9 ”.3.
am.“ .1

$4...

fa
’ziig

; V
.4 a. a. xn. any“?

:23; .T.
i. fihhz .e .

v»:-

ummu .

.3 ‘fiillzu.
Jam“. cwwmwwirmuvatr. .
£5... ‘ m. ‘nfi... ,

5r

 

‘. .1 ..
haLwnrfht}. WSNV
5‘ fvvhs«\1uI««v|V|MH
‘3...ulzil(lcx...~n

. . .. I 1‘

12$...»

““v ‘1‘?

. .thv u

. ‘ 2...:

 

 

Hwfltw .

npu’v‘ ..

LIBRARY

I Michigan State ’
University

This is to certify that the
dissertation entitled

STATUS AND DISTRIBUTION OF FROGS AND TOADS IN
SOUTHERN MICHIGAN: POPULATION TRENDS AND THE
INFLUENCE OF HABITAT AND LANDSCAPE
CHARACTERISTICS

presented by
KRISTEN S. GENET

has been accepted towards fulfillment
of the requirements for the

Ph.D degree in Zoology and Ecology,
Evolutionary Biology and
Behavior

 

 

Jéflm 272 gN/J;

Major Professor’ 5 Signature
’4'“ 5 out %' 073 07 50V
0 / 7

Date

MSU is an Affirmative Action/Equal Opportunity Institution

 

 

 

.UBRARY
Michigan Slate
University

 

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

 

DATE DUE DATE DUE DATE DUE

 

WEIR-032%

 

JUL 1 9 ZUUY

 

TAtt ”‘1
on“ “)5.

'di

 

 

 

 

 

 

 

 

 

 

 

 

6/01 cJCIRC/DateDue.p65-p.15

STATUS AND DISTRIBUTION OF FROGS AND TOADS IN SOUTHERN
MICHIGAN: POPULATION TRENDS AND THE INFLUENCE OF HABITAT AND
LANDSCAPE CHARACTERISTICS
By

Kristen S. Genet

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

2004

ABSTRACT

STATUS AND DISTRIBUTION OF FROGS AND TOADS IN SOUTHERN
MICHIGAN: POPULATION TRENDS AND THE INFLUENCE OF HABITAT AND
LANDSCAPE CHARACTERISTICS

By
Kristen S. Genet
Declining amphibian populations in many disturbed and pristine habitats worldwide have
been the source of great concern and research in amphibian biology and conservation. Research
conducted at multiple spatial scales, life history stages, and over long periods of time (at least 10-
15 years) needs to be conducted and synthesized to hilly understand amphibian population
dynamics so that effective management and conservation strategies can be formulated. I
mvestigal ed some of these issues for 12 species of anurans native to southern Michigan. My
objectives were to: (1) assess the occurrence, distribution, and status of each species using
multiple methods to analyze data collected by volunteers in the Michigan Frog and Toad Survey
(MFTS) from 1996 to 20012, (2) evaluate the quality and consistency of data collected by
observers in the MFTS and assess implications for population trends analyses of volunteer-
collected data, (3) investigate relationships between anuran abundance and distribution and land
cover adjacent to wetland breeding sites and within potential dispersal distances using GIS
analyses, and (4) determine the potential influences of habitat characteristics on growth,
development, and survival of lava] Pseudacris crucifer in natural wetlands. Populations of all
Michigan anurans were variable over the seven-year duration of the study, but no major declines
were detected. Seven species did show statistically significant trends in site occupancy or
abundance (assessed by calling intensity of breeding males), but trends were not consistent in
direction across species and were small in magnitude. These trends need to be evaluated over
longer time periods to determine whether they represent significant long-term p0pulation trends

or short-term changes related to climatic variation. Volunteer observers in the MFTS were

reliable more than 80 percent of the time in identifying all species of frogs and toads combined
from their species-specific mating calls, but there was considerable variability in assessing
abundance (i.e., assigning of categorical call index values to calling males). Volunteer
background or prior experience in wildlife biology had little influence on data quality. Given
variability among volunteers (and years) in abundance estimation, the poor understanding of
empirical relationships between call index values and true breeding population size, the most
robust analyses of MFTS data are those that use presence/absence (i.e., detection/non-detection)
data. Food availability, hydroperiod, canopy cover, and predators affected P. crucifer tadpole
development, survival and growth in wetlands in southwestern Michigan. Growth, deveIOpment,
and survival were greatest at sites with intermediate hydroperiods, partial canopy cover, and few
(if any) fish predators. Land cover types influenced presence and abundance of anurans at
wetland breeding sites. Land cover types indicating habitat alteration or 3055 (e.g., roads, urban)
negatively influenced presence and/or abundance of anurans while land cover types representing
important foraging and breeding habitats (e.g., open land, wetlands) represented positive
influences. Species richness was not correlated with either the amount of adjacent forest cover or
the amount of forest cover in the landscape within 1000 meters of breeding sites. Associations
between anuran presence and abundance at breeding sites and land cover adjacent to and within
1000 meters of the breeding sites combined with data from population trends analyses provide
valuable insights that can be used to identify critical habitats for management and conservation of

these species.

TABLE OF CONTENTS

LIST OF TABLES

LIST OF FIGURES

INTRODUCTION

CHAPTER ONE

Short-term population trends and status of frogs and toads in southern Michigan (1996—

2002)
Abstract
Introduction
Materials and Methods

Michigan Frog and Toad Survey

Climatological Variables
Data Analyses

Results
Climatological Variables
Patterns of Species Richness

Site Occupancy and Calling Intensity

Route Regression Analyses

Repeated Measures Analyses

Discussion
Species Accounts

Analyzing Population Trends from Call Surveys
Management and Conservation Implications

Acknowledgements

CHAPTER TWO

Evaluation of methods and data quality from a volunteer-based amphibian call survey

Abstract

Introduction

Materials and Methods

Results
Volunteer Demographics
Species Identification
Abundance Estimation

Discussion

Acknowledgements

CHAPTER THREE

The Influence of Landscape Characteristics on the Abundance and Distribution

of Frogs and Toads in Southern Michigan

iv

10
10
12
l3
16
16
17
17
19
I9
20
21
29
32
34

61
62
65
68
68
69
70
71
79

Abstract
Introduction
Materials and Methods
Michigan Frog and Toad Survey
Landscape Variables
Landscape Context and Temporal Patterns
Statistical Analyses
Results
Principal Components Analysis
Species Presence
Species Abundance
Landscape Context and Temporal Patterns
Discussion
Acknowledgements

CHAPTER FOUR
The Influence of Local Habitat Characteristics on the Growth, Development, and
Survival of Spring Peeper (Pseudacris crucifer) Tadpoles in Southwestern Michigan
Abstract
Introduction
Materials and Methods
Study Sites and Habitat Characteristics
Field Study of Tadpole Growth
Statistical Analyses
Results
Resident Anuran Community
Habitat and Water Chemistry Characteristics
Tadpole Response Variables
Tadpole Growth and Development Profile Analysis
Discussion
Acknowledgements

SUMMARY AND CONCLUSIONS
APPENDIX

LITERATURE CITED

9O
91
94
94
96
98
99
100
101
101
104
107
107
117

134
135
138
138
140
142
143
143
144
145
146
147
152

167

174

177

1.1

1.2

1.3

1.4

1.5

1.6

1.7

1.8

1.9

1.10

LIST OF TABLES

Wetland categories assigned by volunteers to MFTS survey sites. Descriptions of
wetland types are from Sargent (2000). 35

Number of sites of each wetland type for which volunteers submitted complete
data to the MFTS. Wetland types and years not satisfying the assumptions of
normality are indicated with an asterisk (*). 36

Summary of parametric ANOVAS and nonparametric Kruskal-Wallis tests
evaluating differences in species richness among wetland types each year, 1996 --
2002. Sample sizes are listed in the bottom row of Table 1.2, and all tests had 6
degrees of freedom. 36

Results of simple linear regressions of the mean number of species present in each
wetland type over the time period 1996-2002. 37

Proportion of all sites surveyed by each species, 1996-2002. The total number of
sites for each species includes only those sites within the species range. Sample
size (i.e., number of sites where present) for each species is given in parentheses
in each year column. 37

Results of simple linear regressions for proportion of sites within ranges occupied
by each species during the period 1996 — 2002. N = 7 for each species,
representing each year of the survey since 1996. 38

Results of simple linear regressions evaluating trends in mean call index values
for each species during the period 1996-2002. N = 7 for each species, representing
each year of the survey since 1996. 38

Correlations between cumulative precipitation during anuran breeding season
(Feb-Jul) or cumulative winter (Nov — Mar) precipitation (rain and snowfall) and
either proportion of sites within range occupied or mean annual call index values
(1996-2002). Values are Spearrnan rank correlation coefficients (rs; *p < 0.05,
**p < 0.01). 39

Correlations between annual total snowfall prior to the anuran breeding season
(Nov — Mar) and each species’ route variables. Values are Speannan rank
correlation coefficients (rs; *p<0.05, **p<0.01). Sample size for each pairwise
comparison is listed in parentheses following the species name. 39

Correlations between cumulative monthly precipitation prior to and during

breeding period and each species’ route variables. Values are Spearman rank
correlation coefficients (rs; *p<0.05, **p<0.01). Sample size for each pairwise

vi

1.12

1.13

1.14

1.15

2.1

2.2

2.3

comparison is the same for all species (except Bufofowlerz‘) for any given month
and is given in the bottom two rows of the table. 40

Correlations between mean monthly temperature during the breeding period and
each species’ route variables. Values are Spearman rank correlation coefficients
(rs; *p<0.05, **p<0.01). Sample size for each pairwise comparison is the same
for all species (except Bufofowleri) for any given month and is given in the
bottom two rows of the table. 42

Results of route frequency regressions. Sample size indicates the number of
routes used for analysis for each species for all route regression dependent
variables. Mean trends (slopes of regressions in units of change in frequency per
year) and associated 95% confidence intervals indicate the direction and
magnitude of the annual rate of change, P values indicate whether trend (slope)
differed significantly from zero. 44

Results of route index regressions. Mean trends (SIOpes of regressions in units of
change in route index per year) and associated 95% confidence intervals indicate
the direction and magnitude of the annual rate of change, P values indicate
whether trend (slope) differed significantly from zero. 44

Results of route adjusted index regressions. Mean trends (slopes of regressions in
units of change in adjusted index per year) and associated 95% confidence
intervals indicate the direction and magnitude of the annual rate of change, P
values indicate whether trend (slope) differed significantly from zero. 45

Results of repeated measures analyses of variance for routes for which the same
volunteers submitted data for each year 1996-2002. Sample size is the same for
all species (N = 20) except Bufofowleri (N = 12). See text for explanation of
dependent variables (route frequency, route index, adjusted route index). 46

Demographic characteristics of survey respondents. Means and SE were
calculated from all volunteers returning questionnaires, and the sample size for
each demographic character (i.e., number of respondents providing answer on
questionnaire) is listed in the last column. 81

Proportion of respondents identifying species on each track of the CD recording.
Species present on each track are indicated with an asterisk. ‘ 82-83

Summary of correct identifications and identification errors averaged over all 12
CD tracks (means and SE reported). All means represent proportion of total
respondents. Correct identifications indicate volunteers who correctly recorded a
species as present when it was calling. Missed identifications indicate volunteers
who recorded a species as absent when it was actually present. Incorrect
identifications indicate volunteers who recorded a species present when it was

vii

2.4

3.1

3.2

3.3

3.4

3.5

3.6

3.7

3.8

3.9

absent fi'om the recording. Species with no SE value were present on only one
track, precluding estimation of variability. 85

Percentage of respondents living within and outside ranges of the three species
with limited distributions within Michigan with correct species identifications.
The relationship between living within a limited range and correct identification
was tested using contingency table analyses. 86

Species of anurans found in southern Michigan and included in this study. All
species except B. fowleri are distributed throughout the study area. Species
ranges were determined from Harding (1997) and Conant and Collins (1998).

118

Summary of land cover and road variables at both spatial scales in southern
Michigan. All land cover types represented the total area (m2) within each buffer;
roads represented the total length of roads within each buffer. Values are means
from all study sites (N=1055), their associated standard errors, and the proportion
of the total buffer area comprised by each land cover type. 1 18

Principal components retained from PCA of variables at 100 m scale. The first
four principal components explained a total of 85.12% of the variation in the
dataset at that spatial scale. 119

Principal components retained from PCA of landscape variables at 1000 m scale.
The first two principal components explained a total of 93.66% of the variation in
the dataset at that spatial scale 119

Significant (p<0.05) landscape factors affecting occurrence of individual anuran
species at 100 m scale determined using logistic regression. 120

Significant (p<0.05) principal components affecting occurrence of individual
anuran species at 100 m scale determined using logistic regression. 121

Significant (p<0.05) landscape factors affecting occurrence of individual anuran
species at 1000 m scale determined using logistic regression. 122

Significant (p<0.05) principal components affecting occurrence of individual
anuran species at 1000 m scale determined using logistic regression. 123

Results of multiple regression analyses of landscape variables on anuran
abundance and species richness at 100 m scale. N=1055 for all species except B.

fowleri (N=651). 124

3.10 Results of multiple regression analyses of Principal Components on anuran

abundance and species richness at 100 m scale. N=1055 for all species except B.

fowleri (N=651). 125

viii

3.11

3.12

3.13

3.14

3.15

4.1

4.2

4.3

4.4

4.5

4.6

Results of multiple regression analyses of landscape variables on anuran
abundance and species richness at 1000 m scale. N=1055 for all species except B.
fowlerr' (N=651). 126

Results of multiple regression analyses of Principal Components on anuran
abundance and species richness at 1000 m scale. N=1055 for all species except B.
fowlerr' (N=651). 127

Temporal patterns in species richness at breeding sites in various landscape
contexts. Trends indicate results from regression analyses 1996-2002. 127

Results of regression analyses for temporal trends in site occupancy at breeding
sites in anthropogenic and natural landscape contexts. N = 7 for all species
(proportion of sites surveyed occupied by each species each year 1996-2002).

128
Results of regression analyses for temporal trends in annual call index values at
breeding sites in anthropogenic and natural landscape contexts. 129
Characteristics of the 10 study sites in southwestern Michigan 153

Resident anuran community composition at field study sites. Hyla versicolor and
H. chrysoscelis were combined and counted as a single species. 154

Water chemistry variables at field study sites, measured monthly during anuran
breeding and larval period (March — July, n = 5 samples for all variables except
chl a, where n = 4). Values represent mean (std. error) for each site, and variables
that differed significantly among sites are indicated by asterisks (Kruskal-Wallis
ANOVA; *p<0.05, **p<0.01, ***p<0.005, ****p<0.001). 155-156

Mean length of larval period (days) for Spring peeper tadpoles at field study sites.
Length of larval period was defined as the number of days between the beginning
of the experiment (2 June 2000, day 0) and the day the tadpoles metamorphosed
(stage 42, forelimb emergence). 157

Mean survivorship of Spring peeper tadpoles at study sites during field study.
Survival represents the mean survival of all enclosures at a site (n=3 for all sites
except Lawrence Lake Marsh and Jackson Hole Outflow, where n=2). Enclosures
where fish or turtles completely depredated tadpoles were eliminated from further
analyses (all enclosures were eliminated from Eagle Pond). 157

Overall rates of growth (mm/day) and development (stage/day) for larval Spring
peepers at field study sites in southwestern Michigan averaged over entire larval
period (enclosure stocking to metamorphosis). Both growth and development
rates differed significantly among sites (Kruskal-Wallis ANOVA, p < 0.05). 158

ix

4.7

4.8

Correlations between habitat and water chemistry parameters and tadpole
response variables. Values are Spearman Rank Correlation coefficients (rs).
Significance in indicated by asterisks ( p<0.05 Tp<O.lO). 159

Correlations between habitat and water chemistry parameters and overall tadpole
grth and development rates averaged over entire larval period. Values are
Spearrnan Rank Correlation coefficients (rs). Significance in indicated by asterisks
(‘p<o.05 Tp<o.10). 160

1.1

1.2

1.3

1.4

1.5

1.6

1.7

1.8

1.9

LIST OF FIGURES

Calling calendar for hogs and toads of southern Michigan. Arrows represent
approximate beginning, ending, and duration of calling period for males of each
species. Breeding effort is largely dependent on weather conditions; dates are
extremely variable and dependent on temperature and precipitation each year.
Figure courtesy of Lori G. Sargent (MDNR, MFTS coordinator). 47

Distribution of MFTS routes in Michigan. The box represents the region
evaluated in detail in this study. Each point represents a route comprised of 10
survey sites at anuran breeding ponds. 48

Location of climate stations used for temperature and precipitation in southern
Michigan. Daily temperature and precipitation data were obtained for these sites
for 1995-2002 and matched to the nearest MFT S route. 49

Year of route establishment. There were a total of 269 routes in southern lower
Michigan. Each bar represents the proportion that was established in each year of
the MFTS. 50

Number of years routes have been surveyed. Bars represent the proportion of the
269 total routes in southern lower Michigan that was surveyed for a given time
period. 50

Patterns of temperature and precipitation during anuran breeding season, 1996-
2002. (A) Monthly mean low temperature, (B) Monthly mean high temperature,
(C) Monthly mean daily temperature, (D) Monthly mean precipitation. All data
represent means : SE. Asterisks represent significance level of statistical tests
(*p<0.05, **p<0.01, ***p<0.005, ****p<0.001). 52

Mean number of species (+/- SE) present in wetland types in each year. Each
panel represents a single year of MFTS data (year given in upper right comer of
each panel). Horizontal and vertical axes are identical on all panels, and axis
labels are provided along the left and bottom margins for all panels. Points with
matching letters indicate significant pairwise differences (Bonferroni corrected
pairwise comparisons, p < 0.05). 54-55

Mean number of anuran species detected calling (+/- SE) in each wetland type
from 1996 — 2002. (A) Woody vegetation dominated wetland communities, (B)
Herbaceous vegetation/open water dominated wetland communities. 56

Site occupancy and mean call index values (+/- SE) for each species. Each panel
represents one species (species indicated in upper right comer of each panel).
Open circles indicate proportion of sites occupied by each species, black circles
indicate mean annual call index value. Mean annual call index values are plotted
on the primary vertical axis, and site occupancy is plotted on the secondary

xi

2.1

3.1

3.2

3.3

3.4

3.5

4.1

4.2

vertical axis. Axis labels are identical for all panels and are provided at the outer
margins. Significant linear relationships are indicated with trendlines. See Tables
1.6 — 1.7 and Results section for statistical analyses. 58-60

Call indices assigned by respondents for each CD track. Lightly shaded bars
indicate call index 1, open bars indicate call index 2, and black bars indicate call
index 3. Y-axis (percentage of respondents) is the same for all figures. Correct
call index values are given at the base to the right of each bar. (a) Track 1, (b)
Track 2, (c) Track 3, ((1) Track 4, (e) Track 5, (I) Track 6, (g) Track 7, (h) Track
8, (i) Track 9, (j) Track 10, (k) Track 11, (1) Track 12. 88-89

Locations of survey routes for the Michigan Frog and Toad Survey throughout
Michigan. Only routes in the outlined portion of the southern lower peninsula
were used in this study. 130

Eigenvalue plots fi'om Principal Components Analysis at 100 m spatial scale. All
three panels share the indentical X-axis. Landscape variables plotted are those
described in Table 3.2. Abbreviations are as follows: ag=agriculture, open=open
land, forwet=forested wetlands, nforwet=nonforested wetlands. Other variables
not abbreviated. 131

Eigenvalue plots from Principal Components Analysis at 1000 m spatial scale.
Landscape variables plotted are those described in Table 3.2. Abbreviations are as
follows: ag=agriculture, open=open land, forwet=forested wetlands,
nforwet=nonforested wetlands. Other variables not abbreviated. 132

Temporal trend (1996-2002) in site occupancy for Acrr‘s crepitans blanchardi.
Open circles indicate proportion of survey sites occupied in natural landscape
contexts, and darkened circles indicate proportion of survey sites occupied in
anthropogenic landscape contexts. Solid line represents significant linear trend in
site occupancy in anthropogenic landscape context. 133

Temporal trend (1996-2002) in site occupancy for Rana pipiens in survey sites in
a natural landscape context. Solid line represents Significant linear trend in site
occupancy. 133

Location of study sites in southwestern Michigan. All study sites were natural
wetlands in Kalamazoo and Barry Counties. 161

Mg2+ concentration and Specific Conductance of the 10 study sites. Both
variables were used as an indicator of water source, with sites having higher
MgZ+ concentrations and Specific Conductance more predominantly
groundwater-fed, and site with lower values for both variables more precipitation-
fed. Values plotted represent mean values of measurements taken monthly during
the anuran breeding season (n=5) and their associated std. errors. 162

xii

4.3

4.4

4.5

4.6

Relative abundance of invertebrates and fish sampled by bottle traps in 10 study
sites. Predatory fish were present in 6 of the 10 sites, but comprised >5% of the
total community in only one site (Eagle Pond). Species of predatory fish included
Bluegill, Green sunfish, Pumpkinseed, and Largemouth bass. 163

Growth rates of tadpoles at field study sites. Growth rates were measured at
approximately weekly (5 -— 8 days) intervals. Sites with open symbols are sites
where mean water temperature was greater than 16°C; sites with darkened
symbols had mean water temperatures less than 16°C. 164

Growth Rate (mm/day) profiles of Spring peeper tadpoles over five weeks of field
study related to (A) Pond permanence, (B) Canopy cover, and (C) Presence of
fish predators. 165

Development Rate (mm/day) profiles of Spring peeper tadpoles over five weeks
of field study related to (A) Pond permanence, (B) Canopy cover, and (C)
Presence of fish predators. 166

xiii

Status and distribution of frogs and roads in southern Michigan:
Population trends and the influence of habitat and landscape characteristics

Introduction

The issue of declining amphibian populations in many locations worldwide has
been the source of both great concern and research over the past two decades. While
declines in some areas have been attributed to relatively unambiguous causes (e.g.,
Johnson 1992, Hayes and Jennings 1986), the situation remains enigmatic in other areas
where habitats remain intact and no obvious sources of disturbance to natural population
dynamics have been identified (e. g., Richards et a1. 1993, Lips 1998). Some issues
concerning anuran amphibians in the midwestem United States have been addressed
(e.g., Lannoo 1998), but the factors affecting population dynamics of many species in this
region remain poorly understood.

The hypothesized causes for widespread amphibian declines include habitat
modification, chemical contamination (including acidification), predator introductions,
global climate change, ultraviolet radiation, disease, parasitic infections, or synergistic
interactions among multiple factors (e.g., Blaustein et al. 1994a, Sarkar 1996, Green
1997b, Alford and Richards 1999). The best documented amphibian declines have been
attributed to habitat modification. Habitat loss is unambiguously related to reduced
amphibian abundance and diversity (Johnson 1992, Hecnar and M’Closkey 1996, Hecnar
1997). Alteration of breeding, foraging, and/or overwintering habitats also has drastically
impacted amphibian populations (e.g., Welsh 1990, Delis et a1. 1996). Environmental

contaminants and acidification negatively affect amphibian populations. The effects of

acidification have been well documented (e. g. Freda and Dunson 1986, Freda et a1.
1991), and environmental contaminants also reduce growth, development, and survival of
anurans (Power et al. 1989). However, the long-term effects of routinely applied non-
point source environmental contaminants (e. g., pesticides) have not yet been determined
(Bishop 1992). Predation plays a very important role in the distribution and abundance
of amphibians at all life history stages (e. g., Alford 1999). Widespread introductions of
predatory game fish into formerly fish-fi'ee habitats, as well as the introduction of
bullfrogs outside their native range, have led to dramatic reductions in amphibian
diversity at some sites (e. g., Bradford 1989, Lannoo et a1. 1994, Fisher and Shaffer 1996).
Ultraviolet radiation reduces survival of eggs and tadpoles, and has been implicated as a
causal factor in high altitude regions with species that breed in shallow, clear water (e.g.,
Blaustein et al. 1994b, Blaustein et al. 1997, Licht and Grant 1997). Global climate
change has the potential to dramatically impact amphibian populations either through
shifts in phenology and/or breeding site use (Beebee 1995, Gibbs and Breisch 2001) or
interacting with other factors such as disease or a pulse of contaminants (Pounds and
Crump 1994).

Anuran distribution is a function of the morphological and hydrological
characteristics of wetlands, habitat heterogeneity within wetlands, community
interactions, and habitat/breeding site availability and suitability for adults (Smith 1983,
Wilbur 1984). Characteristics of both the local habitat and landscape setting affect the
distribution and abundance of anurans. Landscape patterns are an important influence on
adult distributions while larval communities and juvenile recruitment respond more to

local habitat characteristics (Bunnell and Zampella 1999). The primary factors

structuring tadpole communities are hydrology, predation, and competition (Smith 1983,
Berven 1990, Skelly 1997). Adult communities are structured through a complex suite of
interacting factors including wetland size, isolation, hydrology, substrate characteristics,
and the distribution of suitable aquatic and terrestrial habitats throughout their life history
(Wyrnan 1988, Laan and Verboom 1990, Skelly et a1. 1999). Anuran communities are
influenced by habitat factors that determine success and fitness at the population level
and landscape factors that determine which habitats are suitable (Lehtinen et al. 1999).
Evaluating responses of frogs and toads to habitat and landscape characteristics
during adult and larval stages of their life history should lead to a greater understanding
of relationships between local habitats and landscape variables for current populations
and result in improvements in the ability to project effects of climate and landscape
changes on future populations. As a result, we will be better able to identify, manage,
conserve, and preserve critically important components of natural ecosystems. The
objective of this study was to provide such data for Michigan frog and toad populations.
This study included the use of large-scale amphibian monitoring data from the statewide
Michigan DNR Frog and Toad Survey, implementation of geographic information
systems (GIS) as a tool to investigate patterns of amphibian distribution relative to
potential influential factors (e.g., climate, land cover), and evaluation of larval amphibian

communities in 10 wetlands in southwestern Michigan.

The specific objectives of this study were to:
(1) Assess the occurrence, abundance, distribution and projected status of each of

Michigan’s species of anurans using data collected by volunteers for the Michigan

Frog and Toad Survey (MFTS). This objective was addressed in Chapter One by
evaluating species’ distribution and site occupancy from1996 to 2002, calculating
population trends for each species, and comparing results using the multiple
analytical methods that have been used in other monitoring programs (e. g.,
Mossman et a1. 1998).

(2) Evaluate the quality and consistency of data collected by multiple volunteer
observers for the MFTS. This objective was addressed in Chapter Two using data
collected from a mail survey of active MFTS volunteers to evaluate how
volunteer background and experience influenced anuran identification and
abundance estimation. Data quality related to observer differences should be
included as covariates in population level analyses if possible (Sauer et al. 1994,
Kendall et al. 1996).

(3) Investigate relationships between anuran abundance and distribution and land
cover adjacent to and within their potential dispersal distance surrounding wetland
breeding sites using GIS analyses. This objective was addressed in Chapter
Three. Hypotheses tested were that amphibians would be (1) positively
associated with land cover types representing necessary habitat during some
portion of their life cycle (e. g., forest, wetlands) and (2) negatively associated
with land cover types representing habitat loss or alteration (e. g., urban, roads).

(4) Determine potential influences of habitat characteristics on growth, development
and survival of larval Spring Peepers (Pseudacris crucifer) in natural wetlands.

Conditions in the aquatic larval habitats can have a profound influence on tadpole

performance and ultimate metamorphosis (Wilbur and Collins 1973, Alford

1999). This objective was addressed in Chapter Four.

Chapter One
Short-term population trends and status of frogs and toads

in southern Michigan (1996-2002)

Abstract

The Michigan Frog and Toad Survey (MFTS), a volunteer-based amphibian
monitoring program, was established in 1996. Volunteers have since collected data
annually on abundance and distribution of Michigan’s 13 species of anurans. The MFTS
used protocols similar to other states and the North American Amphibian Monitoring
Program, so these results are comparable to other regional surveys. Volunteers surveyed
routes, each consisting of ten sites, after dusk on three dates each spring and recorded the
identity and index of calling intensity of breeding males. Population trends and status of
each species in southern Michigan were assessed using the MFTS data from 1996-2002.
Trend analyses indicated that populations of all species were highly variable through
time. Significant declines were not detected in Michigan. However, there were some
significant but slight trends detected. Rana catesbeiana increased in site occupancy,
while Rana palustris decreased. Mean annual call index increased over the study period
for Rana sylvatica, Pseudacris crucifer, H. versicolor/chrysoscelis, and Rana clamitans,
while Bufo americanus decreased. Rana catesbeiana also increased in route frequency
and abundance, while Hyla versicolor/chrysoscelis decreased. Rana palustris and Acris
crepitans blanchardi were too rare to be evaluated with statistical methods. This study
provides a six-year baseline record that can be used to track future anuran population

trends and identify declines in time to implement appropriate conservation measures.

Introduction

Since the early 1990’s, researchers have recognized that amphibian populations
worldwide have been suffering seemingly inexplicable declines (e.g., Blaustein and
Wake 1990, Pechmann et a1. 1991, Alford and Richards 1999). This is disturbing as
amphibians play important roles in many ecosystems. Many of their characteristics render
them good biological indicators of habitat quality, such as complex life cycles, trophic
relationships, permeable skin, use of both terrestrial and aquatic habitats, and sensitivity
to water chemistry during early life history stages (Stebbins and Cohen 1995, Bowers et
a1. 1998, US. EPA 2002). The hypothesized causes for widespread declines include
habitat modification, environmental contamination (including acidification), predator
introduction, global climate change, ultraviolet radiation, disease, parasitic infections or
synergistic interactions among multiple factors (e. g., Blaustein et al. 1994a, Sarkar 1996,
Green 1997b, Alford and Richards 1999). The primary obstacle for evaluating amphibian
declines is separating effects due to anthropogenic influences from natural population
fluctuations (Pechmann and Wilbur 1994, Travis 1994). To do this, long-term data from
extensive areas are needed (Blaustein et al. 1994a). Regional amphibian monitoring
programs can contribute such information. The values of data from long-term
monitoring programs include identification of distributional changes and the ability to
distinguish regional and long-term trends from local or short-term events not
representative of population level phenomena.

Surveys of calling anurans (hereafter call surveys) at wetland breeding sites are
being used for amphibian monitoring in many states and Canadian provinces (e.g., Huff

1991, Bishop et al. 1997, Lepage et al. 1997, Mossman et al. 1998). These surveys have

potential to provide valuable information about population size and status (Zimmerman
1994, Shirose et a1. 1997, Driscoll 1998). Call surveys are an efficient and effective
method of surveying anurans over large geographical regions (Scott and Woodward
1994), and have successfully allowed collection of data on presence and abundance of
frogs and toads in many areas. Call survey protocols are easy for volunteers to learn and
provide the additional benefit of promoting education about wetland ecosystems and
amphibian conservation.

