999 numbers (and proportions) of antlered and antlerless deer
checked from private and public lands. The chi-squared tests compare the private
land harvest composition to that of public lands.

 

Region Harvest Private Public x2 d.f. p-value
Upper Antlered 3448 (0.67) 2154 (0.86) 309.964 1 <0.001
Peninsula Antlerless 1681 (0.33) 345 (0.14)
Northern Antlered 7410 (0.49) 3538 (0.53) 23.99 1 <0.001

LP Antlerless 7676 (0.51) 3174 (0.47)
Southern Antlered 4652 (0.51) 724 (0.50) 0.135 1 0.713
LP Antlerless 4531 (0.49) 720 (0.50)

 

 

 

 

 

 

 

 

 

 

66

Question 7: Does the composition of checked deer differ between highway and field

check stations?

Significance

Most of Michigan’s check stations are field stations located at MDNR field
offices, and state parks, game areas, and recreation areas. Traditionally, however, the
MDNR has maintained three highway check stations at Birch Run, Alma, and Big
Rapids. In 1999, a fourth highway check station was added at the Mackinac Bridge.
These four check stations are located along major southbound arteries in Michigan
(Figure 2) and are meant to be a convenient place for hunters who hunt in the UP or
Northern LP to stop and check their deer on their way home to southern Michigan.
Unlike the field stations, the highway check stations are only open during several days of
the firearm season, but they check thousands of deer every year. The Alma highway
check station will close after the 2000 hunting season due to the Michigan Department of
Transportation’s closing of the rest area and the off-ramp (H. Hill, MDNR, personal
communication). The significance of the highway check stations to the collection of the
firearm biodata and the impending loss of the Alma check station suggested I should
examine any differences between the biodata collected at the highway check stations and

the data collected at the field check stations.

Methods

Although not every check station has an individual code in the ‘station’ variable

of the SPSS biodata file, each highway check station does have its own designation

67

(Appendix 2). I was therefore able to determine whether each record in the biodata was
collected at a highway station or a field station. The highway check stations are open
only during the firearm season, so in comparing the field data to the highway data I chose
to compare only firearm season data. All other seasons were eliminated from the
analyses. I then divided the data among the three regions and compared the composition
of the biodata collected at the field stations to the data collected at the highway stations
among the three regions since 1987. All comparisons were made using Pearson’s chi-

square test.

Results and Discussion

The data collected at the highway check stations has made up from 16.0% to
28.5% of firearm season biodata and from 13.8% to 25.8% of all biodata collected over
the past 13 years (Figure 18). The decrease in the percent of deer checked at the highway
check stations over the past several years may reflect the recent increase in the number of
field stations. The highway check stations collect data primarily from deer harvested in
the UP (Mackinac Bridge station), the Northeastern LP (Birch Run station), and the
Northwestern LP (Alma station and Big Rapids station) (Figure 19). Although some
people may live and hunt in the counties immediately surrounding each station and
simply come to the highway station because it is nearby, many people appear to be
checking their deer at the highway stations as they travel home from their hunting
grounds. As a consequence, more of the data are collected from public lands at highway
stations than at field stations (Table 14). Many hunters from southern Michigan, where

there is little public land available for hunting, travel to the Northern LP or the UP to hunt

68

30

 

25 =8

 

 

N
O
\

15

 

 

Percent of Biodata

10

 

 

 

 

0 Y I T T I

 

 

1987 1988

Year

I

1989 1990 1991 1992 1993 1994

I

1995

l

1996

if

1997 1998 1999

Figure 18. The percent of the firearm (solid line) and total (dashed line) biodata collected at
the highway check stations: Ahna, Birch Run, Big Rapids, and the Mackinac Bridge (added in

l 999).

69

é, Birch Run

 

 

 

 

 

Figure 19. Harvest locations of the deer checked at the four highway check
stations in 1999. Approximate station locations are marked as white (Alma,
Birch Run, Big Rapids) or black (Mackinac Bridge) circles.

70

 

Table 14. The number (and proportion) of deer checked from public and private
lands at highway and field stations. The chi-squared tests compare the land-type
composition of deer checked at the highway stations to that of those checked at field

check stations during the firearm season. All tests have 1 degree of freedom.

 

 

 

 

 

 

Highway Field
Year Region Public Private Public Private x2 p-value
1998 UP 139 (0.43) 185 (0.57) 1256 (0.36) 2267 (0.64) 6.747 0.009
NLP 1531 (0.38) 2509 (0.62) 3448 (0.27) 9517 (0.73) 189.989 <0.001
SLP 40 (0.09) 395 (0.91) 899 (0.11) 6993 (0.89) 1.987 0.159
1999 UP 552 (0.48) 589 (0.52) 1662 (0.29) 4110 (0.71) 167.849 <0.001
NLP 1601 (0.39) 2458 (0.61) 4680 (0.29) 11377 (0.71) 159.971 <0.001
SLP 48 (0.10) 415 (0.90) 1069 (0.14) 6812 (0.86) 3.855 0.050

 

7]

 

on public land. The land-type composition of the harvest seen at the highway check
stations could be explained if fewer hunters from southern Michigan travel north to hunt
on private land than they do to hunt on public land.

The highway and field check stations differed not only in who checked the deer
there, but in what deer were checked. In regions and years with significant differences,
hunters tended to check a greater proportion of antlered deer at highway check stations
than at field check stations (Table 15). Hunters must, therefore, have checked a greater
proportion of antlerless deer at field check stations than at highway stations (Table 15).
Although there is a bias against checking antlerless deer (Question 1) hunters who hunt
on private land may be more interested in checking all deer they harvest to get accurate
data on the deer from their lands. The greater tendency of private land hunters to check
their deer at field stations could lead to the difference in the composition of antlered and
antlerless checked deer at highway and field check stations.

The age composition of the checked deer also differed between highway and field
check stations. The differences in age composition were found primarily among the
checked antlered deer (Table 16) while the antlerless deer age compositions tended to be
more similar (Table 17). Among the years and regions where there were significant
differences, a greater pr0portion of yearling antlered deer tended to be checked at
highway stations than at field stations. The difference in age composition is more
difficult to explain than the difference in harvest-type composition, because the pattern is
opposite from what I would expect based on the land-type composition (Question 6).
There are, however, a few possible explanations. If private land owners were practicing

self-imposed Quality Deer Management (QDM) on their own land, they would have

72

Table 15. The number (and proportion) of antlered and antlerless deer checked at
highway and field stations. The chi-squared tests compare the harvest-type
composition of deer checked at the highway stations to that of those checked at
field check stations during the firearm season. All tests have 1 degree of freedom.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Highway Field
Year Region Antlered Antlerless Antlered Antlerless x2 p-value
1987 UP 285 (0.93) 22 (0.07) 3615 (0.59) 2544 (0.41) 142.387 <0.001
NLP 3250 (0.73) 1180 (0.27) 5075 (0.68) 2425 (0.32) 42.862 <0.001
SLP 222 (0.76) 71 (0.24) 2943 (0.62) 1778 (0.38) 21.374 <0.001
1988 UP 549 (0.87) 79 (0.13) 3789 (0.82) 825 (0.18) 10.883 0.001
NLP 3856 (0.64) 2209 (0.36) 4786 (0.64) 2722 (0.36) 0.041 0.840
SLP 256 (0.63) 152 (0.37) 3172 (0.58) 2301 (0.42) 3.580 0.058
1989 UP 437 (0.80) 111 (0.20) 3991 (0.78) 1106 (0.22) 0.610 0.435
NLP 3264 (0.61) 2108 (0.39) 3789 (0.60) 2546 (0.40) 1.093 0.296
SLP 245 (0.54) 206 (0.46) 3550 (0.56) 2737 (0.44) 0.785 0.376
1990 UP 363 (0.78) 101 (0.22) 3809 (0.79) 1040 (0.21) 0.026 0.873
NLP 3597 (0.60) 2375 (0.40) 4720 (0.62) 2913 (0.38) 3.636 0.057
SLP 157 (0.49) 163 (0.51) 3049 (0.57) 2280 (0.43) 8.175 0.004
1991 UP 358 (0.80) 87 (0.20) 3981 (0.78) 1145 (0.22) 1.846 0.174
NLP 3311 (0.76) 1057 (0.24) 4347 (0.67) 2126 (0.33) 93.989 <0.001
SLP 196 (0.61) 125 (0.39) 3087 (0.59) 2104 (0.41) 0.318 0.573
1992 UP 326 (0.70) 143 (0.30) 3636 (0.74) 1253 (0.26) 5.250 0.022
NLP 2254 (0.73) 851 (0.27) 3082 (0.64) 1767 (0.36) 69.950 <0.001
SLP 167 (0.73) 63 (0.27) 2740 (0.61) 1719 (0.39) 11.562 0.001
1993 UP 745 (0.88) 104 (0.12) 3676 (0.90) 404 (0.10) 4.190 0.041
NLP 2578 (0.81) 610 (0.19) 2981 (0.76) 930 (0.24) 22.306 <0.001
SLP 196 (0.74) 68 (0.26) 2726 (0.63) 1623 (0.37) 14.327 <0.001
1994 UP 808 (0.84) 154 (0.16) 6081 (0.88) 864 (0.12) 9.588 0.002
NLP 3328 (0.93) 264 (0.07) 5815 (0.91) 545 (0.09) 4.571 0.033
SLP 284 (0.78) 81 (0.22) 4567 (0.66) 2354 (0.34) 21.772 <0.001
1995 UP 651 (0.63) 390 (0.37) 4644 (0.66) 2415 (0.34) 4.239 0.040
NLP 3863 (0.81) 885 (0.19) 4614 (0.79) 1214 (0.21) 7.896 0.005
SLP 288 (0.75) 95 (0.25) 4336 (0.64) 2478 (0.36) 21.104 <0.001
1996 UP 190 (0.57) 141 (0.43) 3135 (0.70) 1330 (0.30) 23.784 <0.001
NLP 2492 (0.69) 1113 (0.31) 5022 (0.63) 2987 (0.37) 44.887 <0.001
SLP 273 (0.76) 88 (0.24) 3901 (0.54) 326740.46) 62.532 <0.001
1997 UP 181 (0.78) 52 (0.22) 2364 (0.79) 612 (0.21) 0.405 0.525
NLP 2529 (0.69) 1127 (0.31) 4890 (0.64) 2754 (0.36) 29.682 <0.001
SLP 249 (0.53) 192 (0.44) 3493 (0.46) 4030 (0.54) 16.830 <0.001
1998 UP 280 (0.89) 36 (0.11) 3039 (0.88) 405 (0.12) 0.038 0.846
NLP 2385 (0.61) 1499 (0.39) 5431 (0.44) 6991 (0.56) 370.795 <0.001
SLP 266 (0.65) 142 (0.55) 3698 (0.48) 4001 (0.52) 45.681 <0.001
1999 UP 936 (0.83) 194 (0.17) 4351 (0.76) 1383 (0.24) 25.774 <0.001
NLP 2324 (0.60) 1574 (0.40) 7501 (0.49) 7942 (0.51) 152.000 <0.001
SLP 277 (0.62) 167 (0.38) 3568 (0.47) 3966 (0.53) 37.930 <0.001

 

 

 

 

73

 

Table 16. The number (and proportion) of yearling and adult antlered deer checked
at highway and field stations. The chi-squared tests compare the age composition
of antlered deer checked at the highway stations to those checked at field check
stations during the firearm season. All tests have 1 degree of freedom.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Highway Field
Year Region 1.5 2.5+ 1.5 2.5+ x’ p—value
1987 UP 142(0.52) 120 (0.46) 1968 (0.58) 1422 (0.42) 1.481 0.224
NLP 2429 (0.76) 769 (0.24) 3368 (0.69) 1521(0.31) 47.533 <0.001
SLP 172(079) 47 (0.21) 1939 (0.68) 917(032) 10.713 0.001
1988 UP 326(0.60) 215(040) 1924 (0.54) 1663 (0.46) 8.310 0.004
NLP 2774 (0.72) 1054 (0.28) 2951(0.65) 1597 (0.35) 55.211 <0.001
SLP 203 (0.80) 51 (0.20) 2148 (0.70) 938 (0.30) 11.984 0.001
1989 UP 153 (0.51) 147 (0.49) 1477 (0.43) 1990(057) 7.933 0.005
NLP 1541(0.64) 865 (0.36) 1863 (0.59) 1287(0.41) 13.832 <0.001
SLP 118(0.57) 89 (0.43) 2346(0.70) 1029 (0.30) 14.208 <0.001
1990 UP 162 (0.46) 187 (0.54) 1551 (0.42) 2102 (0.58) 2.041 0.153
NLP 2356 (0.67) 1180(0.33) 2781(0.62) 1734 (0.38) 21.762 <0.001
SLP 117(077) 35 (0.23) 1990 (0.67) 969(0.33) 6.251 0.012
1991 UP 190 (0.54) 165 (0.46) 1709 (0.46) 2000(054) 7.212 0.007
NLP 2349 (0.72) 917(0.28) 2533 (0.61) 1600 (0.39) 91.939 <0.001
SLP 135(070) 59 (0.30) 1956 (0.66) 1014(034) 1.130 0.288
1992 UP 161 (0.50) 158 (0.50) 1228 (0.37) 2091(0.63) 22.376 <0.001
NLP 1539 (0.69) 693 (0.31) 1759 (0.62) 1100 (0.38) 30.301 <0.001
SLP 129 (0.77) 38 (0.23) 1730 @65) 921(035) 10.055 0.002
1993 UP 335 (0.47) 380 (0.53) 1428 (0.41) 2055 (0.59) 8.346 0.004
NLP 1612(0.64) 924 (0.36) 1588 (0.55) 1275 (0.45) 36.535 <0.001
SLP 129 (0.68) 60 (0.32) 1629 (0.62) 999 (0.38) 2.952 0.086
1994 UP 476 (0.59) 326(0.41) 3278 (0.57) 2477 (0.43) 1.646 0.199
NLP 2227 (0.69) 1023 (0.31) 3486 (0.64) 1969 (0.36) 19.258 <0.001
SLP 199 (0.71) 81(0.29) 2801 (0.65) 1509 (0.35) 4.297 0.038
1995 UP 395(0.61) 248 (0.39) 2203 (0.52) 2028 (0.48) 19.658 <0.001
NLP 2899 (0.77) 863(0.23) 2987(O.69) 1322 (0.31) 60.945 <0.001
SLP 205 (0.72) 78 (0.28) 2698 (0.67) 1327 (0.33) 3.518 0.061
1996 UP 75 (0.40) 114(0.60) 822 (0.29) 2054 (0.71) 10.558 0.001
NLP 1680 (0.68) 799 (0.32) 2919(0.61) 1868 (0.39) 32.426 <0.001
SLP 212(079) 57 (0.21) 2615 (0.70) 1133 (0.30) 9.838 0.002
1997 UP 76 (0.44) 97 (0.56) 602 (0.27) 1663 (0.73) 24.106 <0.001
NLP 1630 (0.65) 865 (0.35) 2932 (0.63) 1729 (0.37) 4.138 0.042
SLP 191(0.77) 58 (0.23) 2244 (0.66) 1142 (0.34) 11.418 0.001
1998 UP 196(0.70) 83 (0.30) 1795 (0.61) 1147 (0.39) 9.213 0.002
NLP 1691(0.71) 679 (0.29) 3310(0.63) 1913(037) 46.135 <0.001
SLP 180 (0.68) 86 (0.32) 2362 (0.66) 1204 (0.34) 0.227 0.633
1999 UP 595 (0.64) 337 (0.36) 2412(057) 1841 (0.43) 15.946 <0.001
NLP 1557 (0.68) 731(032) 4348 (0.60) 2877 (0.40) 45.726 <0.001
SLP 194(071) 80 (0.29) 2346 (0.68) 1083 (0.32) 0.671 0.413

 

74

 

Table 17. The number (and proportion) of fawn, yearling, and adult antlerless deer
checked at highway and field stations. The chi-squared tests compare the age
composition of antlerless deer checked at the highway stations to that of antlerless
deer checked at field check stations during the firearm season. All tests have 2
degrees of freedom.