The Wisconsin Department of Natural Resources (WDNR) was the pioneer in
developing protocols for call surveys at breeding sites. The Wisconsin Frog and Toad
Survey (WFTS), initiated in 1981, was the model for many state and national amphibian
monitoring programs (Mossman and Hine 1984, Mossman et al. 1998, Weir and
Mossman, in press). Developed in response to known or suspected declines in some
anuran species, the WFTS was based on the successful North American Breeding Bird
Survey (BBS, Robbins et a1 1986, Peterjohn et a1 1994). The original WFTS protocols
were subsequently modified and extended to the national level for the North American
Amphibian Monitoring Program (NAAMP) that was established in 1994. NAAMP
develOped a unified protocol for volunteer-based call surveys that has been implemented
in 24 states so far (Weir and Mossman, in press).

In 1996, the Michigan Department of Natural Resources (MDNR) established an
annual statewide frog and toad survey that used a network of volunteers to monitor
breeding anuran populations (Sargent 2000). The Michigan Frog and Toad Survey
(MFTS) protocol was based on the successful WFTS. Michigan also contributes data

from several survey routes to NAAMP. The goals of the MFTS are fourfold: (1)

determine abundance and distribution of anurans in Michigan, (2) identify areas of
concern where more intensive research efforts should be allocated, (3) educate local
citizens and raise awareness of anurans and wetlands ecosystems, and (4) promote a
positive relationship between citizens and the MDNR as a result of this volunteer-based
program (Sargent 2000).

Call surveys contribute useful information for amphibian monitoring programs
provided several assumptions are met and/or acknowledged (Link and Sauer 1997).
These assumptions include: (1) call index values accurately represent numbers of
breeding males, (2) breeding (calling) males accurately reflect population size of that
species (i.e., sex ratio approximately equal or proportion of males and females in the
population is stable), (3) proportion of calling males heard and estimated remains
constant over time, (4) observers are reliable, and (5) patterns of change in call index
values generally track patterns of change in actual population sizes. Shirose et a1. (1997)
demonstrated that call counts of some species of anurans are linearly related to chorus
size. Thus, methods that base call index values on estimates of numbers of breeding
males should provide meaningful approximations of abundance. Quality of MFTS
observers was addressed in Genet and Sargent (2003, see also Chapter Two).

Michigan is home to 13 species of frogs and toads (see Harding 1997, Conant and
Collins 1998). These species breed according to a predictable phenology beginning in
early spring and extending into summer (Figure 1.1). One species, Blanchard’s cricket
frog (Acris crepitans blanchardi) is a species of special concern in Michigan, and has
been suffering recent distributional and population declines (Lee 1998, Lehtinen 2001).

Other than anecdotal reports, not much is known about the status and population trends of

other species. Data from the MFTS are intended to identify species or areas of concern in
need of further research and/or conservation efforts.

The MFTS currently has data for hundreds of survey routes (approximately 350
routes statewide) collected from 1996 through 2002. I investigated patterns of abundance
and distribution for Michigan’s anurans in the area where survey routes were most
heavily clustered, the southern lower peninsula. Specifically, my objectives were to use
the MFTS data over the time period of 1996-2002 to assess: (1) distribution of each
species and changes in site occupancy over time, (2) population trends for each species,
(3) effectiveness of multiple analytical techniques implemented by other amphibian
monitoring pro grams (e. g., Mossman et al. 1998), and (4) effectiveness and contributions

of the MFTS to long-term amphibian monitoring programs.

Materials and Methods
Michigan F rag and Toad Survey

The Natural Heritage Program in the Wildlife Division of MDNR established
survey routes and recruited volunteers for the MFTS beginning prior to the anuran
breeding season in 1996. Data are collected annually by volunteers; there are currently
seven years of data available for approximately 350 frog and toad survey routes
throughout Michigan. The routes are most heavily clustered in southern Michigan
(Figure 1.2). Detailed protocols for the MFTS (Sargent 2000) are summarized below.
Volunteers establish routes by submitting a map and descriptions of a series of 12
potential survey sites identified outside the anuran breeding season (i.e., without

consideration of habitat suitability or frog presence). Volunteers classify sites along

10

survey routes into one of six wetland type categories: (1) vernal pond, (2) wet meadow,
(3) bog or fen, (4) marsh, (5) wooded swamp, (6) pond (Table 1.1, Sargent 2000). The
state coordinator evaluates route location and suitability of each site along the route and
mails data sheets and instructions to volunteers. Each accepted route consists of 10
wetland sites separated by at least 400 m, so that origin of calls can be unambiguously
determined for each site.

The volunteers survey routes three times each spring and record the identity of
each species and an index of calling intensity. The three suggested survey periods are
early spring, mid to late spring, and summer, when minimum air temperatures are
approximately 7°, 13°, and 18° C, respectively. Michigan anurans breed according to a
predictable phenology that is largely dependent on weather conditions (Figure 1.1), and
the three survey periods encompass significant breeding effort by each species.
Volunteers conduct surveys beginning one-half hour after sunset and finishing before
midnight under favorable conditions of appropriate temperatures and little or no wind.
They wait at each site for 3-5 minutes before recording data to allow frogs to acclimate to
their presence, and then listen for an additional 5 minutes to identify all calls and assess
their intensity. Intensity of calling males is rated 0 through 3, with O = no individuals
calling, 1 = few individuals with non-overlapping calls (1-5 individuals), 2 =
distinguishable individual calls that overlap (6-12 individuals), and 3 = full chorus with
indistinguishable individual calls (>13 individuals).

After the three survey runs have been completed, volunteers return data sheets to
the MF TS state coordinator by 15 August each year. Verification in the form of

recordings or photos is encouraged for rare species or reports from outside a species’

11

documented range. Documentation is especially needed for A. c. blanchardi, and strongly
recommended for Hyla chrysoscelis and Rana palustris.
Climatological Variables

Temperature and precipitation data were obtained from the state climatolo gist in
the Geography Department at Michigan State University for 61 weather stations
(contained within three climate divisions) distributed throughout southern Michigan
(Figure 1.3). These data included daily average temperatures and precipitation for each
station for 1996-2002. Daily data were condensed into monthly mean values
corresponding to the anuran breeding season (February — July). Cumulative winter
precipitation (rain and snowfall, Nov — Mar), which influences spring wetland
inundation, was also determined for each climate station. These variables were
statistically evaluated to determine differences among years, and examined for
correlations with MFTS data using Spearman rank correlation analyses (Zar 1998).
Correlations between climate variables and anuran abundance were determined for those
months of spring wetland inundation and during the breeding season of each species
(Figure 1.1). MFTS routes were matched to their nearest climate station using a GIS
overlay of Figures 1.2 and 1.3. Route data were omitted from these analyses if no
climate station data were available within the nearest county. Climate station data were
omitted if more than 10% ofdays Within a month (i.e., 3 days) were missing for the
cumulative variables (monthly precipitation, total winter snowfall) or more than 25% of

days (i.e., 8 days) were missing in a month for mean daily temperature.

12

Data Analyses

To be included in statistical analyses, each route needed to satisfy the following
criteria: (1) all three survey runs were completed within a single year, and (2) routes were
surveyed at appropriate times according to anuran breeding phenology. Data were
condensed by determining the maximum calling index for each species for each site (i.e.,
merged data fi'om all three survey runs for each year), and combining the two gray
treefrog species (Hyla versicolor and H. chrysocelis). The calls of these two
morphologically identical species are difficult to distinguish, and temperature can affect
the pulse rate and duration of their calls, making them very difficult to identify in single
species choruses (Harding 1997). Verification for Hyla chrysoscelis (in the form of
recording or expert opinion) became a requirement for the MFTS in 1998, but prior to
that the identities of the two treefrog species were not formally determined. Therefore,
some records for H. chrysoscelis prior to 1998 may represent H. versicolor and vice
versa. All data were graphically assessed for normality using box and normality plots
prior to statistical analyses. Sample size was not equivalent in all wetland types or years
(Table 1.2). As a result of this and some variables not satisfying assumptions of
normality, both parametric and nonparametric analyses of variance were used in addition
to linear regression.

Patterns of species richness in different wetland types in each year were evaluated
with parametric and nonparametric analyses of variance (ANOVA and Kruskal-Wallis
tests). Following these analyses, post hoc tests were utilized for pairwise comparisons;

Bonferroni corrections were applied to all pairwise tests (Miller 1981). Trends of number

13

of species in each wetland type over the seven years of this study were also evaluated
with linear regression.

Several methods were utilized to analyze population trends. In analyses of
WFTS data, Mossman et a1. (1998) used techniques modified from trend analyses for
BBS data (e. g., Geissler and Sauer 1990, Link and Sauer 1994). Four techniques were
used to evaluate trends from the MFT S data: (1) site occupancy, (2) route frequency
regression, (3) route index regression, and (4) adjusted route index regression. Site
occupancy was determined by calculating the proportion of sites within each species’
range (Harding 1997, Conant and Collins 1998) occupied in each year. The total number
of sites where a species was present was summed and expressed as a proportion of the
total number of sites surveyed each year. These annual proportions were arcsine
transformed (Zar 1998) and regressed on year.

The other three route regression methods (frequency, index, and adjusted index)
were similar to one another in that dependent variables were regressed on year, trends
were determined for each route, and then averaged for all routes (Link and Sauer 1994,
Mossman et a1. 1998). Only routes where volunteers had submitted complete data for all
10 sites each year and that had been surveyed for at least three years were used in route
regression analyses. Slopes for all route trends were statistically compared to zero with t-
tests. The number of sites along each route where a species was present was summed for
route frequency regression (range 0-10). The call index values were summed for all sites
along a route for route index regression (range 0-30). For adjusted route index
regression, a number was arbitrarily assigned to the call index that may better

approximate the number of individuals present at each site (1 = 3 individuals, 2 = 9

14

individuals, 3 = 25 individuals). Mossman et al. (1998) arbitrarily assigned 3, 25, and 50
individuals for call index values 1, 2, and 3, respectively. However, given the
instructions provided to volunteers, my approximations probably more accurately reflect
relative abundance of calling males at a breeding site. The first two values represented
midpoints of the range of numbers of individuals given as guidelines in survey protocols
for call index values 1 and 2, and the third value represented the approximate number of
individuals above which enumeration was not practical. These adjusted index values
were then summed for all sites along a route (range 0 — 250). Data were log(x + 0.5)
transformed before statistical analysis (Geissler and Sauer 1990, Zar 1998). Trends were
determined and averaged across all routes for each species in southern Michigan and
reported as slopes from regression analyses with associated 95% confidence intervals.
Data from routes that were surveyed by the same observers for all seven years
were used for repeated measures analysis (von Ende 1993). This technique was used to
evaluate changes in abundance and site occupancy along routes sampled repeatedly from
1996-2002. Only routes that were completely surveyed (i.e., all 10 sites each year, all
three runs in each year, appropriate sampling dates, all seven years by same observers)
were kept for repeated measures analyses, as the same variables calculated for route
regressions were used (route frequency, route index, and adjusted route index). A total of
20 routes was included in the repeated measures analyses for all species except Bufo
fowleri, for which 12 routes were within its range. Univariate repeated measures analysis
was used, as it is generally considered more powerful than multivariate methods, and the
degrees of freedom for the F-test were adjusted to accommodate for time as the within-

subject factor (von Ende 1993). Statistical analyses were performed using SYSTAT

15

(version 8.0; SPSS, Inc. 1998). Unless otherwise reported, P < 0.05 is considered

statistically significant.

Results

During the first seven years of the MFTS, volunteers submitted data in at least
one year, yielding a total of 269 routes in southern lower Michigan. With the exception
of 1997, new routes were initiated each year, although the majority of routes was
established in 1996 (Figure 1.4). Thirty-three percent of routes were surveyed for only
one year, and > 60% of routes were surveyed for three or more years (Figure 1.5).
Volunteers submitted data for all seven years for 61 routes (22.7% of total). Complete
data for all 10 sites and three runs in each year for each route were submitted for 20 of
those 61 routes.

Climatological Variables

Monthly means of air temperature and precipitation during each month of the
anuran breeding season differed significantly among years during the time period 1996-
2002 (Figure 1.6). April 1997 and 1999 were particularly dry, as were February 1997
and 2001 and March 1998 (P < 0.001). There was low precipitation in May 1998 and
1999, and in June 1998 and 2002. High precipitation occurred in May 2000 and 2001 and
July 2000 (P < 0.005). The lower precipitation during May 1998 and 1999 (F = 27.106, P
< 0.001 , df = 6) corresponded to the peak breeding period for many Michigan anurans.
During 1996, February and March temperatures were significantly lower than

temperatures in these months in all other years (P < 0.001). Significantly higher low,

16

”NIHJIHIUBQ .IDL {IV In;

high, and daily mean temperatures occurred in February compared with February
temperatures in all other years (P < 0.001).
Patterns of Species Richness

There were no consistent patterns of species richness among wetland types (Table
1.3), although vernal ponds had significantly fewer species than some other wetland types
in 1998, 1999, and 2002 (Figure 1.7). The mean number of species in each wetland type
also varied from 1996 through 2002 (Figure 1.8). Although there were significant
differences in species richness within a given wetland type among years (Figure 1.8), no
significant linear trend in mean number of species present over the seven years was
detected for any wetland type (Table 1.4). All wetland types had the fewest species
present during one or more years between 1998-2000. Linear trends did not fit the data
well for any wetland type, explaining less than 1% of the variance in mean number of
species over seven years (Table 1.4).

Site Occupancy and Calling Intensity

Bufofowleri, Rana palustris, and Acris crepitans blanchardi were the least
common species in southern Michigan, occurring at fewer than 4% of survey sites in their
respective ranges (Table 1.5). Conversely, Pseudacris crucifer and Hyla
chrysoscelis/versicolor were the most common species, occurring at more than 67% of
sites (Table 1.5). Rana sylvatica, Pseudacris triseriata, Bufo americanus, and Rana
clamitans were common, occuning at 30-60% of the survey sites (Table 1.5). Rana
pipiens and Rana catesbeiana were less common, occuning at less than 20% of survey

sites (Table 1.5). While most species fluctuated with no consistent trend in site

17

occupancy over the seven year period, R. palustris declined significantly and R.
catesbeiana increased significantly (Figure 1.9, Table 1.6).

Mean call index values varied among years for most species (Figure 1.9).
Pseudacris cruczfer and H. chwsoscelis/versicolor called at the highest abundance, with
mean annual call index values > 2. Mean annual call index values of the remainder of
species were either much more variable (e. g., B. fowleri, A. c. blanchardi) or less than 2.
Although five species showed significant linear trends in call index values over time (R.
sylvatica, P. crucifer, B. americanus, H. versicolor/chrysoscelis, and R. clamitans), these
trends were not universal in direction nor did they explain much variability in the data (5
17%) (Table 1.7, Figure 1.9). Bufo americanus showed a trend of decreasing calling
intensity, while the other four species increased in calling intensity over the same time
period (Figure 1.9).

Site occupancy and calling intensity were significantly correlated with cumulative
precipitation during the breeding season for R. pipiens (Table 1.8). Acris crepitans
blanchardi call index values were marginally correlated with cumulative breeding season
precipitation (P < 0.10, Table 1.8). The proportion of sites occupied by Acris crepitans
blanchardi was also marginally correlated with cumulative winter precipitation (P < 0.10,
Table 1.8). Neither site occupancy nor calling intensity was correlated with cumulative
precipitation for any other species. Total annual snowfall (Nov — Mar) was significantly
correlated with anuran route variables for P. crucifer, H versicolor/chrysoscelis, A. c.
blanchardi, R. clamitans, and R. catesbeiana (Table 1.9). Significant correlations
between cumulative monthly precipitation and all anuran abundance route variables were

present for R. palustris, and R. clamitans in February, R. sylvatica in March, R. clamitans

18

in April, and R. clamitans in July (Table 1.10). Significant correlations between mean
monthly temperature and all anuran abundance route variables were present for R.
sylvatica in February and March, P. triseriata in April, and B. americanus and H.
versicolor/chrysosclelis in May (Table 1.11). Other species’ individual route variables
were also correlated with monthly precipitation or temperature, but were not consistent
across all three route variables (Table 1.10-1 1).
Route Regression Analyses

Trends evaluated with the three route regression methods were similar, with most
species showing no significant directional changes in presence and abundance over time
(Tables 1.12 - 14). The two exceptions were Hyla versicolor/chrysoscelis and Rana
catesbeiana. Hyla versicolor/chrysoscelis showed a slight but significant negative trend
for all route regression dependent variables (Tables 1.12 — l4). Rana catesbeiana showed
a slight but significant positive trend for all three route regression methods (Tables 1.12 —
14). The magnitude of the trend was small for both species, and the mean proportion of
variance accounted for by time was also relatively small (< 40% and < 30% variance
accounted for by all three route regression analyses for H. versicolor/ chrysoscelis and R.
catesbeiana, respectively).

Repeated Measures Analyses

Rana palustris and Acris crepitans blanchardi were not reported in sufficient
numbers along the 20 routes included in the repeated measures analyses, and were
omitted from these analyses. Five species showed significant trends over time using
route frequency, compared to three species for route index, and two species for the

adjusted route index regression (Table 1.15). The two species with consistent significant

19

trends for all three dependent variables were Rana sylvatica and Rana clamitans (Table
1.15). Rana sylvatica showed a significant 2nd order (quadratic) polynomial trend over
the seven year period for route frequency, and 5th order (quintic) trend for route index and
adjusted route index (Table 1.15); quadratic trends were marginally significant for this
Species for route index and route adjusted index (P = 0.053 and P = 0.070, respectively).
Rana clamitans showed a significant 5th order trend for route frequency (linear trend
marginally significant, P = 0.083); route index and adjusted route index showed linear
trends (Table 1.15). Other species that showed significant trends over time included
Pseudacris crucifer, Hyla versicolor/chrysoscelis, and Rana catesbeiana (Table 1.15).
These three species were all characterized by either linear or quadratic trends for both

route frequency and route index (Table 1.15).

Discussion

Volunteer-based anuran call surveys have the potential to provide long—term data
on anuran distribution and abundance over broad geographical scales. Amphibian
populations, however, can fluctuate dramatically under natural circumstances (i.e.,
Semlitsch et a1. 1996), and the challenge is to identify trends in amphibian population
dynamics that are potentially human-induced outside the range of natural variability
(Pechmann et a1. 1991, Pechmann and Wilbur 1994).

The mean number of species among different wetland types and within a single
wetland type fluctuated over the duration of the study, but showed no consistent trends.
Temperature and precipitation played roles in anuran presence and abundance at breeding

sites, as has been well documented (e.g., Duellman and Trueb 1986, Stebbins and Cohen

20

1995). Total annual snowfall was most strongly correlated with Pseudacris crucifer
occurrence and abundance, an early spring breeder. Snowfall was also significantly
correlated with the three species that breed primarily in permanent ponds, A. c.
blanchardi, R. clamitans, and R. catesbeiana, perhaps indicating the importance of
seasonal recharge of these ponds via either meltwater or groundwater inputs. Most of the
species showed significant correlations with temperature and/or precipitation during the
months of peak breeding activity. Surprisingly, R. sylvatica occurrence and abundance
was negatively correlated with both temperature and precipitation during its breeding
season. This is an explosively breeding species that may be often overlooked in call
surveys (Crouch and Paton 2002), and unless the spring thaw is abnormally delayed,
volunteers may often miss their peak breeding effort before conducting the first survey
run.
Species Accounts

Rana sylvatica is typically one of the first species to emerge and breed each
spring, usually concurrently with or following the onset of Pseudacris triseriata and P.
crucifer calling (Figure 1.1). The unpredictability of suitable weather conditions,
combined with its explosive breeding strategy (Berven 1990, Harding 1997) and
tendency to call only when air temperatures exceed 9°C (i.e., during warm afternoons,
Crouch 1999), make this species a challenge for observers to sample adequately in night
call survey monitoring programs. Rana sylvatica was only reported from 23-41% of sites
surveyed over the seven years of the study period. This may reflect low detection rates
rather than real trends in distribution. To account for relatively low detection rates and

temporal variability in calling behavior, NAAMP suggested adding an additional

21

sampling period to target this species (Weir and Mossman, in press). The MFTS
recommends that volunteers either conduct the first sampling period in conjunction with
R. sylvatica breeding activity or conduct an extra, separate sampling period during warm
afiemoons in early spring specifically to target this explosively breeding species (L. G.
Sargent, pers. comm).

Rana sylvatica has a broad distributional range that extends farther north than any
other North American amphibian species, and is considered common in suitable habitat
virtually throughout its range (Harding 1997, Conant and Collins 1998). In this study, R.
sylvatica typically called at intermediate calling intensities (mean annual calling index
varied between 1.5 —- 2), and showed a significant linear trend of increasing calling
intensity over the study period. Route regression analyses using all routes did not show
significant trends for wood frog populations. Repeated measures analyses, however,
indicated significant population trends over time, although the higher order polynomial
trends were difficult to interpret. These analyses suggest that R. sylvatica populations are
variable, but provide no evidence that there are population declines in Michigan. This
species is considered to be common and under no declines in the Great Lakes region
(Pentecost and Vogt 1976, Weller and Green 1997, Casper 1998, Moriarty 1998). The
WFTS data indicated a significant trend of population increase for this species, which is
at least partially due to observers improving their survey techniques over time (Mossman
et a1. 1998). The MFTS data are consistent with these patterns observed in the WFTS.

Pseudacris triseriata was heard calling in 49-54% of sites surveyed each year of
the study period and appears to have a stable, if not slightly increasing, pattern of site

occupancy. This species consistently called at intermediate calling intensities (mean

22

annual cal] index between 1.5 — 2). Over the duration of the study period, P. triseriata
populations fluctuated, but there was no indication of declines. The lower frequency and
acoustical carrying capacity of this species’ call make it difficult to detect at sites where it
is calling in a mixed chorus with more boisterous species (i.e., Pseudacris crucifer, see
Chapter Two). Compared to other species, Varhegyi et a1. (1998) found that the call of
P. triseriata attenuated the most rapidly with distance, perhaps affecting its detectability
from roadside listening sites. Pseudacris triseriata also frequently calls during daylight
hours (Harding 1997). If their calling activity is reduced at night, this species may also
be missed more frequently with nighttime call surveys. Perhaps in addition to R.
sylvatica, P. triseriata could also be targeted with an earlier survey run before sunset.

My results are congruent with results for p0pulations of P. triseriata in nearby
states. There have been no reports of declines on a statewide basis for any nearby state
(Pentecost and Vogt 1976, Brodman and Kilmurry 1998, Casper 1998, Mierzwa 1998,
Moriarty 1998). However, the WFTS suggested that P. triseriata may be declining in
some regions (Mossman et a1. 1998). Others have also noted population declines for this
species, especially in areas where habitat has been lost or altered (Daigle 1997, Weller
and Green 1997, Harding 1997). Pseudacris triseriata is highly dependent on open
marshes as well as wooded swamps, and is strongly influenced by climatological
variables that determine both the timing of breeding events and suitability of habitat (i.e.,
meltwater and spring rains that fill wetlands) (Harding 1997); therefore any changes to
their habitat could be translated to population level impacts.

Pseudacris crucifer is the most widespread and abundant anuran in southern

Michigan, occurring at 79-89% of all sites surveyed, and calling at or above a mean

23

annual calling index of 2.5. Based on these data, there is no indication that this species is
suffering any population declines in southern Michigan. Similarly, throughout its range,
this species is widespread and abundant (Harding 1997, Weller and Green 1997).
However, significant but small declines have been noted in Wisconsin for P. crucifer
despite a frequency of occurrence greater than all other species (Mossman et al. 1998).
Pseudacris crucifer populations are well suited to monitoring with call surveys because
of their high detection rates and consistent calling behavior (Mossman et a1. 1998,
Crouch 1999).

Rana pipiens was found at 11-17% of survey sites over the study period. This
species typically calls at lower intensities or in smaller choruses than some other species
(mean annual call index 1 — 1.5). The relative rarity of this species and the paucity of
records for full choruses in call surveys could be due to the low volume of its call
(Bishop et al. 1997) and/or could reflect decreasing distribution and abundance. Rana
pipiens was generally very abundant in the Great Lakes region until the 19605 (Pentecost
and Vogt 1976), and populations have since declined in many areas (Hine et a1. 1981,
Vogt 1981, Moriarty 1998, Mossman et a1. 1998). Although R. pipiens can be locally
abundant in open marshes (e. g., Harding 1997, Mierzwa 1998), several factors have
contributed to declines in distribution and abundance throughout its range including
drainage and conversion of wetlands for agriculture or aquaculture, exposure to
pesticides, collection for the biological supply trade and bait, and introduction of
bullfrogs in the western parts of its range (Lannoo et al. 1994).

Rana palustris has many similarities to R. pipiens, including similar calls and

timing of breeding events; however, these species do not typically breed in the same

24

habitats. In southern Michigan, R. palustris is poorly detected in call surveys, occurring
at 1-3% of survey sites. Its breeding call is low and subtle, and it was not frequently
heard calling above a call index value of 1. The low level of site occupancy of R.
palustris precluded evaluation of population trends over time with repeated measures
analysis. Rana palustris is generally uncommon, perhaps as a result of its preference for
cool, clear waters and intolerance to pollution (Harding 1997). The limited data for R.
palustris make evaluation of population trends difficult, although it did show a significant
decrease in site occupancy over the study period (Figure 1.9, Table 1.6). The MFTS data
suggest that this species is either rare, undersampled and/or confused with the similar
looking and sounding R. pipiens (Genet and Sargent, 2003; see Chapter Two), but is
persisting in suitable vx etlands.

Bigfofowleri reaches the edge of its range in Michigan and is limited to the
western and southem parts of the lower peninsula (Harding 1997). This species is either
very rare or not detected well by the MFTS, as it was present at only 1—4% of all sites
within its range over the study period. Presence at survey sites and mean annual call
index values fluctuated over the study period, but did not show significant trends.
Although B. fowleri is at the edge of its range in the Great Lakes region, others have
reported that it is relatively widespread and locally abundant (Harding 1997, Brodman
and Kilmurry 1998). Detailed investigations of demographic and life history
characteristics of this species have provided information on population status in certain
regions (e.g., Breden 1988, Green 1997a). Green (1997a) found that age structure and
sex ratio of this species are unstable in Ontario populations, and call surveys are not

likely to accurately reflect population size if only chorusing males are sampled. Similar

25

detailed studies in Michigan would be of value to monitoring and conservation of B.
fowleri.

Bufo americanus is common in southern Michigan, occuning at 34-42% of sites
during the study period. This species typically called at intermediate abundances (mean
annual call index 1.5 — 2). The prolonged trill of B. americanus makes it difficult to
compare its call index values to other species with shorter calls (e.g., P. crucifer, Rana
clamitans). The mean annual call index of B. americanus declined significantly over the
study period, although little variance was explained by this relationship. No other
significant changes in abundance or distribution (i.e., frequency of detection) were
detected over the past seven years. This species is generally common throughout the
Great Lakes region, but has experienced local population declines in some regions
(Harding 1997). A review of the status of B. americanus in nearby states indicated that
populations are stable, common and widespread, and that populations survive in altered
habitats and during drought years (Brodman and Kilmurry 1998, Hemesath 1998,
Mierzwa 1998, Moriarty 1998, Mossman et al. 1998).

Hyla versicolor/chrysoscelis is one of the most abundant and frequently
encountered anurans in southern Michigan, occurring at 68-80% sites and frequently
calling in small to large choruses (mean annual call index >2). This species showed a
significant increase in calling intensity over the past seven years, although this
relationship accounted for little variance in the data. This species also showed
significantly decreasing population trends in all three route regression methods, as well as
a significant polynomial trend in the repeated measures analyses. Although their calling

intensity increased slightly over the duration of the study, route regression analyses

26

indicated negative trends for all three dependent variables. This may indicate that there
are more individuals calling from fewer sites. Another potential problem is that the data
represent two treefrog species. These two morphologically identical species have similar
calls that are influenced by temperature and difficult to distinguish (Gerhardt et al. 1994,
Bertram and Berrill 1997). In southern lower Michigan, Hyla versicolor is more
common than H. chrysoscelis (Harding 1997). This is also the case in Wisconsin
(Mossman et al. 1998). Hyla versicolor and H. chrysoscelis appear to have stable
p0pulations and are common, and widespread (Weller and Green 1997, Brodman and
Kilmurry 1998, Mossman et al. 1998).

Acris crepz'tans blanchardi is classified as a species of special concern by MDNR.
Many populations have vanished from historical locations, although some isolated sites
appear to have healthy populations in southern Michigan (Lee 1998, Lehtinen 2001). In
southern Michigan, A. c. blanchardi was very rare, occurring at 03-32% of sites, and
calling at variable intensities over the past seven years. The low frequency of occurrence
precluded evaluation of population trends with repeated measures analyses. Route
2';- gression analyses did not indicate any significant declines. This species suffered
drastic declines in northern portions of its range during the late 19703 am! 19803, and
many previously healthy populations are now either greatly reduced or extirpated in the
Great Lakes region (Harding 1997, Hay 1998, Mierzwa 1998). This is the only anuran
with protected status in Michigan. Additional studies at specific sites should be
conducted periodically to assess population viability, as the MFTS is unlikely to
document significant population trends (other than extirpations) as a result of low

frequency of occurrence in wetlands throughout the region.

27

Rana clamitans is a common and widespread species in southern Michigan,
consistently occurring at 54-68% of sites surveyed, but calling at relatively low
intensities (mean annual call index < 1.5). Although route regression methods did not
indicate a significant population trend, repeated measures analyses indicated significant
trends, usually in an upward direction. Although R. clamitans usually called in small
numbers, their short distinct calls make it unlikely that they will be heard in full choruses
of indistinguishable individuals even though many individuals may be actively calling
from a breeding pond. Harding (1997) reported that R. clamitans is one of the most
conspicuous and abundant anurans in the Great Lakes region, and others have also noted
its abundance and ubiquity (Weller and Green 1997, Brodman and Kilmurry 1998,
Mierzwa 1998, Mossman et al. 1998). The high frequency of occurrence of this species,
coupled with its consistent calling pattern, make it well suited to population monitoring
using call surveys (Crouch 1999).

Rana catesbeiana is the largest and longest-lived anuran in North America
(Conant and Collins 1998). This species is relatively uncommon in southern Michigan,
occurring at 14-20% of survey sites and calling at low intensities (mean annual call index
<1 .5). Despite its relatively low frequency of occurrence, R. catesbeiana showed
significant pepulation trends including increasing site occupancy and increasing trends
for all route regression analyses. This species is native to eastern North America, but has
been widely introduced to western regions, often with drastic consequences for native
amphibian populations (e.g., Hayes and Jennings 1986, Fisher and Shaffer 1996).
Generally, R. catesbeiana is either widespread (e. g., Mierzwa 1998) or irregularly

distributed (e.g., Mossman et al. 1998). No evidence of significant declines of this species

28

within its native range has been reported. Rana catesbeiana requires permanent water
bodies as breeding ponds, as tadpoles overwinter and typically do not metamorphose
until their second or third summer (Harding 1997). Perhaps the increasing trends noted
for this species in southern Michigan reflects a change in the distribution of permanent
waters in the region (Dahl 2000); a landscape level analysis of habitat change over the
past decade would contribute to understanding these patterns.
Analyzing Papulation Trends from Call Surveys

Protocols and data analysis methods for amphibian call surveys have largely been
adapted from BBS methodology (e.g., Geissler and Sauer 1990, Mossman et al. 1998).
Even though anuran call surveys have been widely implemented in the USA. and
Canada, few programs have accumulated data long enough for meaningful analyses of
population trends. Shirose and Brooks (1997) suggested that populations should be
monitored for a minimum of the generation time of the longest-lived species surveyed.
In Michigan, the longest lived species is R. catesbeiana, which lives at least 7-8 years in
the wild (Zug 2001); its lifespan is potentially much longer, based on captive animals
(Staniszewski 1995, Harding 1997). Optimally, species should be monitored over
multiple generation times since they typically show lags between environmental impact
and population level response (Shirose and Brooks 1997). Wisconsin is the only state
that has accumulated call survey data over long enough time intervals to allow rigorous
analyses (Mossman et al. 1998). I used the methods of Mossman et al. (1998) to
facilitate comparisons of population trends in Michigan with trends for Wisconsin.

Several methods to analyze population trends from count data could potentially be

modified to accommodate call index data, such as route regression and rank trends

29

analyses (Thomas and Martin 1996), but consensus on the most accurate or reliable
method is lacking. Thomas and Martin (1996) found that three common methods for
analyzing population trends in BBS data generally produced similar results, but the
number of significant declines in bird populations differed depending on the method
used. They suggested that additional research on population trends analysis methods was
warranted, as the ability to detect significant population declines and prioritize species for
conservation efforts depend strongly on the method used.

The route regression analyses used here and in Mossman et al. (1998) are based
on standard route regression methods for BBS data (Geissler and Sauer 1990). They
involved evaluating trends over time for each route and averaging those trends for each
region of interest. Three dependent variables were calculated (route frequency, route
index, and adjusted route index), representing presence/absence (i.e., detection/
nondetection) and abundance data. Using a number of techniques to evaluate population
trends appears prudent, as results can be compared among methods and redundancies
used to select the most appropriate and useful methods. When significant trends are
detected by multiple methods, confidence in the results also increases. Mossman et al.
(1998) found route frequency regression to be the most useful technique because it
accounted for differences in sampling intensity among regions and allowed analyses of
routes not run every year. In my analyses, the three route regression analyses produced
similar results, as did analyses of the proportion of sites occupied over time. The
direction of each species’ trend was the same for all route regression methods, except for
Rana sylvatica. However, the only significant trends noted were a slightly decreasing

trend for Hyla versicolor/chnzsoscelis and a slightly increasing trend for R. catesbeiana.

30

Call index values were difficult to interpret, since adjusted route index regression
involved arbitrary approximation of number of breeding males in a chorus. The call
characteristics of each species influence perceived abundance and assignment of call
index categories. In effect, it takes fewer individuals with a prolonged call such as Bufo
americanus to be perceived at higher call index values of 2 or 3 than a species with a
shorter call such as Rana clamitans. Sargent (2000) provided a numerical guide for call
index values in addition to the subjective determination of the degree of overlap and
distinction of individual calls as a means of making assignment of index values more
uniform among volunteers (see Methods section). Even so, the relationship between the
calling index and actual numbers of breeding males needs to be evaluated for each
species before call index data can be reliably used to track changes in abundance.
Researchers and managers generally agree that detection/nondetection data are the most
valuable contributions of call surveys at the current time (Bishop et al. 1997, Bonin et al.
1997, Green 1997a, Lepage et al. 1997, Mossman et al. 1998, Weir and Mossman, in
press). Volunteers should continue to record call index values so that these data can be
used to evaluate changes in abundance once empirical relationships are developed (Weir
and Mossman, in press).

Repeated measures analyses of variance (i.e., von Ende 1993) are widely used and
provide a statistically rigorous means to evaluate changes at sites repeatedly sampled
over time. This method allows evaluation of polynomial trends. Polynomial trends such
as those seen in this study for R. sylvatica and R. clamitans are difficult to interpret and
may be of little use or biological significance. Lower order trends were marginally

significant (P < 0.10) for these species. If such trends are biologically meaningful, results

31

could still be useful in planning for conservation and management of these species and
should be reported. Repeated measures analyses are most useful when all sites have been
surveyed for long periods of time. MFTS volunteers join and/or drop out of the surveys,
so the number of sites included only represent a very small fraction of the total that can
be used in other population trends analysis techniques. In this study, data from only 20
routes could be used in repeated measures analysis (12 routes for B. fowleri). The low
sample size limits the power to detect significant trends, and the trends detected by
repeated measures but not by other procedures (i.e., route regression) could be an artifact
of this low sample size.

I agree with the suggestion of Mossman et al. (1998) to continue evaluation of
multiple analytical techniques to evaluate trends in anuran call survey data. The MFTS is
still a relatively young monitoring program. Although it is currently in its eighth year and
has achieved the duration recommended by Shirose and Brooks (1997), the value of data
for the identification of population trends increases over longer time scales. Perhaps ten
years is a more realistic minimum for meaningful population trends analyses. Call
surveys provide valuable information, but the data must be analyzed and interpreted
within the limitations of the survey methodology. These limitations include data
reliability, interobserver variation, dependence on volunteers for long-term data,
subjective selection of survey routes, and poor understanding of how call index values are
related to actual population size.

Management and Conservation Implications
Prior to the establishment of the MFTS, reports of anuran status, abundance, and

distribution in Michigan were largely anecdotal or unknown. Trend analyses of the first

32

seven years of data on the abundance and distribution of 13 species of anurans indicated
that populations of all species fluctuated between years, but no significant declines were
detected in Michigan. However, slight but significant trends in site occupancy, annual
call index values, frequency and abundance that emerged from statistical analyses for
Rana sylvatica, Rana palustris, Rana catesbeiana, Pseudacris crucifer, Bufo americanus,
Hyla versicolor/Chrysoscelis, and Rana clamitans need to be monitored closely, and if
they continue, should be evaluated to determine potential influential or causal factors.
Several species were poorly detected by call surveys in southern Michigan, and these
species warrant additional attention at specific sites to determine if populations are stable
or declining.

Anurans are dependent on wetlands for breeding, and many species require
wetlands for their entire life cycle. Unfortunately, wetlands are also among the nrost
endangered ecosystems in the world. Wetland habitats in the midwestem United States
have suffered losses in excess of 75% for areas dominated by agricultural, industrial, and
urbanized activity (Detenbeck et al. 1999). The landscape of southern lower Michigan is
characterized by agriculture and urban development, and this region has experienced the
greatest loss of wetlands in the state since the 1800’s (Comer 1996). Conservation of
Michigan’s anurans will require conservation of their wetland habitats. Most wetland
losses occurred from the late 1800’s and early 1900’s (Comer 1996). Thus, current
surveys document status in remaining wetlands and highlight the need to protect and
manage these ecosystems.

The MFTS collects valuable information concerning abundance and distribution

of native frogs and toads. The value of this dataset increases as annual data are

33

accumulated, providing a long-term record of species presence and abundance. Given
reports of widespread declines worldwide and for some species in the Great Lakes region,
these data provide evidence that Michigan frog and toad populations are relatively stable
at present and also provide a baseline for determining species status and protecting
species from future losses. The results are a conservative estimate of species trends;
given the relatively low resolution of calling index data and the fact that species
undetected by call surveys may not necessarily be absent from a site. The MFTS should
continue to monitor frogs and toads indefinitely, as the overall population trajectory

needs to be distinguished from the short-terrn fluctuations.

Acknowledgements

Correspondence with J. R. Sauer, M. J. Mossman, and L. Bailey aided in data analyses
and interpretation. T. M. Burton, S. K. Hamilton, D. J. Hall and J. H. Harding provided
valuable comments on earlier drafis. Support for this research was provided by the
Michigan Coastal Management Program of the Michigan Department of Environmental
Quality, National Oceanic and Atmospheric Administration, US. Department of
Commerce, Geological and Land Management Division of the Michigan Department of
Environmental Quality with funding from the State and Tribal Wetland Development
Program, Region V of the US. Environmental Protection Agency, Great Lakes Wetland
Research Consortium of the Great Lakes Commission with funding from the Great Lakes

National Program Office, and the US. Environmental Protection Agency.

34

Table l .1

Wetland categories assigned by volunteers to MFTS survey sites. Descriptions of
wetland types are from Sargent (2000).

 

Wetland Type

Description

 

Vernal Pond

Vernal ponds are small bodies of standing water that form in the spring
from meltwater and are often dry by mid-summer or may even be dry
before the end of the spring growing season. Many vernal ponds occur in
depressions in agricultural areas, but may also be found in woodlots.
Wetland vegetation may become established but are usually dominated by
annuals.

 

Wet Meadow

Wet meadows usually look much like a fallow field except that they are
dominated by water-loving grasses and sedges. They will contain nearly
100% vegetative cover with very little or no open water. Any surface
water present is temporary or seasonal and only during the growing season
in the spring. Wet meadows often form a transition zone between aquatic
communities and uplands with soils that are often saturated and mucky.

 

Bog or Fen

Bogs are found on saturated, acid peat soils that are low in nutrients. They
support low shrubs, herbs and a few tree species on a mat of Sphagnum
moss. Some bogs are totally overgrown and some consist of open water
surrounded by floating vegetation. Acid-tolerant plants found in and
around bogs include woody plants such as labrador tea, poison sumac,
tamarack, and black spruce. Many species of orchids prefer bog habitats,
as do insect-eating sundews and pitcher plants. Bogs are usually only
found in the northern part of Michigan. Fens are similar to bogs except that
the soils are more alkaline because they result from water passing through
calcareous deposits. Fens have a higher plant diversity than bogs due to
higher nutrient levels. Fens can be found in the southern part of Michigan.

 

Marsh

Marshes have standing water from less than an inch up to 3 feet deep. The
amount of water can fluctuate seasonally or from year to year. They are
dominated by soft-stemmed emergent plants such as cattails and rushes.
Vegetative cover is usually around 50%. In Michigan, marshes can be
found at the edge of some rivers and lakes, in lowlands and depressions,
and in swales between sand dunes.

 

Wooded
Swamp

Wooded swamps are aptly named because they are dominated by woody
plants such as shrubs and/or trees. The soil is saturated throughout the
growing season. Some may become dry during the summer months. In
Michigan, trees and shrubs found in wooded swamps include red and
silver maple, cedar, balsam, willow, alder, black ash, elm and dogwood.
They often occur along streams or on floodplains, in flat uplands or
shallow lake basins.

 

Pond

 

 

Ponds are open bodies of water that are less than 20 acres in size and that
do not dry up during summer months. There is little emergent vegetation
but some floating vegetation may occur around the edges.

 

35

 

Table 1.2

Number of sites of each wetland type for which volunteers submitted complete data to the
MFTS. Wetland types and years not satisfying the assumptions of normality are
indicated with an asterisk (*).

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Wetland Type 1996 1997 1998 1999 2000 2001 2002
Vernal Pond 165 111 142 150 123 134 73
Wet Meadow 110 80 106 91 71 68 42
Bog or Fen 34 14* 25 26* 18 15* 10*
arsh 457 280 384 385 293 275 135
Wooded Swamp 388 285 355 327 265 258 127
Pond 434 296 363 370 267 270 138
Total 1588 1066 1375 1349 1037 1020 525
Table 1.3

Summary of parametric ANOVAs and nonparametric Kruskal-Wallis tests evaluating
differences in species richness among wetland types each year, 1996 — 2002. Sample
sizes are listed in the bottom row of Table 1.2, and all tests had 6 degrees of freedom.

 

 

 

 

 

 

 

 

 

 

 

 

ANOVA ANOVA K—W K—W
Year F ratio P value H P value
1996 1.763 0.103 11.455 0.075
1997 1.599 0.144 8.312 0.216
1998 2.888 0.008 14.705 0.023
1999 5.203 <0.001 32.836 <0.001
2000 1.896 0.079 12.107 0.060
2001 0.562 0.760 2.651 0.851
2002 2.711 0.013 15.704 0.015

 

 

 

36

Table 1.4

Results of simple linear regressions of the mean number of species present in each
wetland type over the time period 1996-2002.

 

 

 

 

 

 

 

 

 

 

 

 

Wetland Type Rr P value
Vernal pond 0.001 0.446
Wet meadow 0.001 0.552
Bog/fen 0.003 0.520
Marsh <0.001 0.783
Wooded swamp 0.002 0.073
Pond <0.001 0.670
Table 1.5

Proportion of all sites surveyed by each species, 1996-2002. The total number of sites for
each species includes only those sites within the species range. Sample size (i.e., number

of sites where present) for each species is given in parentheses in each year column.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Species 1996 1997 1998 1999 2000 2001 2002
Bufofowleri 0.031 0.029 0.029 0.012 0.035 0.021 0.018
(36) (19) (21) (10) (25) (8) (5) ——+
Rana sylvatica 0.335 0.256 0.260 0.318 0.226 0.356 0.408
(5 64) ( 289) (3 81) (45 7) (248) (3 84) (226)
Pseudacris triseriata 0.486 0.505 0.489 0.485 0.488 0.524 0.538
(817) (571) (716) (697) (536) (565) (298
Pseudacris crucifer 0.834 0.851 0.822 0.790 0.799 0.871 0.892
( 1402) (962) (1203) (1136) (874) (939) (494L
Rana pipiens 0.170 0.138 0.156 0.146 0.113 0.126 0.152
(286) (156) (229) (210) (124) (136) (84)
Rana palustris 0.021 0.013 0.014 0.014 0.012 0.003 0.009
(35) (15) (20) £20) (13) (3) (5L
Bufo americanus 0.420 0.407 0.347 0.344 0.374 0.366 0.421
(706) (460) (508) (494) (411) (395) (233)
Hyla versicolor/ 0.752 0.763 0.693 0.707 0.676 0.788 0.800
chrysocelis (1265) (863) 1014) (1017) (742) (849) (443)
Acris crepitans 0.023 0.032 0.026 0.006 0.003 0.013 0.018
blanchardi (3 8) (36) (3 8) (9) (3) (14) (10)
Rana clamitans 0.573 0.637 0.617 0.577 0.542 0.662 0.675
(963) (720) (903) (830) (595) (714) (374)
Rana catesbeiana 0.141 0.154 0.152 0.163 0.173 0.158 0.202
(238) (174) (223) (235) (190) (170) (l 12)

 

 

 

37

Table 1.6

Results of simple linear regressions for proportion of sites within ranges occupied by
each species during the period 1996 -— 2002. N = 7 for each species, representing each
year of the survey since 1996.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Species r2 P value
Bufofowleri 0.204 0.309
Rana sylvatica 0.199 0.315
Pseudacris triseriata 0.497 0.077
Pseudacris crucier 0.186 0.366
Rana pipiens 0.235 0.270
Rana palustris 0.592 0.043
Bufo americanus 0.014 0.802
Hyla versicolor/chrgoscelis 0.088 0.5 18
Acris crepitans blanchardi 0.236 0.269
Rana clamitans 0.200 0.314
Rana catesbeiana 0.691 0.021
Table 1.7

 

Results of simple linear regressions evaluating trends in mean call index values for each
species during the period 1996-2002. N = 7 for each species, representing each year of
the survey since 1996.

 

 

 

 

 

 

 

 

 

 

 

 

 

Species r2 P value
Bufofowleri 0.10 0.277
Rana sylvatica 0.10 <0.001
Pseudacris triseriata 0.01 0.194
Pseudacris crucifer 0.01 0.010
Rana pipiens 0.01 0.359
Rana palustris 0.02 0.665
Bufo americanus 0.04 <0.001
H la versicolor/chrysocelis 0.01 0.002
Acris crepitans blanchardi 0.17 0.119
Rana clamitans 0.02 0.004
Rana catesbeiana 0.01 0.257

 

 

 

 

38

Table 1.8

Correlations between cumulative precipitation during anuran breeding season (F eb-J ul)
or cumulative winter (Nov — Mar) precipitation (rain and snowfall) and either proportion
of sites within range occupied or mean annual call index values (1996-2002). Values are
Spearrnan rank correlation coefficients (rs; *p < 0.05, **p < 0.01).

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Cumulative Breeding Cumulative Winter

Season Precip. Season Precip.

Species Prop. Call Prop Call
occupied index occupied index
Bufofowleri 0.270 0.179 -0.306 -0.536
Rana sylvatica -0.607 -0.143 0.036 0.321
Pseudacris triseriata -0.286 -0.450 0.536 0.090
Pseudacris crucifer -0.357 -0.536 0.429 0.071
Rana pipiens -0.786* -0.964** 0.071 0.214
Rana palustris -0.180 -0.487 -0.162 0.252
Brfo americanus -0.107 0.643 -0.107 -0.571
Hyla versicolor/chrysoscelis -0.286 -0.107 0.393 -0.036
Acris crepitans blanchardi -0429 0.714T 0.750T -O.536
Rana clamitans -0.429 -0.500 0.679 0.643
Rana catesbeiana 0.286 -0.214 -0.286 0.643

P < 0.10
Table 1.9

Correlations between annual total snowfall prior to the anuran breeding season (Nov -
Mar) and each species’ route variables. Values are Spearrnan rank correlation
coefficients (rs; *p<0.05, **p<0.01). Sample size for each pairwise comparison is listed
in parentheses following the species name.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Species Route Route Index Adj. Route
Frequency Index
Bufofowleri (236) 0.090 0.092 0.092
Rana sylvatica (459) 0.025 0.031 0.037
Pseudacris triseriata (459) —0.050 -0.032 -0.021
Pseudacris crucifer (459) 0.267** 0.257** 0.227"
Rana pipiens (459) 0.000 0.002 0.003
Ranapalustris (459) -0.043 -0.043 -0.043
Bufo americanus (459) -0.021 -0.010 -0.002
Hyla versicolor/Chrysoscelis (459) 0092* 0.077 0.069
Acris crepitans blanchardi (459) 0.123” 0.127“ 0.129“
Rana clamitans (459) 0.139“ 0.173** 0.172**
Rana catesbeiana (459) 0.183“ 0.196** 0.198**

 

39

 

Table 1.10

Correlations between cumulative monthly precipitation prior to and during breeding
period and each species’ route variables. Values are Spearman rank correlation
coefficients (rs; *p<0.05, **p<0.01). Sample size for each pairwise comparison is the

same for all species (except Bufofowleri) for any given month and is given in the bottom
two rows of the table.

 

 

 

 

 

 

 

 

 

 

Februagy March Agil May June July
Bufofowleri
Frequency 0.032 0.053 -0.014 -0.051 -0.014
Index 0.031 0.053 -0.011 -0.054 -0.014
Adjusted Index 0.030 0.053 -0.010 -0.054 -0.013
Rana sylvatica
Frequency -0.009 -0.1 1 3 * 0009
Index 0.010 -0.113* 0.005
Adjusted Index 0.017 -0.115** 0.007
Pseudacris triseriata
Frequency 0.105 * 0.060 0.05 1 0.022
Index 0096* 0.047 0.056 0.049
Adjusted Index 0.087 0.033 0.058 0.067
Pseudacris crucifer
Frequency 0.103* 0.025 -0.019 0.038
Index 0.073 0.010 0.034 0.052
Adjusted Index 0.055 0.007 0.051 0.046
Rana pipiens
Frequency -0.075 0.001 0.044 0065
Index -0.083 0.005 0.043 0075
Adjusted Index 0086 0.004 0.043 -0.078
Rana palustris
Frequency -0.093 * 0.008 0.013 -0.023
Index -0.093* 0.008 0.013 0025
Adjusted Index -0.093* 0.008 0.014 -0.025
Bufo americanus
Frequency 0.048 0.035 -0.042 0089*
Index 0.020 0.019 -0.023 0.082
Adjusted Index 0.002 0.014 -0.013 0.072
Hyla versicolor/
chrysoscelis 0.102* -0.063 -0.048 0.078 0.036
Frequency 0.081 -0.066 -0.043 0.114* 0.049
Index 0.067 -0.049 -0.045 0.110* 0.054
Adjusted Index
Acris crepitans
blanchardi 0.082 0.087 -0.053 -0.027 -0.035 0084
Frequency 0.083 0.086 -0.053 -0.026 -0.033 -0.084
Index 0.084 0.085 -0.054 -0.026 -0.032 -0.084
Adjusted Index

 

 

 

 

 

 

 

 

 

40

 

 

 

 

 

 

 

 

 

 

 

 

Rana clamitans 0.135" 0.036 -0.097* 0.021 -0.087 -0.126**
Frequency 0.173” 0.059 -0.108* 0.003 -0.106* -0.137**
Index 0.176" 0.051 -0.093* -0.004 -0.101* -0.132**
Adjusted Index

Rana catesbeiana
Frequency 0.056 0.024 -0.050 0.053 -0.007 0.029
Index 0.063 0.019 -0.008 0.045 0.011 0.059
Adjusted Index 0.058 0.019 -0.006 0.040 0.009 0.061

N; Bufofowleri 253 253 256 261 259 256

N; all other species 575 578 582 583 575 566

 

41

 

Table 1.11

Correlations between mean monthly temperature during the breeding period and each
species’ route variables. Values are Spearrnan rank correlation coefficients (rs; *p<0.05,
**p<0.01). Sample size for each pairwise comparison is the same for all species (except
Bufofowleri) for any given month and is given in the bottom two rows of the table.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

February March April May June July
Bufofowleri
Frequency 0.016 -0.105 -0.042 0106
Index 0.015 -0.105 -0.040 -0.106
Adjusted Index 0.015 -0.105 -0.040 -0.107
Rana sylvatica
Frequency -0.176** -0.268** 0008
Index -0.l65** -0.274** 0.021
Adjusted Index -0.158** -0.273** 0.035
Pseudacris triseriata
Frequency 0.062 0.063 0.108“ 0.020
Index 0.035 0.039 0.099“ 0.016
Adjusted Index 0.017 0.023 0.096* 0.019
Pseudacris crucifer
Frequency -0.066 -0.095* -0.027 0071
Index 0050 -0.098* 0.008 -0.045
Adjusted Index —0.036 -0.087 0.020 -0.031
Rana pipiens
Frequency -0.019 -0.032 -0.002
Index 0026 0036 -0.002
Adfled Index 0031 -0.037 -0.003
Rana palustris
Frequency -0.041 -0.070 -0.033
Index 0041 -0.070 -0.033
Adjusted Index 0041 -0.070 -0.033
Bufo americanus
Frequency -0.082 -0. 140* *
Index > -0.095* 0141"
Adjusted Index 0090* -0.129**
Hyla versicolor/
chrysoscelis -0.039 -0.138** -0.109*
Frequency 0013 -0.147** 0080
Index -0.004 -0.137** -0.062
Adjusted Index
Acris crepitans
blanchardi -0.005 0.000 0.032
Frequency -0.008 0.000 0.032
Index 0009 0.000 0.032
Adjusted Index
Rana clamitans
Frequency 0.001 -0.062 0.068
Index 0.000 -0.037 0.107*

 

42

 

 

 

 

 

 

 

 

 

 

 

 

Adjusted Index 0008 -0.018 0.120**
Rana catesbeiana
Frequency -0.090* -0.002 0.007
Index -0.046 -0.051 0.034
Adjusted Index 0049 -0.053 0.035
N; Bufofowleri 237 242 246 244
N; all other species 524 526 534 540 533 519

 

43

 

Table 1.12

Results of route frequency regressions. Sample size indicates the number of routes used
for analysis for each species for all route regression dependent variables. Mean trends
(slopes of regressions in units of change in frequency per year) and associated 95%
confidence intervals indicate the direction and magnitude of the annual rate of change, P
values indicate whether trend (slope) differed significantly from zero.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Species N Mean Lower CI Upper CI P value
trend
Bifofowleri 25 -0.037 -0.084 0.010 0.120
Rana sylvatica 132 -0.007 -0.032 0.017 0.560
Pseudacris triseriata 133 0.009 -0.004 0.022 0.184
Pseudacris crucifer 130 -0.002 -0.009 0.005 0.612
Rana pipiens 1 15 -0.017 -0.038 0.004 0.1 l l
Rana palustris 35 -0.039 -0.083 0.004 0.077
Bufiamericanus 130 -0.002 -0.021 0.017 0.819
Hyla versicolor/chrysoscelis 131 -0.015 -0.026 -0.004 0.010
Acris crepitans blanchardi 21 -0.018 -0.046 0.009 0.173
Rana clamitans 132 0.000 -0.010 0.009 0.920
Rana catesbeiana 109 0.017 0.002 0.033 0.028
Table 1.13

Results of route index regressions. Mean trends (slopes of regressions in units of change
in route index per year) and associated 95% confidence intervals indicate the direction
and magnitude of the annual rate of change, P values indicate whether trend (slope)
differed significantly from zero.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Species Mean trend Lower CI Upper CI P
value
Bufofowleri -0.043 -0.098 0.013 0.124
Rana sylvatica 0.006 -0.021 0.032 0.677
Pseudacris triseriata 0.009 -0.008 0.026 0.293
Pseudacris cmcjer -0.002 -0.010 0.006 0.597
Rana pipiens -0.018 -0.041 0.005 0.116
Rana palustris -0.039 -0.089 0.012 0.127
Bufo americanus -0.004 -0.027 0.019 0.721
Hyla versicolor/chrysoscelis -0.017 -0.031 -0.004 0.014
Acris crepitans blanchardi -0.017 -0.05 0.016 0.304
Rana clamitans 0.003 -0.008 0.014 0.591
Rana catesbeiana 0.019 0.003 0.035 0.012

 

44

 

Table 1.14

Results of route adjusted index regressions. Mean trends (slopes of regressions in units
of change in adjusted index per year) and associated 95% confidence intervals indicate
the direction and magnitude of the annual rate of change, P values indicate whether trend
(slope) differed significantly from zero.

 

 

 

 

 

 

 

 

 

 

 

 

 

Species Mean trend Lower CI Upper CI P value
Bufofowleri -0.076 -0. 163 0.011 0.937
Rana sylvatica -0.003 -0.048 0.041 0.894
Pseudacris triseriata 0.013 -0.013 0.040 0.324
Pseudacris crucifer -0.001 -0.013 0.01 1 0.873
Rana pipiens -0.027 -0.062 0.008 0.136
Rana palustris -0.064 -0.146 0.018 0.121
Bufo americanus -0.009 -0.044 0.026 0.604
Hyla versicolor/chrysoscelis -0.028 -0.049 -0.008 0.007
Acris crepitans blanchardi -0.028 -0.083 0.026 0.289
Rana clamitans 0.005 -0.011 0.021 0.561
Rana catesbeiana 0.033 0.007 0.058 0.012

 

 

 

 

 

45

 

Table 1.15

Results of repeated measures analyses of variance for routes for which the same
volunteers submitted data for each year 1996-2002. Sample size is the same for all
species (n = 20) except Bufofowleri (n = 12). See text for explanation of dependent
variables (route frequency, route index, and adjusted route index).

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Species F P Value Polynomial Trend P Value
1. Route Frequency

Bufofowleri 1.370 0.260

Rana sylvatica 2.232 0.045 2 0.047
Pseudacris triseriata 1.070 0.381

Pseudacris crucifer 2.574 0.037 2 0.011
Rana pipiens 0.599 0.721

Bufo americanus 1.460 0.198

Hyla versicolor/chrysoscelis 2.997 0.025 2 0.048
Rana clamitans 2.681 0.031 5 0.025
Rana catesbeiana 1.902 0.090 1 0.025
2. Route Index

Bufofowleri 1.575 0.186

Rana sylvatica 2.614 0.021 5 0.011
Pseudacris triseriata 1.410 0.222

Pseudacris crucifer 2.743 0.028 2 0.001
Rana pipiens 0.924 0.476

Bufo americanus 1.254 0.284

Hyla versicolor/chrysoscelis 2.321 0.056

Rana clamitans 4.375 0.002 1 0.046
Rana catesbeiana 1.645 0.141

3. Adjusted Route Index

Bufofowleri l .670 0.166

Rana sylvatica 2.565 0.023 5 0.015
Pseudacris triseriata 1.569 0.172

Pseudacris crucifer 2.107 0.074

Rana pipiens 0.915 0.482

Bufo americanus 1 .212 0.305

Hyla versicolor/chorsoscelis 1.958 0.108

Rana clamitans 4.256 0.004 1 0.046
Rana catesbeiana l .530 0.174

 

 

 

 

 

46

 

Fi

gure1.1

Calling calendar for frogs and toads of southern Michigan. Arrows represent

approximate beginning, ending, and duration of calling period for males of each

species. Breeding effort is largely dependent on weather conditions; dates are
extremely variable and dependent on temperature and precipitation each year.
Figure courtesy of Lori G. Sargent (MDNR, MF TS coordinator).

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

MARCH APRIL MAY JUNE JULY]
‘— Rana sylvatica —’
*-——Pseudacris triseriata >
4 Pseudacris crucifer >
‘ Bufojbwlefi 5
‘l"'_ Rana pipiens ——>
*— Rana palustris ——’
‘— Bufo americanus—>
""_Hyla versicolor——'>
‘—_Hyla clfrysoscelis —’
‘— Acris crepitans blanchardi —’
I 4 Rana clamitans >
Approximate ‘—‘L Rana catesbeiana ——*
Breeding Period

47

Figure 1.2

Distribution of MFTS routes in Michigan The box represents the region evaluated
in detail in this study. Each point represents a route comprised of 10 survey sites at
anuran breeding ponds.

 

 

 

 

- h...

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

O
O O
a: . '
, . J’.‘
O. o: '0
L a . ' '

 

 

 

 

 

 

 

48

Figure 1.3

Location of climate stations used for temperature and precipitation in southern
Michigan. Daily temperature and precipitation data were obtained for these sites for
1995-2002 and matched to the nearest MFTS route.

/

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

49

Figure 1.4

Year of route establishment. There were a total of 269 routes in southern
lower Michigan. Each bar represents the proportion that was established in
each year of the MFTS.

 

0.7
m
L; 0.6

8

=: 0.5“

a
“a .
a 0.4

.9.
t: 0.3‘
§-
; 0.2-
0.1‘
0.1

 

 

 

11.4...

1996 1997 1998 1999

 

Figure 1.5

Number of years routes have been surveyed. Bars represent the
proportion of the 269 total routes in southern lower Michigan that was
surveyed for a given time period.