 

Year Region

Highway
0.5 1.5

2.5+

0.5

Field
1.5

2.5+

1'2

p—value

 

1987

UP
NLP
SLP

9(0.45) 0(0.00)
461 (0.39) 227 (0.19)
23 (0.32) 14(020)

11 (0.55)
485 (0.41)
34 (0.48)

1077 (0.43)
869 (0.37)
778 (0.44)

368 (0.15)
510 (0.22)
429 (0.25)

1040 (0.42)
976 (0.41)
542 (0.31)

3.792
3.188
9.070

0.150
0.203
0.01 1

 

1988

UP
NLP
SLP

30 (0.39) 11(0.14)
842 (0.38) 439 (0.20)
62 (0.41) 37 (0.24)

36 (0.47)
917 (0.42)
53 (0.35)

254 (0.33)
912 (0.35)
978 (0.43)

157 (0.21)
642 (0.25)
584 (0.26)

351 (0.46)
1042 (0.40)
707 (0.31)

2.056
15.959
0.910

0.358
<0.001
0.634

 

1989

UP
NLP
SLP

41 (0.59) 5(007)
773 (0.46) 233 (0.14)
74 (0.41) 36 (0.20)

24 (0.34)
660 (0.40)
70 (0.39)

351 (0.39)
865 (0.39)
1139 (0.43)

139 (0.15)
473 (0.21)
690 (0.26)

407 (0.45)
886 (0.40)
821 (0.31)

10.888
40.584
5.880

0.004
<0.001
0.053

 

1990

UP
NLP
SLP

38 (0.38) 16(O.16)
889 (0.38) 474 (0.20)
64(0.40) 36 (0.23)

47 (0.47)
984 (0.42)
60 (0.38)

306 (0.32)
1017 (0.36)
920 (0.41)

219 (0.23)
607 (0.22)
567 (0.26)

437 (0.45)
1199 (0.42)
733 (0.33)

2.962
2.329
1.519

0.227
0.312
0.468

 

1991

UP
NLP
SLP

24 (0.29) 18(0.21)
403 (0.39) 203 (0.20)
44(035) 38 (0.30)

42 (0.50)
435 (0.42)
43 (0.34)

381 (0.37)
842 (0.41)
897 (0.43)

157 (0.15)
350 (0.17)
440 (0.21 )

501 (0.48)
868 (0.42)
727 (0.35)

3.455
3.268
6.345

0.178
0.195
0.042

 

1992

UP
NLP
SLP

48 (0.34) 29 (0.20)
274(032) 183(0.22)
24 (0.38) ”(02$

66 (0.46)
388 (0.46)
22 (0.35)

377 (0.34)
655 (0.39)
741 (0.44)

209 (0.19)
364 (0.22)

518 (0.47)
642 (0.39)

411 (024L533 (0.32)

0.149
14.631
0.853

0.928
0.001
0.653

 

1993

UP
NLP
SLP

43 (0.41) 24 (0.23)
216(0.35) 123 (0.20)
30 £1.45) 12(0.l8)

37 (0.36)
271 (0.44)
25 (0.37)

161 (0.42)
350 (0.39)
677 (0.42)

68 (0.18)
172 (0.19)
384 (0.24)

156 (0.41)
366 (0.41)
533 (0.33)

1.787
2.527
1.398

0.409
0.283
0.497

 

1994

UP
NLP
SLP

65 (0.42) 32 (0.21)
104(0.40) 48 (0.18)
24 (0.30) 24 (0.30)

57 (0.37)
110 (0.42)
33 (0.41)

309 (0.37)
191 (0.37)
964 (0.42)

177 (0.21)
107 (0.21)
540 (0.24)

343 (0.41)
221 (0.43)
781 (0.34)

1.458
0.863
5.114

0.482
0.650
0.078

 

1995

UP
NLP
SLP

163(0.42) 74 (0.19)
286 (0.33) 208(0.24)
30 (0.32) 26(0.27)

150 (0.39)
372 (0.43)
39 (0.41 )

777 (0.36)
341 (0.30)
923 (0.39)

429 (0.20)
278 (0.25)
576 (0.24)

927 (0.43)
508 (0.45)
882 (0.37)

4.699
1.775
2.000

0.095
0.412
0.368

 

1996

UP
NLP
SLP

52 (0.37) 23 (0.16)
329 (0.30) 242 (0.22)
36M1) 17(019)

65 (0.46)
533 (0.48)
35 (0.40)

338 (0.28)
790 (0.28)
1305 (0.41)

232 (0.19)
629 (0.22)
712 (0.22)

649 (0.53)
1432 (0.50)
1191 (0.37)

5.442
1.846
0.483

0.066
0.397
0.786

 

1997

UP
NLP
SLP

20 (0.40) 8(0.16)
363 (0.33) 203 (0.18)
70 (0.36) 36 (0.19)

22 (0.44)
550 (0.49)
86 (0.45)

217 (0.37)

104 (0.18)

263 (0.45)

769 (0.29) 518 (0.20) 1368 (0.52)
1543 (0.39) 933 (0.24) 1494 (0.38)

0.196
4.815
4.523

0.907
0.090
0.104

 

1998

UP
NLP
SLP

17(0.47) 5(0.14)
550 (0.37) 304 (0.20)
57 (0.40) 33 (0.23)

14 (0.39)
638 (0.43)
51 (0.36)

134 (0.34)

84 (0.21)

174 (0.44)

2118 (0.31) 1331 (0.20) 3276 (0.49)
1578 (0.41) 865 (0.22) 1429 (0.37)

2.717
20.132
0.092

0.257
<0.001
0.955

 

 

1999

UP
NLP
SLP

 

71 (0.37) 45 (0.23)
517(033) 328 (0.21)
67 (0.40) 36 (0.22)

77 (0.40)
721 (0.46)
64 (0.38)

 

352 (0.27) 233 (0.18) 731 (0.56)
2154 (0.28) 1639 (0.21) 3857 (0.50)

1546 (0.40) 861 (0.22) 1455 (0.38)

 

16.628
15.736
0.057

<0.001
<0.001
0.972

 

75

 

harvested a greater proportion of larger, older bucks than hunters hunting on public land.
(QDM is a program in which hunters take fewer yearling bucks and more antlerless deer
to try to balance the sex ratio of the deer population [MDNR 2000b].) Private
landowners may also have been more familiar with the deer available on their land and so

could be more selective in choosing the deer they harvest.

76

Question 8: How does aging deer as ‘A’ or ‘AA’ affect the biodata?

Significance

Although MDNR personnel are trained in aging deer prior to the start of the
season, determining the exact age of a deer can be difficult due to variability in wear and
replacement patterns. Check station agers may also not be able to determine the age of a
deer because the head is frozen and the jaw cannot be examined. When agers cannot
determine the age to a specific year, they record the age as either ‘A’ (not a fawn) or

‘AA’ (older than a yearling). I examined the effect this practice has on the biodata.

Methods

I examined the true ages, as determined at the Rose Lake Wildlife Research
Station (see Question 1) of deer aged as ‘A’ by field personnel. The sample size was too
small to conduct similar analyses on deer aged as ‘AA.’ 1 determined the true age
structure of the ‘A’-aged deer in both sexes and compared it to the age structure of all
known-age checked deer. I also tracked the change in the percent of deer aged as ‘A’ or
‘AA’ over time and examined how it differed between highway check stations and other

check stations.

Results and Discussion
Annually, an average of 1500 deer are placed in the ‘A’ age category and almost
2000 are placed in the ‘AA’ age category, leaving an average of greater than 10% of the

checked deer without known ages (Table 18). The loss of these data may be unavoidable.

77

Table 18. Number and percent of checked deer aged as 'A' or 'AA' each year.

 

 

 

 

Year A AA Total Deer % A or AA
1987 967 1954 27948 10.45
1988 1071 2023 29264 10.57
1989 4026 3639 29022 26.41
1990 1068 2297 2961 l 1 1.36
1991 1 172 2282 27224 12.69
1992 1302 1889 22966 13.89
1993 923 1585 21213 11.82
1994 1701 2188 32853 11.84
1995 2287 2096 32780 13.37
1996 1418 1721 30267 10.37
1997 1153 1427 29351 8.79
1998 1407 1327 35718 7.65
1999 1570 1455 42769 7.07
average 1543 1991 30076 12.02

 

78

 

The only cause for concern would be if the true age distribution of the unaged deer was
different from that of the true age distribution of the known-age deer. The Rose Lake
aging procedure identified almost 6% of the ‘A’-aged deer as fawns, which should not be
included in this age category at all. These results are somewhat surprising, as the fawn
age category was the only category in which the field personnel achieved 100% accuracy
in the Rose Lake aging analysis (Table 2). The age composition of the remaining deer
(Figure 20) matched fairly closely with that of the true age structure of the checked deer
(Figure 21). The ‘A’-aged male deer’s true age distribution matched more closely to the
actual age distribution than that for females probably because agers can use the antlers to
help determine the ages of the younger males. Although the true age distribution of the
‘A’-aged deer did not match exactly with that of the aged deer, the differences were not
great and do not seem cause for concern. If the differences had been larger, the loss of
information resulting from unaged deer could have been affecting certain age categories
more than others, causing the aged deer to present a false age distribution. The minor
differences and small sample sizes involved do not appear to cause a problem, however.
I can only assume the same is true for the ‘AA’-aged deer, but a similar analysis is not
possible due to small sample sizes of ‘AA’-aged deer sent to Rose Lake.

Recently, the percentage of deer categorized as ‘A’ or ‘AA’ has decreased (Table
18). Although the percent of deer aged as ‘A’ has remained constant (Figure 22a), the
percent of deer aged as ‘AA’ has decreased over the past several years, leading to the
overall decrease in the percent of deer with unknown age (Figure 22b). The decrease
may indicate that agers are becoming more confidant in their aging ability due to

experience or increased training. Highway check station agers tend to categorize deer as

79

I1 212 133 E4 [1115 56+

70.26

Percent of A-aged Deer

232; 204 0.23 0.11

 

Females Males

Figure 20. Age structure of the female and male deer classified as 'A' and
aged at Rose Lake (in 1999), excluding fawns.

80

'1 B2 E313 154 IlllS 86+

Percent of Aged Deer

Females

68.10

8.57

335555232; 1-52 0.29 0.18

 

Figure 21. Age structure of all female and male deer checked during 1999,

excluding fawns.

81

(a) deer aged as 'A'
30

 

 

: A
.5 /\
/\

10

 

 

 

 

Percent of Checked Deer

 

 

 

 

 

 

 

0 r j I 1 l l T T l l

1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999
Year

(b) deer aged as 'AA'
14

 

 

 

12

/ \

10 -

3r

 

 

 

J

. r\ “'~
.I
,l

1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999

 

 

Percent of Checked Deer

 

 

 

 

 

T r

Year

Figure 22. Percent of checked deer aged as (a) 'A' or (b) 'AA' during all seasons at
highway check stations (solid line) and other check stations (broken line).

82

‘A’ or ‘AA’ with less frequency than do agers working at other check stations (Figure
22). The only exception to these trends was in 1989 when weather conditions resulted in
a higher than normal proportion of frozen deer and aging was difficult (H. Hill, MDNR,
personal communication). Highway station agers may age fewer deer as ‘A’ or ‘AA’
because there are generally more personnel working at highway stations at any given
time than at field stations. If questions arise as to the age of a particular deer, a highway

ager can ask the opinions of others while the field ager may not have that opportunity.

83

Question 9: At what level do sample sizes provide the desired level of precision for

statistical analyses?

Significance

When the biodata are collected, information is recorded on where each deer was
harvested. The location information is recorded as township and range, deer management
unit, and county. Based on the county designation, the data can then be combined into
the eight wildlife division management units and the three ecoregions (see Figure 1 for
boundaries). The location information can then be used to stratify the data for statistical
analyses based on several different spatial scales. Statistical analyses are most valid
when the data fulfill minimum sample size requirements. I examined several subsets of

the biodata to determine what scales provide the minimum required sample sizes.

Methods

In most cases, the biodata are used to determine proportions (proportion of
antlered vs. antlerless, fawns vs. does, lactating vs. non-lactating, etc.). In such binomial
situations, the minimum required sample size can be approximated using the following
equation:

n =(22a/2P(1'P))/52

where n is the required sample size, za/z is the value on the standard normal curve that
corresponds to (I-a9 percent confidence, p is the proportion of interest, and E is the
desired margin of error (Wackerly et al. 1996). The margin of error can be quantified

using either relative error or absolute error. Relative error is expressed as some percent

84

of the proportion of interest. For example, a proportion of 0.30 with a relative margin of
error of 20% would have a confidence interval of 0.30 :1: 0.06, because 0.06 is 20% of
0.30. Absolute error is expressed as a number of percentage points. For example, a
proportion of 0.30 with an absolute margin of error of 20% would have a confidence
interval of 0.30 i 0.20. The values for relative error change with each estimated
proportion, so I used absolute error in all calculations.

Assuming maximum variance (p = 0.5), to achieve a 20% margin of absolute
error with 95% confidence, at least 24 deer must be sampled from the area of interest,
whether it is a county, management unit, or region. The required sample size for a 10%
margin of error with 95% confidence is 96 deer. The required sample size for a 5%
margin of error is 385 deer. These numbers reflect the worst case scenario. If
information is already available on the proportion of interest, the preliminary data can be
used to provide an estimate of p, and the resulting required sample size will be smaller
than those described above.

I divided the biodata into several different categories based on season harvested,
age, sex, and harvest type, and on several different levels (county, management unit,
region, and, in one case, township and range). Although the DMUs may provide the
most useful divisions for management purposes, they are generally too small and too
variable to be of practical use for statistical comparisons. I first evaluated the data based
on the above selected sample sizes. I then calculated several ratios that may be of
greatest interest to MDNR managers to determine whether or not the data are sufficient to

provide 5%, 10%, or 20% margins of error.

85

Results and Discussion

Although the biodata are collected throughout all deer seasons, the data are
sometimes sub-divided and examined separately by season. The most common divisions
are between the archery season and the firearm season. The number of deer checked
from archery season has doubled since 1987, but the archery season generally provides
fewer data than the firearm season in all categories (Figure 23). The sample size of the
archery data could be especially significant for analysis of the lactation data. The archery
season provides the most accurate data for lactation analyses because does start to wean
their fawns during the fall and generally stop lactating sometime during the fall or early
winter (Scanlon and Urbston 1978). Therefore, the archery data may provide lactation
data before a large proportion of does have stopped lactating. The archery season
provides much smaller sample sizes than the later firearm season, however, providing
less precise estimates. The appropriate scale for analysis of the lactation data is discussed
more completely in Chapter 3. The rest of this section will focus on the sample sizes
from firearm season only or all seasons combined.

The UP has the smallest sample sizes among the regions, although all regions
contain at least 100 records in almost all categories (Table 19). The UP has the lowest
harvest levels (Figure 24) and the fewest check stations (Figure 2) among the regions,
leading to the fewest checked deer. The regional sample sizes are sufficient across
almost all categories, however, to calculate several different ratios with a margine of
absolute error of less than 5% (Table 20). The ratio of lactating to non-lactating does
using the October data is the only ratio whose absolute error exceeds 5% in two regions

(Table 20).

86

Number of Deer Checked

 

40000

 

35000
p

30000 /

 

 

I— -l\ /
25000 rial!- "!-.

 

20000 \\~ 4

15000

 

l 0000

0

I T I T 1

1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999

 

 

 

 

Year

Figure 23. Number of deer checked from archery (solid line) and firearm
(broken line) seasons from 1987 through 1999.

87

 

 

 

mowm wcmm .w: 39.. 03... ammm 3% m9... £355..
.53.... 2.55:5
3.0 28 owmm RS mmom mac: 3... ~33 «.355.—
.539. F5537.
8N. chem wwm ommm m . m omom anm «can 2.5.5.5..
5......—
Efifiuaa. 55G .36... 82.5.2.4 .5535. 5:32.... 85.5.25. 8.25 .55: .33. 5&3.

+m.N wag—«8w ”5.2a“;

 

 

 

.8305 35. So... m5~.m 5.2.88 .223.— 588 .0 8.9538”. .3 53a...

88

 

# deer harvested
D < 500

500 - 1000
1000 - 4000
4000 - 8000
> 8000

IIID

 

 

 

  

Figure 24. Estimates of the number of deer harvested by county from the

1999 mail survey.

89

.MZDS. 2.. .3 .55.:23 8.68. 8:85» 6\6 55 8 85.23.85 6. 2.8 6.2.... .23

 

90

 

 

 

 

 

m m m w m m S E on o 5882.2 3.3362 6.262
m m m w 6 6 66 8 2 2. 2 32663 <3 3+ 82.382
m m 2 w 6 2 62 2. o 222 28860 3+ 82238..
m m m w 22 6 226 mm 6 N 22... 3+ 82238.2
m m m w m w R 66 6m 6 .2582... .33: 62.2228»
m m m w 22 w 8 E 8. m 22.... 3232 52.22.32,
m m m w w 6 63 t. 2 2 2. 22882.2 8:365...
m m .6. 22 22 n t. 226 2 N 22... 8:35....

2.223122 2.222": $612.2 .2231: 2.22218 $612.. .5223": $222": 2.6": 32. 622 828m 2...:

8172. 2228: 3172. 2225 2.6222 .8172. 322255

 

 

 

823.6 3m. :0 .568. 5.5 652.88 ..< .6\6om
2o .6\oo. .23m .0 Am. 885 8282.23 .8 8868 a 8.3 60.28 .636... 2.2 86.25.25 8 85.05226 58 63.6 5.2.8266 562.3 6:6.on .683 .38:

858586868 $5.885 m: 238:: 2.2 .686 .886: .56.. 2.. 86.3.8 9 88658: 86.6 w:.mm.8 65.8228 .0 52.828 2.... .3 538,—.

The Southeastern LP and the Eastern UP contain the fewest records among the
management units (Table 21). The Southeastern LP is geographically the smallest
management unit, and the greater metropolitan area of Detroit makes up a large fraction
of the management unit, leaving little area available for deer harvest. The Eastern UP has
the second smallest area of the management units and contains none of the counties with
the highest harvest levels (Figure 24). There are also very few check stations in the
Eastern UP compared to other areas of the state (Figure 2). Other than the Southeastern
LP and the Eastern UP, the management units have recently contained at least 100
records in almost all categories (Table 21). The number of checked deer has fluctuated
greatly over the past 13 years, and most of the management units saw dramatic increases
in 1999 (Figure 25). The recent increases in the number of checked deer are probably
due primarily to the increased interest in checking deer for TB, but they provide greater
sample sizes for greater statistical precision. In general the management units do provide
sufficient data to analyze the data with margins of error less 10%, although many
management units do not have sufficient data to calculate ratios with absolute margins of
absoluate error of less than 5% (Table 20).

Although the regions and management units provide large samples with which to
statistically analyze the deer harvest and population, such analyses may be most useful
when conducted by county or even smaller units. Analysis at such fine scales may not
always be practical, however. When plotted by township and range coordinates, the buck
data reveal that very few units contained more than 40 checked bucks (Figure 26). The
distribution of the checked bucks suggests that analysis on the township level would not

be statistically valid, especially outside the Northern LP. Analysis on the county level

91

 

 

 

 

nmm wwm mm . Nam EN new can now. .3 586.3555
emu Nmm .mm mmo. cow :3. o. m . ow.m .3 .9550 5.6m
mmo. man. a; men. awn mwm. m3. .65..V .3 5383.....5
Nov. «3. wmm .No. So 83 .mwm Ewm .3.— 3.3.95
camm 3.0 .o. m. 3mm mg. NS» 3% mmnm. .3 53635:}.
Now . .wmm mmv mwvm own oamm whom 38 .3 aback—.25.
.o. Mme E e... no u. m omm. awe. .5 586a...—
ovo. bvwm v. m v9.6. w:V NE. v.3 38 m: 5883
5:3an .509 .33. 82.5.23. 69.2.5. 652.... 662.8..=< 8.25 been .39... 3.5
+m.~ ”—5.58% ”5.23? «sauna—82

 

 

.8360... mom. 802.. 63.6 0.9566 :52 EoEomewE 088 .o 8.955.. AN 2...;

92

—)(—West UP - fl - East UP - - O - -Northwest LP
- + . Northeast LP — O- - Saginaw Bay - A - Southwest LP
- - O - - South Central LP + Southeast LP

 

 

 

 

 

 

    

I
I
12000 ‘ i
I
I
10000 I.
I
k I
/ \4'

 

'0‘ ~'~.o
¥§A__\

4000 ~ ..':,.¢—"
/'G.':G- “.. ,/'/ B--—{

V’ °"--o----o--"

Number of Deer Checked

 

 

 

00......'

 

4

 

\*_-—

0 I I T l I Y I l T l T l l

1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999

 

Year

Figure 25. Total number of deer checked from each management unit between
1987 and 1999.

93

    
 

 

# bucks checked
< 5

5 - 10
10 - 40
4O - 80
> 80

IIIEID

 

 

 

Figure 26. Number of bucks checked in 1999 by township and range
coordinates.

94

may be more feasible, however, and may be gaining more significance for managers,
even though the county boundaries are not ecologically meaningful. The DMU
boundaries, on which harvest regulations are based, have begun to match county
boundaries, especially in the Southern LP, setting a precedent for county level analysis.

Several counties have shown consistently high or low sample sizes of checked
deer. In almost all cases, the TB surveillance counties and Lake, Osceola, Clare,
Gladwin, Allegan, and Menominee counties provide the highest samples sizes, while the
Detroit area counties (primarily Monroe, Wayne, and Macomb) and Brand, Berrien, and
Keweenaw counties have the smallest sample sizes (Table 22). The TB surveillance
counties have high numbers of checked deer due to high harvest levels (Figure 24) and
especially because of the emphasis that has been placed on checking deer from these
counties to test for TB. The high sample sizes in Lake, Osceola, Clare, and Gladwin are
probably due to their proximity to the TB surveillance area and high harvest levels
(Figure 23). Menominee County also has a large harvest (Figure 24), but the high
number of checked deer in Allegan County cannot be fully explain by harvest numbers,
which generally do not fall within the highest levels. Allegan County has abundant land
for public hunting, however, which could increase the percent of deer checked. The
Detroit area counties, Berrien County, and Keweenaw County have especially low
harvest numbers (Figure 24), resulting in low checking numbers. The low sample sizes
of Branch County are probably due to the lack of check stations in the immediate area,
rather than to especially low harvest numbers (Figure 2).

Due to the great variability in the number of deer checked from each county,

county-level analyses should be treated with caution. In several cases, counties contain

95

 

 

 

62 662 6. 662 66 66 666 666 66.3.8666
66 v6. 62 666 66 66 666 666 6666.662
6 66 6 :2 6 6 662 R2 85
66 .66 66 666 66 6: 666 666 63666.5
ficflauoad noon— —a~¢.~. macro—«=4 “5.5—=36 magah mum—ho—u=< axon—m— uowa— 739—. bun—50

 

 

 

 

 

 

+m.N w:...:3> uni—66>
6.85:6... 62.6.3 886%. A...
66 662 62 666 _ 2 66 66.. 666 6666635
6:. 666 262 666 626 666 666 66: 86:66:62
66 666 62 666 62 .6 666 666 8.2.6::
2 66 6 62 6 6 66 66 3656332
662 666 3 666 66 666 666 666 :22
66 .6. . 2 662 6 66 666 666 636666:
66 6.. 62 66 62 66 662 662 636660
66 626 26 .62 66 662 6. 6 63. 68.56.:
666 666 .6 666 66 666 666 2 2 22 6:6:
62 6.62 6 662 6 66 666 62 6 6626:
_ 66 6 62 2 6 6 6: .62 .662
Gama-30am— homn— 32.9 mac—503:4 Echo—=36 magnum mmO—hQ—fl—Aw 9.05m 500A— —3¢H. man-50

+m.~

”5...:on ”—5.88%

 

 

 

6.33:6... .2 .5 8260? A3

6.6.5... 3:. :8... 60.6.6 0.6.866 .358 0:86 .0 626.86%. .NN «Bah.

96

 

 

 

66 66. 66 666 26 66. 666 666 6.6.5.5
666 666 .6. 666 66. 666 666 666. 6.68.5
66 66 66 66 66 66 66 .6. 6668:
66. .66 66 666 66 666 666 666 66663672
66. .66 66 666 66 666 666 666 8.66.6.2
66. .66 66 666 66 666 6.6 666 3.862
66 66. .6 66. 66 66. 66. 666 66.62
66 66. 6. 666 66 62 2 6. 6 666 8.6.662
62 66 6. 66 66 .6 66 66. 6666.63
666 .66 66 666 66 666 666 666 3.3
62 2 66. 66 .66 66 66. 666 666 5.3.3.
26 .6 6. 66 6. 66 66. 66. 8.62:6 .5
66 .6 6 66 6. 66 66 66. 6.65..
:cflauoaw— noon— ?aQH. mac—.33: fiQHO=E< 3.33m me—hO—u—Jw mac—um noon— ?«OH. .0550

+m.~

”5.....on «£168»

 

 

 

6.35:6... .36.. 826635.87. .6.

.6668. 66 6.66.6

97

 

 

 

66. 666 .6 666 66 666 666 666 66.6588...
666 666 6.. 666 66. 666 666 666. 6.6. 6:68...
66. 666 66 66. 66 666 666 666 666.5
666 666 6.6 66. :6 .66. :6 6.6. 6685
666 666 6. . 666 6. . 666 666 .66. 36.560
.66 666 66. 666 6.6 666. 666 666. 6666.65.52
66. . 666 66 66. 66 666 666 .66 866.
66 66. 6. 6. . 66 66 66. 666 .25....
66. 666 66 66. 66 666 .66 666 6.6.5.0
66 66. 66 66. 66 66. .66 666 66662.65
66 66. 66 66 66 66. 66. 666 6.68.5.0
66. 66. 66 66. 66 .66 666 666 55.2
666 666 6.6 .66 6.6 .66 666 666. 686:6
666 666. 666 666 666 666. 666 66.6 2.8.6
flown-336..— 5009 Each. amaze—«Ed. fluke—«:4 mutah mac—ho—u=< mac—um .30: Each. mun—50

+m.~ 32.68» ”—5.58%

 

 

 

6.355.. .53.... 886605.87. A...

.6668. 66 6.66.6

98

 

 

 

66. .6. .6 66. 66 666 666 666 6.82.6
66. 66. 66 66. 66 6.6 666 666 86:66
66. 66. 66 .66 66 666 666 666 3.3.666
66 66. 66 .6 .6 666 66. 666 65.6.2
66 66. 6. 66. 66 66. 66. 666 6:662.
66 66 .6 66 66 66. 66 666 :26:
6 66. 6 66. 6 66 .66 666 6.3.5.0
666 666 66. 666 66. 666 666 666. 266
66 .6 6. 66 6. .6 66 66. 66m
6. . 66. 66 6.6 .6 666 666 666 852

.533an human Quah. amaze—2:6 gnu—ufldV magah macro—«:4 axon—m haun— _auG.—. 63:ch

+m.~ mum—63> mum—.63?

 

 

 

6m Bafiwmm o

.6683 66 636,6

99

 

 

 

 

66. 66. 66 66. 6.. 666 .66 666 8...: 66>
.6 .6 66 66 66 66. 66. 666 6668.. .66
66 66 6. 66 6. 66 66 6. . 6:35
66 66 6. 66 6. 66 66 66. 6668.65,.
66 66 .6 66 66 66 66. .66 :5.

66. 66. 66 .6. 6. . 666 666 666 86656.3.
66 66. 66 66 66 66. .6. 666 6.6.6
66 .6 6. 66 66 66 .6 66. 6666.5
6. 66 6 66 6. 66 66 66 62.2:

. . 66 6. 6. 66 .6 66 66 86:66.

66. 666 66 666 66 666 6.6 .66 6:66.
666 666 66 666 66. 666 666 666 6666.2

:Oflauoafl noon— —30H. auto—236. tune—«~36 agar..— mmG—ho—u:< 395m— .309 Each. 60:50.0
+m.~ mam—.88» ”um—63>

 

 

3:65.566 6033 58603833 QV

6.6.6686 66 636,6

100

 

 

 

.6 66 66 66 66 66. 66. 666 366666663
66. 66. 66 66. 66 66. 66. 666 866636.66
66 66 66 . .. 66 .6. 66. .66 8.86662
66 66. 66 .6. 66 666 66. 6.6 6866635
66 66 6. 66 .6 66 66 6. . 85:3
66. 66. 66 66. 66. 666 666 666 59.8..
.6 66 66 66 66 66. 66 666 6.66.
66 66 66 66 66 66. 66 666 82.6...
66 66 6. 66 66 66 66 66. 2666...:
66 66 6. 66 66 .6 66. 66. .2220
66 66 6. 66 66 66 66 66. 66.66.
66 66 .6 . . . 66 66. 66. 666 .8250

aoflaaoaa— Loon— —a«¢.—. 32.—03:4 3.5—«=4 mug-6h mum—503:4 axon-m— yoca— Euch. 63:ch

+m.N mum—63> «65.30%

 

 

 

Esmficom .833 .8650 £65m @

.6668. 66 6.666

101

 

 

 

6 6. 6 6 6 .. 6. 66 2.663

mm. 66. .6 8. .w .66 mo. 696 :30 am

66 mm mm mm 6.6 66. 66. EN 2:63.60

2 a m m. 2 mm w. 6.6 8.532

o. c. o m. 3 mm mm .6 6.883).

66 .6 66 .6 66 66. 66. 666 52.3

mum 6:6 co. mam mm. 6.66 mmm 6N3 362.00
683325 63: .33. 662.6222. coho—=26 .6.—Sam 666—3.6:«6 66.25 .63: :35. b.5660

+m.~ mam—6.63» $5.3“;
636.5666 60304 586855 Ev

 

 

 

66.6686 66 6.6a...

102

no records in a particular category or the data is insufficient to provide estimates with
absolute error of less than 5%, 10%, or even 20% (Table 20). Less than half of the
counties provide sufficient sample sizes to calculate any of the ratios in Table 18 to
within 5%. These problems are not as significant on the management unit or regional
levels. On all levels, although the data may appear to provide an accurate picture of the
deer that are checked, the accuracy does not extend to a picture of the total harvest or the
total deer population. As discussed earlier, the population of checked deer is not

necessarily similar to the population of harvested deer.

103

Question 10: What errors does the database contain?

Significance

The MDNR field personnel record the biodata on datasheets (Appendix 1), which
are then collected at the Lansing offices and transcribed into the computerized database
(Appendix 2). Although several checks are run to check for logical errors (such as
inconsistencies in the location information), the final SPSS data file may contain errors or

inconsistencies that make data analysis difficult.

Methods

While working with the biodata, I discovered several problems that made some
analyses difficult without additional cleaning of the data. Some of my observations are
of errors or inconsistencies within the data. Others, however, are comments on missing
information whose inclusion into the database may make possible several analyses that

cannot currently be performed.

Results and Discussion

Quality checks are run on the biodata immediately after they are entered to check
for inconsistencies in the location data (for example, checking for township and range
coordinates that do not fall within the reported county). The checks, however, do not
identify inconsistencies between the date reported in the “remar ” column and the
season in which the deer was harvested. The date in the “remarks” column identifies the

date on which the hunter or ager observed the lactation status of the deer. Although the

104

date of harvest is not recorded in the biodata, the lactation date provides a close estimate
of the harvest date for a small subset of the biodata. The lactation date should fall within
the boundaries of the season in which the deer was harvested, or a few days thereafter, as
it most likely reports the date on which the hunter dressed the deer. A comparison of the
dates of the firearm and archery seasons and the date reported in the “remarks” column
reveals that 2.87% of the regular firearm deer and 1.58% of the archery deer fall outside
the bounds of their respective season (a total of 871 deer from 1993-1999). Although a
few of these inconsistent records are probably caused by hunters not checking the deer
until after the season ends, it is unlikely that this is true for all the deer, especially those
reported several weeks after the season ended. This explanation also could not apply to
deer reported before the season in which they were supposedly harvested. The hunters
may be reporting the date wrong, or the recorder may be writing the date down wrong.
These results suggest that a more accurate date to record for lactation status would be the
date on which hunters harvest their deer, not the date they field-dress their deer.

Many of the variables contain inconsistencies that make comparisons among
different categories within the variable or comparisons between different variables
difficult. The most common is the notation used to indicate missing or illegible data. For
many of the variables, ‘99’ indicates missing data, but blanks are also used, and in the
“tb” variable, ‘0’ is used to indicate no data could be collected. Several different
annotations for missing data make it difficult to determine exactly how many records are
missing data. Compounding the problem is the practice of leaving the “cseason” column
blank to indicate the deer was harvested during firearm season. When examined in the

context of the rest of variables, the blank would appear to indicate the season is unknown.

105

In the comparison of the season dates and “remarks” dates, the firearm season had a
greater percentage of inconsistencies than the archery season. The difference may be due
to the fact that some “cseason” entries were lefl blank if the season was not reported or
was unknown. The blanks would automatically place the deer in the firearm season,
causing greater inconsistency in the firearm season than the archery season.

One easily remedied inconsistency is found in the annotation of dates. Dates are
recorded in both the “remarks” column and the “date” column. In both cases, numbers
less than 10 are recorded as both single and double digits (with a leading zero). For
example, the date ofJanuary 1 may read as ‘1/1,’ ‘01/01,’ ‘01/1,’ or ‘1/01.’ Some dates
in both columns may also include the year while others do not. These inconsistencies
make recoding these dates into the “sepldays” column (number of days since September
1) and picking out individual dates difficult and tedious.

Since 1997, the MDNR has held late firearm seasons, generally sometime during
the last two weeks of December through the first week of January. During the first year,
data collected during the late season was recorded as the ‘H’ season (holiday season)
while the season was recorded as ‘L’ (late season) during the following years. The
different symbols made the data appear to have come from different seasons. In a similar
case, some (fewer than 50 records) of the original data under the “prv_pub” variable were
coded as ‘ l 1’ and ‘22’ rather than ‘PB’ or ‘PT’ with no indication as to which code stood
for private land and which stood for public land.

Other problems I observed in the database concerned the omission of data that
may be necessary to include. One significant omission is the check station location for

each deer. Each datasheet gives the name of the check station from which the data were

106

collected. When the data are transcribed in the computer database, however, the check
stations are lumped according to the district (1987-1997) or management unit (1998-
1999) in which the station is located. Even the combinations are not consistent, however.
Several check stations are recorded individually, including all highway check stations,
the Marquette, Roscommon, and Lansing Field Offices, Drummond Island, and the
Houghton Lake, Cusino, and Rose Lake Wildlife Research Stations. Identifying the
individual stations at which each deer was checked would allow the MDNR to determine
patterns of check station usage and identify those areas that need additional check
stations or check stations that may be underused. Knowing the check station location for
each deer could also aid in determining how many heads each station receives for TB
testing to aid in planning to meet TB testing quotas. Check stations could also be
individually evaluated for quality control. The original datasheets also include individual
page and line numbers, which provides each entry with an individual code. Although
these page and line numbers are included in the individual yearly databases, the
comprehensive database of all years does not include these variables. This omission
makes it impossible to track an entry back to the original datasheet if a question arises
about the data.

A third type of data problem is the coding used to determine the antler status of
the deer. Antlered deer are those deer with at least one antler greater than 3 inches long.
All other deer are antlerless. Although this definition appears straightforward, two
different variables code the deer into the ‘antlered’ and ‘antlerless’ categories and are not
always consistent with one another. The original variable, “killtype,” first lists antlered

deer as all males that are not fawns, A, or AA. It then determines that all antlerless deer

107

are all females (not age A or AA), all male fawns, and all males with spikes. All others
(including only all A and AA deer) are considered unknown. 1 developed a second
variable, “antler,” which first lists antlered deer as all males with points but not short
spikes. Antlerless deer are then listed as all females, all male fawns, and all males with
spikes. The unknowns are then all non-fawn males missing points data (including those
coded as ‘99’ in the “points” column) and all males with a ‘B’ in the “points” column.
The primary inconsistencies in the two variables lie in the ‘unknown’ category. If
“antler” and “killtype” described antler status in the same manner, Table 23 would have
totals along the diagonal and zeros elsewhere. As they stand, the variables are not
consistent with one another. The “killtype” variable lists five deer as ‘bucks’ which the
“antler” variable lists as ‘antlerless,’ but the primary inconsistencies in the two variables
lie in the ‘unknown’ designation. The “killtype” variable describes only deer of unknown
age as having unknown antler status (Table 23). The “antler” variable codes deer with
unknown age in the same manner as deer with known ages, and describes only males

with broken antlers as having unknown antler status.

108

Table 23. Crosstabulation of records between the "antler" variable and the "killtype"

variable.

 

"Antler"
Unknown Antlered Antlerless Total
Unknown 1625 32061 12176 45862
Bucks 14553 203432 0 217985
Antlerless 1 84 5 126965 127154

Total 16362 235498 139141 391001

 

 

 

 

"Kiutype"

 

109

Conclusions
The check station biodata can be used to estimate population parameters and to
develop population indices of Michigan’s deer population. Direct counts of the total deer
population are impossible and impractical due to the size of the state and the number of
deer in the population. Hawn and Ryel (1969) compared a direct count survey with a
sampling survey of harvest estimates in Michigan and found that the sampling survey
was more efficient and more accurate than the direct count. Measurements of such
aspects of the deer population as antler dimensions or lactation status are likewise
impossible on every individual deer. The check station surveys can therefore save
valuable financial and personnel resources while providing accurate data. The biodata
allow wildlife managers to draw inferences about the true population based on data
collected from a fraction of the population. The biodata, however, cannot be used
confidently unless their biases and limitations are understood.
This evaluation identified several compositional, spatial, and seasonal biases in
the biodata.
o Hunters tend to check a greater proportion of the harvested antlered deer than of
the harvested antlerless deer.
o Counties with at least one check station tend to have a greater number of checked
deer than those without check stations.
0 The TB surveillance counties check a greater percentage of their harvested deer
than other counties.
0 Deer harvested during the firearm season are more likely to be checked than deer

harvested during the archery season.