 

O O
i: o i:
.
till W Ur

Proportion of total routes
o
_ o
u. N

.o o
H

 

 

   

1 2 3 4 5 6 7
Number of years surveyed

50

.2899... .2599: 4999.... .895 ass seesaw? :32 asses? ease: 382.... .mm H .52:
38052 See =< .cousamooa 538 35:02 ADV 658388 Ewe—u :38 3552 AB .2383an :3: :35 36:22
EV .238388 32 :38 35:32 As .mooméofl .eommom wficoofi 9228 maize sousamooa new 888388 .8 £533

3 05mm

51

.3 s: a: a< 3: an .2 5. a: 5. a: a;

 

 

 

 

 

 

 

 

 

 

 

 

 

{ii . **i c **§§ u {i}... . §§§§ r {**i O {fiifi . **§* . {iii _ iii} . **** _ **** m-
I, . U
.. m s
T0 .w r W
U
I” T
m 2
- c. .m
x. .2
.2 m.
m
r: D. van
m
. 2 [mt . mm
L:
8 -8
.2 s: a: a... a: so...
’53.... _ **i . **ii . **li e §*§§ _ iii! m- o—
ro I
m a
m
.m .m
m s
S x
m .n
.m
-2 m
m .2
)
-8 O
-2
n

 

 

 

 

 

 

(3) annexadmer qfltq Amman: (team

52

(3) annexadurar M01 Klqruour anew

. .306 v a £53388 03ng 38880

Eotomcomv Baggage 35:8 28333 8865 Bozo. manage 53, 35cm .2053 =e 5m mime?“ Boson use «2 2: mac?
none/ea 0.8 .223 mg 98 .233 2m no 18352 8m Boa .592, new .Seoutom .9052. some .«0 .858 Em: Ban: 5 Guam
.890 See WEE .8 3% Ech a 85852 3:3 comm can» some 5 89¢ 95.83 5 882m Amm ..\+V 860% we 89:8: .822

S as»;

53

 

 

 

 

 

 

@593 5m 3838 BSA 9536 new 3838 ESQ
98m @2595 58E \won 83 HE? neon n38? 538 \won 63 RE? Wm
M ,m
can . M
V W m w a .2
m m m . a m
o n a w -v
on
-mé
32 33
m
p . L _ p g . . L . b r W.N
.m
M w
m m W
M M W m m ,.
fimé
32 03— m

 

 

 

 

 

sagoads }o Jaqumu ueaw

sayoads J0 Jaqumu ueaw

54

vaoa

aEmBm
3953 538

com

AB

3838 anon
83 BE?

_

98m 3803 5.88 \won

 

 

P—o—l

Ncom

 

 

 

 

_ ocm

 

 

ooom

-mé

 

 

 

Wm

sagoeds 5o laqumu ueaul

sayoods 5o .19qu mm

55

Figure 1.8
Mean number of anuran species detected calling (+/- SE) in each wetland type from

1996 — 2002. (A) Woody vegetation dominated wetland communities, (B)
Herbaceous vegetation/open water dominated wetland communities.

:5
M
|

&

Mean numba' of species (+/- SE
n. ‘3.

.N
Ln

 

 

 

 

 

 

   

Mean number of species (+/- SE

 

 

    

 

 

wetmeadow
01996 ll997 I1998 .1999 I2000 l2001 [32002

56

dog—mam 32393 new

5308 338m 28 n; I e; maid. com .8565: 53, 3:86.: 2m 32328:: 80:: 58¢in .mEmhwE 3:5 05
an 3255 2m 98 20:3 =e 8% 28:52 one £23 32 .mca =35?» F8288 05 :o cote—m fl 5:338 86 use
.38 Boat? began 05 no @883 0.8 $29 5?: :8 3:58 €02 62$ 5?: :8 .353 :38 8865 mpg—88
x83 .86on :08 an @2968 8% mo 59¢qu 8365 83:“. :25 .933 some .8 .888 EN: Ban: 5 3822:
382.3 86on 28 $5852 ESE comm .8825 some he.“ Amm .\+v 829, 5?: =8 :35 98 @2338 02m

3 03E

57

paydnooo sons JO uoguodmd

poydnooo sous JO uopxodoxd

cod

NOON Sou coca §~ mama baa ga—

 

Nod ..
vcd .,
cad .,

de A

2.0

gaexoxzm

 

 

 

8d

 

26..

and L.

5.6 .,

{Illl/

«\4\.I.|IYI<.

agar» arguam

 

cod

 

 

Noon SON coon 89 mag hag 063

 

 

 

 

 

 

 

o and » ~
who:
rWe mg
. 36: w
. m
3
_ 35 m
7 . N m.
w.
X
cad M
M; . .
hmfi
mad: (
RN 53:09.: Mn.
N 8; x .83 m m
~ cod L _
E 2.? 2
r w U W
8.? ~\+\w\~\ o
m W
,N N m.
omd: W
m
ON MN .
f 3.? m
89889.?
333%,. 33% x35 :3 .o.
m omd m

 

 

58

Noon Eon son 33 mag 33 gm:

 

 

*

Qmumoabfiisgmflg :3:

w

*

n-g-o

 

 

and

m 8.3

w.

u o; J,

m.

wows

m

3

m.

We?
8;
cod

.m .85

W

m. 23

O

u.

m 86

m.

m. mod
25

 

L
I

 

;

fithaE 33%

 

 

m.—

 

 

ovd r

ooom 32 M33 3m—

h h b by P

Noam Son 39

 

 

I m
.3

hazautmfiu gum

N V.
V...

V!
N

(38 -/+) xapm [res Imam

 

 

3.0 -r

Ed:

26.,

and

 

E]

5?: =8 o
59889 a

Emfii 33x

 

‘0.
.—1

N
-/+) xaptn 1180 mm

tnmm

 

 

59

NOON _OON OOON an mag 50% com—

 

 

 

 

 

 

 

ood _
.m 86
.m At mN._
o . w
m. m
m 2.9. o
W .3 W
m 2.? o .. w
w . x
m . m: H
m. ONO; . n.b
m G
GKQE hung» 3&6
o mNd . a x N
NooN _OON OOON oaa_ wam_ Nam_ cao_
9o _ . . . L > H 86 _ P P _ . P _
8.9, .
m 8.0 MN; ,2 m
w m w
w. \EJ No.9. W
U
wow: * W ,2 N m.
w. 8.? _ m
m N s
m . .3 G
m. 86 N P r36 \
D. .
4 334253 :oanoaé.
ESNESQ 38% QSEEQ 223‘ k. 3?: =8... . m
85 N 3°

 

 

 

 

 

 

6O

Chapter Two

E valuation of methods and data quality from a volunteer-based amphibian call survey

Abstract

The Michigan Frog and Toad Survey (MFTS) is an annual volunteer-based
anuran census. One major concern with data collected by volunteers is its quality and
consistency. The goal of this study was to evaluate the effect of observer experience on
data quality. Questionnaires and an audio CD with a simulated anuran survey route were
mailed to all active volunteers. We were able to determine volunteer demographics and
commitment to program; species characteristically missed, misidentified, over or
underestimated; and influence of volunteer background on data quality. Volunteers were
reasonably reliable in their abilities to determine species presence, but there was
extensive variability in abundance estimation. Some species were characteristically
confused by volunteers, and additional species were frequently recorded even when
absent from a site. Prior experience and background had little influence in the ability to
identify or estimate abundance of calling frogs. Our results indicate that such survey
approaches are easy for volunteers to learn and provide reasonable estimates for species’
presence, but do not necessarily estimate abundance well. These results will be used to
improve data collection protocols for the MFTS and better analyze and interpret the data

collected, and could also be beneficial for other regional amphibian monitoring programs.

61

Introduction

In the last decade, there has been increased interest, debate, and research
concerning the apparent global declines of amphibian populations, although the causes
have yet to be unambiguously determined (Blaustein et al. 1994a, Sarkar 1996, Green
1997b, Alford and Richards 1999). The primary obstacle that must be overcome in
evaluating potential amphibian declines is separating effects due to anthropogenic
influences from natural population fluctuations (Pechmann and Wilbur 1994, Travis
1994). To do this, long-term data from extensive areas are needed (Blaustein et al.
1994a), and regional amphibian monitoring programs can contribute such information.

Surveys of calling anurans (hereafter call surveys) are being used for amphibian
monitoring in many states and Canadian provinces (e. g., Huff 1991, Bishop et al. 1997,
Lepage et al. 1997, Mossman et al. 1998), and have the potential to provide valuable
information about population size and status (Zimmerman 1994, Shirose et al. 1997,
Driscoll 1998). The North American Amphibian Monitoring Program (NAAMP) has
developed a unified protocol for volunteer-based call surveys that has been implemented
in 24 states (Weir and Mossman, in press). This technique provides a fairly efficient and
effective method of surveying anurans over large geographical regions (Scott and
Woodward 1994), and has successfully allowed collection of data on presence and
abundance of frogs and toads in many areas. Call survey protocols are easy for
volunteers to learn, and also provide an excellent means for promoting education about
wetlands and amphibian conservation.

Annual monitoring data for Michigan’s 13 species of frogs and toads can be used

to evaluate short-term trends in population dynamics and help to guide research and

62

conservation efforts for these species. The Michigan Frog and Toad Survey (MFTS)
protocol was originally modified from that developed for the Wisconsin Department of
Natural Resources (WDNR) frog survey (Mossman et al. 1998). The first surveys in
Michigan were done in 1988 on a limited basis and were discontinued a few years later
due to lack of personnel to coordinate the survey. In 1996, the Michigan Department of
Natural Resources (MDNR) initiated a statewide annual frog and toad survey that uses a
network of volunteer observers to monitor breeding anuran populations (Sargent 2000).
NAAMP developed survey protocols in 1995, but these were not approved for national
use until 2000 (Weir and Mossman, in press). Since routes had already been established
in Michigan, and data had already been submitted for five years by the time the NAAMP
protocols were officially available, the MFTS continued to use the original protocols.
However, some MFTS routes adhere to NAAMP protocols, and these data are submitted
for use at state and national levels. In the MFTS, volunteers establish routes by first
submitting a map and descriptions of a series of 12 potential wetland survey sites
identified without knowledge of wetlands or presence of amphibians (i.e., sites identified
in winter or otherwise outside the breeding season when anurans are inactive); upon
approval of the MDNR state coordinator the route is driven and sites are established
(Sargent 2000). NAAMP routes use randomly generated driving maps (Weir and
Mossman, in press). The volunteers survey routes 3 times each spring, corresponding to
peak breeding times of anurans, and record the identity of each species and an index of
their calling intensity (0 = absent, 1 = few individuals with non-overlapping calls, 2 =
many individuals with overlapping but distinguishable calls, 3 = full chorus with

individual calls indistinguishable). MFT S protocols instruct volunteers to begin

63

surveying their routes 30 minutes after sunset and to listen for 5 minutes at each site
(Sargent 2000).

Volunteer training is an important component that contributes to the success of a
regional monitoring program. Prior to onset of data collection in 1996 and again in 2001,
training workshops were held throughout the state. These training workshops provided
information on general frog biology and habitat requirements, how to distinguish species
by sight and sound, and instructions for establishing a route and conducting the survey
(Sargent 2000). In addition to these initial training workshops, instructional packets that
included information on protocols and a training cassette with examples of all species’
calls were distributed to interested participants.

Several factors (e. g., prior experience or hearing ability) may influence observers’
abilities to correctly identify anurans and estimate their abundance. Previous studies
suggest that novice observers provide reliable data on species’ presence, but their
assignment of call index values is more variable (Bishop et al. 1997, Shirose et al. 1997,
Hemesath 1998). Differences among observers could influence survey data quality, and
should be incorporated into analyses of population changes (Sauer et al. 1994).

Because data for the MF TS are collected each year by hundreds of volunteers
with varying expertise (range of 198-293 routes submitting data during 7 years of
program; most routes had at least 2 volunteers), we initiated a study to identify factors
that influenced data collection so that we could incorporate modifications to deal with
these factors into data analyses. Although the problems of observer bias and subsequent
data analyses have been investigated for monitoring programs such as the North

American Breeding Bird Survey (BBS, Sauer et al. 1994, Link and Sauer 1996, Kendall

64

et al. 1996), differences among volunteer observers and their implications for data
analyses need to be more thoroughly addressed for amphibian monitoring programs (but
see Bishop et al. 1997, Shirose et al. 1997, Hemesath 1998). Methods for assessing
observer experience have been proposed (Bishop et a1. 1997, Shirose et al. 1997,
Hemesath 1998), but have not been widely implemented.

The overall goal of this study was to evaluate observer experience and data
accuracy and precision for the MFTS. Based upon this goal, the primary objectives were
to evaluate how volunteer background and experience influenced anuran identifications
and assignment of call index values, determine implications of observer differences on
analyses of population trends, and establish an observer evaluation process that could be
implemented by other states with similar monitoring programs. This chapter represents
collaboration with Lori G. Sargent, the State Coordinator for the MFTS (Genet and

Sargent 2003).

Materials and Methods

We administered a mail survey between July and December of 2001. The survey
was evaluated for ethical appropriateness by the Michigan State University Committee on
Research Involving Human Subjects (UCRIHS IRB# 01-324). In mid-July we sent a
questionnaire and audio CD to all MFTS volunteers who had submitted data in 1999
and/or 2000 (n = 355). We sent a second mailing of reminder/thank you postcards in
early September, and we also provided additional questionnaires and/or CDs to
volunteers who requested them at that time. Finally, we mailed the annual MFTS update

in early December, which included a preliminary report of the survey data and

65

encouragement to send in responses if volunteers had not already done so. We did not
accept any responses afier 31 December 2001. We removed from the sample surveys
that were returned as undeliverable by the United States Postal Service or in cases where
the recipient was deceased. We included surveys that were returned blank or with a note
indicating no interest in participating in the survey in the sample size, but considered
these non-responses (counted in final corrected sample size, but data omitted from
analyses). Respondents provided information about basic demographics, participation,
and experience with the MFTS and wildlife in general, and commitment to this and other
wildlife monitoring programs on their questionnaires (specific questions from the
questionnaire are found in Appendix A).

We also enclosed an audio CD with 12 tracks with the survey. Each track
represented a wetland breeding site, typical of those visited by volunteers as they collect
data for the MFTS. We instructed the volunteers to listen, and using data forms identical
to those used for annual surveys, record each species and assign an abundance index for
each CD track in the same manner as for the sites they monitor annually. Each track was
5 minutes in length, the recommended listening time for MFTS protocol (Sargent 2000).

If respondents failed to follow survey instructions or only submitted a partial
questionnaire or datasheet, we edited the data as follows. If the respondent gave a range
when asked for a numeric response, we assigned the arithmetic mean. In cases where
respondents indicated they would submit data as long as possible in response to question
3 (Appendix A), we assigned a value of 10 years. We omitted non-responses on
individual questions from analysis. We categorized observers as novice (n = 18),

intermediate (n = 46), or expert (n = 90) based on the number of years they had submitted

66

data to the MFTS (1 — 6 yrs.) and their perceived level of expertise (S-point rating scale,
range 2 — 5). We summed these two values (summed score range: 3 - 11) and divided
the resulting scores equally into the three experience categories: novice = 3 — 5,
intermediate = 6 — 8, expert = 9 — 11. The call index values assigned by respondents
were compared to the values independently predetermined by experts (both authors and
two additional experts in the field) for each species on each of the 12 tracks. Statistical
analyses were performed using appropriate statistical methods described in the following
paragraphs (Zar 1998) using SAS statistical software (SAS Institute, Cary, NC).

We used chi-square analyses to test whether species identifications deviated from
expected (i.e., all volunteers correctly identified all species on each track) and to
determine whether respondents assigned the expected call index values (predetermined
by authors and two other experts) for each species on each track. Volunteers in other
regional call surveys have difficulty discriminating between the two highest call index
values (L. B. Johnson, University of Minnesota Duluth, personal communication). Thus,
2 separate hypotheses were tested for each track of the CD: 1) respondents correctly
assigned 4-category call index values (0, 1, 2, 3), and 2) respondents correctly assigned 3—
category call index values (0, 1, 2+3 combined). For the three species with restricted
distributions within Michigan (Fowler’s toads, mink frogs, and Blanchard’s cricket
frogs), we used contingency table analyses to determine whether there was a significant
association between living within a species’ range and correct identification. We also
used contingency table analyses to determine if observer experience level (novice,
intermediate, or expert) influenced abundance estimation of each species (correct, over-

or underestimation of abundance, based on authors’ predetermined values). We used

67

both parametric and non-parametric (Kruskal-Wallis) analysis of variance (AN OVA) to
determine whether observer experience influenced correct identification of species.
Nonparametric tests were implemented in the cases where data did not meet assumptions
of parametric tests (i.e., small sample size). Data are reported as means : SE unless

otherwise noted, and a P—value < 0.05 is considered statistically significant.

Results
Volunteer Demographics

Of the 355 questionnaires sent to volunteers, 4 were undeliverable, 1 recipient
was deceased, and 179 were returned, yielding a 51.1% response rate (corrected sample
size, n = 350). Volunteers indicated extensive experience with wildlife and commitment
to this program (Table 2.1). The typical respondent was just over 50 years old, and 26%
of respondents were retired. Slightly more females (53.7%) than males (46.3%)
participated in this project, although the difference was not significant (le = 0.5746, P =
0.464). Over 70% of respondents were also avid birders, and 44% were involved in at
least one additional wildlife monitoring program.

The typical respondent felt he or she had improved approximately 2 points on a
rating scale from 1 (low) to 5 (high) fi'om beginning in the MFTS to their current level of
expertise (beginning: 1.66 i 0.07, current: 3.56 i 0.05). With respect to training,
volunteers predominantly listened to the training tape (97.7%) and attended training
workshops (66.5%). The majority of volunteers participated in more than one type of
training activity (82.4%), but fewer practiced in the field (11.4%) or participated in some

other type of training activity (33.0%). On average, two pe0ple participated in each

68

survey route, and in > 85% of the cases, the number and identity of those people was
consistent from year to year. Most respondents (80%) indicated that a primary observer
had been designated for their route (if > 1 observer on that route). The primary observer
was responsible for the data forms and establishing consensus when there were
discrepancies among observers. Thirty-five percent of the respondents indicated that
there had been discrepancies among observers in terms of species heard or call index
values to be assigned. The most common methods for resolving these discrepancies were
to discuss among observers until consensus was reached (58.3%), listen longer at the site
(51.7%), check calls against the training tape (28.3%), and/or allow primary observer to
make decision (8.3%).
Species Identification

Volunteers were able to identify species by their calls (Table 2.2, 2.3), but correct
identifications varied from 60% for Fowler’s toads (Bufofowleri) to > 98% for northern
green frogs (Rana clamitans melanota) and bullfrogs (Rana catesbeiana). The majority
of species on a track were correctly identified by _>_80% of volunteers (Table 2.2, 2.3).
Many respondents confused northern leopard frogs (Rana pipiens) and pickerel frogs
(Rana palustris) (> 10%, Table 2.2). Fowler’s toads were correctly identified by only
60% of the respondents (Table 2.2), and a large proportion of respondents indicated the
presence of one (or both) of the gray treefrog species (Hyla versicolor, H. chrysoscelis)
either instead of or in addition to Fowler’s toads. Several respondents also confused the
calls of mink frogs (Rana septentionalis) and wood frogs (Rana sylvatica) (Table 2.2).
No other species appeared to be characteristically misidentified or confused with a

similar-sounding species.

69

On 6 of the 12 tracks, volunteers’ responses significantly differed from correct
identifications (P < 0.005), indicating that they missed or misidentified a significant
number of species. For the remaining tracks, volunteers correctly identified all species
present, however, additional species not calling on the CD were also recorded (Table 2.2,
2.3). For the three species with restricted ranges, living within the species’ range did not
affect the observers’ abilities to correctly identify these species (Table 2.4, P > 0.05).
Novice, intermediate, and expert observers did not differ significantly in their abilities to
correctly identify species’ presence based on their calls (H2 = 1.468, P = 0.480).
Respondents in all three experience categories recorded correct identifications of all
species combined (i.e., all species calling in concert at a breeding site) in > 85% of cases.
Observer experience level also had no significant influence on respondents’ abilities to
correctly identify individual species (P > 0.05).

Abundance Estimation

There were discrepancies in assignment of call indices for some species even
among respondents who correctly identified the species on each track of the CD
recording. In many cases, there was consensus among respondents, but in others,
respondents were relatively equally split among different call index values (Figure 2.1).
For example, spring peepers (Pseudacris crucifer) were almost unanimously assigned
call index 3 on track 1 (Figure 2.1a), while Blanchard’s cricket frogs (Acris crepitans
blanchardi) and green frogs were almost equally split between call index values 2 and 3
on tracks 2 and 8 (Figures 2.1b and 2.1h). For 11 of the 12 tracks, call index values
assigned by respondents differed from those predetermined by the authors (P < 0.05).

Combining call index values 2 and 3 only improved the respondents’ abundance

70

estimation on one track. In most cases, the deviation from the expected call index values
was due to either not recording a species as present (assigning a call index of 0) or
discrepancies among volunteers in abundance estimation when a species was recorded as
present (Figure la-l). Track 8 was an exact repeat of track 2, and 75.2% of respondents
identified the same species in both tracks. However, of those respondents that identified
the same species in both tracks 2 and 8, only 43.8% of them assigned the same call
indices to those species in both tracks. In 3 x 3 contingency table analyses of volunteer
experience level (novice, intermediate, or expert) and abundance estimation (correct,
over- or underestimation) for each species, only two tests produced significant
associations (6.7% of comparisons). Novice observers tended to overestimate northern
leopard frogs on track 9 while intermediate and expert observers underestimated calling

intensity (x24 = 51.0047, P < 0.001), and correct index assignment was positively

associated with observer experience for mink frogs on track 12 (x24 = 10.8867, P =

0.0279).

Discussion

The MFTS was modified from protocols originally developed by the WDNR and
NAAMP. NAAMP supplies randomly generated route maps to interested volunteers, and
the volunteer then chooses sites based on an equidistant or stratified-by-habitat method
(Weir and Mossman, in press). The MFTS instructs volunteers to select the area where
they first establish a route, then choose sites without regard to anuran presence (Sargent
2000). While there are likely some volunteers who are knowledgeable about wetland

locations and who have paid attention to the frogs calling at these wetlands before

71

involvement in the MFTS, most volunteers have no prior experience with the wetlands or
the anurans in an area before establishing a survey route (L.G. Sargent, personal
communication). As such, sites are chosen haphazardly, only omitting areas that are “too
dangerous or too noisy to hear” calling fi'ogs. In addition, the ability to survey a route
near an observer’s home significantly increases the likelihood of data submission and
volunteer retention. Furthermore, the longevity and the large number of routes
established for the MFTS dilutes any bias site selection may inherently introduce. The
MFTS, NAAMP, and other large-scale amphibian monitoring programs are intended to
provide a meaningful and relatively inexpensive method to track changes in distribution
and abundance of species with applicability at a variety of scales. Volunteer observers
are an integral part of these goals.

The volunteers participating in the MFTS were able to correctly identify
Michigan’s frogs and toads by their distinctive breeding calls relatively consistently. All
species were correctly identified by _>_ 60% of respondents, and all but four species were
correctly identified by more than 80% of the respondents. Species that were missed or
misidentified were most likely due to confusion with other species with similar calls (e.g.,
northern leopard frogs and pickerel frogs); inability to determine spatial location of call
from CD recording and whether that species was present within the wetland site or farther
away (e.g., gray treefiogs in tracks 2 and 8, spring peepers in track 7); inability to hear
less conspicuous calls masked by more prominent species (e.g., western chorus frogs,
Pseudacris triseriata, calling in the presence of a full chorus of spring peepers); or
unfamiliarity with species not present in volunteers’ survey areas (e.g., mink frogs,

Fowler’s toads, Blanchard’s cricket frogs). In general, we view our results as positive

72

affirmation of the data quality of the MFT S; however, there are several concerns that we
will address in future analyses and protocol revisions. Additionally, just over half of the
active volunteers in this study returned questionnaires, potentially limiting our
conclusions and recommendations to the most conscientious observers.

The cases of mistaken species identification probably represent a worst-case
scenario. The use of the CD recording has obvious limiting factors that need to be
considered when comparing this study with field investigations. The volunteers in this
study were not given any background information about the CD recordings. Had that
information been available (i.e., date, geographic location, habitat type), identification
problems may have largely been avoided. Even so, results indicate the need to review
training materials to decrease misidentifications. Many commonly misidentified species
pairs do not breed at the same time (e. g., mink frogs and wood frogs) or have differences
in habitat preferences (e.g., leopard frogs and pickerel frogs). Ensuring that volunteers
know the basic biology of each species in addition to being trained to recognize
distinctive breeding calls could greatly improve data accuracy. The relatively low
proportion of respondents who correctly identified Fowler’s toads is also a concern; it
appears that many observers are not trained to document potential range expansions or
isolated populations. Additionally, the CD was produced from field recordings, and truth
is not positively known for the recordings. The authors and additional experts
independently and unanimously determined species presence and their call index values,
but volunteer responses are being compared to authors’ identifications and index value

assignments and not to true population sizes.

73

The breeding habits and call characteristics may also affect the probability of
detection of a species. Some species simply have a boisterous, unmistakable call (e. g.,
spring peepers) that is much easier to distinguish than a call that is much more subtle and
lower in volume (e. g., northern leopard frogs). Bishop et al. (1997) attributed the paucity
of records for breeding leopard fiogs in Ontario to its subtle call and lack of concentrated
choruses. Our results also indicate that some species may be missed as a result of not
being heard over the din of louder species in greater concentrations. Western chorus
fi'ogs frequently call at the same time and location as spring peepers, but it can be hard to
discern the calls of western chorus frogs within the deafening chorus of spring peepers.

Volunteers sparmed a wide range of ages and backgrounds. The mean age was
50.52 yrs., and although some hearing loss is expected with age, the 13 species of
Michigan frogs typically call between 300 - 3200 Hz (T. O. Matson, Cleveland Museum
of Natural History, personal communication), well within the normal range of human
acoustical sensitivity. Detection of frog calls also depends on volume, and 95% of men
aged 50.4 (average age of men in this study) can detect 250 Hz at 11.78 dB, and 3000 Hz
at 25.22 dB (comparable values for women aged 50.6, average in this study, are 10.73
and 18.76 dB for 250 and 3000 Hz, respectively) (G. A. Flamme, University of Iowa,
personal communication). The volume at which a frog calls and the distance between it
and the observer are very important factors, as well as additional external noise (e. g.,
vehicle traffic). We acknowledge that the average age of the MFTS volunteer is middle-
aged, and there may be issues related to hearing loss. We will continue to investigate the

potential influences of age-related hearing loss on our data quality.

74

Evaluation of volunteer collected data for other amphibian monitoring programs
has indicated that inter-observer agreement on presence and abundance estimation is
generally high, but experience also plays a role in data quality (Shirose et al. 1997,
Hemesath 1998). Contrary to results from other regional amphibian monitoring programs
using call surveys, observer experience played only a minor role in the MFTS. Perhaps,
this reflects the fact that our volunteers had an average of 24 years of wildlife experience
and had participated in the MFTS for more than 4 years of participation (Table 1).
Observer experience level was not a significant influence in either the identification or
abundance estimation for any Michigan anuran. Other investigators have found that
inter-observer agreement on species presence was very high and not influenced by
experience (>96% agreement regardless of experience level, Shirose et al. 1997), but
agreement on calling intensity varied with experience level (47 -83% in Ontario, Shirose
et al. 1997; 56-83% in Iowa, Hemesath 1998). Novice observers tended to underestimate
calling intensity relative to experts (Shirose et al. 1997). Similarly, for the North
American Breeding Bird Survey (BBS), observer differences were related to experience
such that observer quality increased over time (Sauer et al. 1994). First-time BBS
observers tended to underestimate species and individuals, and population analyses of
BBS data include observers as covariates in order to avoid confounding observer quality
with population trends (Erskine 1978, Kendall et al. 1996). Since experience level of the
MFTS volunteers did not influence abundance estimation, observer experience level
should not affect analyses of changes in abundance over time.

Although observer experience was not an influential factor, there were dramatic

differences among volunteers in the assignment of call index values. In many cases,

75

almost the same proportion of survey respondents assigned two different call index
values. In general, observers assigned a call index value of 1 with reliability and
consistency, but had difficulty distinguishing between call index values 2 and 3. The
subjective interpretation of overlapping calls being distinguishable as individuals or not is
the difference between call index values 2 and 3. The character of the individual species’
calls also influenced observers’ ability to assess abundance and assign the correct
abundance index. Shirose et al. (1997) also found that inter-observer variation in
estimation of calling intensity depended on the species considered. Species with
prolonged calls (e.g., American toads, Bufo americanus) do not appear to overlap as
much as species with shorter calls (e.g., spring peepers). In other states’ amphibian
monitoring programs, volunteers seemed able to reliably determine the call index of 1,
but often had trouble discerning between call indices of 2 and 3 (LB. Johnson,
University of Minnesota, Duluth, personal communication). One solution would be to
translate the data into a three-level abundance index: absence (0), low abundance (l), and
high abundance (2). In addition to inter-observer differences, we also need to consider
intra-observer differences in species detected and call index values assigned. With the
duplicate tracks present on the CD, we found that there was relatively low agreement in
call index values assigned by the same volunteer. While the combination of call index
values 2 and 3 would improve the agreement in abundance estimation, this is a concern
that we need to address in our training packet and communications with volunteers.

A related concern is the relationship between the calling index and the actual
population size for each species. As a result of the different call characteristics of the 13

species of Michigan anurans (e. g. the long trill of east. American toads compared to the

76

short call of the spring peeper), the same call index value recorded for different species
will translate to very different population sizes. The relationship between call counts and
population size has been investigated for some species (Shirose et al. 1997), although
until empirical calling index — population size relationships can be developed for all
species involved, call index values may be best translated into presence/abundance data
that can be used to track changes in populations over time (Weir and Mossman, in press).
An inevitable component of a large-scale regional monitoring program is
differences among observers. There is a trade-off between the amount and extent of the
data and its reliability and consistency. While some monitoring programs have reported
significant differences among volunteer observers, in other programs it does not appear to
be a major concern (Mossman et al. 1998, Kline 1998). Observer bias is considered
minimal in some programs as a result of a combination of their volunteers’ experience,
scientific evaluation of the data, and number of observations (C. M. Francis and A.
Chabot, Long Point Observatory, unpublished report, Mossman et al. 1998, Kline 1998).
For amphibian call surveys in Wisconsin, increasing the number of observations (i.e., the
number of routes surveyed) reduced variability in the dataset more than increasing
volunteer training (Kline 1998). Similarly, power to detect significant population trends
in the Marsh Monitoring Project increases as more stations are surveyed (C. M. Francis
and A. Chabot, Long Point Observatory, unpublished report). It appears that widespread
species that call frequently can be adequately monitored with roadside call surveys;
species that call infrequently may require more effort (i.e., more routes surveyed) to track
meaningful population trends (Crouch and Paton 2002). Although there were some

differences among observers in abundance estimation, we have data from more than 400

77

survey routes statewide (with 10 wetland sites along each route). With such a great
number of sites surveyed each year, we should have the power to track significant
population trends over time. However, a formal power analysis should also be
conducted.

With respect to the MFTS, we recommend that volunteers refresh their skills prior
to each survey season. Being familiar with the basic biology, phenology, range, and
habitat requirements of Michigan’s frogs and toads would help avoid some simple
identification mistakes found in this study. Additionally, NAAMP is currently
developing an online frog survey quiz for volunteers. When operational, the MFTS will
consider volunteers’ quiz scores when determining the inclusion of their data. We also
encourage the volunteers to review and adhere to MFTS protocols. There was substantial
variability in amount of time spent listening at survey sites. Although most species are
heard within the first minute of a survey, and 3 minute stops have been recommended for
call surveys (Shirose et al. 1997), volunteers should be strongly encouraged to follow
MFTS written protocols to ensure standardization of time spent at stops.

Concerning analyses of sites occupied by each species and changes in occurrence
over time, presence is a reliable observation by volunteers, but absence at a site is not
guaranteed by undetected calling males. For analyses of abundance trends, we propose
combining call index values 2 and 3. While some resolution will be lost by eliminating
one abundance category, there is so much variability among observers in assigning call
index values 2 and 3 that any apparent difference between the two abundance categories
may not be biologically meaningful. We also recommend that verification in the form of

a photo, recording, or expert observation be required for rare and hard-to-identify species

78

(Blanchard’s cricket frogs and Cope’s gray treefrogs). In addition to these two species,
we also encourage verification for pickerel frogs and may make this a requirement in the
future. Finally, the training and refresher workshops offered in the past have been very
successful and well attended. We recommend that funding and personnel be made
available to continue to offer these workshops periodically.