110

o In the LP, hunters are more likely to check deer harvested on public land than
those harvested on private land.

Although some biases are inherent in the voluntary checking system, the biodata
can still provide valuable information about Michigan’s deer. The lactation data or beam
diameter data could be used to provide indices of herd health. Other data could be used
to compare the harvest composition between different years or geographic areas. The
geographic distribution of the biodata could also be examined in conjunction with
ecological data to delineate ecologically significant areas that contain sufficient sample
sizes for statistical analyses. Such areas could take the place of the current DMUs, which
are too variable and sometimes too small for statistical analyses. Burgoyne (1981) argues
that biased data can still be used in developing estimates or indices of population size.

The quality of the biodata can also be improved, and the effects of biases can be
reduced. Estimates derived using the biodata may have greater precision if the values are
weighted by the harvest levels of the particular geographic area or by the seasonal
distribution of the data. Further study is necessary to determine if such estimates would
be more accurate than those calculated without weights. The distribution of the check
stations could be examined and reorganized to increase the percent of deer checked and
to decrease the effects of spatial biases. The MDNR could educate hunters on the
importance of checking their deer and the value of the data collected. Such an education
program could increase participation and decrease compositional or seasonal biases.
Increased efforts in personnel training and in planning could also decrease the occurrence

of errors in the collection and transcription of the data. The biodata are a valuable

lll

resource for Michigan’s deer managers and every effort should be made to maintain or

improve the quality and utility of these data.

112

CHAPTER 3

THE LACTATION SURVEY EVALUTION

Introduction

At voluntary deer check stations, the MDNR has collected data since 1993 on the
number of female white-tailed deer who show evidence of lactation. Ideally, the data
could be used as an indicator of the state of the herd based on the assumption that higher
reproductive rates, especially within the yearling population, indicate a healthy herd.
Doe fawns will not breed at 6 months old unless they have reached a minimum body size;
the quality of the habitat and the density of the herd determine the percentage of doe
fawns who reach puberty during their first year (Jacobson 1994). Thus, high yearling
reproductive rates would indicate sufficient high quality food for the deer population.
Antler measurements are also an indicator of herd health (Severinghaus et al. 1950,
Richie 1970). Correlations between average buck antler size and doe lactation rates
could support the use of lactation rates as an indicator of herd health. A second projected
use of the data is as an estimate of annual recruitment, assuming that only those does
whose fawns have survived through the entire summer will still be lactating during the
hunting season, when the lactation data are collected. I examined the distribution and
abundance of the lactation data, the accuracy of the data, how the data could be used in
deer population management, and on what temporal and spatial scales the data provide

the most accurate and useful information.

113

Methods

The source of the data examined in this study are check-station records collected
between 1993 (the year lactation data were first recorded in Michigan) and 1999. The
records are stored as an SPSS file that contains the biophysical data from all check station
records from 1987 to 1999. I eliminated all records that did not contain lactation data, all
data from fawns and any deer whose age was recorded as ‘A’ (not a fawn but cannot be
aged), and any record where the date of observation of lactation was recorded as later
than the date the form was completed (due to some transcription or typographical error).

I sorted the data to determine how it was distributed among several different
categories including age, year killed, geographic location, and season the lactation data
were collected. For most analyses the deer were divided into 3 age categories: 1.5 years
old, 2.5 years old, and 3.5 years old and older. The division excluded the data from deer
classified as ‘AA’ (2.5 years old or older), which made up less than 10% of the data in
each year. Geographic divisions were based on Wildlife Division management units and
regions, first designated in 1998 (Figure 1). Data collected before 1998 were assigned to
the appropriate management unit based on the county in which the deer was reported
killed. The regional boundaries were based on the ecoregions defined by the MDNR
(Figure 1). Seasonal divisions used were October 1-31, November 1-14, November 15-
22, and November 23-30. Seasonal divisions were assigned based on the date that
evidence of lactation was or was not observed, recorded in the “Remarks” column of the

Deer Physical Data sheet (Appendix 1).

114

I chose the divisions used in the final analyses because they contained a sufficient
number of records. Minimum required sample sizes were calculated using the following
equation:

n =(z’6p(1-p))/52
where za/z is the value on the standard normal curve that correlates to (l-a) percent
confidence, p is the proportion of lactating does, and E is the desired margin of error
(measured as an absolute percent) (Wackerly et al. 1996). In a binomial distribution, if p
is unknown, a p value of 0.5 is used to calculated minimum required sample size because
it provides maximum variance. Data are already available on the lactation rates,
however, and they can provide an estimate of p, which can be used in the above formula
to provide smaller required sample sizes.

Afier determining the appropriate scales on which to analyze the lactation data, I
examined the trends in lactation rates associated with age and geographic location. I also
tracked annual trends in lactation rates. Finally, I determined the correlation between
annual lactation rates of yearlings and adults and the average annual beam diameter of

yearlings and adults.

Results

Lactation data are available on approximately 76% of all mature does (1.5 years
old and older) that came through the check stations between 1993 and 1998, resulting in a
final sample size of 37,100. Annual sample size increased 503% from 1993 (n=1 ,923) to
1999 (n=9,665), largely due to an increase of 363% in the number of female deer checked

during the same years. Hunters’ observations of the presence or absence of milk when

115

field dressing the deer provided 98.1% of the lactation data. Check station workers
collected the remaining data while completing the physical data sheet entry on the
animal. The quality of the lactation data is therefore dependent on the hunters’ ability to
correctly identify whether or not a deer is lactating and accurately report their
observations at the check stations.

If the hunters assign the deer into one category or another without sufficient
observation to determine the true state, the data would not adhere to basic biological
patterns. The dates of weaning should vary among the deer due to differences in birth
date, experience of the mother, and environmental conditions. The variation should
cause the percent of deer reported lactating to decrease as the hunting season progresses,
if the lactation status is not assigned randomly. The percent of lactating deer does
decrease from October to December (Figure 27). Ozoga et al. (1994) report that only 5%
to 60% of doe fawns breed each year while more than 95% of older does breed. The
available lactation data reflect this situation and show that the percent of lactating
yearlings during the hunting season is lower than that of the older does (Figure 28). The
above evidence suggests the hunters do not randomly assign lactation status; they appear
to be able to determine accurately the lactation status.

Assuming maximum variance, a sample size of 385 gives an estimate accurate to
within 5% with 95% confidence. A sample size of 96 gives an estimate accurate to
within 10% with 95% confidence. When the available lactation data are used to estimate
the true variance, smaller sample sizes are required and vary depending on the estimate

used. I looked at both the optimum sample sizes and the estimated required sample sizes

116

 

70%

 

60%

 

50% \

 

 

 

50
Q
E \
3 40%
a
A \
fl
5 30% -
O
in
.. \i
a.
20%

 

10%

 

 

 

0%
October November December

Month

Figure 27. Percent of adult females lactating in Michigan from October to
December, averaged from 1993-1999. Error bars are :t 1 standard error.

117

 

70%

60% - ‘%

40%

 

 

 

 

 

 

30%

Percent Lactating

 

20%

 

10%

 

 

 

0% 1 I I
1.5 yrs 2.5 yrs 3.5 yrs 4.5 yrs 5.5+ yrs

Age

Figure 28. Percent of females in Michigan lactating in each age class (averaged
from 1993-1999 using data from all seasons combined). Error bars are d: 1
standard error.

118

to determine whether the data are sufficiently abundant when divided into different
geographical areas, years, and seasons.

The earliest available lactation data should provide the most precise and
representative estimate of true reproductive rates because the percent of lactating deer
decreases as the season progresses (Figure 27). Data from October and early November
would thus be the most useful, but the sample sizes are generally much smaller than those
from the November firearm season (November 15-30). The first 2 weeks in November
do not provide sufficient data to be statistically useful. A chi-square test demonstrates
that there is a significant decrease in the lactation rates from the first to the second week
of the firearm season (Table 24). The October data and the first week of firearm season
therefore provide the most potentially useful lactation data. The largest sample sizes, and
therefore the smallest margins of error, are found in the regional Firearm Week 1 data
(Table 25). Among the management units, the Eastern UP and the Southeastern LP
sample sizes are generally too small to provide accurate estimates of lactation rates
(Table 25).

The October data can be used to estimate the minimum proportion of lactating
does in each region. Percent of lactating does increases with increasing age within the 3
regions and decreases with increasing latitude within the state (Figure 29). The lactation
rate for 1.5 year old does in the Southern LP is twice that of the 1.5 year old does in the
northern regions of Michigan. While the difference is not as dramatic in the older age
categories, the percentage of lactating does in the northern regions is consistently lower

than in the Southern LP. Between the first and second years of life, reproductive effort

119

Table 24. Results of a chi-squared test comparing the proportion of lactating does
in the first week of firearm season to that of the second week of firearm season.

 

Ream Age x2 value df p-value

 

Upper Peninsula 1.5 0.212 1 0.645
2.5 0.547 0.460
3.5+ 11.313 0.001

 

Northern Lower 1.5 2.706 0.100
Peninsula 2.5 3.637 0.057

 

 

Southern Lower 1.5 1 . 181 0.277
Peninsula 2.5 44.254 <0.001
3.5+ 22.27 <0.001

 

1
l
1
l
3.5+ 10.036 1 0.002
1
l
l

 

120

Table 25. Average annual number of lactation records and the average annual margin
of error associated with the estimates of the percent of does lactating in each region
and management unit. The margins of error are based on absolute percentages. All
values are averaged from 1993-1999.

 

 

 

 

 

 

October Firearm Week 1
Region/U nit Age 11 Error 11 Error
UP 1.5 23.9 12.95 92.6 6.44
2.5+ 71.0 11.40 293.1 6.53
Northern LP 1.5 144.9 7.57 389.7 4.33
2.5+ 377.6 6.00 1174.0 3.70
Southern LP 1.5 106.6 9.58 407.3 4.55
2.5+ 154.3 6.82 654.6 3.91
Western UP 1.5 35.0 16.08 138.3 6.58
2.5+ 68.7 12.89 332.0 6.75
Eastern UPll 1.5 48.3 11.74 207.9 15.74
2.5+ 125.6 28.42 515.4 21.89
Northwestern LP 1.5 82.9 15.46 166.1 8.75
2.5+ 185.6 10.42 390.9 8.55
Northeastern LP 1.5 32.0 10.47 137.5 5.69
2.5+ 59.4 8.77 299.3 4.90
Saginaw Bay 1.5 48.9 15.01 216.3 8.94
2.5+ 132.5 11.02 543.5 7.00
Southwestern LP 1.5 85.8 15.71 155.2 6.37
2.5+ 201.4 9.23 400.1 5.49
Southcentral LP 1.5 29.4 17.53 136.8 8.97
2.5+ 58.0 15.45 306.9 7.69
Southeastern LP 1.5 53.7 24.66 242.4 20.71
2.5+ 147.3 20.18 616.6 15.80

 

 

a. No lactation records were recorded from the Eastern UP during the firearm seasons of
1997 and 1998 so these years are not included in the averages of n and the margin of error.

121

131.5 yrs .25 yrs U3.5+yrs

\O
O

 

OO
O

76.76 78.35

 

71.53
I...

\l
C

 

 

 

Q
C
J

 

 

 

 

 

M
O
1

 

40.00

.b
o
r

 

 

w
o
l

 

 

 

Percent Lactating

 

 

 

N
O
l

 

 

 

 

 

 

 

 

6

 

 

 

 

Upper Peninsula Northern Lower Peninsula Southern Lower Peninsula

Region

Figure 29. Percent of females lactating in each age class and in each region
(averaged from 1993 -l999 using October data). Error bars are + 1 standard error.

122

apparently doubles in the Southern LP and triples and quadruples in the Northern LP and
UP respectively.

Although the large categories are useful for observing general trends, smaller
categories, such as by year and management unit, are useful for observing annual
fluctuations and developing area specific management plans. Annual division by region
provides larger datasets than the management units and provide smaller margins of error
(Table 25). October data can also be used for better minimum estimates. Plots of trends
in the percent of lactating does over the past 6 years, as measured during October, reveal
annual variation, especially in the UP (Figure 30, Table 26). The greater variation in the
UP may be due to the smaller sample sizes in the region, or it may reflect greater effects
of climatic fluctuation on the deer population. The fluctuations in annual proportion of
lactating does of the different age categories track each other fairly closely, indicating
that the deer of different age classes are generally affected in the same manner by
environmental variations.

When divided by age class, the sample sizes the first week of firearm season
provide margins of error of less than 10% in all 8 management units except the Eastern
UP and Southeastern LP units (Table 25). The October data provides margins of error of
less than 20%, again except in the Eastern UP and the Southeastern LP (Table 25). The
firearm data of the management units reflect the same pattern as the regions of increasing
lactation rates with increasing age (Table 27). The northern management units also tend
to have lower lactation rates (Table 27).

Among the yearling population, there were no significant correlations between

lactation rates and antler beam size. Contrary to expectations, the UP and Southern LP

123

(a) Upper

Peninsula

- 9' 1.5 yrs —I—2.5 yrs +3.5+ yrs

 

90
80
70
60
50
40
30

Percent Lactating

10
0

1993

 

 

 

{

\

‘I’

%A

H
\

¥
‘

 

 

\

 

 

20¢“

 

.-.'

 

 

...’

 

T

1994

1995

(b) Northern Lower Peninsula

___,.

80

05%
CO

1996 1997 1998 1999

 

~

*—

 

 

W‘

.6

-—::L\1

 

Ur
O

\1

 

.5
O

 

U)
C

Percent Lactating
N
O

————4—Q—-

 

O
‘ ‘

---4L----9._

._._,.-‘----?

 

.—
O

0

 

 

 

j

1

1993 1994 1995 1996 1997 1998 1999

(c) Southern Lower Peninsula
90

80

 

 

\l
O

 

ON
C

 

Jim
CO

 

Percent Lactating
b)
O

N
O

 

.—a
O

 

 

 

 

0 T r l l 1

1993 1994 1995 1996 1997 1998 1999

Figure 30. Percent of females lactating from 1993 to 1999 in each age class and
in each region (using October data). (Exact values and sample sizes can be
found in Table 26.)

124

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

ow oodh mm m fl . mm a N~ .06. mm 2.466. we vméh mm Q6666. mm eBNw +m.m

ca mmJK :. dew 02 3.65 mo 5.06. no vWNw :6 E . m6. 36 mvdw m.N and
m: cmév ON: oofim #2 ©m6m co omfim ca: mm._m we amdm no N936 m4 Eon—25m
w—V _ five hmw Nwfie aom aNAh vw mméo vw coax. fin amdw om whfih +m.m
3N oodm 5mm :48 o6: wm.m© an $0.50 R vaéc mm Ede an hqwo m.N m1—
mMN _m.©~ Ohm 6%.: cm. VN.M_ mm. Nm.—N ca: OBAN ow oodm ov OWN" m.— 6..—05.32
6.0 we. :. om Gown cm whdm mv 966nm mm nqwh mm wmdh NN 35¢ +m.m

56 $.06. w 09mm 2 _N.vw mm dem 3 6.0.00 VN : .Vm c— omdc m.N

wv wwél NN MBNN 2 cod _N de om hoém NN ood : 2.2 m4 an:

a 23 .x. a 25 .x. = and .x. a and 6\e a and ..\e .— 25 .x. a and .x. ow< uemwom

gag waa~ ban eaa— mam vaa— mama
6.3860

Eat 86 869 666.6 29566 366.8666 “6:6 Somme some ... mom. 8 mom. 86E 66.65826. moo—6 60 6.880.” .3 26.6%

125

 

w. NN.N6 NN 00.0w N. 60.00 a 66.66 6 code. 6 .6.0w N code. +0.0
wN 0N.60 w. NN.N6 m . 06.100 .6.. 0w.m0 6 code. 0. 0N.m0 0 oodo. 0.N a...
.6 0a..N Nm 06.06 mm 00.00 m. wo.0N .. w. .w. N. 00.x 0 mm.mm 0.. ...—Smaofiscm

 

0.. 0. d0 6N. .600 6N. 60.60 6@ o6d0 cm 9.00 .6 .600 Nm 0600 +0.0
.. . 0. .00 mm S00 60 6060 06 N000 .0 0 0 .N6 N6 mN.60 6N 0Nd0 0.N 6.1—
.6. oodN 60. 6de N6. 0. .mm o. . 0060 6a 00.6N a6 0w.6N 6 00.60 0.. .6.—6:00 6.23m

 

mm. mN.00 00N 00.60 mm. 00.00 06. 00.60 60 . N600 . . . 66.66 06 66.0 +0.0
O6. 00.60 ow. 66.66 NM. 60.00 06. 00.00 06. 0w.60 66 60.60 66 0000 0.N m6.
8N N060 O6N N660 m6N wN.6N 0mN 06.NN 6ON ma.wN O6. oNdN 6m. 6o.0m 0.. ...—38.55:.6m

 

0cm 0. .00 6a. 5.60 00. 0m. .0 6N. 0o.w0 Na 06.00 .6 0.60 mm N006 +0.0
6.N 0. . . 0 N6. 06.60 N0. .060 6w 5.06 06 66.60 06 0600 w. .560 0.N
aNN a0 .0. 0w. o0.mN m6. 0. .ON 06 60.x. w0 0o.NN .0 06.0 . 0N Na.0N 0.. 66m 36.—3.6m

 

6N. . ow.w6 60m N6.o0 .06 0w.00 wen 66.60 .6. 06.60 66 NN.00 m . N .a.N0 +0.0
mN0 mN.66 0G6 366 O6. 0w.N6 00. 06.06 mm a. .w6 00 $00 0.. 6N.66 0.N mu—
w60 wmd. N66 0N.6. 66. N66. w6. NN.0. w6 NmN. am wm.0 . 6w 00.6 0.. ...—36.35.87.

 

606 00.60 NNm .060 w6m 00.00 0aN 0w..0 m6. mN.o0 w 00.60 am mN..0 +0.0

 

NON wmd0 m0. .060 0w. 0w.66 00. 6w.m6 06 0.00 0 00.00 0N 60..0 0.N 6.x— .
w0N 3.0. .ON 06.6. 06. 6. .6. 06. 0wd. 66. 00.0. o. codN .N 06.6 0.. ...—36935.87.
ON 8.00 o 0 N0 mN.m0 0N 806 o. 0900 6 oo.0N +0.0

6 60.wN o o .6 .6. . m .. 06.06 . cod N .560 0.N

6 00.0 o o .N 06.6 a NN.NN 6 oo.0N . cod 0.. 6.: ...—366m—

 

amN 8.0 Na 6060 ac. .N.m0 0a. 6m.06 Nwm N660 0N. 60.6.0 m0 60.60 +0.0
0N. 66.66 00 NN.66 60 ...N6 N0 N0. . 0 06. 00.66 N wN.w6 00 60.06 m.N
6.. 00d 66 $6. 00 06.0 ow 0N0 ma. 3.0. 60 06.6 06 00.6 0.. ..D ...—88>?

 

.— 056.— a\e fl Gad c\e ~— ufiwm ..\e E van— AX. - UG‘H e\c : 051— 9\9 .— 0&1— a\e 0m< DE

 

 

02.. «a. 6.3. 03. 000. 600. 003

 

 

 

 

 

 

 

 

.. 6.83 8.88.... 88..
86 8.5 666.6 0.6.866 086.8666 .6.... ...... .coEomaamE some 6.. mam. o. mam. 80.0 mass—om. 68.6 00 .5806. .6N 0.6.6...

126

 

had weak negative correlations (Figure 31). Among the adult (2.5+ years old)
population, only the UP had a significant (p = 0.0028) correlation between mean beam
diameter and the proportion of does lactating (Figure 32). Although the Northern LP and

Southern LP had only weak correlations, both correlations were positive.

Discussion

One of the proposed uses of the lactation data is to estimate annual recruitment,
the number of fawns who survive from birth until hunting season, by using the
assumption that the number of lactating does reflects the number of surviving fawns.
Unfortunately, several unknowns and wide variability prevent the use of the lactation
data in this way. Does that give birth to 2 or 3 fawns will continue to lactate even if they
have lost 1 or 2 fawns, so long as at least one fawn still lives. The lactation data cannot
distinguish among these possibilities. One significant unknown is how long a doe
continues to produce milk, even after she has lost all her fawns. Does that are lactating in
October or November may not have a surviving fawn but may still be producing milk.
We also do not know when fawns are weaned. If a doe is not lactating during the hunting
season, it could be because she lost her fawn several months earlier, or because her fawn
survived to hunting season and has been weaned. The lactation data are therefore not
useful as an estimate of deer recruitment.

The data appear to accurately reflect the lactation status of the deer that come
through the check stations (see Results), so they may be useful as a minimum estimate or
an index of reproductive success. In several categories of the data, the significant

decrease in the proportion of lactating deer between the first and second week of firearm

127

0 Upper Peninsula X Northern LP A Southern LP

 

M
#

 

S

o
q
9‘?
D”

8
X

 

—
oo

 

.—
O\

 

—.
h

 

.—
N

 

Average Beam Diameter (mm)
[>
[>

 

 

10 Y T T l l I l’
0.00 0.05 0.10 0.15 0.20 0.25 0.30 0.35 0.40
Proportion Lactating

 

Figure 31. Correlations between the annual average beam diameter and
proportion of lactating does (from October data) among yearlings. (UP:

r2=0.3668, p=0.1495; NLP: r2=0.l929, p=0.324l; SLP: r2=0.1044, p=0.4796).

0 Upper Peninsula X Northern LP
34

 

32

 

3O

 

28

 

26

 

 

24

 

 

22

 

Average Beam Diameter (mm)

 

 

 

20 l T l I T l T l
0.40 0.45 0.50 0.55 0.60 0.65 0.70 0.75 0.80 0.85
Proportion Lactating

Figure 32. Correlations between the annual average beam diameter and
proportion of lactating does (from October data) among adults (2.5+ years old).

(UP: r2=0.8561, p=0.0028; NLP: r2=0.0971, p=0.4964; SLP: r2=0.3909,
p=0.1332).

129

season suggests that the first week of firearm season will provide the more precise
estimates (Table 24). The October and Firearm Week 1 data provide sufficient sample
sizes to be useful (Table 25). The October data are most useful as a minimum estimate of
reproductive success, because they provide the earliest available estimates of the number
of lactating does. Lactation rates drop dramatically after the minimum duration of
lactation of 4 months for does with living fawns (Scanlon 1978). Most Michigan does
give birth in late May or early June (Ozoga et al. 1994) so lactation rates would start to
decline during October. The November data provide larger sample sizes for a shorter
period of time, but by mid November the observed lactation rates no longer accurately
reflect the true proportion of does whose fawns survived through the summer. The
November data therefore do not reflect even the minimum number of reproductively
successful does, but could be used as an index of reproductive success.

The October data show that almost 80% of does 2.5 years old and older are
reproductively active in the Southern LP (Table 26). The Southwestern and Southeastern
LP deer populations have especially high lactation rates (Table 27). What may be more
significant is the high levels of yearling reproductive effort, which reaches a maximum of
almost 40% in the Southern LP and at least 15% to 20% throughout the UP and Northern
LP (Figure 30, Table 26). Yearlings generally only reproduce when resources are
abundant and environmental conditions are favorable (Jacobson 1994). Such high levels
of yearling reproductive effort are a sign of a healthy deer herd. The annual variation in
the percent of does lactating at both the regional (Figure 30, Table 26) and management
unit (Table 27) levels suggests that the health of the herd may be susceptible to climate,

p0pulation density pressures, or other variable factors. The sudden drop in lactation rates

130

in the fall of 1996, especially in the UP (Figure 30a, Table 26), may reflect the severe
winter of 1995-1996 (Langenau 1996). The lower lactation rates in the Northeast LP
management unit as compared to the Northwest LP management unit, especially among
the. 1.5 and 2.5 year old deer (Table 27), may reflect the higher population densities in the
Northeast management unit.

Ozoga et a1. (1994) suggested using both productivity and antler size as indicators
of herd health. Although both productivity rates, as measured by lactation rates, and
antler size, as measured by beam diameter, should increase as habitat conditions improve,
the 2 indicators are not significantly correlated with one another at all among yearlings
(Figure 31) and only in the UP among adults (Figure 32). The same habitat factors that
affect the reproductive success of does may not have similar effects on the antler
development of bucks. Once a doe has lost her fawn due to harsh winter conditions, she
is not able to become pregnant again, so high quality summer habitat and climate
conditions will not increase her chance of reproductive success. On the other hand, a
buck who has experienced harsh winter conditions can benefit from higher quality
summer conditions and still develop large antlers, regardless of the previous winter’s
conditions. These differences may be contributing to the lack of correlation between
lactation rates and beam diameters.

Although the lactation data are not useful as an estimate of deer recruitment, they
do provide some indication of the variation of reproductive effort across age, geographic,
and annual divisions. More data need to be collected in October from the UP, especially
the Eastern UP, and the Southeastern LP to provide more accurate estimates of lactation

rates. Data recorded earlier in the autumn would provide more exact estimates of

131

reproductive success, but collecting such data would be expensive and time-consuming.
Using the October regional data as a minimum estimate and the Firearm Week 1
management unit data as an index of reproductive success are the most reasonable uses of
the available lactation data. These uses would provide sample sizes large enough to

examine both annual and regional differences

132

CHAPTER 4

THE WINTER SEVERITY INDEX EVALUATION

Introduction

White-tailed deer survive harsh northern winters by taking shelter in cedar
swamps and hemlock stands and by supplementing their stored fat reserves with winter
browse (Ozoga et al, 1994; Langenau, 1996). Even so, white-tailed deer populations
suffer annual winter losses, especially during long or especially severe winters. Cold
weather and windy conditions increase body heat loss, and deep snow covers browse and
causes deer to expend more energy when searching for food (V erme, 1968). Verme
( 1968) recognized these conditions as the major factors contributing to winter deer
mortality and devised a winter severity index (WSI) to measure the harshness of the
winters on a scale that would reflect the conditions actually experienced by the deer.

Using Verme’s (1968) index, the MDNR has measured the WSI for the UP and
Northern LP since the winter of 1969-1970 (henceforth winters will be identified by the
first of the 2 years) and in the Southern LP since 1988. Currently the WSI is used
primarily to explain observations of high winter mortality and low reproductive success,
especially of yearlings. The WSI could also be useful as a predictive tool of harvest
numbers, lactation rates, proportion of yearlings in the population, doe to fawn ratios, and
winter losses. Such uses for the WSI are possible only if the data are reliable and
collected in a consistent and statistically valid manner. This study examined the methods
used to collect the WSI data, the accuracy of the data, and how the data may be useful in

predicting various measures of the white-tailed deer population of Michigan.

133

Methods

The MDNR collects WSI data using the technique devised by Verme (1968). A
chillometer measures atmospheric chill, and a compaction gauge measures the potential
of the snow for supporting a deer. Weekly measurements from both devices are
combined to form a weekly value, and a final cumulative value of all weekly values is
determined at the end of the winter. As of 1999, the MDNR collected WSI data from 10
stations in the UP, 8 stations in the Northern LP, and 12 stations in the Southern LP
(Figure 33). The exact number and location of stations varies slightly from year to year,
however (Figure 34). The data are collected and summed weekly for each station to
provide that station's cumulative WSI value. At the end of the winter, the cumulative
WSI values from each of the stations in a region are averaged to provide the final WSI
value for that region. The MDNR stores the weekly cumulative station values and
regional averages in Excel files that contain the WSI data from all years and in individual
files separated by year.

The MDNR stores only the cumulative values, so I first created a non-cumulative
weekly WSI value for each station. I subtracted from each weekly value the cumulative
value of the previous week. In some cases, large negative values resulted from the
subtraction; in such cases I eliminated that station’s data from that year’s average total
WSI value because the data must have been gathered or recorded incorrectly. I numbered
each data collection date by the number of days since November 1. A plot of the range of
data collection days for all years in all regions (Figure 35) shows that the data most
consistently cover the time frame from day 42 (December 13) to day 168 (April 18 or

April 17 in leap years). I then created new cumulative WSI values (henceforth referred to

134

 

 

 

 

 

 

Legend:

01 Allegan Forest
02 Allegan Farm
03 Atlanta

04 Baldwin

05 Baraga

06 Cadillac

07 Cass City

08 Crane Pond
09 Crystal Falls
10 Cusino

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

26
1 6. 10 . 1 l
= ' l8
102 ~01 1
/ l 1 ° 33
‘ 28
J4!!!
ll Dunbar 20 Kenton 30 St. Charles
12 Escanaba 21 Lapeer 31 Thompson
13 Flat River 22 Manistique 32 Wakefield
l4 Gaylord 23 Mio 33 Waterloo
15 Gladwin 24 Muskegon
16 Gwinn 25 Naubinway
l7 Houghton Lake 26 Newberry
(Porter Ranch) 27 Platte River
18 Holly 28 Pointe Moillee
19 Kalkaska 29 Rose Lake

Figure 33. The WSI station locations.

135

.886 88 088:8 8: 883 «80
888.08 80% 0.8“... < .883 .03 08888 8: 8 08: 8. 8: 0.38 80 088:8 883 860 888.08 «80 8888 8.?
8.83 < .829» .m? 08888 80 8 .88 08 088:8 885 800 888.08 «80 808: 5.3 808.0. < .00 8.88.... 88..

808 80 8 082.888 808 80me 8.... .wao. 8 000. 80.0 8.88 .03 88 88... «80 .8 68:50.38 2.... .60 98”....
80>

och waa. 0mm. 6mm. Noo. cam. wwa. 0wa. 6wa. Nwm. owm. 06o. 06a. 66a. N6o. 06¢. w0a.

P F — . 1F . 6 p . . P

 

T

v-‘OONOONOWV'MN—‘O

'XXXXr
X'XXX
*IXII
XX'XX
XXXXX~
XXXXX
XX'XXT
XXXXX
XXXXX~
XXXXX
XXX
XXX
XXX
XXX
X'X

II
II
I
I

.
I
*II
.
xII
we
I
.
I
I
I
.
.
Ix

1 T I I I I 1’ 1 r

XXXX
XXXX
XXXX
XXXX
'XXX
IIII
IIgI
IIII
IIII
xlll
xlll
IIII
IIII
IIII
I
I
I
I
I

XX
XX
XX
XX
II
II
II
II
II
xI
xI
I
I
I
I
I
I
I
I
éééééééézé

'XX'XXXX'XXXX'
apoo norms

'XX'XXX'XXXXXX

XXX XXXX XX XXXXXXXXXXXXXX
'XX XXXX XX XXXXXXXXXXXXXX

XXX

I
I
I
I
I
I
I
I
I
r
8

XX XXXXXXXXXX'XXX
X
X
X
X
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
933

'X
rfi
WV
NN

we
I

we
I
I
I
I
.
I
I
I
I
I
.
I

XXX XX XXXXXXX'XXXXXX
XX
XX
XX

XX'
XXX XXXX XX XXXXXXXXXXXXXX

Il*x

'XXX'XX"'X X XX'X""XXXXX XX'XX
F2

X'X"XXXXXX X'XXXX'XXXXXXX'
XXX XXXX XX XXXX X"XXXXXX

XX X XX

X

XX

xI
I
I
I
I
I
I
I
I

X

X X X
x x I x I I

'XX
XXX XXXX XX

 

 

XXX XXXX XX XXXXXXXXXXXXXX
XXX XXXX XX XXXXXXXXXXXXXX

XXI

 

r I 1 T T 1 I
VMN CONWK‘Q
MM") MNNNN

80885 I 8383 X

1%

1969
1970
1971
1972
1973
1974
1975
1976
1977
1978
1979
1980
1981
1982
1983
1984
1985
1986
1987
1988
1989
1990
1991
1992
1993
1994
1995
1996
1997
1998
1999

Year I

 

Region

\UP

- SLP

28 56 84 1 12 140 168 196
Days (since November 1) in WSI

Figure 35. The time period WSI data were collected in each region fi'om 1969 to 1999.
The bold lines mark days 42 (December 13) and 168 (April 18).

137

as the corrected WSI) by summing the weekly values beginning from approximately day
42 until approximately day 168. (All sums include data from the 19 weeks most closely
matching the days 42-168 period, although some sums may include a few additional days
due to leap year or data collection periods longer than 7 days. Note that although the
recorded start date is day 42, the data collected on day 42 covers the week immediately
preceding it.) I did not calculate a station’s corrected WSI value if data were missing
from several weeks during a single year. I also calculated a modified WSI value for each
region by summing the weekly values of only days 42-63 (first-month WSI) and days
147-168 (fourth-month WSI).

After finding the annual corrected WSI value for each station, I calculated the
average WSI values and the coefficient of variation (CV) for each management unit and
region. I regressed several population parameters on the corrected WSI values by region
and by management unit and on the first- and fourth-month WSI values for each region.
The population parameters include: (from the succeeding autumn’s harvest) the buck
harvest (uncorrected and corrected for hunter effort); total harvest (uncorrected and
corrected for hunter effort); proportion of lactating yearlings, 2.5 year olds, and 3.5+ year
olds during the first week of firearm season; proportion of yearling bucks in the entire
harvest; and average yearling beam size of deer checked during the following fall’s
hunting season.

To explore possible alternatives to the current WSI, I developed an alternative
index. Minimum temperatures and daily snow depth values from December 1986
through March 2000 were aquired from the National Oceanic and Atmospheric

Administration website (www.ncdc.noaa.gov/ol/climate/stationlocator.html). The data

138

were collected from 3 UP stations (Marquette, Manistique, and Ironwood) and 4 Northern
LP stations (Alpena, Cheboygan, Houghton Lake, and Traverse City). From the
minimum temperature data, the variable Degree<32 was created, which is the sum of the
daily differences between minimum temperature and 32 °F. A monthly average was
calculated from the snow depth data. A weighted value was then assigned to the monthly
average based on the system developed by Leckenby and Adams (1986). I was then able
to calculate the B-WSI as follows: Monthly B-WSI = (Monthly Sum Degree<32)*(Snow
Depth Weighted Value). The yearly B-WSI is the sum of the Monthly B-WSI values for
December through March. The regional values are the average of the individual station
values. See Appendix 3 for an example B-WSI calculation. Afier developing the B-
WSI, I examined the correlation between the B-WSI and the MDNR WSI and the

correlation between the B-WSI and yearling beam diameter.

Results

The WSI data collection stations are evenly distributed throughout the entire state
(Figure 33). Each management unit contains at least 3 stations, although Cass City and
St. Charles in the Saginaw Bay management unit only began collecting data in 1998. In
the Southwestern management unit, however, 2 of the stations (Allegan F arm and
Allegan Forest) are located within a few miles of one another. Allegan Farm and Forest
apparently experience similar climatic pressures, creating almost identical trends in
annual WSI values (Figure 36), but the trends are usually more than 25 points apart.

WSI values are available for the UP and the Northern LP for the last 30 years and

for the last 11 years in the Southern LP. Over the years, however, the time frame for the

139

140

120 A
100 / \ 1
80 \V/ \/

60. ' I if.

 

 

 

 

ws1
/>

 

4O

 

20

 

 

 

 

0 I I T I I I I l

1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999

Year

Figure 36. The annual corrected WSI values recorded at Allegan Farm (solid
line) and Allegan Forest (broken line).

140

collection of the WSI data varied (Figure 35). For example, during some years collection
did not begin until late December or early January, while recently, collection began in
early November in the UP. The length of the collection also varied from 83 days (in
1972 in the Northern LP) to 173 days (in 1992 in the UP). Even for those years when
values covered the same number of days, the starting and ending dates were frequently
different. Also, the WSI rarely covered the same time frame in the 3 different regions in
any single year. Occasionally, data collection did not begin and end on the same date
even for different stations within the same region. Such inconsistencies in the current
uncorrected data make comparisons between years and regions impossible because the
WSI is cumulative, and its final value depends on the number of days and the period for
which the data were collected. The time frame covered most consistently was between
days 42 and 168 (days since November 1). The corrected WSI values (described above)
were used for all of the following analyses.