The observer evaluation process presented here is a valuable tool that could be
implemented by other regional anuran monitoring programs. One caveat is that similar
studies should provide information on time of year and site characteristics with the CD
recording. If we had provided that information, the misidentifications may have been
reduced. Even so, our results indicate that data from the MFT S can reliably be used to
track trends for most species of frogs and toads. Data quality is a major concern in large-
scale long-term monitoring programs, and documenting differences among observers is
essential for the analysis and interpretation of the data. The detection and abundance
estimation of frogs and toads are likely influenced by a variety of factors, and confidence
in the data and any resulting trends is greatly improved when the influences of observer

bias on the data collection process are understood and documented.

Acknowledgements

C. A. Lepczyk, T. M. Burton, S. K. Hamilton, P. Paton, G. C. White, I. Kiyak, A. Souza,
M. J. Sanregret, K. B. Joldersma, D. J. Hall, J. Liu, and an anonymous reviewer provided
valuable comments on earlier drafts. J. McGrath provided frog call recordings. J. H.
Harding and J. McGrath provided expert identifications and abundance estimation for the

CD recording. We thank the hundreds of volunteers in the Michigan Frog and Toad

79

Survey for their continuing efforts. Funding was provided by the MDNR Wildlife
Division, Michigan State University chapter of Sigma Xi, and the Ecology, Evolutionary

Biology and Behavior graduate program at Michigan State University.

80

Table 2.1

Demographic characteristics of survey respondents. Means and SE were calculated from
all volunteers returning questionnaires, and the sample size for each demographic

character (i.e., number of respondents providing answer on questionnaire) is listed in the
last column.

 

Demographic character Mean SE Range n

 

Involvement with program to date (yrs.) 4.67 0.13 1-6 175

Total anticipated involvement (yrs.) 10.04 0.32 3-30 142
Wildlife experience (yrs) 24.01 1.28 1-65 168
Age (yrs) 50.52 0.86 17-81 177

 

81

 

 

 

Nd 035.

N66 8.6V 5.6V 6 v~6 _66 2
_66V 6 66.6 _66 6 666 2
.666 3.6 .366 6_ .6 _66 86 6_

6 _ .6 .56 .666 86 66.6 8.6V 6
3.6V 6 86 36 N66 6 6
_66V 6 .666 m6 .6 8.6 .666 b
S .6 .3 .6 :66 V66 36 N66 6

m _ .6 sow .6 .666 .36 .56 6 m
_66v ~66v .666 36 ~66v N66 v

6 6 ~66v 6 6 8.6V m

6 6 V66 86 86 36 m

6_ .6 *666 .64 .656 3 .6 _6.6V _

web 35on wet 6.3662 .55: common macaw wot 38:0 .683 wot 683 38 mac—36m 8.8:.
.xmtouma

an 5.; 330%.: 2a Jug :08 no 59.26 86on 65688.. Q0 05 no x8e“ :08 no 3609. 6:592: $53668“ no 6668695

82

83 868% 36.6. .wEEooE DU 886 62668686 on 60:58 6266 86m 6256 028:0 6566306 886 M66668 on ~38 mwoaoob 680.“

.368: 3266 60.6 896698 6363686 88.6 686680

 

 

 

V66 {.66 _._ 66.6 m66 66.6v 66.6v 6 N6
6 6 666 N66 :66 366 666V 66
66.6 666 66.6 66.6 6 N66 6 66
6 6 6 6 .6 666v 6 N66 *666 6
$66 *666 666 *666 m66 :m 6 .6 666 w
666v 6 6 6 mm .6 66.6 666 6
N66 66.6 V66 N66 36 *666 .666 o
6 66.6 66.6 m66 6 ~66 66.6v m
6 *666 6 6 66.6 36.6 66.6v v
6 66.6v 6 6 666v m66 *666 m
*666 *666 66.6 $666 66.6v 3m 6 .6 6 N
6 66.6 6 66.6 66.6v m66 N66 6
wow wot 68606.8 wotoob 6:36

woman: :0on 666.8: web .658 96566055 zflw 92600 668.686 bfim 536.6262

693 6686.6.

 

P6680 .N.N 06.65.

83

Table 2.3

Summary of correct identifications and identification errors averaged over all 12 CD
tracks (means and SE reported). All means represent proportion of total respondents.
Correct identifications indicate volunteers who correctly recorded a species as present
when it was calling. Missed identifications indicate volunteers who recorded a species as
absent when it was actually present. Incorrect identifications indicate volunteers who
recorded a species present when it was absent from the recording. Species with no SE
value were present on only one track, precluding estimation of variability.

84

 

 

Correct Missed Incorrect

Species Mean SE Mean SE Mean SE
Fowler's toad 0.602 0.398 0.015 0.005
wood frog 0.864 0.136 0.055 0.023
west. chorus frog 0.701 0.059 0.299 0.059 0.041 0.013
spring peeper 0.874 0.052 0.126 0.052 0.033 0.013
north. leopard frog 0.795 0.080 0.205 0.080 0.029 0.027
pickerel frog 0.688 0.313 0.051 0.019
east. American toad 0.957 0.031 0.043 0.031 0.010 0.004
gray treefi'og 0.917 0.014 0.083 0.014 0.099 0.077
Cope's gray treefrog 0.873 0.127 0.043 0.028
Blanchard's cricket

frog 0.858 0.006 0.142 0.006 0.014 0.005
mink frog 0.705 0.295 0.034 0.011
north. green frog 0.984 0.004 0.016 0.004 0.009 0.003
bullfrog 0.981 0.006 0.019 0.006 0.008 0.004
all species combined 0.831 0.034 0.169 0.034 0.034 0.007

 

85

Table 2.4

Percentage of respondents living within and outside ranges of the three species with
limited distributions within Michigan with correct species identifications. The
relationship between living within a limited range and correct identification was tested

using contingency table analyses.

 

 

Species Within range Outside range X21 P value
mink frog 75.0% 69.9% 0.2216 0.638
Fowler’s toad 63.8% 59.2% 0.3265 0.568
Blanchard’s cricket frog 89.5% 78.6% 3.5559 0.059
(track 2)

Blanchard’s cricket frog 87.2% 84.3% 0.2711 0.603
(track 8)

 

86

Figure 2.1

Call indices assigned by respondents for each CD track. Lightly shaded bars indicate call
index 1, open bars indicate call index 2, and black bars indicate call index 3. Y-axis
(percentage of respondents) is the same for all figures. Correct call index values are given
at the base to the right of each bar. (a) Track 1, (b) Track 2, (c) Track 3, ((1) Track 4, (e)
Track 5, (i) Track 6, (g) Track 7, (h) Track 8, (i) Track 9, (j) Track 10, (k) Track 11, (1)
Track 12.

87

l 00%
80%
60%
40%
20%

0%

Spring Leopard
frog peeper frog

(e)

100% ~

 

 

 

80%

 

60%

40%

20%

 

 

 

 

0%.

 

elj U

 

 

Wood
frog frog

Chorus Spring

peeper frog

[:1 1

Leopard

[:1 2

88

(b)

 

 

l 3 2 2
E. gray Blanchard's Green
treefrog cricket frog frog

 

 

 

 

 

 

 

Bullfrog

(d)

 

 

 

 

 

 

 

E. Gray

 

 

 

 

 

 

 

 

 

 

 

 

Spring Green
Peeper treefrog flog
__ ___4 _
_. _— —l —.
, 1 1 ' 2 2
Spring Leopard America E. gray
peeper frog n toad treefrog

I 3

 

 

    

 

 

 

 

 

 

 

 

100%
-—-1
80% _
60% W —
40% —— _—-r
’ a
20% 9
1 I, 2
0% ' '
Fowler‘s Spring E. gray Blanchard's Green Bullfrog

toad peeper treefrog cricket frog frog

0)

l 000/0

80%

60%

40%

20%

   

0%
Spring Leopard American Spring Pickerel
peeper frog toad peeper frog

1 00%
80%
60%
40%

20%

   

0%
E. gray Cope's gray Mink Green
treefrog treefrog frog frog

[:11 D2 I3

89

Chapter Three

The Influence of Landscape Characteristics on the Abundance and Distribution
of Frogs and Toads in Southern Michigan

Abstract

A variety of environmental factors operating at multiple spatial scales are likely
responsible for shaping anuran populations and communities. Increased interest in and
debate concerning the causes of global amphibian declines is driving research with
respect to factors that govern amphibian population dynamics. In this study, land cover
variables adjacent to and in the landscape surrounding breeding sites within potential
anuran dispersal distances were examined to assess their influence on anuran presence,
abundance, and site species richness in southern Michigan using GIS analyses of call
survey and land cover data. Short-term temporal patterns (1996-2002) of species
richness and individual species’ occurrence and abundance at breeding sites in differing
landscape contexts were also assessed. Species richness at survey sites was positively
related to open land and nonforested wetlands, but lacked the negative associations with
urban land and roads and positive associations with forest cover typically reported in the
literature. Species richness declined slightly at survey sites in natural contexts, but
showed no significant temporal patterns in anthropogenically influenced landscapes.
Anuran presence and abundance were influenced by many landscape variables that were
generally related to their habitat preferences. The presence and/or abundance of most
species was negatively influenced by land cover types representing habitat alteration or
loss and positively influenced by land cover types representing potential habitat in the

landscape surrounding breeding sites. Species showed few consistent patterns of

90

presence and/or abundance in breeding sites of differing landscape contexts, although
Acris crepitans blanchardi declined in abundance in anthropogenic context breeding sites
and increased in calling intensity in natural context breeding sites. Anuran diversity in
southern Michigan appears to be enhanced by a mosaic of land use types surrounding
their wetland breeding habitats. The associations I found between anuran occurrence or
abundance and land cover types at the landscape level provide valuable information for

managing and conserving habitats for these organisms in areas where populations may be

low or in decline.

Introduction

Amphibians are an important component of forested ecosystems, comprising a
major portion of vertebrate biomass in these systems (Burton and Likens 1975). They are
also potentially sensitive environmental indicators due to use of multiple habitats
throughout their life cycles, permeable skin, sensitivity to water chemistry during the
larval stage, trophic linkages, and metapopulation structure around discrete wetland
breeding habitats (Stebbins and Cohen 1995, Bowers et al. 1998, US. EPA 2002). As a
result of amphibians’ environmental sensitivity the herpetological community has been
focused on understanding recent population declines around the world (Houlahan et al.
2000). A host of factors has been identified as probable causes contributing to the
declines, including ultraviolet radiation (e. g., Blaustein et al. 1997, Licht and Grant
1997), predation (e. g., Fisher and Shaffer 1996, Kiesecker and Blaustein 1997), habitat
modification (i.e., loss, alteration, and fragmentation) (Johnson 1992, Hecnar and

M’Closkey 1996), chemical contaminants (e.g., Freda et al. 1991, Carey and Bryant

91

1995, Home and Dunson 1995), disease (e.g. Berger et al. 1998, Muths et al. 2003), and
climate change (e. g., Carey and Alexander 2003). Synergistic interactions among
combinations of these factors further complicate the issue of declining amphibian
populations (Kiesecker et a1. 2001).

Although the causes of these amphibian declines have not been unambiguously
determined, habitat modification, including habitat loss and fiagmentation, is widely
considered to be the most important factor affecting amphibian distribution and
abundance (Blaustein et al. 1994a, Pechmann and Wilbur 1994). Indeed, F ahrig (1997)
has suggested that the effects of habitat loss far outweigh fragmentation in determining
species’ risk of extinction. Given the common requirement among amphibians for more
than one habitat during different life history stages or seasonally for adults, habitat loss in
multiple ecosystem types is an especially dire concern for these organisms (Pope et al.
2000).

Most amphibians are dependent on wetlands ecosystems for at least part of their
life cycle, and many species are wetland-associated throughout their entire lives.
Wetlands were historically distributed discontinuously across the landscape (Hollands
1987, Winter 1988); amphibian populations are likewise patchy and fit well within the
framework of metapopulation dynamics (e. g., Sjogren 1991, Bradford et al. 1993, Alford
and Richards 1999). Wetland habitats in the Midwestern United States have suffered
losses in excess of 75% for areas dominated by agricultural, industrial, and urbanized
activity (Detenbeck et al. 1999). The loss of wetlands increases isolation and reduces the

probability of movement among wetland habitats, which can subsequently interfere with

92

metapopulation dynamics characteristic of many species (Gibbs 1993, Sjogren-Gulve
1994, Semlitsch and Bodie 1998).

Aside from habitat loss, landscape-level factors are very important in structuring
amphibian communities. The best documented effects of landscape variables on
amphibian distribution and abundance are associated with wetland isolation, land cover
types and landscape context. Fewer anuran species are likely to be present in isolated
wetlands, due to reduced colonization when potential breeding sites lie outside the
maximum migration distance (Gibbs 1993, Vos and Stumpel 1995, Findlay and Houlahan
1997, Lehtinen et al. 1999). Altered land uses (i.e., agriculture, urban) also negatively
affect anuran abundance by reducing habitat availability and suitability (Brodman and
Kilmurry 1998, Knutson et al. 1999). Amphibian distribution and abundance are
dependent on landscape connectivity and available forested habitat surrounding wetland
breeding sites (Findlay and Houlahan 1997, Hecnar and M’Closkey 1998, Lehtinen et al.
1999, Knutson et al. 1999, 2000).

Road networks are another feature of human-dominated landscapes with profound
effects on amphibian populations. Roads are both a structuring force on the landscape and
a significant barrier to anurans (Forrnan and Alexander 1998, Trombulak and Frissell
2000). For example, road mortality poses a significant threat to anurans migrating to and
from breeding sites, and amphibian abundance is reduced in areas with high road
densities and/or traffic volume (Reh and Seitz 1990, F ahrig et al. 1995, Ashley and
Robinson 1996, Vos and Chardon 1998, Carr and Fahrig 2001).

Evaluation of the influence of landscape characteristics on anuran communities is

vital for management and conservation of these sensitive species. The documented

93

 

 

-4]
111

 

importance of landscape characteristics on amphibian distribution and abundance in other
midwestem regions (e.g., Minnesota, Lehtinen et al. 1999; Wisconsin and Iowa, Knutson
et al. 1999, 2000), suggested the need to investigate landscape influences on frogs and
toads in southern Michigan. The objective of this study was to determine the influence of
surrounding landscape on amphibian presence, abundance, and species richness by
investigating the importance of different land use types and roads on amphibian breeding
sites surveyed annually in the Michigan Frog and Toad Survey (MET S). I evaluated the
following hypotheses: (1) anuran species associated with forested areas for at least a
portion of their life cycles should be sensitive to the total area of forest surrounding
breeding sites, compared with species that can utilize a wider variety of habitat types; (2)
since all Michigan anurans are dependent on wetland breeding sites for annual
reproductive events, they should be sensitive to the total area or proportional coverage of
aquatic habitats (forested and nonforested wetlands, water) surrounding breeding sites;
and (3) vagile species that are potentially capable of moving long distances through
different habitat types should be less sensitive to altered land cover types than species
that have limited movement capabilities and are dependent upon specific habitat types in

a much smaller area.

Methods and Materials
Michigan Frog and Toad Survey
In 1996, the Michigan Department of Natural Resources (MDNR) established an
annual frog and toad survey that uses a network of volunteer observers throughout the

state to monitor breeding anuran populations (Sargent 2000). Twelve species of anurans

94

are found in southern Michigan (Table 3.1). The MFT S currently has annual data for
approximately 350 routes surveyed statewide since 1996 in Michigan (Figure 3.1).
Detailed protocols for the MFTS can be found in Sargent (2000), but are summarized
below (see also Chapter One). Each route consists of 10 wetland sites selected by
volunteers; these sites are separated by at least 400 m such that the origin of calls can be
unambiguously determined for each site. The volunteers then survey their routes three
times each spring and early summer (according to minimum temperature guidelines and
with at least two weeks between successive survey dates) and record the identity of each
species and an index of calling intensity. Volunteers are instructed to conduct surveys
after dark, beginning one-half hour after sunset and finishing before midnight, and under
favorable conditions of appropriate temperatures and little or no wind. Intensity of
calling males is rated 0 through 3, with 0 = no individuals calling, 1 = few individuals
with non-overlapping calls (1-5 individuals), 2 = distinguishable individual calls that
overlap (6-12 individuals), and 3 = fiill chorus with indistinguishable individual calls
(13+ individuals).

The call index data were combined for the two gray treefrog species (Hyla
versicolor and H. chrysocelis) prior to analyses. The calls of these two morphologically
identical species are difficult to distinguish, and temperature can affect the pulse rate and
duration of their calls, making them very difficult to identify in single species choruses
(Harding 1997). Verification for Hyla chrysoscelis in the form of recording or expert
opinion became a requirement in 1998, but prior to that, the identities of the two treefrog
species were not formally determined. Therefore, some records for H. chrysoscelis prior

to 1998 may represent H. versicolor and vice versa.

95

In 2003, there were 255 active MFTS routes in southern Michigan (L.G. Sargent,
MDNR, pers. comm.; Figure 3.1). Data were included for a species if surveys were
complete on an annual basis (i.e., completed all three survey runs during appropriate
times and weather conditions each year) and data had been submitted for that route at
least three years during the entire MFTS period (1996-2002). There were 1055 survey
sites representing 127 routes in southern Michigan that satisfied the selection criteria.
All sites were included for those species with distributions throughout southern
Michigan, and a subset was included if species had a restricted distribution (Bufofowleri;
651 sites representing 74 routes). The maximum call index value assigned to a species in
any of the surveyed years was defined as the abundance index of that species at that site.
This value represents the highest abundance of a species during the survey period and
provides a good estimate of the wetland breeding site’s potential given the highly
variable nature of amphibian p0pulation dynamics over time (Pechmann and Wilbur
1994, Semlitsch et a1. 1996).

Landscape Variables

MF T S listening points were located on digital USGS 7.5 minute quadrangle
maps, and coordinates were displayed in ArcGIS (ArcMap 8.2, ESRI, Redlands, CA).
Land cover data were obtained from the Michigan Department of Natural Resources
(MDNR) and the Michigan Geographic Data Library. Land cover data for the southern
lower peninsula were derived from classification of Landsat Thematic Mapper (TM; 30
m resolution) images from 1997-2000. Land cover classifications of the data were
simplified to the following seven categories that generally correspond to the major

classes on the statewide land cover map used by the MDNR: (1) urban (airports, parking

96

lots, low to high intensity urban), (2) agriculture (non-vegetated farmland, row crops,
forage crops, orchards, vineyards, and nurseries), (3) open land (parks, golf courses,
herbaceous open land, upland shrubs, low density trees), (4) forest (upland deciduous and
coniferous forests), (5) forested wetlands (lowland forests with >25% tree cover), (6)
nonforested wetlands (<25% tree cover including floating aquatic plants, emergent
wetlands, and lowland shrubs, and (7) open water (permanent ponds and lakes). The road
network coverage was obtained from the Michigan Geographic Data Library. This data
set is part of the Michigan Geographic Framework, which serves as the digital base map
for State of Michigan government agencies. The geographic framework was created from
multiple data sources from the time period 1997-2002.

Two buffers were created around each survey site to characterize the area
surrounding breeding sites and quantify the landscape variables. The center of each
buffer was the MFT S observer’s listening point. A buffer of 100 m was determined to be
the smallest size to characterize the land use directly adjacent to the survey point given
the resolution of the land cover data, and a buffer of 1000 m was used to determine the
potential influence of landscape characteristics on anuran presence, abundance, and
species richness. This distance also represents the dispersal range of many anuran
species (Merrell 1977, Berven and Grudzien 1990, Sinsch 1990, Stebbins and Cohen
1995) and has been found to be an optimal distance for landscape characterization in
other amphibian landscape studies (e.g., Vos and Stumpel 1995, Knutson et a1. 1999,
Pope et al. 2000, Guerry and Hunter 2002). For each land cover category, total area (m2)
of each land cover type surrounding the anuran survey site was calculated at each scale

(100 m and 1000 m). Total road length was also calculated at each scale. There was a

97

total of seven land cover variables (total area of each land cover type within buffer) and
one road variable (total road length within buffer) at each scale.

Variables describing the area of land cover types and road length were chosen
over other landscape indices for several reasons. Other studies have found area of land
cover types to be most strongly related to amphibian variables (e.g., Vos and Chardon
1998, Lehtinen et al. 1999, Knutson et al. 1999). The nature of both the land cover data
(original coverages were raster and converted to polygon) and MFT S data (listening
points were along roadsides and did not necessarily plot within a wetland on the land
cover map) also led to the selection of the suite of variables in this study. The metrics
chosen in this study were intended to characterize the potential influences of different
land cover types and roads on anuran dependent variables.

Landscape Context and Temporal Patterns

MFTS survey sites were classified into four landscape contexts based on the
proportional coverage of each land cover type in a 1000 m radius surrounding the
breeding site. Each landscape context consisted of sites that were primarily (>50% of
surrounding land cover) urban, agricultural, forested, or wetlands (both forested and
nonforested wetlands combined). The other land cover types (open land and water) did
not comprise the majority of total land area surrounding any of the breeding sites. Short-
terrn temporal patterns (1996-2002) of species richness were evaluated in these four
landscape contexts, as well as in a broader anthropogenically influenced landscape
context (urban and agriculture combined) and natural landscape context (Open land,
forest, wetlands, and open water combined). Individual species occurred at too few sites

in the four landscape context categories (urban, agriculture, forest, and wetlands), and

98

temporal patterns of occurrence and abundance were assessed in anthropogenically
influenced vs. natural landscape contexts (Lepczyk 2002). Occurrence in
anthropogenically influenced and natural landscape contexts was assessed by evaluating
temporal changes in the proportion of sites sampled annually occupied by each species
from 1996-2002. Calling intensity of breeding males was used to assess abundance;
patterns in mean annual call index values were used to evaluate temporal patterns in
abundance in each landscape context.
Statistical Analyses

After examination of descriptive statistics and normal probability plots for
original variables and potential data transformations, all landscape variables were log (x
+ 0.5) transformed prior to statistical analyses. Principal components analysis (PCA) was
used to summarize the variation in landscape characteristics separately at each scale
(Morrison 1990). The resulting principal components (PCs) represented a reduced
number of independent variables that were then used to investigate relationships between
landscape variables and anuran presence, abundance, and species richness. Only the PCs
that explained a significant amount of variation in the original dataset were retained for
further analyses; total variation explained at each of the spatial scales was _>_85%. All
data analyses described below were carried out using both the original (log-transformed)
landscape variables and PCs at each buffer scale to determine which variables explained
a higher proportion of total variation in the models, and would thus have the highest
potential for application to management and conservation strategies.

Multiple regression analyses were used to examine how individual landscape

variables and principal components were associated with anuran abundance and species

99

richness (Jongrnan et al. 1995, Lehtinen et al. 1999, Knutson et al. 1999, 2000). Logistic
regression was used to determine the explanatory power of landscape variables and
principal components on species presence (Jongman et al. 1995, Vos and Chardon 1998,
Guerry and Hunter 2002). Both original (log-transformed) land cover variables and
principal components were used to construct regression models to identify which
variables (land cover or principal components) were most influential and practical for
future management and conservation applications. Temporal patterns of species richness
and individual species’ occurrence and abundance were evaluated using simple linear
regression techniques. All data analyses were performed using SYSTAT (version 8.0,

SPSS, Inc. 1998) and SAS (release 8.02, SAS Institute 2002).

Results

Three land cover types dominated the landscape surrounding MFTS sites:
agriculture, forest, and nonforested wetlands. At both buffer scales, these three variables
comprised >62% of the total buffer area (Table 3.2). These three land cover types had
the greatest total area compared to all other land cover types, although their rankings
differed slightly between the two buffer scales (Table 3.2). Within 100 m adjacent to the
survey site, forest was the dominant land cover (24.2%), followed by agriculture (20.5%)
and nonforested wetlands (17.3%) (Table 3.2). In the landscape up to 1000 m from the
survey site, agriculture was the dominant land cover type (32.2%), followed by forest
(26.3%) and nonforested wetlands (11.4%) (Table 3.2). At both scales, water was the

least abundant land cover type, covering < 3% of the buffer area (Table 3.2). There was

100

an average of 21 1.42 m (std. error = 2.19 m) of roads within 100 m of survey sites, and
7689.12 m (std. error = 157.58 m) of roads within 1000 m (Table 3.2).
Principal Components Analysis

Principal components analyses of landscape variables at the 100 m scale yielded a
reduced set of four independent variables that explained approximately 85% of the total
variation in the original data set (Table 3.3). Examination of eigenvalue plots fi'om the
retained PCs provided insights into the original variables that were highly influential in
the dataset and contributed to the ordination of the PCA (Figure 3.2, Table 3.3). The first
PC represented a contrast between agricultural and both forested and nonforested wetland
sites. The second PC represented a contrast between open or forested sites and those in
agricultural or wetlands context. The third PC represented a weighted average of all
variables. However, three variables (agriculture, open, and forest) were most influential
in the ordination of the third PC, and two variables (roads and urban) contributed very
little. The area of open water surrounding survey sites dominated the fourth PC.

At the 1000 m buffer scale, PCA produced two PCs that accounted for >93% of
the total variation (Table 3.4). The first PC was heavily dominated by water, and the
second PC was heavily dominated by agriculture (Table 3.4, Figure 3.3). All other

variables contributed very little to the ordination.
Species Presence

Logistic regression analyses at both spatial scales revealed several landscape
variables that significantly influenced species presence at survey sites. Immediately
surrounding survey sites (100 m buffer scale), the variables that were most often

significantly related to species’ abundances and site species richness were agriculture,

101

water, urban, and nonforested wetlands (Table 3.5). Agricultural land cover was most
frequently associated with anuran occurrences, as indicated by significant relationships in
8 of the 11 cases (Table 3.5). These relationships were evenly split between both positive
and negative. The second most frequent land cover association was found between water
and anuran occurrence, with significant relationships seen for 7 of 11 species (Table 3.5).
These relationships were also approximately evenly split between negative (4 species)
and positive (3 species) associations (Table 3.5). Urban land was significantly associated
with 6 of 11 species (Table 3.5). All but one species was significantly negatively
associated with urban land (Table 3.5). Six species were also significantly associated
with the area of nonforested wetlands (Table 3.5). These relationships were generally
positive associations, although two species were negatively related to area of nonforested
wetlands adjacent to survey sites (Table 3.5). All other land cover variables were

significantly associated with fewer than half of the species in the study area.

At the 100 m scale, principal components 1 and 4 each were significantly
associated with the occurrence of three species, while PC 2 was significantly related to
the occurrence of four species (Table 3.6). PC 1, a contrast between agricultural and
wetland sites, was significantly positively related to the occurrence of R. clamitans and R.
catesbeiana, and was negatively associated with B. americanus presence (Table 3.6). PC
2, a contrast between open or forested sites and agricultural or wetland sites, was
consistently negatively related to the presence of P. triseriata, R. pipiens, R. palustris,
and H. versicolor/Chrysoscelis (Table 3.6). PC 3 was a weighted average of all land
cover types, but was not significantly related to any species’ presence. PC 4, area of

water surrounding survey sites, was negatively related to the occurrence of R. sylvatica

102

and P. triseriata, but positively associated with the presence of A. c. blanchardi (Table
3.6). Neither P. crucifer nor B. fowleri occurrence was significantly associated with any

of the principal components.

At the broader landscape scale (1000 m radius around survey site), several
variables were significantly related to the occurrence of anurans. Urban land was
generally significantly negatively related to the presence of anurans (R. sylvatica, P.
crucifer, R. pipiens, B. americanus, and A. c. blanchardi), but positively associated with
the occurrence of R. catesbeiana (Table 3.7). Forest cover in the surrounding landscape
was generally negatively related to the presence of anurans (R. sylvatica, P. triseriata, R.
pipiens, B. fowleri, and B. americanus), but was positively related to the presence of P.
crucifer and R. catesbeiana (Table 3.7). Open land was consistently positively related to
the occurrence of R. sylvatica, P. triseriata, R. pipiens, B. americanus, H. versicolor
/chrysoscelis, and R. catesbeiana (Table 3.7). Although total road length was only
significantly associated with the occurrence of two species (P. triseriata and H.
versicolor/chrysoscelis), both relationships were negative (Table 3.7). Forested and
nonforested wetlands each predicted the occurrence of five species, but the relationships
were generally negative although not consistent among all species (Table 3.7). Similarly,
water in the surrounding landscape was significantly associated with the occurrence of
four species; two species had positive relationships (R. pipiens and R. catesbeiana) and
two had negative relationships (A. c. blanchardi and R. clamitans) (Table 3.7).
Agriculture was significantly negatively associated with the occurrence of R. catesbeiana

(Table 3.7).

103

Principal components were significantly associated with the occurrence of j ust
over half of the species present in the study area at the 1000 m buffer scale. The
occurrences of five species (P. crucifer, R. palustris, A. c. blanchardi, R. clamitans, and
R. catesbeiana) were not significantly related to any of the PCs (Table 3.8). PC 1,
strongly dominated by water, predicted the occurrence of five species; B. americanus had
a positive relationship while R. sylvatica, P. triseriata, R. pipiens, and H. versicolor/
chrysoscelis were all negative (Table 3.8). PC 2, strongly dominated by agriculture, was
significantly positively related to the occurrence of R. sylvatica and B. fowleri (Table
3.8).

Species Abundance

Multiple regression models fit the data relatively poorly, explaining less than 10%
of the variation in the data accounted for by landscape variables at either spatial scale
(Tables 3.9 — 3.12). Regression models built with the original variables explained
slightly more variance than those using principal components at either spatial scale.

Directly surrounding survey sites, urban land was significantly negatively related
to four species’ abundance (R. sylvatica, R. pipiens, H. versicolor/Chrysoscelis, and R.
clamitans; Table 3.9). Nonforested wetlands were also significantly positively related to
three species’ abundance (R. pipiens, R. clamitans, and R. catesbeiana) as well as site
species richness (Table 3.9). Agriculture was a significant factor in the regression
models for five species’ abundance and overall species richness; however, the direction
and magnitude of the trends were not consistent among all dependent variables (Table

3.9). Agriculture was negatively associated with B. fowleri abundance, and positively

104

associated with species richness and the abundances of R. sylvatica, P. triseriata, R.
pipiens, and H. versicolor/ chrysoscelis.

Multiple regression models using PCs yielded similar results to those presented
above. Six species’ abundances were significantly related to PC 4; however, the
direction and magnitude of these relationships were also not consistent among species
(Table 3.10). Bufo americanus, A. c. blanchardi, and R. catesbeiana were positively
associated with PC 4 (area of open water surrounding survey sites), while R. sylvatica, P.
triseriata, and H. versicolor/chrysoscelis were negatively associated with that PC. The
fourth PC also accounted for the least total variation in the dataset (< 9%, Table 3.3). PC
1, a contrast between agricultural and wetland sites, was significantly related to four
species’ abundances; the relationship was positive for three of those species (P. crucifer,
R. clamitans, and R. catesbeiana; Table 3.10). PC 2, a contrast between open or forested
sites and agricultural or wetland sites, was significantly negatively related to R. pipiens
and R. palustris abundance and species richness, and positively related to B. fowleri

abundance (Table 3.10).