Calculating the non-cumulative weekly WSI values made it possible to examine
trends in winter severity as it fluctuated throughout the season. Each winter has its own
pattern of fluctuating WSI values. An example of 3 winters in the UP with similar final
WSI values is shown in Figure 37. Currently, the MDNR creates figures similar to
Figure 37a to compare the severity of several winters. Such figures may obscure the true
differences between the weather patterns of these 3 winters. After the first week, the
1998 winter was more severe than the others in the beginning of the season (Figure 37b).
The 1994 winter peaked above the others in the middle and end of the season (Figure
37b). The 1990 winter had a large drop in severity in the middle of the season (Figure

37b). These patterns are not visible in Figure 37a.

141

(a)
80

70

ON
C

M
0

Cumulative WSI
8 8

N
O

p—I
O

0

(b)
10

*1994-95 " ‘- ' 1990-91 —i-l998-99

 

 

 

 

 

 

 

 

 

 

 

 

 

 

-—LI I I T fl *I fl T T 'I 77 f I I T I T I

 

7- 14- 21- 28- 4- ll- 18- 25- l- 8- 15- 22- l- 8- 15- 22- 29- 5- 12- 19- 26-
Dec Dec Dec Dec Jan Jan Jan Jan Feb Feb Feb Feb Mar Mar Mar Mar Mar Apr Apr Apr Apr
Date

+1994-95 (total=68.62) ' I ' 1990-91 (total=7l.57) q—1998-99 (total=73.48)

 

 

 

l/ \\ I

 

    

\ 1‘
\

 

 

Weekly wsr

 

 

 

 

 

O

 

 

 

 

 

I I T T I T I I T l I I I l

7- 14- 21- 28- 4- ll- 18- 25- l- 8- 15- 22- l- 8- 15- 22- 29- 5- 12- l9- 26-
Dec Dec Dec Dec Jan Jan Jan Jan Feb Feb Feb Feb Mar Mar Mar Mar Mar Apr Apr Apr Apr

Date

Figure 37. The (a) cumulative and (b) weekly WSI values from three selected
years with similar final WSI values in the Upper Peninsula.

142

Restricting the WSI data to those years that covered the day 42 to day 168 period,
reduced the available data to 20 years in the UP, 16 years in the Northern LP, and 11
years in the Southern LP (Figure 38). From the available data, the worst winter on record
for the Northern LP and the Southern LP was the winter of 1993. The 1993 winter was
of average to below-average severity for the UP, however, where the worst winters were
in 1978 historically and 1995 recently. In all 3 regions, the mildest winter on record was
the 1997 winter. The Northern LP and Southern LP winters are more similar to one
another than either’s winters are to the UP, but the WSI values for all 3 regions track one
another closely (Figure 38).

The WSI values averaged across the entire region, may not be an accurate
reflection of the true winter conditions of the region, especially in the Northern LP and
Southern LP. In the UP, when the data were divided into the separate management units,
the lines rarely deviated from one another (Figure 39), suggesting that the severity of the
winters across the UP was fairly homogeneous. The similarity was also reflected in the
CV values (Table 28). The variation of the WSI values for the entire region was no
greater than the variation of values for the separate management units. In the Northern
LP, however, the WSI values of the Northeastern management unit were generally larger
than those of the other 2 management units (Figure 40). But note that Gladwin was the
only station that had data from more than one year in the Saginaw Bay management unit
(Figure 34), so the Saginaw Bay management unit values were not representative of the
entire management unit. The CV values varied across the years within both the average
regional and average management unit WSI values (Table 29), perhaps indicating that

winter weather varied widely across the region, and was not adequately represented by

143

coom 32 mag 82 03. m3: 08: mg— mg: _

p _ h p r _ p l—

.c2m8 :08 E 32 8 32 Eat 82? 53 0358 080880 .wm 953:.

58>
3: O8— mwa wwa: 5o: cam: 32 vwa $2 $3 $2 8o: 33 mg: RE 03—

 

 

- x /
..-Ii|-b.wl.
\ —
‘
s o
OI. —
a
. .............. a. uuuuuuu Starr's. IIIIII
.............. w-...-.m--..-u..---.-
a i
0 ~
I
I .
O

 

 

_ p n L h p b p _ 1P — p h p P o

ION

r ow
ow8>a mum

 

 

I'll'l Ill

own—3a .52 1 cc

 

 

Adz
38-838823: . ow
.1 ~ —
.......... 1.-....1..-.q-...-.-----....-.---.---.-.-......-....
OI — , 882:5 roo-
— \ I .
9‘ o .
e [ON—
m: .
Q ~
«. .o:

 

 

oc—

ISM

144

WSI

—O—Western UP ' 'I ' Eastern UP

 

160

 

.-

 

100

80

60

140 /
120

81 m
\f

M

5*

 

40

3

 

20

 

 

O I

 

\qflb\9’\'\\9’\

‘i:

\9

4°)

\

I

_I T—I —I—

\

I

9%“\9%\9°°'K9%°’ 9%K9‘b‘ifib 9%\9%°°\9%9 99°\99\99'K993\99“\995\99

I I

\

fi'

f I If I I j

\

Year

fir

I

I

6

T

i I

\991\99<13\999L®o

Figure 39. Corrected annual WSI values of the Western UP and the Eastern UP

145

 

Table 28. The mean corrected WSI values of the UP, averaged over the stations
in the Western UP and Eastern UP management units.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Year Unit/Region N Mean WSI CV
1969 Western UP 3 114.28 12.7
Eastern UP
UP 3 114.28 12.7
1977 Western UP 4 108.5 19
Eastern UP
UP 4 108.5 19
1978 Western UP 5 145.02 13.5
Eastern UP
UP 5 145.02 13.5
1981 Western UP 3 128.42 19.4
Eastern UP 3 135.4 14.4
UP 6 131.91 16.7
1983 Western UP 6 109.6 18.8
Eastern UP 4 115.23 12.4
UP 7 112.81 14.1
1984 Western UP 6 100.77 18.2
Eastern UP 4 107.75 17.7
UP 10 103.56 17.3
1985 Western UP 6 118.33 16
Eastern UP 4 106.58 7.2
UP 10 113.63 14.1
1986 Western UP 4 55.13 27
Eastern UP 4 64.58 14.7
UP 8 59.85 21.1
1987 Western UP 6 100.18 23.2
Eastern UP 4 90.9 10.2
UP 10 96.47 19.4
1988 Western UP 6 106.48 21.8
Eastern UP 4 90.48 15.5
UP 10 100.08 20.8
1989 Western UP 6 86.17 22.7
Eastern UP 1 107.3
UP 7 89.14 21.9
1990 Western UP 6 69.87 23.9
Eastern UP 4 74.13 18.4
UP 10 71.57 20.8

 

146

Table 28 (cont'd.)

 

 

 

 

 

 

 

 

 

 

1991 Western UP 6 75.2 19.2
Eastern UP 4 79.18 25.1
UP 10 76.79 20.6
1992 Western UP 6 76.9 21.4
Eastern UP 4 78.85 24.2
UP 10 77.68 21.3
1993 Western UP 6 82.67 17
Eastern UP 4 93.29 15.7
UP 10 86.92 16.7
1994 Western UP 6 69.23 22.3
Eastern UP 4 67.7 19.1
UP 10 68.62 20
1995 Western UP 6 122.25 18.4
Eastern UP 4 107.51 16.8
UP 10 116.35 18.2
1996 Western UP 6 109.13 18.6
Eastern UP 4 97.41 19.8
UP 10 104.45 18.9
1997 Western UP 6 56.18 16.4
Eastern UP 4 54.06 29
UP 10 55.33 20.6
1998 Western UP 3 73.63 21.2
Eastern UP 5 73.38 17.8
UP 8 73.48 17.6
1999 Western UP 4 76.42 11.4
Eastern UP 3 70.4 26.4
UP 7 73.84 17.3

 

 

147

 

—O—-Northwestern LP " C ' Northeastern LP —A" Saginaw Bay

 

100
90
80
70
60
50
4O
3O
20
10

0

\9‘9 \9‘6“ 0%" @558 9%" \q‘b‘i’ 9‘19 99° \99‘ 99" 993 \99“ \99" \qu .99“ 99% \ng r1000

WSI

 

Year

Figure 40. Corrected annual WSI values of the Northwestern LP, the
Northeastern LP, and Saginaw Bay.

148

Table 29. The mean corrected WSI values of the Northern LP, averaged over the
stations in the Northwestern LP, Northeastern LP, and Saginaw Bay management
units, and the coefficient of variation.

 

 

 

 

 

 

 

 

 

 

Year Unit/Regifli N Mean WSI CV
1984 Northwestern LP 2 67.5 7.12
Northeastern LP 2 69.3 12.4
Saginaw Bay 1 75.1
Northern LP 5 69.74 8.4
1985 Northwestern LP 2 66.9 27.9
Northeastern LP 2 82.2 16.2
Saginaw Bay 1 60.1
Northern LP 5 71.66 21.2
1986 Northwestern LP 2 39.15 23.7
Northeastern LP 2 41.5 26.6
Saginaw Bay 1 34.7
Northern LP 5 39.2 19.7
1987 Northwestern LP 2 63.65 11
Northeastern LP 2 38.25 29.4
Saginaw Bay 1 42.1 .
Northern LP 5 49.18 30.2
1988 Northwestern LP 4 58.6 21.7
Northeastern LP 3 73.63 17.9
Saginaw Bay 1 52.8
Northern LP 8 63.51 21.9
1989 Northwestern LP 4 54.73 16.1
Northeastern LP 3 92.67 24.2
Saginaw Bay 1 46.7
Northern LP 8 60.45 25.2
1990 Northwestern LP ' 4 46.25 14.4
Northeastern LP 3 69.2 21.3
Saginaw Bay 1 47.3
Northern LP 8 54.99 26.9
1991 Northwestern LP 4 43.15 12.7
Northeastern LP 3 69.9 25.7
Saginaw Bay 1 41.9
Northern LP 8 53.03 32.7

 

 

149

 

Table 29 (cont'd).

 

 

 

 

 

 

 

 

 

1992 Northwestern LP 4 46.25 20.4
Northeastern LP 1 56.6
Saginaw Bay 1 50.8
Northern LP 6 55.28 16.5
1993 Northwestern LP 4 70.75 10.7
Northeastern LP 3 91.53 25.9
Saginaw Bay 1 69.7
Northern LP 8 78.41 22.2
1994 Northwestern LP 4 46.25 16
Northeastern LP 3 62.33 27
Saginaw Bay 1 43.5
Northern LP 8 51.94 25.8
1995 Northwestern LP 4 70.45 19
Northeastern LP 1 67.9
Saginaw Bay 1 67.4
Northern LP 6 69.52 15.1
1996 Northwestern LP 3 65.12 17.4
Northeastern LP 3 74.77 21
Saginaw Bay 1 50.8
Northern LP 7 67.21 21.1
1997 Northwestern LP 2 30.85 13.5
Northeastern LP 2 39.55 5.9
Saginaw Bay
Northern LP 4 35.2 16.3
1998 Northwestern LP 4 48.13 8.3
Northeastern LP 3 55.27 22.9
Saginaw Bay
Northern LP 7 51.19 17
1999 Northwestern LP 3 43.67 21.3
Northeastern LP 1 39.9 0
Saginaw Bay
Northern LP 4 42.73 18.3

 

 

150

 

the chosen stations. In the Southern LP, the WSI values of the Southwestern
management unit were consistently several points higher than those of the South Central
and Southeastern management units, although all follow the same general pattern (Figure
41). The regional CV values were greater than the management unit CV values in almost
every year (Table 30). The Southwestern LP appeared to experience more severe winters
than the rest of the Southern LP and its WSI values were inflating the regional average.

The results of the regression analyses varied drastically and showed few obvious
patterns on either the regional or management unit scales (Table 31). Generally the p-
values were lower and the r'2 values were higher for the UP and its management units than
for the other regions and management units. No population parameters were significantly
correlated with the WSI for all regions. The slopes of the regression lines even varied in
sign in several of the categories. Correcting the buck and total harvest for hunter effort
generally decreased the significance and the fit of the regression line. In a few cases,
using the WSI values summed for only the first and fourth month improved the fit of the
regression line, but generally there was little change from the regressions using the total
WSI values (Table 31).

The B-WSI correlated well with the MDNR WSI in the UP (r=0.9067) and only
moderately well with the MDNR WSI in the Northern LP (r=0.8108). Regressions
against yearling beam diameters using the yearly B-WSI values provided similar results
to the regressions using the MDNR WSI (Table 32 as compared to Table 31). Using just
the UP February and Northern LP March monthly B-WSI values, however, provided
stronger correlations with yearling beam diameter than using the yearly B-WSI values

(Table 32).

151

WSI

90
80
70
60

' '0 " Southwestern LP

—I— South Central LP

fl' Southeastern LP

 

+

 

I \

 

,4.
’A‘ I

 

\‘r

12%.

 

50 ,

 

40

\ r \
"‘ I

 

20

 

10

 

 

O T T T T v I

 

 

I I I I I I

1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997
Year

Figure 41. Corrected annual WSI values of the Southwestern LP, South
Central LP, and the Southeastern LP.

152

1998 1999 2000

Table 30. The mean corrected WSI values of the Southern LP, averaged over the
stations in the Southwestern LP, South Central LP, and Southeastern LP, and the
coefficient of variation.

 

 

 

 

 

 

 

 

 

 

Year Unit/Region N Mean WSI CV
1988 Southwestern LP 1 41.7
South Central LP 4 41.1 23.7
Southeastern LP 3 43.7 11.4
Southern LP 8 42.16 16.7
1989 Southwestern LP 3 68.2 24.8
South Central LP 4 37.3 19
Southeastern LP 3 46 23.3
Southern LP 10 49.17 36.6
1990 Southwestern LP 3 57.5 29.3
South Central LP 4 37.5 6.8
Southeastern LP 3 31.4 33.6
Southern LP 10 41.68 37.8
1991 Southwestern LP 3 53.9 33.2
South Central LP 4 32.6 6.9
Southeastern LP 3 40.6 14.8
Southern LP 10 41.37 34.5
1992 Southwestern LP 3 63 33.6
South Central LP 3 43.6 3.7
Southeastern LP 3 43.1 38.4
Southern LP 9 49.91 36.6
1993 Southwestern LP 3 88.5 33.1
South Central LP 3 56.8 1.3
Southeastern LP 3 63.2 28.1
Southern LP 9 69.49 35.6
1994 Southwestern LP 3 59.9 32.1
South Central LP 4 43.8 18.1
Southeastern LP 3 34.7 21.3
Southern LP 10 45.88 35.5
1995 Southwestern LP 3 75.3 26.9
South Central LP 4 50.4 15.3
Southeastern LP 1 34.1
Southern LP 8 57.71 36.5

 

 

153

 

Table 30 (cont'd).

 

 

 

 

1997 Southwestern LP 3 48.7 22.1
South Central LP 4 30.2 27.5
Southeastern LP 3 21.2 33.2

Southern LP 10 33.05 43.3

1998 Southwestern LP 3 65.3 23.3
South Central LP 3 43.6 10.4
Southeastern LP 3 33.3 28.5

Southern LP 9 47.41 37.4

1999 Southwestern LP 3 49.73 32.7
South Central LP
Southeastern LP 3 28.97 33.2

Southern LP 8 45.18 39.4

 

 

154

 

Table 31. Results of the regression analyses of total annual WSI values and the first
and fourth month annual WSI values against several dependent variables.

 

 

 

 

 

 

 

Total Annual ws1 1't & 4‘" Month Annual wsr
Dep. Var. Region/U nit N Int. Slope p 1'2 Int. Slope p I‘2
Buck UP 16 78376 -348.23 0.062 0.262 67894 -761.11 0.038 0.312
Harvest Western UP 16 63557 -294.62 0.011 0.356
Eastern UP 16 15675 -88.956 0.003 0.467
Northern LP 13 107927 -193.47 0.387 0.058 93221 207.84 0.668 0.015
Northwest LP 13 44899 -23.017 0.849 0.003
Northeast LP 13 39265 43.564 0.658 0.016
Southern LP 8 95414 122.7 0.859 0.004 106208 -341.03 0.813 0.007
Southwest LP 8 23246 149.85 0.305 0.131
South Central LP 8 30312 172.52 0.564 0.043
Southeast LP 8 16684 -l71.64 0.133 0.259
Total UP 16 121544 -587.73 0.076 0.240 101587 -1200.3 0.069 0.249
Harvest Western UP 16 98103 —490.84 0.023 0.298
Eastern UP 16 20080 -114.91 0.015 0.336
Northern LP 13 238920 -1339.8 0.041 0.285 167448 -375.18 0.805 0.005
Northwest LP 13 91569 -339.34 0.411 0.053
Northeast LP 13 82666 -186.83 0.380 0.060
Southern LP 8 272155 -1662.4 0.410 0.087 223154 -2104.6 0.624 0.032
Southwest LP 8 61426 18.719 0.965 0.000
South Central LP 8 96741 -598.86 0.543 0.048
Southeast LP 8 37707 -486.26 0.079 0.336
Buck UP 11 37.512 -0.023 0.886 0.002 38.0602 -0.1016 0.763 0.010
Harvest Western UP 11 41.041 -0.0331 0.812 0.006
per Eastern UP 11 34.215 -0.1117 0.367 0.082
Thousand Northern LP 10 23.328 -0.024 0.748 0.011 21.288 0.0397 0.795 0.007
Hunter Northwest LP 10 19.503 0.0503 0.541 0.039
Hours Northeast LP 10 22.123 0.0036 0.947 0.001
Southern LP 8 14.607 0.031 0.723 0.017 15.784 0.0211 0.909 0.002
Southwest LP 8 13.02 0.0475 0.279 0.145
South Central LP 8 12.883 0.1026 0.401 0.090
Southeast LP 8 15.12 -0.1124 0.266 0.152
Total UP 11 55.124 -0.0323 0.891 0.002 53.339 -0.0388 0.938 0.001
Harvest Western UP 11 64.023 -0.0799 0.717 0.014
per Eastern UP 11 41.388 -0.1067 0.472 0.053
Thousand Northern LP 10 54.965 -0.2962 0.127 0.217 38.84 -0.054 0.898 0.002
Hunter Northwest LP 10 39.838 -0.0234 0.926 0.001
Hours Northeast LP 10 49.023 -0.157 0.220 0.146
Southern LP 8 41.187 -0.2227 0.419 0.083 32.85 -O.1593 0.787 0.010
Southwest LP 8 32.615 -0.0327 0.828 0.006
South Central LP 8 40.942 -0.1958 0.634 0.030
Southeast LP 8 35.047 -0.3843 0.143 0.248
%1.5 yrs UP 6 20.473 -0.1284 0.143 0.376 16.529 -0.2825 0.112 0.426
Lactating Northern LP 6 5.7661 0.1388 0.320 0.196 8.5992 0.3309 0.339 0.182
Southern LP 5 35.528 -0.1495 0.439 0.156 37.681 -0.7117 0.213 0.354

 

 

 

155

 

Table 31 (cont'd).

 

 

 

 

 

%2.5 yrs UP 6 62.108 -0.2216 0.023 0.675 56.244 -0.5273 0.001 0.894
Lactating Northern LP 6 43.639 0.0722 0.435 0.126 48.62 -0.0479 0.839 0.009
Southern LP 5 68.516 -0.0959 0.637 0.061 72.074 -0.6159 0.310 0.253
%3.5+ yrs UP 6 81.523 -0.3007 0.004 0.840 69.685 -0.5529 0.026 0.664
Lactating Northern LP 6 46.871 0.1595 0.229 0.273 55.291 0.0562 0.874 0.006
Southern LP 5 60.956 0.0803 0.545 0.098 69.749 -0.3467 0.397 0.183
%Yearling UP 11 74.245 -0.3791 0.024 0.449 63.092 -0.861 0.010 0.540
Bucks in Western UP 11 69.13 -0.3261 0.017 0.450
Harvest Eastern UP 11 72.151 -0.3097 0.010 0.500
Northern LP 12 61.869 -0.0766 0.549 0.037 61.69 -0.2592 0.334 0.094
Northwest LP 12 72.779 -0.1517 0.464 0.055
Northeast LP 12 56.665 -0.0391 0.722 0.013
Southern LP 9 55.504 0.0515 0.383 0.110 55.269 0.1879 0.105 0.331
Southwest LP 9 52.224 0.1367 0.121 0.308
South Central LP 9 56.241 0.0207 0.796 0.010
Southeast LP 9 54.483 -0.101 0.258 0.178
Average UP 15 -1815.4 24.051 0.004 0.478 -1815.4 24.051 0.004 0.478
Yearling Western UP 13 18.431 -0.0133 0.021 0.397
Beam Size Eastern UP 13 19.055 -0.017 0.072 0.264
Northern LP 13 18.651 -0.01 0.149 0.179 18.452 -0.0224 0.128 0.197
Northwest LP 13 19.101 -0.0079 0.323 0.089
Northeast LP 13 17.911 -0.0065 0.177 0.159
Southern LP 10 22.996 -0.0266 0.017 0.533 21.981 -0.0176 0.523 0.053
Southwest LP 9 23.386 -0.0283 0.000 0.851
South Central LP 10 22.955 -0.0177 0.217 0.183
Southeast LP 10 22.61 -0.0147 0.444 0.075

 

 

 

156

 

Table 32. Results of regression analyses of monthly and annual B-WSI values against
yearling beam diameter. Monthly and annual B-WSI values were divided by 10 and
25, respectively, to place them on a scale similar to that of the MDNR WSI.

 

 

 

 

 

 

 

 

B-WSI Refln N Int. Slope p-value r2
December UP 13 17.509 0.0042 0.4295 0.0576
Northern LP 13 18.167 0.0055 0.2702 0.1096
January UP 13 17.654 0.0042 0.3557 0.0779
NorthemLP 13 18.151 0.0017 0.69 0.022
February UP 13 18.299 0.0139 0.0064 0.5061
Northern LP 13 18.197 0.0031 0.3936 0.0669
March UP 13 17.71 0.0093 0.1122 0.2132
Northern LP 13 18.403 0.0183 0.0018 0.6039
Annual UP 13 18.1 0.0091 0.043 0.3222
NorthernLP 13 18.302 0.0052 0.1379 0.1888

 

157

 

Discussion

The quality of the WSI data depends on how well the data represent the true
winter conditions across Michigan and the consistency of the data collection process. On
a geographic scale, the data present few concerns. The stations are scattered evenly
throughout the management units (Figure 33). Although historically only Gladwin has
represented the Saginaw Bay management unit, the recent addition of St. Charles and
Cass City will provide a more accurate average WSI of this management unit (Figure 33).
The average WSI value of the Southwestern management unit is heavily biased in favor
of the winter conditions within the Allegan area. Although the Allegan Farm and
Allegan Forest stations provide consistently different WSI values (Figure 36), probably
due to higher snowfall at Allegan Farm, they may not provide independent data because
of their close proximity to one another.

Although the geographic distribution of the WSI data is generally sound, the
inconsistencies in the dates of data collection detract from the quality of the data. The
WSI is a cumulative index, and annual values cannot be compared unless each station
collects the data for the same time period each year. I had to calculate corrected annual
WSI values because of the variation in the collection times (Figure 35). Using the
corrected values reduced the number of years in the historical record in some regions by
almost half (Figure 34). The choice of the day 42 to day 168 time period was based
solely on the condition of the available data, and I do not recommend using this time
period for all future WSI surveys. Future data collection time periods should be
determined by MDNR personnel familiar with Michigan’s winter conditions and should

cover all months during which winter conditions are expected. The time period used in

158

the corrected value eliminated the early winter values from the index. Winter arrives
early in the UP, and including data from November and early December would increase
the accuracy of the index. If early winter data are used in the UP, statewide comparisons
cannot be made unless the UP WSI values are corrected for the time period covered by
the Lower Peninsula data. Data must also be collected on the same dates at every station.
If a station is missing data from even just a few weeks throughout the season, that
station’s data must be eliminated from the regional average for the entire season.

Once each station’s data are standardized, the final values can be averaged to
determine a WSI value for each management unit or region. Averaging over such large
geographical areas may not make sense, however. The UP has fairly homogeneous
winter weather patterns (Figure 40, Table 28), but the Northern LP (Figure 40) and the
Southern LP (Figure 41) each contain a management unit whose winter severity differs
from the other two management units in the region. These results imply that using the
WSI on the management unit level would be more accurate than on a regional level. The
regression analyses do not reflect this, however; the regressions against the management
unit WSI values do not generally provide better fits than those against the regional values
(Table 31). In several cases, regressions against the management unit values were not
possible due to insufficient data (in the dependent variables). Whatever spatial division is
used in the WSI, other data must be available on the same scale for such correlational
studies.

Verme (1968) suggested that a critical WSI value can be determined for each area
to mark when winter losses will reach significant levels. He observed that in the UP, a

WSI value of greater than 100 led to moderate to severe losses in deer as reported by

159

field biologists. When the index did not reach 100, winter losses were insignificant.
Based on Verme’s (1968) critical value of 100 for the UP (which covers the same time
period as the corrected WSI of this study), few recent winters should have resulted in
severe winter losses (Figure 38). Similar critical values can be determined for the
Northern LP and Southern LP based on climate, habitat carrying capacity, and population
density (Verme, 1968). The MDNR would then be able to predict moderate to severe
winter losses when a region’s WSI value passed its critical value.

In its present form, the WSI can produce similar final values for years whose
winter weather patterns are different from one another (Figure 37). Severinghaus (1947)
found that severe weather during the early months of spring has a greater adverse effect
on the deer than a similar pattern occurring earlier in the year. As winter progresses, deer
gradually use up their fat stores and will not survive the winter if spring and its new food
sources come too late (Mautz, 1978). Early onset of winter will also decrease the length
of time deer have to accumulate their fat supplies. In the example presented in Figure 36,
I might expect higher deer mortality following the 1994 winter. The winter began and
ended more severely than the others (Figure 37b). The current index gives the 1994
winter the lowest WSI value of the 3 winters presented, however (Figure 37a).

Verme and Ozoga (1971) recognized the greater effect the beginning and end of
winter has on deer and compared a WSI of the first and fifth months of winter weather to
the total WSI value. They found that the first and fifth month index provided much better
correlations with deer physical condition and fawn mortality than the total WSI value.
Using the first and fourth month (probably comparable to Verme and Ozoga’s (1971)

second and fifth month) WSI value increased the quality of the fit of the regression in a

160

few cases, but did not demonstrate a consistent pattern of better fit (Table 31). The index
might have provided a better fit if the first month included data from November or early
December, rather than mid December to January.

An accurate winter severity index that reflects the true winter conditions of
Michigan could be used to predict over-winter deer losses and the percentage of
pregnancies that are carried to term in spring. The current WSI cannot be used in this
manner. This evaluation highlights several options for the future of the WSI. The
current system could be maintained with improvements made in the collection process.
The dates of collection must be standardized across the state and across years. Additional
evaluations may also be necessary to determine what combinations of months will
provide the best index. The number of collection stations could be increased to reduce
the influence of any one station on the average and decrease the variance of the regional
means. The location of the stations could also be evaluated to determine if they are
placed in the areas with the highest deer population levels.

A final option would be to explore other methods of collecting similar data.
Other severity indices for use in big game management have been developed based on
daily maximum temperatures and snow depth (Picton and Knight, 1971); the number of
days of winter stress (Roper and Lipscomb, 1973); the deviation from monthly
temperature and precipitation means (Picton, 1979); weekly air temperature, air
movement, snow cover, and snow depth (Leckenby and Adams, 1986); and previous
winters’ snow (Mech et a1, 1987). Data collected by the weather service such as daily
snowfall, average daily temperature, and daily minimum and maximum temperatures

may provide such a solution. The data would be available from more stations throughout

161

the state to increase the number of replicates, would not require MDNR personnel to
collect the data, and would be more easily standardized.

As an example of the possibilities for such an alternative WSI, I developed a
preliminary alternative index, which I named the B-WSI. The B-WSI is not strongly
correlated with the current WSI, but the differences appear to make it a stronger predictor
of yearling beam diameters (Table 32). Single monthly values of the B-WSI (Table 31)
are more strongly correlated with yearling beam diameter than either the yearly MDNR
WSI or the first and fourth month MDNR WSI (Table 31). The B-WSI is also calculated
from a reliable and standardized data source that is easily accessible. Further exploration
into the development of a new winter severity index may be the best alternative to the

current WSI system.

162

CHAPTER 5
CONCLUSION

The voluntary check stations, the lactation survey, and the winter severity index
can all provide valuable data to the MDNR for the management of Michigan’s white-
tailed deer. The check stations provide data that can be used in the SAK estimates of
population size, in deriving indices of herd health, and in tracking herd or harvest
composition. The lactation data are perhaps the least affected by hunter-derived bias and
provide useful data on the proportion of lactating does. The winter severity index can be
used to predict antler development, a possible indicator of herd health. All three surveys
have not reached their full potential, however. All can be improved to increase the
quality of the resulting data and the manner in which those data are used. The biodata
also faces possible significant changes as the MDNR discusses mandatory deer
registration.

The evaluation of the check station data suggests several management
recommendations that could improve the quality of the biodata:

0 Increasing the number of check stations, especially in the Southern LP where
checking rates are low, could increase the number of deer checked and reduce
geographic biases. As the number of check stations increases, the convenience of
checking deer will increase, encouraging more hunters to check their deer. Checking
convenience can also be increased by opening check stations during the evening or
weekend hours. Longer hours of operation may be especially important in collecting
additional data from the archery season when the convenient highway check stations

are not open.

163

If the number of check stations are increased, however, the number of qualified. agers
will have to also increase through more intensive training methods. Current aging
practices lead to acceptable error rates if deer are classified as fawns (0.5 years),
yearlings (1.5 years), or adults (2.5+ years). Divisions into older age classes severely
reduces accuracy. If individual age classes are deemed necessary, more intensive
training could increase the number of people qualified to age deer, making more
check stations possible, and could increase the accuracy of the aging deer into older
age categories and decrease the number of deer aged as ‘A’ or ‘AA.’
While the lactation and beam diameter data follow expected trends and could be used
as indices of herd health, additional research should be conducted to determine
exactly how these indices could be calculated and used.

Additional research could also be conducted to explore the possibility of weighting
the biodata by the distribution of the biodata records across a geographic region or a
the different seasons to provide more precise estimates. Such weights could counter
the effect of observed geographic and seasonal biases.

Until sample sizes are increased through the addition of check stations or other
means, most analyses should not be conducted on the county level. Many counties do
not have sufficient sample sizes to make calculations within an acceptable margin of
enon

The data transcription process could be revised to eliminate inconsistencies or to

include data that could be useful in the analysis of the biodata.

Most of these recommendations are valid only for the current voluntary checking system.

If Michigan institutes mandatory deer registration, the biodata could change drastically.

164

The results of the biodata evaluation do not suggest that a mandatory registration system
is necessary to collect accurate and useful data on Michigan’s harvested deer. The
current data have several valuable uses, and implementing the above recommendations
can only improve their quality and value.

The lactation data, as a subset of the biodata, will also benefit from many of the
recommendations listed above for improving the biodata. The evaluation of the lactation
survey found that, while the data may not be useful as an estimate of annual recruitment
or even as an estimate of the number of reproductive does, they may provide a useful
index of reproductive success. Additional research could answer questions as to when
fawns are weaned and how long a doe continues to lactate after her fawn dies. Such
information could be used in conjunction with the lactation data to develop estimates of
recruitment. By increasing the total number of deer checked, the MDNR should also be
able to increase the number of does checked during October to provide more accurate
estimates of the number of reproductively active females on smaller geographic scales,
such as by management unit or possibly county.

The WSI is entirely separate from the check station data and would not be
affected by improvements to the biodata. The evaluation of the WSI did suggest several
management recommendations that could improve the quality of the winter severity data.
The MDNR must first decide whether the current WSI should be maintained with greater
quality control measures, or whether an alternate WSI should be developed. The current
WSI could be vastly improved by standardizing the data collection process, but such
improvements may not necessarily increase the value of the WSI. By exploring

alternative WSIs, the MDNR could develop an index that incorporates easily collected

165

data but that provides valuable predictive power for winter mortality, beam diameters, or
other measures of herd health.

By improving and standardizing data collection and analysis processes and by
increasing the geographic and temporal coverage of each field survey, the MDNR will be
able to collect more accurate and precise data on Michigan’s white-tailed deer herd.
These data can then be used to track the success of current management practices or to
determine the necessity for changes in management practices. The MDNR will also be
able to justify their management decisions to the public by providing high quality data as
supporting evidence. Improving the quality of the field surveys can increase the
confidence the MDNR has in their management practices and the confidence the public

has in the MDNR.

166

APPENDICES

167

Appendix 1. The data sheet used to record the 1999 biodata.

 

 

 

 

 

MICHIGAN DEPARTMENT OF NATURAL RESOURCES - WILDLIFE BUREAU

1999 DEER PHYSICAL DATA

Please pnnl clearly with a number 2 penal!

 

 

 

 

 

 

   
 
 

 

(A\ I fir . I I l . T \ ,”‘A fl
13‘8”; (3;: (3; ; LOCATION TAKEN ; (Q, :3. 5-..,DAITLERS ' (a. . 63 I If; le @ : I
l ‘ "P 4: ‘ V ,3 3 f .l :R. I)! '“m I 15'

I. E 'POIVATE 0n- MG. ”5‘; l (5‘ C‘ I SEX AG: ' 7". . 7" 1 A, I SPIKES I . ‘ ‘.
I cl , 11 3 JR 1 UNIT 1 V V I V’ I N I'msl; 4:" i \“J 1 ~13! , LESS 1 I LAC’ATION I ‘ "i
I c a E swaps? .0110. g courlrv . ; or , , I IOTA. . THAN? lucmlsn DA'ECP I ASEF ...
L j 1 1 110.13 I 2355 L TOM.‘ I RANGE . F . 11.37.101.81 POINTS . w 1 :00 I REMARKS HS 113 I f
5 I :PVTI I I I ‘ “ i I [ I

     

:1 PUB' 1 '. l
I: 9VT:
"3 Pue‘

:1 PVT!

msmgmg

E PVT

m2

PVT

gmzmzmzmzmg

  

:3:- PVT

 

 

w
E
w
81 E .
1': PVT N, w
' s E
PVT N' w!
5‘ E 3
PVT N w
S E
El PVT w
"‘ E
:3 PVT w
E
3 PVT w!
E I
:3 PVT w
pUBI E
Ia PVT W
;-— s E
PVT; N w
P031 5 E a
13 PVT N w
8 E 1
:3 PW, N w
"'7 S E
I: PVT, N w
s E
D PVT N w
5 E
[3 PVT N w
. S E
@STATTON STATION NUMBER 1 (13) DATE ® UNIT OR STATION SUPERVISORS SIGNATURE 1
I
I I
4- Mo Day Yr i I

 

 

MAIL COMPLETED FORM PROMPILY TO.
WlIdITIe Fueld Surveys - Michigan Demnmenl 01 Natural Fesowces
8011 30030 - 1.808109, Mlcmgar 48909-7330

R-2064 (Rev 771971999,

168

 

 

Appendix 1 (cont’d).

INSTRUCTIONS FOR DEER PHYSICAL DATA SHEET

EXPLANATION OF DATA ENTRIES - Each line on the form is for recording information ab0ut one deer.

1

h)

6-7.

10-11.

VISIBLE or CONCEALED; Highway check
stations only. Write “V" it the deer would be
vusrble to an observer standing ott the road on
the passenger side of the vehicle Write “C" it it
IS nOt vrs:ble from that location.

B. M. E or L: Leave blank .t the deer was killed
during the regular firearm oeer season. Otherwrse
write "B" for bow season. "M" for muzzleloader
season. ”E“ tor early season. or "L' tor late
antlerless season.

PRIVATE or PUBLIC LAND‘ Check “PVT" fOr
private land or “PUB" for public land.

DEER MANAGEMENT UNIT: For eacn deer
Checked. ass:gn a three-digit Deer Management
Unit (DMU). The DMU shOUld reference the unit
where the deer was tagged. not where the hunter
had a permit.