At the larger landscape scale (1000 m buffer), the amount of forested wetlands,
open land, water, and urban land were the four most consistent influential variables, with
significant trends consistently in the same direction with anuran abundance and species
richness (Table 3.11). There were also more significant relationships between land cover
types and individual species’ abundance at the landscape scale (1000 m buffer) compared
to the local scale (100 m buffer). Forested wetlands were significantly positively related
to four species’ abundances (R. sylvatica, P. triseriata, R. pipiens, and B. americanus),

and open land was also significantly positively related to species richness and the

105

abundances of P. triseriata, R. pipiens, and B. americanus (Table 3.11). The area of open
water in the landscape surrounding survey sites was positively related to the abundances
of R. pipiens, B. americanus, R. clamitans, and R. catesbeiana (Table 3.11). Urban land
was significantly negatively associated with the abundances of R. sylvatica, R. pipiens, B.
americanus, and H. versicolor/chrysoscelis (Table 3.11). Agriculture was a significant
variable in the regression models for six species and species richness, but the direction
and magnitude of the trends were not consistent among all dependent variables (Table
3.11). Somewhat surprisingly, species richness and four species’ abundances (R.
sylvatica, P. triseriata, P. crucifer, and H. versicolor/Chrysoscelis) were significantly
positively related to the amount of agricultural land in the 1000 m buffer around the
survey site; significant negative relationships were found for the abundances of R.
clamitans and R. catesbeiana (Table 3.11). The remaining landscape variables (roads,
forest, and nonforested wetlands) were significantly associated with individual dependent
variables, but the direction and magnitude of the trends were not consistent among

species (Table 3.11).

Both principal components were significantly related to site species richness and
many species’ abundances, with the direction of the trends consistent across all
significant relationships. PC 1 was significantly negatively associated with the
abundances of R. sylvatica, P. triseriata, P. crucifer, R. pipiens, H. versicolor/
chrysoscelis, and R. clamitans, while PC 2 was significantly positively associated with
site species richness and the abundances of R. sylvatica, P. crucifer, R. pipiens, B.

fowleri, R. clamitans, and R. catesbeiana (Table 3.12). However, all of these

106

relationships explained very little of the total variance in the dataset (R2 range: <0.001 —

0.068; Table 3.12).
Landscape Context and Temporal Patterns

There were no significant temporal patterns in species richness in any of the four
landscape contexts (urban, agriculture, forest, or wetlands), however, species richness
decreased slightly over the 1996-2002 time interval in survey sites in natural contexts (R2
= 0.019, P = 0.026; Table 3.13). At survey sites in anthropogenic contexts, A. c.
blanchardi declined significantly in site occupancy (R2 = 0.667, P = 0.025; Table 3.14,
Figure 3.4), while R. sylvatica, P. crucifer, R. palustris, and R. clamitans showed
significant increases in calling intensity over the same time period (P < 0.05, Table 3.15).
At sites in natural contexts, R. pipiens declined in site occupancy (R2 = 0.851, P = 0.003;
Table 3.14, Figure 3.5). Mean annual call index values for five species showed
significant trends over the time period at survey sites in a natural landscape context; four

species increased significantly, and B. americanus decreased (Table 3.15).

Discussion

Foremost among the findings was that the occurrence and abundance of anurans,
as well as site species richness, exhibited significant relationships with land cover types
and roads regardless of the methodology used. The principal components and regression
analyses provided qualitatively similar results. Regression analyses using the individual
landscape variables will be more useful for the development of future predictive models,

while PCA was able to reduce the data to a smaller set of independent variables that

107

provided a composite picture of anuran-habitat relationships. The regression models
using original landscape variables explained a higher proportion of variance than those
using principal components, which was also the case for a similar study of anuran-habitat
associations in Wisconsin and Iowa (Knutson et al. 1999). Using both methods provided
complementary information for a more comprehensive understanding of anuran-habitat
relationships in southern Michigan.

The regression models accounted for only a small proportion of the total variance
(<10%). In other similar studies, the proportion of variance explained by landscape-level
factors was greater, yet also relatively small (5 20% in Knutson et al. 1999, 5 35% in
Bonin et al. 1997 and Hecnar 1997). These studies differed slightly in the methods used
to measure anuran abundance, which may have been influential. Knutson et al. (1999)
and Bonin et al. (1997) used call index values from regional monitoring programs, as I
did in this study. Hecnar (1997) limited his analyses to landscape-level associations with
species richness. Previous studies of anuran-land cover associations in southeastern
Michigan also yielded weak but significant relationships (K. S. Genet, unpublished data).
In that case, the paucity of strong associations was attributed to a relatively small
geographic area (seven counties in southeastern Michigan) with less variation in the
distribution of land cover types (highly populated and urban developments) than the
larger geographic area considered in this study.

Species richness at wetland breeding sites is generally reduced at sites in urban
contexts with more roads and where land cover is dominated by urban or other
anthropogenically-modified areas (Lehtinen et al. 1999). Conversely, regional forest

cover has been found to be a very important explanatory variable for anuran species

108

richness (Hecnar and M’Closkey 1998). In this study, urban area and roads were not
significantly associated with species richness, but agriculture and Open land were. Many
anurans depend on a complex of habitats throughout their life cycles, and open land
including both natural and managed recreational areas provides important foraging
habitats for many species. The lack of significant relationships between species richness
and urban land or roads was surprising; perhaps future examination of additional
landscape characteristics related to habitat fragmentation would clarify these
relationships. Some species have adapted quite well to altered habitats in urbanized areas
(e. g., R. clamitans; Harding 1997, Hecnar and M’Closkey 1997b). Forest cover was also
not a significant influence at the 1000 m scale, but perhaps this reflects the diverse habitat
requirements of southern Michigan anurans that benefit from a variety of habitat types
and include species that are not wholly dependent on woodland habitats.

Individual species varied greatly in their response to landscape variables.
Although urban land was not significantly associated with species richness at survey
sites, it was negatively related to occurrence and/or abundance for nine of the eleven
species in the study area. Urbanization negatively influences anurans as a result of the
associated land use changes, including loss of naturally vegetated habitats, industrial land
uses, fragmentation of formerly continuous populations and/or dispersal corridors, and a
broad range of pollutants and contaminants. Urban land was negatively related to anuran
presence and abundance in other studies (Knutson et al. 1999, Lehtinen et al. 1999).
Roads were also negatively related to four species. Roads reduce habitat connectivity,
and reduced anuran presence and/or abundance in urban areas or those with a high

density of roads can be attributed to habitat loss, isolation from neighboring habitats, and

109

dispersal baniers (Mader 1984, Fahrig et al. 1995, Findlay and Houlahan 1997, Lehtinen
et a1. 1999). Roads are significant barriers to dispersal, and juvenile dispersal is among
the most important life history movements linking populations in fragmented landscapes
(Brown and Kodric-Brown 1977, Berven and Grudzien 1990).

Forests provide important habitat for many species that spend all or part of their
nonbreeding season in trees, shrubs, or litter. Positive associations between forest cover
and anuran abundance is one of the most consistent landscape scale habitat relationships
reported in the literature (e. g., Strijbosch 1980, Laan and Verboom 1990, Bonin et al.
1997, Findlay and Houlahan 1997, Hecnar 1997, Mitchell et al. 1997, Knutson et al.
1999). In southern Michigan, the species that would be expected to be most sensitive to
forest cover are those that spend a considerable amount of time in woodland habitats,
including R. sylvatica, P. crucifer, and H. versicolor/chrysoscelis (Harding 1997).

I did not. find consistent positive relationships with forest cover for most species
in this study. Pseudacris crucifer presence was positively influenced by forest cover at
the landscape scale, but negatively associated with forest directly surrounding the
surveyed breeding sites. This species uses a wide variety of temporary and permanent
wetlands for breeding, but disperse to woodlands after the breeding season (Harding
1997). The occurrence of R. sylvatica, the species with the strongest expected sensitivity
to forest cover, was surprisingly negatively associated with forest cover at the landscape
scale and unrelated to forest cover directly surrounding breeding sites. However, this
species was positively influenced by forested wetlands, and lowland forests likely
provide more valuable habitat for this species than upland forests. I did not differentiate

between different types of forest cover in this study (i.e., deciduous vs. coniferous or

110

natural vs. planted/managed), but these variables may be important for some species and
have been more thoroughly investigated in northeastern North America (e. g., Waldick et
al. 1999). The presence and/or abundance of other species (i.e., P. triseriata, R. pipiens,
B. fowleri, B. americanus, and R. clamitans) were negatively associated with forest cover,
which is likely due to their preference for more open or permanently aquatic habitats
(Pais et al. 1988, Koloszvary and Swihart 1999).

Many open habitats provide important foraging habitats for anurans, and several
species in southern Michigan use open areas for at least part of their life cycles. 1 found
the presence and/or abundance of R. sylvatica, P. triseriata, P. crucifer, R. pipiens, B.
americanus, H. versicolor/Chrysoscelis, and R. catesbeiana to be positively associated
with open land. Two of these species, R. pipiens and B. americanus, are considered to be
species with open and generalist habitat affinities, respectively (Harding 1997, Hunter et
al. 1999, Guerry and Hunter 2002). Although P. crucifer is typically most abundant in
wooded areas, they also use a wide variety of open habitats during the terrestrial phases
of their life cycle (DeGraaf and Rudis 1990). The open land cover type in this study
included areas of herbaceous and woody open land, as well as recreational areas that are
suitable habitat for these species.

I expected to find negative associations between anuran presence or abundance
and agricultural land. Some agricultural land use practices have the potential to impact
aquatic habitats (including anuran breeding sites) by altered nutrient regimes, sediment
accretion, changes in water temperature and oxygen content, and increases in pollutants
(Abramovitz 1996). These types of habitat changes are likely to negatively affect

mortality, reproductive success, growth, and behavior of organisms that live in these

111

habitats (Moyle and Leidy 1992, Saunders et al. 2002). More specifically, modern
farming practices are associated with habitat loss, drainage of wetlands, conversion of
important habitats to intensively managed monocultures of annual crops, soil compaction,
and disturbance to anurans that may spend time in underground habitats (Bonin et al.
1997).

Some species were negatively related to agriculture (B. fowleri, A. c. blanchardi,
R. clamitans, and R. catesbeiana), however, other species were either positively related to
agriculture or had mixed responses at the two spatial scales (R. sylvatica, P. triseriata, P.
crucifer, R. pipiens, H. versicolor/Chrysoscelis). Although others have found a negative
relationship between agriculture and anurans (e. g., Brodman and Kilmurry 1998),
Knutson et al. (1999) also failed to find consistent or strong negative associations
between anurans and agriculture. Agricultural landscapes often include small remnant
forest patches, which may provide refugia for some species. Although intensively
managed and annually cultivated row crops are probably inhospitable to most anurans,
other types of agriculture (e. g., forage crops) may offer suitable habitat for anurans
during some parts of their life cycles. Distinguishing among these different agricultural
practices may be useful in evaluating agricultural land impacts on anurans, but fiequent
crop rotations make this very difficult. The land cover data used in this study are
composite images from multiple years, and some agricultural land was likely rotated
between forage and row crops.

All species in this study depend on wetlands for breeding and larval development.
Therefore, I expected all species to be positively influenced by the water and wetland

areas directly adjacent to and in the landscape surrounding breeding sites. Many species

112

rely on temporary wetlands or shallow, vegetated edges of permanent water bodies for
reproduction, and the area of open water adjacent to or in the surrounding landscape of
survey sites represented a negative influence for most species (including R. sylvatica, P.
triseriata, P. crucifer, R. palustris, H. versicolor/Chrysoscelis, A. c. blanchardi, and R.
clamitans). Most open 'water habitats are permanent ponds or lakes, and predatory fish
limit the distributions of many anurans to seasonal freshwater habitats with reduced
predation pressure (Wellbom et al. 1996, Hecnar and M’Closkey 1997a). Rana pipiens,
B. americanus, and R. catesbeiana were positively influenced by open water areas. R.
pipiens and B. americanus are open or generalist habitat species that benefit from a
variety of land cover types, while R. catesbeiana is restricted to breeding only in
permanent water bodies as a result of the extended larval development period (Harding
1997)

The area of wetlands (forested and/or nonforested) positively influenced the
occurrence and/or the abundance of all but four species (R. palustris, B. fowleri, H.
versicolor/charsoscelis, and A. c. blanchardi). Those species that were unaffected by
wetland area (R. palustris and H. versicolor/chorsoscelis) are likely more sensitive to
other landscape factors, perhaps representing important habitats outside the breeding
season. Species that were negatively influenced by wetland area (B. fowleri and A. c.
blanchardi) may be less dependent on areas classified as wetlands and able to use a wider
variety of land cover types. Another consideration for wetland habitats is their dynamic
nature; climate strongly influences depth and duration, and these characteristics are not
reflected in the relatively coarse spatial and temporal resolution (multiple seasons and

years) of the land cover data. Those species with positive relationships to wetland area

113

were typically associated with the types of wetlands that comprise their characteristic
breeding habitats (e. g., R. sylvatica was related to forested wetlands, R. clamitans was
related to nonforested wetlands). The positive influence of wetlands for most species’
presence and abundance probably reflects not only requirements for this habitat type for
breeding, but also the importance of connectivity among wetland habitats for
metapopulation structure (degren 1991, Gibbs 1993, Semlitsch 2000, 2002).

Over the duration of this study (1996-2002), Acris crepitans blanchardi, listed as
a species of special concern in Michigan, declined significantly in site occupancy in
anthropogenically-influenced and marginally declined in natural landscape contexts.
Additional detailed studies on this species need to determine if distributional declines are
related to habitat or landscape level factors (Lee 1998, Lehtinen 2001). Calling intensity
of most species increased significantly both in anthropogenically-influenced and natural
landscape contexts. Rana sylvatica and R. clamitans increased in abundance (assessed by
mean annual call index values) in both landscape contexts. Improvements in survey
timing and species detection by volunteer observers probably account for at least a
portion of this trend, particularly for R. sylvatica.

Anurans in southern Michigan appear to benefit fiom a mosaic of land use types
surrounding their wetland breeding habitats. Some land uses that were hypothesized to
be negatively associated with anuran communities may instead provide suitable habitat.
As discussed above, some types of agricultural land may provide foraging habitat and not
impede movements between other habitat patches. Similarly, low intensity urban areas
may also provide suitable habitat for some species, as many areas incorporate small

wetlands, constructed ponds, and intermittent woodlots into urban planning. For

114

example, R. clamitans has been very successful at colonizing new ponds, including
constructed ponds in residential areas (Hecnar and M’Closkey 1997b).

Anurans respond to environmental factors at multiple spatial scales, including
both landscape (as investigated here) and within-habitat characteristics. Anuran
communities are influenced by habitat factors that determine success and fitness at the
population level and landscape factors that determine habitat suitability and connectivity
(Lehtinen et al. 1999). Although landscape variables have been shown in other studies to
be better predictors of species occurrence than pond water chemistry (Beebee 1985),
landscape variables alone have explained _<_ 35% of the variance in other studies (Bonin et
al. 1997, Hecnar 1997, Knutson et al. 1999). A combination of local (i.e., habitat) and
regional (i.e., landscape) variables best accounted for patterns of species richness in
Ontario landscapes (Hecnar and M’Closkey 1998). The potential contributions of many
factors to the abundance and distribution of anurans at both local habitat and regional
landscape scales indicate the importance of examining variables at multiple spatial scales
to understand the factors important in structuring anuran communities.

In this study, the relatively small proportion of total variance explained by the
significant relationships is an important limitation of these results. Studies such as this
may be limited both by the data collection methods (multiple volunteer observers,
nonrandomly selected survey sites, and categorical call index values) and insufficient
power to detect significant associations with the statistical methods used (van Dorp and
Opdam 1987, Lehtinen et al. 1999). The study sites represent listening points for the
MFTS, and these roadside locations may not be precisely located within the breeding

pond fiom which anurans are calling. Some of the survey sites may also be from very

115

small wetlands. While these small wetlands are important habitats for anurans (e.g.,
Gibbs 1993, Semlitsch and Bodie 1998, Snodgrass et al. 2000), they may be too small to
be represented in the land cover data, given its resolution. An additional consideration in
the models developed to predict species occurrence is an inherent limitation of the call
survey data from the MFTS: while a species documented as calling definitely represents
presence at a survey site, species not documented as calling remain undetected as they
may be present but unheard by the volunteer observer(s). As a consequence of the
limitations of this study, the results should perhaps be viewed as exploratory, and
refinement of sampling methods (e. g., intensive quantitative sampling of populations at
individual breeding sites) may improve the predictive power of subsequent models.
However, data such as these could possibly be used to develop spatial models (e.g.,
Halley et a1. 1996) that can be used to predict the persistence of anuran populations at the
landscape level, as long as the limitations of call surveys and inherent low statistical
power are addressed.

The associations I found between anuran occurrence or abundance and land cover
types at the landscape level provide valuable information for managing and conserving
habitats for these organisms in areas where populations may be low or in decline.
Population trends determined fi'om monitoring programs (see Chapter One) need to be
combined with results such as these to achieve a more comprehensive understanding of
anuran-habitat associations. Anuran presence and abundance in southern Michigan are
associated with land cover that reflects habitat availability and suitability in this region.
Future research should incorporate anuran-habitat associations such as these into the

development of practical management and conservation strategies.

116

Acknowledgements

Land cover data were provided by the MDNR, and CR. Steen facilitated the conversion
and transfer of this data. T.M. Burton, S.K. Hamilton, D]. Hall, J. Liu, C.A. Lepczyk,
and W. Gordon provided valuable editorial comments. Support for this research was
provided by the Michigan Coastal Management Program of the Michigan Department of
Environmental Quality, National Oceanic and Atmospheric Administration, US.
Department of Commerce, Geological and Land Management Division of the Michigan
Department of Environmental Quality with funding from the State and Tribal Wetland
Development Program, Region V of the US. Environmental Protection Agency, Great
Lakes Wetland Research Consortium of the Great Lakes Commission with firnding from
the Great Lakes National Program Office, and the US. Environmental Protection

Agency.

117

Table 3.1

Species of anurans found in southern Michigan and included in this study. All species

except B. fowleri are distributed throughout the study area. Species ranges were

determined from Harding (1997) and Conant and Collins (1998).

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Species Distribution
Rana sylvatica Statewide
Pseudacris triseriata Statewide
Pseudacris crucifer Statewide
Rana pipiens Statewide
Rana palustris Statewide
Bufofowleri Southern and Western Lower Michigan
Bufo americanus Statewide
Hyla versocolor Statewide
Hyla chlsoscelis Statewide
Acris crepitans blanchardi Southern Lower Michigan
Rana clamitans Statewide
Rana catesbeiana Statewide
Table 3.2

Summary of land cover and road variables at both spatial scales in southern Michigan.
All land cover types represented the total area (m2) within each buffer; roads represented

the total length of roads within each buffer. Values are means from all study sites

(N=1055), their associated standard errors, and the proportion of the total buffer area
comprised by each land cover type.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Variable 100 m 1000 m

Mean SE Prop. Mean SE Prop.
Roads (m) 211.42 2.19 7689.12 157.58
Urban (m2) 4941.55 119.96 0.157 279134.94 9447.15 0.089
Agriculture (m2) 6448.53 254.63 0.205 101073878 23003.6 0.322
Open Land (m2) 3607.66 107.13 0.115 345940.05 5713.59 0.110
Forest (m2) 7608.18 215.59 0.242 826547.71 14206.76 0.263
Water (m2) 469.63 63.15 0.015 88833.27 5529.91 0.028
Fogested Wetlands 2749.66 114.24 0.088 216686.72 5657.77 0.069
(m )
Nopfor. Wetlands 5423.66 168.86 0.173 357936.29 7333.39 0.1 14
(m )

 

118

 

Table 3.3

Principal components retained from PCA of variables at 100 m scale. The first four
principal components explained a total of 85.12% of the variation in the dataset at that
spatial scale.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

PC 1 PC 2 PC 3 PC 4
Eigenvalue 6.310 3.971 2.380 1.419
Prop. Total Variance 0.3814 0.2400 0.1439 0.0858
Eigenvectors of
Original Variables
Roads -0.00006 0.00194 0.00594 -0.00001
Urban -0.05374 0.01013 0.02366 0.12487
Agriculture -O.56738 0.61456 0.50623 0.14574
Open Land -0.00562 -0.39858 0.56344 0.24481
Forest 0.10894 -0.46158 0.53646 -0.05227
Water 0.07234 -0.05914 -0.20277 0.88056
Forested Wetlands 0.60096 0.33863 0.26543 -0.20611
Nonforested 0.54488 0.36348 0.16220 0.28747
Wetlands

Table 3.4

Principal components retained from PCA of landscape variables at 1000 m scale. The
first two principal components explained a total of 93.66% of the variation in the dataset
at that spatial scale

 

 

 

 

 

 

 

 

 

 

 

 

 

PC 1 PC 2
Eigenvalue 5.772 2.314
Pr0p. Total Variance 0.6686 0.2680
Eigenvectors of
Original Variables
Roads 0.03 153 006492
Urban 0.03622 -0.08214
Agriculture -0.21743 0.96620
Open Land 0.01687 -0.04394
Forest 0.02376 -0.04946
Water 0.97305 0.22029
Forested Wetlands 0.03361 0.03266
Nonforested Wetlands 0.04014 0.03890

 

 

 

 

119

Table 3.5

Significant (p<0.05) landscape factors affecting occurrence of individual anuran species
at 100 m scale determined using logistic regression.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Species Landscape Regression Chi- P-
Variables Coefficient srmare Value
Rana sylvatica Urban -0.255 7.532 0.006
Agriculture 0.229 12.377 <0.001
Nonforested wetlands 0.163 4.719 0.030
Pseudacris triseriata Urban 0.323 4.3 15 0.038
Agriculture 0.320 13.418 <0.001
Water 0289 21.693 <0.001
Forested wetlands 1.477 15.479 <0.001
Nonforested wetlands 0.622 12.330 <0.001
Pseudacris crucifer Urban -2.686 13.828 <0.001
Agriculture -1.152 6.096 0.014
Forest 0290 4.055 0.044
Water -0.221 4.477 0.034
Forested wetlands -2.079 9.765 0.002
Rana pipiens Roads -1.835 6.464 0.011
Urban -0.185 6.192 0.013
Agriculture 0.140 15.168 <0.001
Water -1 .878 3.962 0.047
Forested wetlands -1.766 9.142 0.003
Nonforested wetlands 0.138 6.025 0.014
Rana palustris Water -4.549 4.726 0.030
Bufofowleri Urban -0.739 6.739 0.009
Agriculture -0.227 5.220 0.022
Nonforested wetlands -0.977 5.61 1 0.018
Bufo americanus Open land -0.397 7.007 0.008
Forested wetlands -0.724 17.244 <0.001
Nonforested wetlands -0.170 6.318 0.012
Hyla Agriculture 0. 1 81 8.714 0.003
versicolor/chrysoscelis
Acris creptians blanchardi Agriculture -0.691 6.574 0.010
Water 0.204 3.895 0.048
Rana clamitans Water 0.221 4.312 0.038
Rana catesbeiana Urban -0.359 4.314 0.038
Agriculture -0.525 8.151 0.004
Water 0.242 18.769 <0.001
Forested wetlands -0.382 4.197 0.041
Nonforested wetlands 0.135 5.346 0.021

 

120

 

Table 3.6

Significant (p<0.05) principal components affecting occurrence of individual anuran
species at 100 m scale determined using logistic regression.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Species Principal Regression Chi- P-Value
Comminents Coefficient square

Rana sylvatica 4 -0.236 13.791 <0.001

Pseudacris triseriata 2 -0.212 8.233 0.004
4 -0.328 20.765 <0.001

Rana pipiens 2 -0.126 6.574 0.010

Rana palustris 2 -0.218 5.1 17 0.024

Bufo americanus 1 -0.122 4.888 0.027

Hyla versicolor/ 2 -0.415 10.610 0.001

chrysoscelis

Acris creptians 4 0.633 1 1.793 <0.001

blanchardi

Rana clamitans 1 0.250 14.264 <0.001

Rana catesbeiana l 0.21 1 18.528 <0.001

 

121

 

Table 3.7

Significant (p<0.05) landscape factors affecting occurrence of individual anuran species
at 1000 m scale determined using logistic regression.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Species Landscape Regression Chi— P-
Variables Coefficient square Value
Rana sylvatica Urban -1 .625 15.088 <0.001
Open land 4.580 10.831 0.001
Forest -3.813 8.206 0.004
Forested wetlands 0.694 7.018 0.008
Pseudacris triseriata Roads -l6..067 8.726 0.003
Open land 1.604 9.064 0.003
Forest -1.044 5.309 0.021
Nonforested wetlands -14.958 9.108 0.003
Pseudacris crucifer Urban -19.788 16.152 <0.001
Forest 2.800 15.417 <0.001
Forested wetlands -38.444 17.123 <0.001
Nonforested wetlands -13.546 7.874 0.005
Rana pipiens Urban -1 .268 9.965 0.002
Open land 1.246 8.460 0.004
Forest -1.377 14.302 <0.001
Water 0.100 9.269 0.002
Bufofowleri Forest -21.535 4.169 0.041
Forested wetlands -7.299 7.000 0.008
Nonforested wetlands -9.845 8.712 0.003
Bufo americanus Urban -20.889 14.354 <0.001
Open land 1.282 7.200 0.007
Forest -20.075 16.568 <0.001
Nonforested wetlands -0.979 5.879 0.015
Hyla Roads -28.035 5.652 0.017
versicolor/chozsoscelis Urban -5 1 .202 12.83 1 <0.001
Open land 28.002 7.891 0.005
Acris creptians blanchardi Urban -38.221 5.747 0.017
Water -4.842 4.473 0.034
Forested wetlands -32.944 4.648 0.031
Nonforested wetlands -7.140 11.630 0.001
Rana clamitans Water -0.969 4.703 0.030
Rana catesbeiana Urban 16.744 8.708 0.003
Agriculture 3.883 7.965 0.005
Open land 14.977 7.481 0.006
Forest 10.227 9.783 0.002
Water 1.91 1 4.497 0.034
Forested wetlands 7.760 5.605 0.018

 

122

 

Table 3.8

Significant (p<0.05) principal components affecting occurrence of individual anuran
species at 1000 m scale determined using logistic regression.

 

 

 

 

 

 

 

 

 

 

 

 

Species Principal Regression Chi- P-Value
Components Coefficient square
Rana sylvatica 1 -0.096 5.204 0.023
2 0.175 13.465 <0.001
Pseudacris triseriata 1 -0.103 4.128 0.042
Rana pipiens 1 -0.132 10.055 0.002
Bufo fowleri 2 0.313 4. 169 0.041
Bufo americanus 1 0.099 4.153 0.042
Hyla versicolor/chrysoscelis 1 -0.254 1 3 .631 <0.001

 

123

 

Table 3.9

Results of multiple regression analyses of landscape variables on anuran abundance and
species richness at 100 m scale. N=1055 for all species except B. fowleri (N=651).

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Abundance and Richness Regression R2 P-
Variables Landscape Variables Coefficient value
Rana sylvatica Urban -0.204 0.045 <0.001
Agriculture 0.058 0.003
Water 0082 0.010
Pseudacris triseriata Agriculture 0.044 0.034 0.015
Water 0130 <0.001
Rana pipiens Urban -0.104 0.037 <0.001
Agriculture 0.032 0.008
Forest -0.039 0.041
Nonforested Wetlands 0.058 0.002
Bufo fowleri Agriculture 0018 0.018 0.01 1
Bufo americanus Water 0.088 0.016 0.002
Hyla versicolor Roads -0.482 0.039 0.011
/chrysoscelis Urban -0.092 0.008
Agriculture 0.072 <0.001
Water 0053 0.041
Rana clamitans Urban -0.091 0.045 0.003
Forest -0.054 0.016
Water 0.070 0.002
Nonforested Wetlands 0.053 0.018
Rana catesbeiana Water 0.125 0.062 <0.001
Nonforested Wetlands 0.039 0.039
Species Richness Agriculture 0.104 0.035 <0.001
Open Land 0.087 0.039
Forest -0.081 0.050
Nonforested Wetlands 0.094 0.023

 

124

 

Table 3.10

Results of multiple regression analyses of Principal Components on anuran abundance
and species richness at 100 m scale. N=1055 for all species except B. fowleri (N=651).

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Abundance and Richness Principal Regression R- P-

Variables Components Coefficient squared value

Rana sylvatica 4 -0.206 0.034 <0.001
Pseudacris triseriata 1 —0.048 0.026 0.048
3 0.085 0.008

4 -0.133 <0.001

Pseudacris crucifer 1 0.038 0.008 0.028
Rana pipiens 2 -0.052 0.016 0.003
3 -0.048 0.027

Rana palustris 2 -0.014 0.005 0.038
Bufofowleri 2 0.022 0.009 0.025
Bufo americanus 4 0.062 0.007 0.046
Hyla versicolor/chrysoscelis 4 -0.122 0.021 <0.001
Acris creptians blanchardi 4 0.024 0.006 0.036
Rana clamitans 1 0.100 0.034 <0.001
Rana catesbeiana 1 0.080 0.055 <0.001
4 0.123 <0.001

Species Richness 2 -0.097 0.009 0.012

 

125

 

Table 3.11

Results of multiple regression analyses of landscape variables on anuran abundance and
species richness at 1000 m scale. N=1055 for all species except B. fowleri (N=651).

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Abundance and Richness Regression R2 P-

Variables Landscape Variables Coefficient value

Rana sylvatica Roads 0.658 0.099 0.024
Urban -1.153 <0.001

Agriculture 0.063 0.017

Forested wetlands 0.481 <0.001

Pseudacris triseriata Roads -1 . 145 0.075 <0.001
Agriculture 0.065 0.010

Open land 0.902 <0.001

Forest -O.387 0.024

Forested wetlands 0.396 0.001

Nonforested wetlands -0.422 0.011

Pseudacris crucifer Roads -O.493 0.081 0.013
Agriculture 0.043 0.018

Forest 0.412 0.001

Nonforested wetlands 0.269 0.025

Rana pipiens Urban -0.413 0.088 0.001
Open land 0.308 0.022
Forest -0.478 <0.001

Water 0.036 0.001

Forested wetlands 0.161 0.034

Bufo americanus Urban -0.507 0.034 0.006
Open land 0.799 <0.001
Forest -0.71 1 <0.001
Water 0.060 <0.001

Forested wetlands 0.283 0.011
Nonforested wetlands -0.602 <0.011

Hyla versicolor/chrysoscelis Urban -0.527 0.080 0.002
Agriculture 0.098 <0.001

Rana clamitans Agriculture -0.066 0.041 0.001
Water 0.045 <0.001

Rana catesbeiana Agriculture 0044 0.046 0.011
Water 0.044 <0.001

Species Richness Agriculture 0.077 0.021 0.035
Open land 0.760 0.011

 

126

 

Table 3.12

Results of multiple regression analyses of Principal Components on anuran abundance
and species richness at 1000 m scale. N=1055 for all species except B. fowleri (N=651).