COUNTY CODE: Record the code number for
the county in which the deer was killed. The
county code shoulc be a two d'git number. Le.
Alcona County should be recorded as 01

TOWNSHIP and RANGE: Record the location
where the deer was killed. Township and range
should be recorded as two digit numbers. Circle
direction, Le. “N“ or “S" for township and "W" Or
“E” for range.

SEX: Record the sex of the deer. Use “F" IOr
temale or “M" for male. and no other symbols.

AGE: Use the fellowing age classes: 1/2. 1-1/2.
2-1/2. 3412. etc.. but record these as 1/2, 1, 2.
3. etc. It the deer is, not a lawn. but cannot be
aged to the nearest year. age as follows: “A"
(adult; not a lawn but can't be aged) or “AA"
(adult-adult; 2-1/2-years-old or older).
lnexpenenced agers should only use the codes
‘1/2" (lawn), “1" (yearling). or “AA” (older than
yearling).

BEAM DIAMETERS: Record the diameter of
each antler measured one inch above the burr.
Each record is the average ot two measurements
taken at the greatest and smallest diameters. If
either antler is less than one inch long. record
“8" (broken). The beam diameter of Spikes should
be measured even it spikes are less than three
inches in length. Beam diameters should be
recorded for all bucks.

169

12.

..A
'(A)

14.

‘15

16.

TOTAL POINTS: Record the total number of boots
on the two antlers. Spike deer Should be included
00 not estimate the number of paints: if one or
both antlers is broken. record "8" =br0keni. Leave
this column blank tor antlerless deer

SPIKES LESS THAN THREE INCHES. Check
(1!) this column it the longest spike is less than
three inches. Measure from the Skull not from
the burr Be sure to record beam diameters of
all spike bucks. DO NOT check this column for
button bucks.

LACTATION CODE: Code “HM+" it the hunter
saw milk. “HM—" it the hunter saw no milk. “M+"
it yOU saw milk and “M-" if no milk was present

LACTATION DATE OR REMARKS Write the date
the milk was or was not seen; the date is critical.
Also use this space to record anomalies. tag
numbers. specrmen bag numbers. ager names. etc.

TB . WILDLIFE BUREAU PERSONNEL ONLY:
Deer that come into any deer cheek station should
have their rib cage examined for TB tubercles
(see lamirated pictures). Record a “+" it pea-
s:zed tan or yellow nodules or lumps are seen
on the inside of the rib cage. If no nodules are
seen. record a “-". It you could not examine the
rib cage. record a “o". It tubercles are present.
collect the head and a section of the rib cage.
and record the hunter‘s name. address and
telephone number in the remarks column.

AGED BY: Fill in the ager's code number. The
ager 5 code number should be tilled in for eacn
deer. Do not use ditto marks. If the ager does
not have a number. leave the number column
blank. but be certain to record the full name ltirst.
middle initial and last) in the column marked
Lactation Date or Remarks (column 15).

16-20. BOTTOM OF PAGE: Record the station name.

number and date checked. Wildlife Management
Unit or station supervrsor should check each form
for errors before signing.

 

 

Appendix 2. Descriptions of the variables contained in the SPSS biodata file ‘8799bio.’

year: year during which the deer was harvested and checked (January harvests are listed
under the previous year with the rest of that year’s harvest)

cseason: categorical code for the season during which the deer was harvested
blank = firearm
B = bow (archery)
M = muzzleloader
E = early season
L = late season

dmu: deer management unit in which the deer was harvested (these boundaries change
each year)

county: numerical code for the county in which the deer was harvested
cage: categorical code for the age of the deer

A = not fawn
AA = not fawn or yearling

00 = fawn
01 = 1.5 years old
etc.

clbeam: categorical code for the diameter (in mm) of left beam at 1 inch above the burr,
recorded for only yearlings through 1991
B = beam is broken (less than 1 inch long)

crbeam: same as above for right beam

cpoints: categorical code for the total number of points on the two antlers
B = one or more antlers broken

staff: ager number

agerdiv: ager division code, recorded through 1998
blank = Wildlife Division
1 = Forest Management
2 = Parks and Recreation
3 = Administrative Services

4 = Fisheries
5 = Law
6 = DEQ

7 = Volunteers
8 = all other DNR divisions
9 = US Fish and Wildlife Service, US Forest Service, or US Park Service

170

Appendix 2 (cont’d).

station: station where deer was checked (district and management unit codes are
conglomerates of all check stations within those areas)

1 = Alma highway check station

2 = Birch Run highway check station

3 = Mackinac Bridge highway check station

4 = Big Rapids highway check station

8 = ?

10 = ?

11 = District 1
12 = District 2
13 = District 3
14 = District 4
15 = District 5
16 = District 6
17 = District 7
18 = District 8
19 = District 9

20 = District 10

21 = District 11

22 = District 12

23 = District 23

25 = Marquette Office

26 = Roscommon Office

27 = ?

29 = Drummond Island

30 = Lansing Office

31 = Houghton Lake Wildlife Research Station
32 = Cusino Wildlife Research Station
33 = Rose Lake Wildlife Research Station
41 = Western UP Management Unit

42 = Eastern UP Management Unit

43 = Northeastern LP Management Unit
44 = Northwestern LP Management Unit
45 = Saginaw Bay Management Unit

46 = Southwestern LP Management Unit
47 = South Central LP Management Unit
48 = Southeastern LP Management Unit
99 = unknown

remarks: date lactation status observed

cspike: marked with ‘1’ if antlers were spikes < 3 inches long

171

 

Appendix 2 (cont’d).

tb: TB status of the deer as observed by the ager, collected since 1996
+ = signs of TB observed in ribcage
- = no signs of TB observed in ribcage
0 = rib cage could not be examined
blank = no observation made

lactate: lactation status of does, collected since 1993
HM+ = hunter saw milk on the date recorded in remarks
HM- = hunter looked but did not see milk on the date recorded in remarks
M+ = ager saw milk on the date recorded in remarks
M- = ager looked but did not see milk on the data recorded in remarks

prv_pub: land type on which the deer was harvested, collected since 1998
PB = public land
PT = private land
Blank = unknown

town and range: town and range where the deer was harvested, collected since 1998 in
the UP and since 1999 in the LP

section: section of township in which the deer was harvested (this data has never been
collected in the general biodata)

vorc: recorded at highway check stations only
v = deer would be visible to an observer standing off road on the passenger’s side
c = not visible from that position (concealed)

season: numerically coded season variable
1 = firearm season
2 = bow season
3 = muzzleloader season

sex: numerically coded sex variable
1 = male
2 = female

 

age: numerically coded age variable
0 = 0.5 years old

1 = 1.5 years old
5 = 5.5+ years old
6 = A

7 = AA

172

Appendix 2 (cont’d).

lbeam: numerical equivalent to clbeam, with B’s eliminated

rbeam: numerical equivalent to crbeam, with B’s eliminated

points: numerical equivalent to cpoints, with B’s eliminated

date: data recorded when datasheet was complete (not date of harvest or check)
avebeam: average of lbeam and rbeam

mgtunit: management unit in which the deer was harvested (determined by county)
1 = Western UP
2 = Eastern UP
3 = Northwestern LP
4 = Northeastern LP
5 = Saginaw Bay
6 = Southwestern LP
7 = South Central LP
8 = Southeastern LP

muregion: region in which the deer was harvested, determined by county not mgtunit
so the Saginaw Bay Management Unit is split between the NLP and SLP)
1 = Upper Peninsula
2 = Northern Lower Peninsula (all counties of MU 3 and MU 4 plus Clare,
Gladwin, and Arenac counties)
3 = Southern Lower Peninsula (all counties of MU 6, MU 7, and MU 8, plus
Isabella, Midland, Bay, Saginaw, Tuscola, Huron, and Sanilac counties)

killtype: antlered and antlerless determination
1 = Bucks = if male and not fawn, ‘A,’ or ‘AA’
2 = Antlerless = if female and not ‘A’ or ‘AA,’ or if male fawn or male with
spikes
0 = unknown = all others

173

 

0:.hv0:oooooccco_m2¢n0>wage—ozto—EEES enun— n
8.2 3 m: m: m: n: m— n: z 3 c: o- 2 2 a m— m: m: m: m— C 3 m- m— m: 3 N.— m_ m: Egon N

 

 

 

 

 

 

 

 

wmfiflimome:N_m_owwhhhwhbhhwooeovvvcnm595—
.N.NNNNNNNNNNNN6444nmmmmmmnneemeNNN588N.
.w>< m an an a" nu on ma VN an «N —N ea 3 an S 2 m— 3 n— n— = c— a a h o n v n _ 0_aatn> 62
><G
/ .<<OZ 8 880.288 0% <2
.8808 8: 888 cc Boom 8: .8 880 8:- \

 

 

 

08808 580 30% 8F

880882 8 8883 $3-83: 8: 80.: 388 03888 08888 8: 8m .Gwms m8m0< 08 3888.: .3 088880
882$ 80 8 80.288 88> 880 385 088883 a 8:08 8: $68 08 8:08 8: .28 580 Beam own-8.8 80 88:88 .082 .m

 

 

 

 

NNNNNVNN_E__TTN-2:N_422282822882_-_-44N_N2N_NE
N McNesz em on 8:28 NNNN22NENTN2N NN.NN NNNNN:2oN an:- N

ea : N N «N NN N2 6 : «N 2 NM «N EN EN 4N NN oN NN 2 NM 8 2 2 8 R 2 N o. N985
.NvNN_N22NN_NN:cNNeN2N_2N_oNN_2NNN.NNN see N
:N 2 3 NN R N N4 N4 2. 38:8 88:8 882322: 2:4 use N n NVfirI.
N2 2 o _- : 2- 2- 2- N_- N- N- a N N e- N- a N 2 22. NN oN : 2 .N N N 2 N N NN as... _

Nu m a N_ z N. N. N N_ 2 : NN NN 2 : N_ «N 2 ..N on NN NN NN N N 2 NN 2 N2 2 _N NVJaoO
NNNNoN2ONanNoN2N4mN__NoNN2NTovn-Ne22222N. ash N.

Esm—ncmonauvnou mnvumu«mafia—2:32:22=3aahomvnn — czar—3’62
><D

 

 

174

 

 

 

 

 

 

 

 

 

.ch TNNuNNveooaoO .
552 e
865 “58882 88MB \ 8:888 £52 :6
:a0.~mnmmv80 on 888888 8882

 

 

 

 

 

 

8:88:82 8 8883 $2.32 8: 808 388. 8:888 8: 8m 8:08 888 80 N8 ~888ch 80 .8 8% 8w 0:: 8::-
888888 880 8888 8: .8 m 8880 mm 82: m8: 8on00 .8 8:88 8: 8 883 898-5, $88on 80 88:88 88..— .N

.:88..88o80888803883888.82086008.333 8 88825 <<OZ 08 3 88:88 0.8 880 888,—.
.882 8:88 88882: .8 880 305 380 80 08 8888888 380 83888 08 08 ENS-m 8: N8 .9888: .80 8:- ._

895-m 8: .8 888888 8:88 :< .M 80887..

 

.6888 .8 888 05. 8805 8888a :28 .8 8:08 5%.-m 05 .8 own-88 8: 08 888 fizz-m 088on 8C- .0

 

.32? b.0808

 

 

 

 

 

2: .8 ea 2: 8 3.8- .255
83> 285-m 8:8 2F v2- 3? man N
NE- 82- NS N
Na. one- :N _
8.2.2- 3o- NNn N2

53.: 5%: .35 NEW»:— .082

 

 

 

.GNNE-umNd
-..NNmuaB-m 85 .88
38888888838

 

 

 

. .88? £888 80 .8 83 80 :80 2 0388882 8% HES-m 8:88 83. .038 8800
305 08:88» 3888 80 B ~988on .8 8% >588 80 w8b8:=8 3 088388 3 880 88 8:8, fizz-m .3888 8C- .m

 

 

 

 

 

 

 

 

8.7 02A
m. .- . I .
one- 2N N o .o- o .N_- 2. o
. . mm o o 0 _ N
8 T 82 N . - . - .
. . S o o N _ c
on 0. mm N. _ m~.o+ od
mNdu :mm N— j
a 8 >5: 8 8 8.
88> 08%7595 .w>< 5.82 :5 0o .— 3 mu 8 _

 

 

/ \
«Nd-

mm 88> 088803 88088888 .Gwi .m8~0< 05 388—8..—

ofi 8 888 :6 .8 580 88% 8800 305 08883
385 own-86 5 08 858809 m80=oamoboo 80 505 588 a
8 08868 8:08 08.303 8E-

 

 

 

 

 

 

 

175

LITERATURE CITED

176

 

LITERATURE CITED

Bull, P. and B. Peyton. 2000. Pilot Study Report: The 1999 Michigan Deer Check
Station Survey to Michigan Department of Natural Resources: Wildlife Division. 62

pages.

Burgoyne, G. E. Jr. 1981. Observations on a heavily exploited deer population. Pages
403-413 in C. W. Fowler and T. D. Smith, eds. Dynamics of large mammal populations.
John Wiley and Sons, New York.

Coe, R. J ., R. Downing, and B. S. McGinnes. 1980. Sex and age bias in hunter-killed
white-tailed deer. Journal of Wildlife Management. 44: 245-249.

Eberhardt, L. 1960. Estimation of vital characteristics of Michigan deer herds.
Michigan Department of Conservation, Game Division Report Number 2282. 192 pages.

Frawley, B. J. 2000. Michigan deer harvest survey report: 1999 seasons. Michigan
Department of Natural Resources, Wildlife Division Report Number 3314. 27 pages.

Hawn, L. J. and L. A. Ryel. 1969. Michigan deer harvest estimates: sample surveys
versus a complete count. Journal of Wildlife Management. 33: 871-880.

Jacobson, H. A. 1994. Reproduction. in Gerlach, D., S. Atwater, J. Schnell, editors.
Deer. Stackpole Books, Mechanicsburg, PA. p. 98-108.

Langenau, E. 1996. The winter of 1995/96: A tough one for white-tailed deer. Michigan
Department of Natural Resources. Wildlife Division Report Number 3251. 4 pages.

Leckenby, D. A. and A. W. Adams. 1986. A weather severity index on a mule deer
winter range. Journal of Range Management. 39: 244-248.

Mattson Hansen, K. M. 1998. Integration of archery white-tailed deer (Odocoileus
virginianus) harvest data into a sex-age-kill population model. MS Thesis. Michigan
State University. 95 pages.

Mautz, W. W. 1978. Sledding on a bushy hillside: the fat cycle in deer. The Wildlife
Society Bulletin. 6: 88-90.

McCullough, D. R. 1982. Antler characteristics of George Reserve white-tailed deer.
Journal of Wildlife Management. 46: 821-826.

Mech, L. D., R. E. McRoberts, R. 0. Peterson, and R. E. Page. 1987. Relationship of

deer and moose populations to previous winters’ snow. Journal of Animal Ecology. 56:
615-627.

177

 

Michigan Department of Natural Resources (MDNR). 1994. Hunting and Trapping
Guide. 31 pages.

Michigan Department of Natural Resources (MDNR). 2000a. 1938-2000 Michigan Deer
Population Figures. http://www.dnr.statemi.us/pdfs/wildlife/deer_popest.pdf.

Michigan Department of Natural Resources (MDNR). 2000b. Hunting and Trapping
Guide. 56 pages.

Ozoga, J. J ., R. V. Doepker, and M. S. Sargent. 1994. Ecology and Management of
White-tailed Deer in Michigan. Michigan Department of Natural Resources, Wildlife
Division Report Number 3209. 73 pages.

Picton, H. D. 1979. A climate index and mule deer fawn survival in Montana.
International Journal of Biometeorology. 23: 115-122.

Picton, H. D. and R. R. Knight. 1971. A numerical index of winter conditions of use in
big game management. in A. O. Haugen, editor. Proceedings of Snow and Ice in
Relation to Wildlife and Recreation Symposium. Iowa State University, Ames, Iowa. p.
29-38.

Richie, W. F. 1970. Regional difference in weight and antler measurements of Illinois
deer. Transactions of the Illinois Academy of Science. 63:189-197.

Roper, L. A. and J. F. Lipscomb. 1973. Estimating winter losses of mule deer in
northern Colorado using a winter severity index. Annual Conference of the Western
Association of State Game and Fisheries Commissioners Proceedings. 53: 142-147.

Roseberry, J. L. and W. D. Klimstra. 1974. Differential vulnerability during a controlled
deer harvest. Journal of Wildlife Management. 38: 499-507.

Ryel, L. A., D. Fay, and R. C. Van Etten. 1961. Validity of age determination in
Michigan Deer. Papers of the Michigan Academy of Science, Arts, and Letters. 46: 289-
316.

Scanlon, P. G. and D. F. Urbston. 1978. Persistence of lactation in white-tailed deer.
Journal of Wildlife Management. 42: 196-197.

Severinghaus, C. W. 1947. Relationship of weather to winter mortality and population
levels among deer in the Adirondack region of New York. Transactions of the North
American Wildlife Conference. 12: 212-223.

Severinghaus, C. W. 1949. Tooth development and wear as criteria of age in white-
tailed deer. Journal of Wildlife Management. 13: 195-216.

178

Severinghaus, C. W., H. F. Maguire, R. A. Cookingham, and J. E. Tanck. 1950.
Variations by age class in the antler beam diameters of white-tailed deer related to range
conditions. Transactions of the North American Wildlife Conference. 15: 551-570.

Verme, L. J. 1968. An index of winter weather severity for northern deer. Journal of
Wildlife Management. 32: 566-574.

Verme, L. J. and J. J. Ozoga. 1971. Influence of winter weather on white-tailed deer in
upper Michigan. in A. O. Haugen, editor. Proceedings of Snow and Ice in Relation to
Wildlife and Recreation Symposium. Iowa State University, Ames, Iowa. p. 16-28.

Wackerly, D. D., W. Mendenhall, and R. L. Scheaffer. 1996. Mathematical Statistics
with Applications. Wadsworth Publishing Company, Belmont, CA.

179

 

     

  
  

      

CHG

12

3

7111llllllllllllllllfllllll
2088 2605

930