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Abundance and Richness Principal Regression R- P-value
Variables Components Coefficient squared
Rana sylvatica 1 -0. 155 0.066 <0.001
2 0.1 19 <0.001
Pseudacris triseriata 1 -0.099 0.023 <0.001
Pseudacris crucifer 1 -0.046 0.040 0.001
2 0.100 <0.001
Rana pipiens 1 -0.095 0.068 <0.001
2 0.075 <0.001
Bufofowleri 2 -0.023 0.012 0.014
Hyla versicolor/Chrysoscelis 1 -0. 1 31 0.052 <0.001
Rana clamitans 1 -0.037 0.019 0.020
2 0.074 <0.001
Rana catesbeiana 2 0.079 0.023 <0.001
Species Richness 2 0.098 0.009 0.004
Table 3.13

Temporal patterns in species richness at breeding sites in various landscape contexts.
Trends indicate results from regression analyses 1996-2002.

 

 

 

 

 

 

 

Landscape Regression

Context N Coefficient R—squared P-value
Urban 52 0.132 0.034 0.194
Agriculture 1604 0.009 0.014 0.674
Forest ~ 454 -0.067 0.073 0.082
Wetlands 140 -0.027 0.013 0.654
Anthropogenic‘ 2134 0.003 0.011 0.891
Natural7 3621 -0.032 0.019 0.026

 

 

 

 

 

 

 

Urban and agricultural landscape contexts combined
2Forest, wetlands, open land, and open water landscape contexts combined

127

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

6666 6mm.6 666.6 36.6 M666 m666 $866868 6:86
636 N656 :66 m 66.6 666.6 m666 Benefit 6:86
6666.6 mmod 6.3.6 $66 36.6. m666- 66.862er 383.26 6.6ro
6.6663356.»

63.6 36.6 66 6.6 m666 666.6- 666.6 \S6oc6flmz 5666
m 6N6 366.6 6666 £6.66 N666- 666.6- Seeotefie o 3%
v.36 $66 6666 6606 666.6- 666.6- isEQKekzm
6m 66 6.606 wmm6 66N6 6666. N666- 36963. 33%
mood 86¢ 53 86 .o Sod- 6.8.6- 22%. as:
wmm6 36.6 666.6 666.6 mood- 666.6 $6.326 E33856
6666 n 666 me 6 .6 m m 6 .6 M666 6666 Seton: drugged
666.6 636 666.6 666.6 6666. N666 8.536.? 336
65.5376 inowemmhfiue. 6.26.376 gauche—66.3. 6.26.376 aeowemeufii 86925

2.69» 1A6 695:6; rm 65696.66er .86me. em

 

 

 

 

$686666 EPA 6608 868% 6680 .3 6066380 60.33% 866m .60 6606:2683 868% 66a .56 n M Z
.3588 0&8656 683g 65 0683626665 :6 866m $666586 3 6056680 86m :6 munch 622688 8.6 mowing 6863296 .60 $6336

666 0666.6.

128

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

nmNd 6.66.6 N666 6.66.6 N666- m 66.6 666 SN 53.56368 33%
666.6V 6666 $6.6 666.6 6N6.6 mN66 66NN m 6N6 4:86:83 6:36
666.6 666.6 NN6 .6 666.6 mm 6 .6 6 66.6 66 6m EnefieeS
6:33.98 6.6ro

666.6 666.6 Nm66 6N66 6N6.6 666.6. 66me 6666 389.3. :66
\uo6ou6wxe> 56.6%

N666 666.6 666.6 666.6 Nm6.6- 666.6. mNm6 mt wzeuotmfie oxmm
Nag 63o ”8.66 :5 8.6.6. wood. s. 2 E63866
666.6 666.6 6 66.6 66 m6 N86- 612 .6 Nm mm fibufiemgex
6N66 666.6 666.6 666.6 6666. N666 66m 66N 36.66% 33%
$6.6 666.6v 666.6 6.6.6.6 666.6. mm66 mwom 6666 Sheet ascended
666.6 86.6 666.6 666.6 666.6. 6666. MN: N66 6 SEEMS 6.665633%
666.6v 666.6 m 66.6 6 66.6 N666 6646.6 6.66 6 666 865.6%. e286

g2 6.56.5. 69.5576 .65.; 6.25.2 665.5. 6.2.6376 6:66.;V leBlom—ml
2.6.; 1A— ghlaradm ry— EfioEer am Z

 

 

m 6 .m 0666.6.

366:8 068356 68:3: 668 Amwfigon 55.668
:6 .8563 60365563 oEoMOQOEEa :6 866m $666095 3 mos—g 56:6 :8 36:5 :6 3:26 680658 60.6 896626 606332 .60 866536

129

Figure 3.1

Locations of survey routes for the Michigan Frog and Toad Survey throughout
MichiganOnly routes in the outlined portion of the southern lower peninsula
were used in this study.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

" 2}
. i-
. A
‘3 o .
I:
O. Q. . O.
L . ’ '

 

 

 

 

 

130

Figure 3.2

Eigenvalue plots from Principal Components Analysis at 100 m spatial scale. All
three panels share the indentical X-axis. Landscape variables plotted are those
described in Table 3.2. Abbreviations are as follows: ag=agriculture, open=0pen
land, forwet=forested wetlands, nforwet=nonforested wetlands. Other variables not

 

 

 

 

 

 

 

 

 

 

abbreviated. 7 . a
6 8
A i nforwet
°\° 3‘ forwet
‘1' .
{‘3 2
8 1 urban road
a. 2+ ° water
-2‘
'3. .open
'4. 'forest
-5
-6
6‘ 0P6“ .
§ 5: ‘ag forest
3 4. forwet
3. 31 .
m 1. urban Dad nforwet
U 0
e. 1.
:2 water
_3‘
-4
-5
-6 - f - a
-9 ' water
"a? 6‘
80 . nforwet
V} 3 open
0:2“ 38 urbano road
2 O forest
_31 forwet
-6

 

 

 

 

-9-7T6i5-Z?2L10 1 2 3 1t 5 b 7
PC 1 (38.14%)

131

Figure 3.3

Eigenvalue plots from Principal Components Analysis at 1000 m spatial scale.
Landscape variables plotted are those described in Table 3.2. Abbreviations are
as follows: ag=agriculture, open=open land, forwet=forested wetlands,
nforwet=nonforested wetlands. Other variables not abbreviated.

 

l e

 

 

38
0.8
0.6“
0.4"
§
3 02 water.
\(9; nforwet
V 0‘ 'forwet
8 open orest
m rgad
_02 ur an
-O.4
-0.6
-00
-1 -

 

 

 

 

-1 -0.8 -0.6 -0.4 -0.2 0 0.2 0.4 0.6 0.8 1
PC1(66.86%)

132

Figure 3.4
Temporal trend (1996-2002) in site occupancy for Acris crepitans blanchardi.
Open circles indicate proportion of survey sites occupied in natural landscape
contexts, and darkened circles indicate proportion of survey sites occupied in
anthropogenic landscape contexts. Solid line represents significant linear trend
in site occupancy in anthropogenic landscape context.

 

 

,0 0.04 .
a O ' Anthropogenic
3 0Natural
0
3 003'
8
a.
>.
D
E 0.02
m
“-4
o
8
'{g 0.01‘ o
o
G.
8
O.‘
0 m ”

 

 

 

1996 1997 1998 1999 2000 2001 2002

Figure 3.5
Temporal trend (1996-2002) in site occupancy for Rana pipiens in survey
sites in a natural landscape context. Solid line represents significant linear
trend in site occupancy.

0.2

 

o
H
09

.9
p—i
ON

9
p—
.p,

 

.9
to

R2: 0.851

 

 

Proportion of Survey Sites Occupied

 

f

1996 1997 1998 1999 2000 2001 2002

.0
hi

133

Chapter Four
The Influence of Local Habitat Characteristics on the Growth, Development, and
Survival of Spring Peeper (Pseudacris crucifer) Tadpoles in Southwestern Michigan
Abstract
Tadpoles are influenced by a wide variety of physical, chemical, and biological

factors specific to breeding ponds during their larval periods. These factors exert
profound influences over growth, development, and survival to metamorphosis, which
subsequently influence adult fitness traits. The grth and development of spring peeper
(Pseudacris crucifer) tadpoles relative to hydrology, water chemistry, and biotic
interactions were monitored in a field study in ten natural wetlands in southwestern
Michigan. Tadpoles in predator-free enclosures were monitored throughout their larval
period for (1) growth and development rates, (2) survival, (3) length of larval period, and
(4) size at metamorphosis. Growth, development, and survival were highest at sites with
intermediate hydroperiods, partial canopy cover, and few fish predators. Few site-
specific water chemistry variables correlated significantly with tadpole response
variables, but chlorophyll a was significantly related to growth and size at
metamorphosis. Tadpole development and metamorphosis are complex processes
influenced to varying degrees by a wide variety of physical, chemical, and biological
factors specific to breeding sites. Growth, development, and survival of Spring peepers
varied depending upon habitat factors at each of the study sites in southwestern

Michigan.

134

Introduction

Anuran amphibians have been widely regarded as model systems both for the
experimental study of small-scale ecological processes structuring larval communities
(e.g., Morin 1983, Wilbur 1987) as well as large-scale ecological patterns of
metapopulation dynamics and amphibian declines (e.g., Blaustein et al. 1994a, Sjogren-
Gulve 1994, Hecnar and M’Closkey 1996). Since the early 1990s, there has been
increased interest, debate, and research concerning the apparent declines of amphibian
populations on a global scale, although the causes have yet to be unambiguously
determined (Sarkar 1996, Green 1997b, Alford and Richards 1999, Houlahan et al. 2000).
Amphibians are generally considered to be sensitive indicators of environmental
conditions as a result of characteristics such as complex biphasic life cycles, cutaneous
respiratory surfaces, food habits, susceptibility to cold and drought, fi'agmented local
population distributions, and vulnerability to environmental contaminants and ultraviolet
radiation in both aquatic and terrestrial habitats (Stebbins and Cohen 1995).

Amphibians have complex life cycles; they are dependent on both aquatic and
terrestrial habitats during distinct life history stages (Wilbur 1980). In the larval stages
anurans have obligate aquatic habitat requirements. As a result, chemical and physical
characteristics of the environment have the potential to exert profound effects on tadpole
characteristics such as growth, development, survival, length of larval period, and
metamorphic size (Alford 1999). F urtherrnore, the aquatic embryonic and larval stages
may be the most vulnerable to mortality as both biotic and abiotic factors can act and
interact to influence growth and survival of these organisms (Dunson and Travis 1991).

Rates of growth, development, survival, and length of larval period also have subsequent

135

effects on life history traits such as timing of and size at metamorphosis and juvenile
recruitment, which are ultimately related to adult traits directly correlated with fitness
(Wilbur and Collins 1973, Collins 1979, Werner 1986, Patterson and McLachlan 1989,
Harris 1999).

Larval amphibians represent an excellent system for investigating the influence of
multiple abiotic and biotic factors on life history traits, and many aspects of tadpole
ecology have been studied extensively. Abiotic factors that affect tadpoles include
characteristics of the local aquatic environment such as hydrology, water chemistry, and
canopy cover. Hydroperiod (the amount and duration of water in a breeding pool) is one
of the most important considerations for tadpoles, and they can initiate rapid development
in drying ephemeral habitats (e.g., Smith-Gill and Berven 1979, Pandian and Marian
1985, Crurnp 1989, Patterson and McLachlan 1989, Tejedo and Reques 1994). Survival,
growth rate, length of larval period, and mass at metamorphosis are all affected by the
amount of water in a pond and seasonal pond persistence (e. g., Wilbur 1987, Rowe and
Dunson 1995). Pond persistence exerts strong influence on community structure in
wetlands where most amphibians breed (Wellbom et al. 1996, Skelly 1997).
Perturbations of water chemistry influence tadpole dynamics. Increased acidity reduces
survival and retards growth (e. g., Saber and Dunson 1978, Dunson and Connell 1982,
Rosenberg and Pierce 1995). Nutrients and dissolved ions can affect tadpoles directly via
osmoregulatory and physiological mechanisms (e.g., Ultsch et al. 1999) or indirectly by
influencing the algal community upon which they graze (e.g., Kiffney and Richardson
2001). Water temperature and dissolved oxygen content also influence growth,

development, and survival throughout larval development (e.g., Lucas and Reynolds

136

1967, Wassersug and Seibert 1975, Harkey and Semlitsch 1988). Canopy cover over
breeding ponds influences the distribution of amphibian larvae, and can reduce growth
rates in ponds underneath closed canopy forests (Skelly et al. 1999, Werner and
Glennemeier 1999, Skelly et al. 2002).

Anurans typically breed at wetland sites with conditions appropriate for larval
development. Which breeding sites are suitable differs among species and generally
represent trade-offs between competition, predation, and pond permanence among
different habitat types (Duelhnan and Trueb 1986, Skelly 1995a, 1995b, 1997, Harding
1997). Some species (e.g., Pseudacris triseriata) prefer ephemeral wetlands that fill with
meltwater early in the spring and completely dry by the end of the summer because of the
rapid nutrient flush in the spring and absence of large predators (e.g., fish) (Smith 1983,
Harding 1997). Other species (e.g., Rana catesbeiana, R. clamitans) have physiological
or behavioral defense mechanisms that allow them to exploit permanent ponds with fish
predators (Harding 1997, Alford 1999).

Spring peepers (Pseudacris crucifer) are able to utilize a variety of temporary and
semi-permanent wetlands for breeding (Skelly 1995, Harding 1997). This species is
common, widespread, and stable throughout Michigan (Harding 1997, see also Chapter
One). Spring Peepers are one of the earliest breeding amphibians in Michigan, and
typically begin calling in March (Harding 1997). Following the breeding season,
metamorphs typically begin to emerge beginning in June, but development and
metamorphosis can take up to 90 days at some sites (Harding 1997). Given their
distribution, status, and breeding site use, Spring peepers are excellent study organisms to

evaluate the influence of biotic and abiotic factors on larval biology.

137

Wetlands in the midwestem United States vary greatly in the physical, chemical,
and biological properties that influence amphibian breeding site selection, larval
performance, and ultimately recruitment of individuals into adult anuran populations.

The purpose of this study was to evaluate Spring peeper growth, development, and
survival in wetlands of differing abiotic (i.e., hydroperiod and water chemistry) and biotic
(i.e., potential predators) conditions. Although tadpoles typically experience significant
high densities and density-dependent factors such as inter- and intraspecific competition
during their larval periods(e.g., Alford 1999), the objective of this study was to examine
abiotic and biotic factors other than competition. Increasing the understanding of factors
that influence amphibian biology at this important larval life history stage will contribute

to the conservation and management of these organisms in wetlands.

Materials and Methods
Study Sites and Habitat Characteristics

Ten natural wetlands containing resident anuran populations were selected in
southwestern Michigan (Figure 4.1) that represented different habitat types,
hydroperiods, and predator communities (Table 4.1). Each site was surveyed
approximately once per week during the breeding and larval periods of most Michigan
anurans (March - August; mean number of site visits was 10). Survey methods used to
determine resident anuran communities at each site included opportunistic visual
encounter surveys for adults (Crump and Scott 1994), trapping of tadpoles using 2L clear
plastic fiinnel bottle traps (Griffiths 1985, Richter 1995, Adams et a1. 1997), and call

surveys (Sargent 2000, Michigan Frog and Toad Survey methodology, see also Chapter

138

One). Anuran larvae were identified using the taxonomic keys of Waterrnolen and
Gilbertson (1996) and Altig et al. (1998). Relative abundance of potential invertebrate
and fish predators was also determined from bottle trap samples

In addition, the following habitat variables were recorded in the field using a YSI
6000 handheld multisensor: pH, dissolved oxygen concentration (DO), percent 0;
saturation, water temperature, and specific conductance. Relative changes in water level
were also recorded. A water sample was collected monthly for laboratory determination
of total alkalinity, sestonic (suspended) chlorophyll a, available nutrients (nitrate [N03-
N], ammonium [NH4-N], total dissolved phosphorus [TDP], soluble reactive phosphorus
[SRP, PO4-P]), cations (Ca2+, Mg2+, Na”, K+), and anions (Cl', SO42'). Upon field
collection, all samples were put on ice. After returning to the lab, samples were
refiigerated (for determination of alkalinity and conductivity), filtered on Gelman Supor
0.47-mm membrane filters and then refrigerated (for determination of anions), or
acidified with 8 N HN03 (cations) until analysis.

In the laboratory, conductivity was measured using an Orion model 135
conductivity meter (Analytical Technology, Inc.). Ca2+, Mg2+, Na+, and K+ were
measured by flame atomic absorption spectrometry. Alkalinity, which generally
represents HCOg' in local waters, was determined by titration with 0.3 N HCl and
calculation of the Gran function (Cantrell et al. 1993). S042} Cl', and N03' were
measured by membrane-suppression ion chromatography. NIL,+ was measured
colorimetrically following an adapted version of the phenylhypochlorite method (Aminot
et al. 1997). Soluble reactive phosphorus (PO4-P) and total dissolved phosphorus (TDP)

were measured colorimetrically following the acid molybdate method (Wetzel and Likens

139

2000); the TDP colorimetric analysis was preceded by a persulfate digestion to
decompose organically bound P (V aldenama 1981). Algal biomass was measured as
chlorophyll-a fluorescence (Welshmeyer 1994) after cold extraction in 95% ethanol.

Canopy cover was recorded at each enclosure (see below) in four cardinal
directions in May (Spring, beginning of leaf-out) and July (Summer, leaf-on) using a
spherical densiometer (Forestry Suppliers, Inc.).

Field Study of Tadpole Growth

Two pairs of amplexed Spring peepers were collected on 5 May 2000 from a
small pond in Meridian Township, Ingham County, Michigan. Frogs were brought into
the laboratory until the eggs were laid and fertilized; adults were then returned to the site
where they were collected. The eggs hatched after approximately one week, and
hatchlings were maintained in the laboratory until they had reached approximately stage
25 (Gosner 1960). At the beginning of the field study, tadpoles from the two clutches
were mixed and randomly divided into 30 groups of 10 tadpoles each. All tadpoles were
measured (snout — vent body length, SVL, mm) and staged (Gosner 1960) at the
beginning of the study.

Three field enclosures were placed at each of the 10 study sites. Enclosures were
constructed from 19 L plastic buckets with screened sides (fiberglass insect screening)
and a flotation device attached to the top. Approximately 3 cm of sediment and detritus
from each study site was added to the bottom of each enclosure to provide a natural
substrate and food resource for the tadpoles. Sediment and detritus were visually

examined before placement in the enclosures to exclude macroinvertebrates.

140

On 2 June 2000, each enclosure was stocked with 10 tadpoles (0.526/L). This
density was low compared to natural densities of tadpoles (Alford 1999) such that growth
and development would not be influenced by density-dependent factors. In natural
temporary pond environments, tadpoles may experience local densities of 25+/L (Alford
1999). Although tadpoles naturally encounter significant density-dependent factors
during their larval period, the objective of this study was to evaluate growth,
development, and survival relative to other environmental influences. Initial size (SVL)
of tadpoles did not differ among the 10 study sites (Kruskal-Wallis ANOVA, H = 13.133,
p = 0.157) and tadpoles in all enclosures were approximately stage 26 (range 24 — 31)
(Gosner 1960). After fish and turtle predation eliminated all three enclosures at one site
(Eagle Pond) and single enclosures at two other sites (Lawrence Lake Marsh and Jackson
Hole Outflow), the tops of the enclosures were screened to exclude predators and retain
froglets as they neared metamorphosis. Since all tadpoles were lost at Eagle Lake, this
site was omitted from all data analyses.

Tadpoles were measured (SVL, mm) and staged in the field at approximately
weekly intervals (5 - 7 days). Measurements were made in the field using a plastic petri
dish marked with a 2 mm grid, which minimized handling of the tadpoles. Upon
reaching metamorphosis (forelimb emergence, stage 42, Gosner 1960), tadpoles were
released at each site. Tadpole response variables included (1) growth rates (mm/day;
determined at weekly intervals), (2) development rates (stage/day; determined at weekly
intervals), (3) total grth (mm; length at metamorphosis — length at beginning of study),
(4) length of larval period (number of days from beginning of study to metamorphosis),

(5) size at metamorphosis (SVL at metamorphosis), and (6) survival (proportion

141

surviving to metamorphosis, adjusted for tadpoles lost to known sources of predation). In
addition, an overall measure of the rates of growth (mm/day) and development
(stage/day) was calculated as the mean increase in length (or stage) per day over the
entire larval period. This study was approved for ethical appropriateness and humane
handling of animals by the All University Committee on Animal Use and Care at
Michigan State University (permit #77585).

Statistical Analyses

All variables were assessed for normality by evaluating goodness-of-fit tests
(Kolmogorov-Smimov procedure) and normal probability plots (Zar 1998). Variables
not meeting assumptions of normality were transformed prior to statistical analyses.
Most variables met assumptions without transformation, but the following variables were
log (x+1) transformed: chlorophyll a, NH4-N, TDP, SRP, K+, Cl', and 8042'. Proportion
variables (i.e., canopy cover, tadpole survival) were arcsine transformed prior to analyses
(Zar 1998). Nonparametric statistics or other methods robust to the violation of
assumptions were used given the small sample size (n = 10 study sites, n = 3 enclosures
per site).

Correlations between species richness and habitat variables (water chemistry,
canopy cover) were evaluated using Spearrnan rank-correlation analyses (Zar 1998).
Similarly, correlations between tadpole response variables (total growth, size at
metamorphosis, length of larval period, and survival) and site-specific habitat/water
chemistry variables were also evaluated using rank-correlation analyses. Tadpole growth
rates and development rates were assessed relative to canopy cover, pond permanence,

and presence of fish predators using Profile Analysis, 3 special case of multivariate

142

analysis of variance (MANOVA) and repeated measures analysis (O’Brien and Kaiser
1985, von Ende 1993). For profile analysis of the influence of canopy cover on tadpole
response variables, sites were classified as either open canopy (<25% canopy cover;
Lawrence Lake March, Cobb-Pifer Marsh, Lux Arbor Pond 28), partial canOpy (26-74%
canopy cover; Jackson Hole Outflow, Douglas Lake Outflow, Loosestrife Pond,
Duckweed Pond), or closed canOpy (>75% canopy cover; Maple Pond, Wood Frog
Pond). All data analyses were performed using SYSTAT (version 8.0, SPSS, Inc. 1998)

and SAS (release 8.02, SAS Institute 2002).

Results
Resident Anuran Community

The species richness of resident anurans ranged from 2 — 8 among the 10 study
sites (Table 4.2). Fowler’s toads (Bufofowleri) and Pickerel frogs (Rana palustris) were
not detected calling or by opportunistic visual encounters during the field study at any of
the sites. The two morphologically identical species of Gray treefi'ogs (Hyla versicolor
and H. chrysoscelis) were combined into one composite treefrog species, as the calls are
difficult to distinguish, and temperature can affect the pulse rate and duration of their
calls, making them very difficult to identify in single species choruses (Harding 1997).
Spring peepers (P. crucifer) were detected at 6 of the 10 study sites; they were absent
from the four largest permanent ponds.

Species richness at the sites was not correlated with any of the water chemistry or

hydrology variables (Spearrnan Rank Correlation, p > 0.05). Canopy cover shading each

143

site varied from <5% - >90% (Table 4.1). Species richness was also unrelated to percent
canopy cover (rs = -0.056; p > 0.5).
Habitat and Water Chemistry Characteristics of Sites

The study sites differed significantly with respect to water chemistry and
hydrology (Table 4.3, Appendix 4.1). The study sites could be arranged along a
continuum representing wetlands that were primarily precipitation-fed to those that were
predominantly groundwater-fed using Mg2+ and specific conductance as indicators of
water sources (Figure 4.2). In this area, Mg2+ and specific conductance strongly covary
because of the influence of dolomite weathering on major ion composition, but Mg2+ is a
better indicator of groundwater (S.K Hamilton, pers. comm). Sites also varied with
respect to water level fluctuations, although Maple Pond was the only site that dried
completely during this study (Appendix 4.1). Bottle trap samples detected the presence
of predatory fish in six of the ten study sites, but only Eagle Pond had predatory fish
comprising >5% of the total community (Figure 4.3). Predatory fish species included
Bluegill (Lepomis macrochirus), Green sunfish (Lepomis cyanellus), Pumpkinseed
(Lepomis gibbosus), and Largemouth bass (Micropterus salmoides). Invertebrate
predators were the dominant predation threat for tadpoles at most sites. Water chemistry
variables differed significantly among sites (Table 4.3). Nitrate (NO3-N) was below
detection limits of ca. 0.01 mg/L in most samples, and was therefore omitted from further
analyses. Many variables also differed significantly among months during the sampling
period (Water Temp., NH4-N, TDP, and SO42; Kruskal-Wallis ANOVA, p<0.05) (Table

4.3). Three of those variables had significant linear trends over time. Water temperature

144

and ammonium (NI-I4-N) increased during the sampling period (water temperature:
3:0.441, p<0.001; NH4-N: r2=0.597, p<0.001) and $0.2“ decreased (r2=0.288, p=0.004).
Tadpole Response Variables

The total length of larval period was marginally different among sites (Kruskal-
Wallis ANOVA, H=15.129, p=0.057). Tadpoles at Jackson Hole Outflow reached
metamorphosis in the shortest amount of time (Table 4.4). At Maple and Wood Frog
Ponds, some tadpoles had not reached metamorphosis before the ponds completely dried,
and tadpoles at Lux Arbor Pond 28 had the longest larval period (Table 4.4). At
metamorphosis (stage 42, forelimb emergence, Gosner 1960), size (SVL) did not differ
significantly among sites (Kruskal-Wallis ANOVA, H=9.625, p=0.292). When tadpole
predation from known sources (i.e., fish and turtles getting into enclosures) was
considered and survival was adjusted for the remaining tadpoles at all study sites,
survivorship did not differ significantly among study sites (Kruskal-Wallis ANOVA,
H=12.212, p=0.142). All enclosures were eliminated from Eagle Pond, and that site was
omitted from further analyses of tadpole response variables. Survivorship was highest in
Cobb-Pifer Marsh, and lowest in the two lake outflow sites, Jackson Hole Outflow and
Douglas Lake Outflow (Table 4.5). Both growth and development rates averaged over
the course of the study differed significantly among study sites (Kruskal-Wallis ANOVA;
growth rate: H=16.640, p=0.034; development rate: H=15.682, p=0.047). Tadpoles grew
and developed most rapidly at Jackson Hole Outflow, and slowest at Maple and Wood
Frog Ponds, respectively (Table 4.6). Generally, growth rates (divided into roughly
weekly intervals) were highest at sites with warmer water temperatures, particularly in

early stages of the larval period (Figure 4.4).

145

There were very few significant correlations between habitat and water chemistry
variables during the months of the field study and tadpole response variables (Table 4.7 —
4.8). Chlorophyll a concentration in July was positively correlated with total growth
(r,=0.967, p<0.001, Table 4.7), size at metamorphosis (r,=0.833, p= 0.010, Table 4.7),
and overall growth rate (rs=0.783, p<0.05, Table 4.8). Canopy cover was negatively
correlated with overall growth rate (rs= - 0.700, p<0.05).

Tadpole Growth and Development Profile Analysis

The slopes of tadpole growth rates (mm/day) over the five-week study period did
not differ significantly among pond permanence categories based on Profile Analysis (F
= 1.03, p = 0.485; Figure 4.5a). Growth rates were relatively constant for the duration of
the study in all pond permanence categories (F = 6.81, p = 0.074). There was a
significant difference among pond types in growth rate between week 1 and week 2 (F =
7.05, p = 0.027); however, no other contrasts of growth rates between successive weeks
of the study were significant. Similarly, canopy cover was not a significant influence on
tadpole growth rates (Figure 4.5b). The slopes of the profiles of tadpole growth rates
over time did not differ with respect to the three categories of canopy cover (F = 0.92, p =
0.543) and growth rates were relatively constant over time (F = 7.35, p = 0.067).
Although grth rates did not differ over the entire duration of the experiment among
canopy cover categories, there was a canopy cover effect on growth rate between the first
two weeks (F = 6.67, p = 0.029). The presence of fish did not affect the slope of the
tadpole growth rate profile over time (Figure 4.5c; F = 0.29, p = 0.871). However,
grth rates of tadpoles did decrease over the duration of the study both in the presence

and absence of fish predators (Figure 4.5c; F = 10.64, p = 0.021).

146

The slopes of the curves of development rate (stage/day) over the five-week
period of the study did not differ significantly with respect to pond permanence category
(Figure 4.6a; F = 3.2, p = 0.059). In all pond types, development rate was relatively
constant over time (F = 2.98, p = 0.197). Although there was no significant time effect
overall, when each interval between sampling periods was considered separately,
development rates differed between week 1 and 2 among pond types (F = 8.78, p =
0.017). Canopy cover also had no significant influence on the slope of tadpole
development rate curves over time (Figure 4.6b; F = 1.33, p = 0.349), and development
rates were constant over the duration of the study (F = 3.83, p = 0.150). Development
rates were slower in closed canopy sites in the interval between weeks 1 and 2 (F =
11.29, p = 0.015), but no other time intervals showed significant differences among
canopy cover classes. The presence of fish did not affect the slopes of the development
rate curves over time (Figure 4.6c; F = 4.43, p = 0.089). However, development rates
differed significantly over the duration of the study both in the presence and absence of
potential fish predators (F = 8.12, p = 0.033). There was a significant difference in
development rates between weeks 1 and 2 with respect to the presence of fish predators

(F = 11.66, p = 0.011).

Discussion

Spring peepers utilize a wide variety of wetland types for successful breeding,
including both ephemeral and permanent ponds. However, this species is typically less
abundant in ephemeral ponds that dry each year relative to other anurans (e. g.,

Pseudacris triseriata, Western chorus frogs) and more abundant in ponds with longer

147

hydroperiods (Skelly 1995b, 1996). In this study, Spring peeper growth and development
rates were lowest and length of larval period was longest in the two ponds with the
shortest hydroperiods (Wood Frog Pond and Maple Pond). On the other end of the
hydroperiod continuum, however, tadpole response variables also did not respond
favorably to conditions in the most permanent ponds, with low growth rates, long larval
periods, and low survival at permanent ponds and lake outflows. Wetlands with
intermediate hydroperiods appeared to represent the balance of physical, chemical, and
biotic conditions that optimized Spring peeper growth, development, and successful
metamorphosis.

The two temporary pond sites with the lowest growth and development rates also
had closed forest canopies shading virtually the entire surface of the breeding pond.
Increased canopy cover indirectly influences tadpoles by reducing other variables
important for growth, development and metamorphosis (e. g., Wassersug and Seibert
1975, Harkey and Semlitsch 1988, Werner and Glennemeier 1999, Skelly et al. 2002).
Shading of breeding ponds also affects tadpole performance, primarily as a function of
temperature and resource differences (Werner and Glennemeier 1999). Water
temperature exerts a particularly strong influence on tadpole growth and development
(e.g., Lucas and Reynolds 1967, Harkey and Semlitsch 1988). Typically, Spring peepers
are only found breeding in open canopy ponds (Harding 1997, Skelly et al. 2002).
Indeed, canopy closure has been hypothesized as an important factor contributing to the
distribution of many species of anurans in Michigan (Werner and Glennemeier 1999,
Skelly et al. 2002). Many factors probably contributed to the low rates of growth and

development and long larval periods at these sites. The predominant influential factor

148

was most likely water temperature, but other factors such as canopy cover, relatively low
oxygen availability, and sestonic chlorophyll a also played a role. Growth and
development are both influenced by temperature, however, development rates are lower
than growth rates at low temperatures, such that tadpoles could grow more with each
developmental stage and ultimately metamorphose at a larger body size (Smith—Gill and
Berven 1979).

Although there were very few significant correlations between habitat/water
chemistry variables and tadpole response variables, sestonic chlorophyll a was
significantly positively associated with most tadpole response variables. Chlorophyll a is
an indication of resource availability. Spring peeper tadpoles ingest both suspended and
attached algae, as well as occasionally supplementing their diets with detritus (Alford
1999). Higher resource levels in the form of more abundant (or higher quality, a variable
not tested in this study) food sources leads to faster growth and development and larger
metamorphs (Steinwascher and Travis 1983, Kupferberg et al. 1994, Kupferberg 1997).

Although the three factors evaluated in the profile analysis (pond permanence,
canopy cover, and presence of fish predators) have been previously reported to influence
tadpole performance (e.g., Kats et al. 1988, Skelly 1996, Loman 2002, Skelly et al.
2002), there were very few significant relationships between those variables and tadpole
response variables in this study. These results are likely due to small sample sizes and
thus reflect the lack of statistical power to empirically determine significant relationships.
The small sample size and unbalanced design also precluded the calculation of
interactions among these three factors. A qualitative examination of the tadpole grth

and development profiles indicated that growth and development rates in the early

149

portion of the larval period were reduced in temporary ponds with closed canopies, and
development rates were higher in ponds with fish predators. These patterns, although not
statistically significant, can be interpreted with respect to their biological significance.
Growth and development of Spring peepers varied depending upon habitat factors
at each of the study sites in southwestern Michigan; however, a few issues need to be
considered in the evaluation of these results. The results of the statistical analyses
coupled with qualitative examination of patterns in the data indicate that I did not have a
sample size large enough to have enough power to detect significant differences in rates
of growth or development using the repeated measures techniques employed here.
Maxwell and Delaney (1990) suggested that these statistical techniques are only
appropriate when sample size is larger than conditions often allow in ecological systems.
If a were set higher (i.e., 0.10) to compensate for low statistical power (Stevens 1992),
differences in grth and development rates among sites over the course of this
experiment would be significant. This indicates the need to evaluate ecological
experiments thoroughly to address issues of both statistical and biological significance.
Although not quantified directly, predation appears to be a factor excluding
Spring peepers from permanent ponds such as Eagle Pond and Jackson Hole Outflow.
Even when predators were excluded from field enclosures, tadpoles in permanent ponds
with predators suffered relatively high mortality and slow rates of growth while
development proceeded relatively rapidly (e. g., Lux Arbor Pond 28, Douglas Lake
Outflow). These tadpoles may be responding to chemical cues in the water from
potential predators (Smith and van Buskirk 1995, van Buskirk 2000). Spring peepers in

other semi-permanent or permanent bodies of water had high survival and relatively rapid

150

rates of growth and development (Cobb-Pifer Marsh, Lawrence Lake Marsh and Jackson
Hole Outflow). Other biotic interactions are also very important in understanding the
larval biology of Spring peepers. This species is considered by many to be a
competitively inferior species (i.e., Morin 1983, Skelly 1995a, 1995b). However,
previous experimental results have been somewhat equivocal. While this study did not
evaluate the effects of any predation or competition interactions, growth and development
appeared to be complex processes that responded to potential biotic factors in addition to
the habitat variables that were measured in this study.

In this study, growth, development and survival were optimized in semi-
perrnanent wetlands. Complex factors (perhaps related to perceived predation threat and
other interactions between biotic and abiotic factors) reduced Spring peeper growth and
survival in permanent ponds, while the tadpoles were unable to complete development
and metamorphose in temporary ponds before they dried. These factors (predation and
pond drying) have been shown to be important in the distribution and abundance of this
species (Skelly 1995a, 1996). The length of their larval period was also optimized in
ponds of intermediate hydroperiods; metamorphosis occurred later in both permanent and
temporary ponds.

Anuran larval dynamics and metamorphosis have been topics of intensive
research for decades. While models that incorporate various factors in the larval
environment have been developed to predict timing of and size at metamorphosis (i.e.,
Wilbur and Collins 1973, Collins 1979, Werner 1986), the results of this study and others
in recent years indicate that tadpole development and metamorphosis are complex

processes influenced to varying degrees by a wide variety of factors specific to breeding

151

sites. Spring peeper tadpoles in southwestern Michigan responded to habitat
characteristics, water chemistry, and biotic factors by altering their growth and
development throughout their larval period. While this species can complete
development and metamorphose from wetlands at both ends of the hydroperiod
continuum, a wide variety of factors (such as temperature, canopy cover, resource
availability, and fish vs. invertebrate predators) contributed to optimal growth and
development at sites of intermediate hydroperiods, where tadpole survival to

metamorphosis was the greatest.

Acknowledgements

Financial support for this project was graciously provided by the Michigan Chapter of
The Nature Conservancy, and NSF Research Training Grant from Kellogg Biological
Station. Funding was also provided by the US. National Science Foundation (DEB-
9701714). I. A. Genet, K.B. Kiehl and W.R. Hilgris assisted with field data collection.
M. Machavaram and S. K. Hamilton assisted with laboratory analyses of water samples.
C. A. Lepczyk and W. Gordon provided valuable comments on earlier drafts. All
animals were handled in accordance with the All University Committee on Animal Use

and Care permit number 77585.

152

much Emmanutso— 98 .335 «ESQ: .32“ 53852 £th wEuBoE Co>oo ooh .xemmvv 6:2 canoe
588288 $05an 5 5E 85308 Ozormmo: .23. E Coho .83 “260 .3050 “520%

mamm— fisofiowfiq can .voomqflmfism £9.33 :85 .5335 @0335 mecca @635 5 580.5 E8305 3:533.
nomuaucsfi Mo 8033 3:038 Eczema

2% 33m 05 8 E0895 £826 womb eo>oo 98$
woficaeuwsaofieo 3wa Banana. mOmD Eoc BEEumo 835208 gamma. Ea owning.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

304:3

@d 92 asfiEEEE seefifieeaeem seem .2238? 23%.? as: 55.38
30$».wa

9&3 mdv 888952: Eocnccom Sodom . even; :25 2&0va 28m outamomooq
30%;“

as e. : essenceefima answered e88 .83 zaavfi mm Fee 82 *3
552.3 an:

an 3 3m enteeeeihea E825; sees :80 .33 213.? 8.3 8:833
308mg Bongo

8.3 Meme 22982525 E0553 93 seem zeemmde 8.3 Ewan
303nm”

Away finm 8.290th EonmESQ-_Eom 620m Zeammdv 95m Bufiosfl
3.268

6; a wee eaSQEEEmE Eggnog seem zeemmae Bee seem

. 83 2533 3850

ANMNV adv ofifioflgfiEmE «accustom mew—=3 EEO Jmonom Zef Wmv 202 5862.
Bommmaw

AN. 3 #5 0353.35 EBOQEQH 620m Zea adv 95m 0332

. 3&2wa
C . : NMQ 0wmh£oflo>£ 159th 50.5% .Ze—NMNV “50% mean @003
WEED .3250 viafifieD .5335 mouaoawfinem unen— mxouaeu «nuances: $035280 35

 

 

5w322 8883538 E 8% beam 2 05 mo 83388350
3‘ 033.

153

 

$0533
momooqm

 

3:53.83u
SEN

 

33.583
38%

 

£35233
32.339
”raw

 

3882b“?
\Lo~ou.~mxm>
a}:

 

X

328.9653
c\:m

 

.Em38\o\am

 

£3.33
38%

 

3&an 38%

 

$593.6
wwxuahammk

 

X

XXX

XXX

X
X
X

Sutmflh
flxucgum

 

X

X

X

X

X

8.39: a
98%

 

ES.—
MP:—

95>?

ES.—
2%:

3 ES.—
.892
a:

 

caca—
£38254

 

£232
8:5
09:23.5

 

Bag—.6
2c:
.5832.

 

each
033—

 

28.—
6095.25

 

335:0
8:3
3.939

 

is:
homun—
.25

 

8.02am

 

154

 

 

 

 

.womoomw BwEm
a mm @8550 can 3:588 803 wmmuwowbfi E 23 x3832»; Sam .85 38m 20m 8 8:63:80 EMS—888 $.55 EoEmoM
N4» 2an

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

155

 

323
85 3.3 as: 83 Ana: 2.»: as 6.8 8.3 9.3 8:50:28
3: mmm we: 0.2m 38 8m 3% 32 3?. as 0583::
335 A38: 63%: 8.5: $.25 :25 A38: 32$ :35 633 $35
8.32 3.3.3 8.32 3.88 38mm 2.83. N32 2.5a 8.2% 2.8m 32:32::
95.8 $98 83: 23.8 5.8 33V 3:: 83: A38 $98
as $5 3..” Saw a.» E N2 N3 8.» 8s mats...
A38 God 3.8 a»: 62V and 5.: $8 and 868 SE:
8.2 gm m3 Rm 9% on: NS. 2.2 2.x 8.2 a .5
83¢ $2: $3: :23 63: 83: 80.3 3%: 9&8 A85 ax;
M32 ”0? v32 N2: $.91 8.2: one: 8.: 25: N5. 3 .am 6::
8on A2: A38 96.3 $2: cm: 3.8 3%.: fine God Shay
N3 Rd 3: 3.2 8.2 Mad «2: m?“ 8.2 m _ a 09...:
A8,: Ga: 8mg avg 2th A38 3:: 69.: A86 82V Gov
3.2 2 .2 0:: 2.2 5.8 2.2 03: 2.: 9.2 8.2 989 833
Euca— mN tack :93: 36530 33.359 :93:
«PE .23.— ..2...< ES.— 8—«4 2:: each ES.— 3—«4 .3?—
183 can: n.1— oabuomcox— 3.3.951— :cioan. 23H 633—25 aim—8n— A390
.28.ovm.......: woodvdz... .dem: .modvfl. m<>oz< 333-35% 353 E 3865 2a 8%. macaw Sauna?

cobwmw 35 332.5, 98 .86 some Sm Coho .Bmv :38 50858 82m> .9.» u a 98:3 .5 Eu 3088 833“”; =a 3m 8383
m u a .33. I :83): “v2.8a BPS. 98 wfivoofi 5.3.8 waist 35:08 @8388 .35 33m 22m 3 833:; Cam‘s—one 58>?
mé 033‘

 

 

 

 

 

 

 

 

 

 

 

 

 

5.8 528 A3. 8 2 _8 Go. 8 A28 5.8 5.8 55.8 598 95.8
5.2 5.5 55.5 5.: 5.5 3.2 5.5 55.2 5.2 OS -65...
588 $8 3.8 65.: 65.8 55.8 c _8 5.8 55.8 5.8 228
o? 55 v: 3.: 2.5 2: 5.2 5.5 3.2 5< -63.:
628 55.8 $8 :58 A38 528 598 5.8 55.8 588 35.8
35 35 So «3 55 3o 28 o; 55 38 +2.5...
95.8 5.8 $8 5.8 $8 :58 A88 5N8 588 5.8 35.8
NS 52 a? 8.5 85 Re 52 52 NS 2.5 521......
:58 A58 598 an: 5.8 55.8 55.8 5.8 55.8 55.8 328
55.5 5.5 5.5 8.2 5.2 8.8 _ 2.8 .8: 5.8 2 .5 $523:
$8 5.5.8 5.8 25.8 55.8 5.8 80.8 95.8 3.8 35.8 35.8
55.2 2.5 8.5 $5 235 m8. a? 8.? 5.55 5.2 +n8...
95.58 3858 5.58 588 :58 88 A58 55.8 55.8 698 358
535 5.5 5.8 5.2 2.5 8.5 3.5 8.55 3.2 5.5 25...:
$28 55.8 5.8 55.8 528 3.88 85.8 3.8 23.8 5.8 358
8.25 8.2 5.5 Rd 5: 5.55 8.5 2.5 :8 8.2 25......
5558 an. 8 538 5.8 A: . 2 8 638 :68 28.8 EV. 8 55.8 Adz 58
5.8 5.5 5.8 8.5 5.2. 5.3 53 2.5 5.5 :l 2.42

 

 

 

 

 

 

 

 

 

156

Table 4.4
Mean length of larval period (days) for Spring peeper tadpoles at field study sites. Length
of larval period was defined as the number of days between the beginning of the
experiment (2 June 2000, day 0) and the day the tadpoles metamorphosed (stage 42,
forelimb emergence).

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Site Days
Mean (std. error)
Cobb-Pifer Marsh 36.7 (1.5)
Duckweed Pond 38.8 (3.1)
Lawrence Lake Marsh 30511.2)
Loosestrife Pond 33.6 (0.6)
Lux Arbor Pond 28 48.1 (3.1)
Wood Frog Pond 46.3 (2.2)
Jackson Hole Outflow 26.7
Douglas Lake Outflow 41 .4(5.9)
Maple Pond 46.0 (1 .5)
Table 4.5

Mean survivorship of Spring peeper tadpoles at study sites during field study. Survival
represents the mean survival of all enclosures at a site (n=3 for all sites except Lawrence
Lake Marsh and Jackson Hole Outflow, where n=2). Enclosures where fish or turtles
completely depredated tadpoles were eliminated from further analyses (all enclosures
were eliminated from Eagle Pond).

 

 

 

 

 

 

 

 

 

 

 

 

 

Site Survival

Mean (std. error)
Cobb-Pifer Marsh 0.726 (0.175)
Duckweed Pond 0.293 (0.136)
Lawrence Lake Marsh 0.639 (0.139)
Loosestrife Pond 0.300 (0.200)
Lux Arbor Pond 28 0.178 (0.097)
Wood Frog Pond 0.500 (0.058)
Jackson Hole Outflow 0.150 (0.150)
Douglas Lake Outflow 0.161 (0.061)
Maple Pond 0.463 (0.067)

 

157

Table 4.6
Overall rates of growth (mm/day) and development (stage/day) for larval Spring peepers
at field study sites in southwestern Michigan averaged over entire larval period
(enclosure stocking to metamorphosis). Both growth and development rates differed
significantly among sites (Kruskal-Wallis ANOVA, p < 0.05).

 

 

 

 

 

 

 

 

 

 

 

Site Growth Rate Development Rate

mm/daL(SE) stage/dayLSE)

Cobb-Pifer Marsh 0.228 (0.009) 0.479 (0.015)
Duckweed Pond 0.213 (0.014) 0.446 (0.030)
Lawrence Lake Marsh 0.225 (0.030) 0.561 (0.019)
Loosestrife Pond 0.204 (0.008) 0.457 (0.039)
Lux Arbor Pond 28 0.171 (0.021) 0.415 (0.012)
Wood Frog Pond 0.169 (0.007) 0.352 (0.008)

Jackson Hole Outflow 0.302 (0) 0.710 (0)

Douglas Lake Outflow 0.163 (0.008) 0.421 @077)
Maple Pond 0.151 (0.002) 0.355 (0.005)

 

 

 

158

 

 

datum RES were 028 35308 3:0 33 ~38 sacs—«Um
dogma-.80 95:88 .9 owfim an 3338 mo AEEV sumac-Av

.328 025 8 5:805 8 82 858—05 :80 8 883:8 Juana-8:508 ES. 358 mo @85on Bob 333mm
.AoocowBEo 98:88 .mv owfimv 38:80:53.8 Ea: beam Eon mo @85on Bow wane QC 538:}

633m mo wgmmon 8 5mg— I 88:90:52: as 5mg: 35% Bow m0 838 S>o Swan: 5 omega _Soh

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

$3- $3- 23- $3- :3 33- 83- 883. am
23 83 83- 83- 83- 83- 23 $3 -5
83 83 83 83 $3- 83 83 t 3 +3
83 t 3 83- :3- 83- $3- :3 8 _ .o mz
883 $3 83- E3- 83- $3- $3 83 mm:
83- 83- $3- 83- 83- :3- 83- 23 :8
23 83- 83 83 83 $3 $3 3.0- me.
.33 S3 83 £3 83- 83 83 83 $5
$3 $3- $3- 83 $3- 83 :3 83. 2.52
83- :3- 183. .83- 83- 83- :3 83 8:88:00 88%
$3- :3 R83- 83- $3- $3- 83 3.3 £5.32
23- 83- 83- 83- $3- 23- 23- 83- E
33 ~03 8_ .o 83 83- :3- .83 83 a EU
83- 9.3- 83 83- a: .o- 3.3. 83- t 3- 885% No
83- 83 83 83- $3- $3- $3- $3 5.85 B835
33 23 M83- 83- 83- 83- 83- 33. 233888. 533
33- mm 3 .o- 83 R3- masou €25
bi. 2:: 33. 05; DE. 2:; .23. 0:3.
255 m Nassau 8ch 285
$8888»: a 3m MESA: 335m :33 8 583 5320 .38.

 

 

 

 

 

.ASdVE modth 8658mm 3 @8885 E oofiocEwE A5 358508

8:38.50 diam 52:8on 2m $252 32353 0&8qu 20333 28 880888 98805 583 28 55m: 5053 macaw—chow

5v Baa-r

159

Table 4.8
Correlations between habitat and water chemistry parameters and overall tadpole growth
and development rates averaged over entire larval period. Values are Spearman Rank
Correlation coefficients (rs). Significance in indicated by asterisks ( p<0.05 Tp<0. 10).

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Growth RateI Development Rate2
June July June July
Canopy Cover3 0700* 0650‘
Water Temperature 0.450 0.333 0.517 0.500
Dissolved Oxygen 0.367 0.200 0.617T 0.383
O; Saturation 0.267 0.200 0.467 0.383
Chl a 0.326 0783* 0.209 0.617
PH -0. 100 0033 0.167 0.233
Alkalinity 0.100 -0150 0.467 0.250
Specific Conductance 0.067 —0.233 0.433 0.150
NH4-N -0.033 0.433 -0.133 0.250
SRP 0.059 0.310 -0.243 0.050
TDP -0.183 -0100 -0.533 -0433
Ca2+ -0.083 -0.383 0.200 -0.017
Mg” 0.283 0.167 0.517 0.500
Na“ 0.300 0.250 0.550 0.533
K* -0.067 -O.167 -0217 0200
CI' 0.217 0.283 0.433 0.500
so; -0.383 -O.467 -0050 -0. 100

 

 

 

 

 

 

Mean daily growth (length) averaged over course of entire larval period (mm/day).
2Mean daily development (stage) averaged over course of entire larval period (stage/day).
3 Canopy cover was only measured once during larval period.

160

Figure 4.1

Location of study sites in southwestern Michigan All study sites were
natural wetlands in Kalamazoo and Barry Counties.

 

 

 

Cobb-Pifer
Marsh

K /

/" Loosestrife Pond
Kal r‘
coamazoo .6‘1 4 Douglas Lake Outflow

 

 

 

 

 

/

Fort Custer Recreation

5 Area
1. Maple Pond

2. Wood Frog Pond

3. Jackson Hole Outflow

4. Eagle Pond

5. Duckweed Pond

 

 

 

161

Figure 4.2

Mg“ concentration and Specific Conductance of the 10 study sites. Both
variables were used as an indicator of water source, with sites having higher
Mg2+ concentrations and Specific Conductance more predominantly
groundwater-fed, and site with lower values for both variables more
precipitation-fed Values plotted represent mean values of measurements
taken monthly during the anuran breeding season (n=5) and their associated
std errors

 

25 450

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

‘ Specific Conductance A
0M3 I 100%
20‘ I; 1. v
i 350 3
2:; = - I ““3
+ L.
"$0 10‘ .1: fl 200%
“15055
5~ I: ”100%
I] 650
Oiv'bi'bl‘b'b brb'V'4V71‘0
a”? 39° 3°° 96% s°° 99° 99° 48¢ 6 ° ‘ o
é o3? \c 0 02> 0 go o 0
"5 ‘6 ‘08 5 9 0 <9 \fi' \° °
3‘ 5“ Y5°° ~17" <9 °° 0 ~29
6‘ ° 6° 9 62
06 «3 V5)- 9’ $3.0 Q9
\3’ 33*“ 9°

162

Figure 4.3

Relative abundance of invertebrates and fish sampled by bottle traps in 10
study sites. Predatory fish were present in 6 of the 10 sites, but comprised
>5% of the total community in only one site (Eagle Pond). Species of
predawa fish included Bluegill, Green sunfish, Pumpkinseed, and
Largemouth bass.

 

 

 

 

 

100% low
”A: % DDrptera
80% _ a IColeoptera
2 50% _ * -‘ ' DCrustacea
; W..- H: A , BMW
3 30% — _ , 3m“?
20% _ - Plamna
10% _ . " ’ INonpredatoryfsh
0% _ 1 1— ‘ I - .IPrechtuyF'sh
‘9 b a .3
383°$§ff§§
.1? 8‘09 Q 3 49° 9% . j \o
3‘ é 7" " 10° 09
Q $0 \Py 0" § @
0° V #9 Q
9 S

163

 

E8 ”-6 60:3 .23... 05.88

 

 

 

 

n .202. V 75.85 m .8603— N .23: _ .28:—
P p _ p . o
- _d
.......... - 3
II OINHHHHHHHHHHHHHUM IM.°
1; ._ [\UK 1111111111111111 \) D
x ................. W W m
1. P M.
., . i m..- m
x - 3
3.825;?-
562 0x308233.-o-- :32menoU.-X-- 4 ed
0x338.-oi 0.02:8x00..i0--
m. 28 $384+ 28 6833+ - 3
28 men 863 lol Boa 0%: ll
wd

 

 

164

.062 55 82 8580988 083 508 was $388 00:00:00

53, 00:0. ”063 :05 080% 003 00880988 e08? 508 2055 008m 08 235$ :03 55, 0.25 2030::

330 w I 3 >300? 30383850 3 380008 0.83 822 5380 30% beam 20: an "”2863 no 838 5380
3. 2:08

Figure 4.5

Growth Rate (mm/day) profiles of Spring peeper tadpoles over five weeks of

field study related to (A) Pond permanence, (B) Canopy cover, and (C) Presence
of fish predators.

0.45

 

89
9%

 

.0
wk

.9
"l

Growth Rate (mm/day)
S
u:

Q
—
u;

 

 

 

 

 

.°
u
U;

 

 

.°
N .
V! .

Growth Rate (mm/day)
p o
N b)

.°
. o—I
. U!

9;:
+
if

 

 

4’" Partial "°" Closed

 

 

.689
a»

 

Growth Rateo(mm/day)
'N

ope
.fl‘

 

+Fishlees
O Y , r #—

 

week 2

week 3

165

week 4

weeks

Figure 4.6

Development rate (stage/day) profiles of Spring peeper tadpoles over five

weeks of field study related to (A) Pond permanence, (B) Canopy cover, and
(C) Presence of fish predators.

 

stage/day)

.° .°

09 «L ""
>

.°.°
at‘l

 

9 .
a

Development Rate (
p c
or U:

 

.09
--:N

 

-°- Temporary 4*" '—Permanent +Pa'manent

O

 

 

r

w

 

9'09“;

‘9 .

Development Rate (stage/day)
o p o o o o o
A 0:

 

be

 

O
y.—

 

 

 

 

 

0.1‘ + fish '°" fishless

 

 

 

week 1 week 2 week 3 week 4 week 5

166

Summary and Conclusions

Amphibians have been the focus of a surge of research activity since initial alarms
were sounded in the early 19903 regarding declining amphibian populations worldwide.
Even though specific causes for declines remain unknown, awareness of these organisms
and their habits and habitats has been raised. With establishment of many state (e. g.,
MFTS), regional (e. g., Marsh Monitoring Program), and continent-wide (e. g., NAAMP)
amphibian monitoring programs, annual data are being collected that will contribute to
the understanding of species’ distributions and trends in long-term population dynamics —
provided data are collected continuously over several generations of the frog and toad
species.

Michigan is home to 13 species of frogs and toads, with 12 occurring in the
southern lower peninsula targeted in this study. I did not detect any major declines for
any of these 12 species that warrant urgent or immediate remediation. However, these
analyses represent only the first seven years of the Michigan Frog and Toad Survey data,
essentially only a snapshot of long-term anuran population dynamics.

Site species richness fluctuated both among different wetland types and among
years, but there were no consistent patterns. Only two species showed significant trends
in site occupancy from 1996 to 2002; the proportion of sites occupied increased for Rana
catesbeiana and decreased for R. palustris. Five species had significant trends in call
index values over time, although these relationships explained little variance in linear
regression models (Table 1.7). Rana sylvatica, Hyla versicolor/chnzsoscelis, Pseudacris

crucifer, and R. clamitans increased calling intensity over the seven years, while Bufo

167

americanus decreased calling intensity. All statistically significant population trends for
these species were small in magnitude and explained relatively little variance in the data.
However, trends that were statistically significant should be further evaluated as MFTS

data are accumulated annually to determine whether they are biologically significant for

long-term population dynamics.

I used multiple analysis methods to evaluate population trends for several reasons.
First, amphibian call surveys were based on protocols of the Breeding Bird Survey
(BBS), and data analyses have been adapted from BBS methodology (Geissler and Sauer
1990). Second, the Wisconsin frog and toad survey has almost 20 years of data, and I
used the methods developed by Mossman et al. (1998) for Wisconsin survey data to
facilitate comparisons of population trends with those in neighboring states. Finally, until
enough data have been gathered to evaluate changes in abundance and distribution over
multiple generations of the longest-lived Michigan species (R. catesbeiana) and crucial
empirical relationships relating calling intensity to actual population size have been
developed, these data need to be evaluated using multiple approaches. If multiple
methods indicate significant changes in distribution or abundance, then confidence in the
results increases. Call surveys provide valuable information, but data must be analyzed
and interpreted within the limitations of the survey methodology. These limitations
include data reliability, inter-observer variation, dependence on volunteers for long-term
data, subjective selection of survey routes, and poor understanding of how call index

values are related to actual population size.

The limitations related to volunteers involved in the survey were addressed in

Chapter Two. Volunteers were acceptably reliable in species identification, but were less

168

so in assignment of call index values. Volunteer background and prior experience had
little influence on data quality and consistency, which differs from the effects of observer
experience seen in the BBS (Sauer et al. 1994, Kendall et al. 1996). These results
indicate that the most robust analyses of population trends using MFTS data will use
presence/absence (i.e., detection/non-detection) data, such as the site occupancy and route

frequency regression methods I presented in the first chapter.

Volunteers are highly committed to the MFTS because of their interest in
conserving native amphibians. Volunteers commit to collecting data on their route for a
minimum of three years before establishing a new route (L.G. Sargent, personal
communication), and active volunteers who participated in observer evaluation indicated
that they intended to collect data for an average of 10 years. However, those who
participated in the observer evaluation process may represent the most conscientious
volunteers, and conclusions about data quality and commitment from volunteers who did
not participate in the observer evaluation study remain unknown. Training materials and
workshops offered by the MDNR are important to prepare volunteers for call surveys and

will be continued given available funding.

Volunteer-based anuran call surveys have the potential to provide long-term data
on anuran distribution and abundance over broad geographic scales. Amphibian
populations, however, fluctuate dramatically under natural circumstances (e.g., Semlitsch
et al. 1996, Alford and Richards 1999), and the challenge is to separate short-term
patterns from long-term trends that are outside the natural wide range of variability.
Shirose and Brooks (1997) suggested that amphibian monitoring programs should last at

least as long as the generation time of the longest-lived species (up to 15 years for Rana

169

W‘”—"——_

catesbeiana in Michigan; Harding 1997). It is generally assumed that at least 10-15 years
of data are needed from a monitoring program to identify patterns that are biologically
meaningful at the population level (Mossman et al. 1998). The MFTS currently has eight
years of data (seven of which are analyzed in the preceding chapters). However, this is
still not long enough to identify population trends that are meaningful over the long run,
as indicated by the relatively few and inconsistent patterns of distribution and abundance
identified in Chapter One. The MFTS annually collects valuable data, and the value of
this data as a contribution to regional management and conservation of amphibians

increases as each year of data is added.

One of the most significant limitations of call survey data is the poor
understanding of the empirical relationship between call index values and actual
population size. This relationship actually needs to be addressed at several levels to fully
understand how an index of the abundance of calling males is related to population
parameters. The relationship between calling intensity (i.e., call index values) of
breeding males and larval abundance should be evaluated to determine whether calling
intensity of breeding males is a reliable predictor of successful reproduction. The
relationship between larval abundance and the production of metamorphs needs to be
determined to assess whether the often extremely high densities of tadpoles are reflected
in numbers of juveniles dispersing into terrestrial habitats. The relationship between the
numbers of metamorphs produced each season and the number that are subsequently
recruited into the breeding adult population also needs to be determined. The sex ratio,
which probably differs among species and among years (e.g., Green 1997a), also needs to

be determined so that call index values of breeding males can be related to actual

170

-. '..,d

 

population size. Currently these empirical relationships are unknown, and assumptions
regarding these issues are built into monitoring programs need to be addressed and tested.
Data analyses such as those presented in the first three chapters will become more

valuable once such relationships are empirically established.

Anurans respond to factors within habitats as well as at the landscape scale
encompassing multiple habitats used during different portions of their life cycles. All of
Michigan’s anuran larvae are obligately aquatic. Habitat factors such as food availability,
hydroperiod, canopy cover and predators affected tadpole performance in natural
wetlands in southwestern Michigan, as reported in Chapter Four. Growth, development,
and survival of P. crucifer tadpoles were highest at sites with intermediate hydroperiods,
partial canOpy cover, and few (if any) fish predators. More detailed manipulative
experiments are needed to identify strongly influential factors in the larval habitat that are

related to recruitment of adults to the breeding population.

Landscape factors, particularly land cover types, influenced the presence and
abundance of anurans at wetland breeding sites, as discussed in Chapter Three.
Generally, land cover types indicating habitat alteration or loss (i.e., roads, urban)
negatively influenced presence and/or abundance of anurans while variables representing
important foraging and breeding habitats (i.e., open land, wetlands) represented positive
influences. Contrary to my expectations, no consistent negative associations with
agricultural land were found in this study; agriculture both adjacent to wetland breeding
sites and in the surrounding landscape was often positively associated with the presence
and abundance of anurans. In this study, “agriculture” included all land that was

intensively managed for vegetation production, including row crops, forage crops,

171

T

r... :6 .a—.—..——
I -

 

orchards, vineyard, and nurseries. While row crops are probably quite inhospitable to
anurans, the other types of agriculture may provide usable foraging habitat or dispersal
corridors, or at least do not inhibit anuran presence or movement through these areas.
Thus, future studies should separate agriculture into subunits (i.e., intensive row crops vs.
other types of agricultural land uses) if cr0p rotations within the time frame of landscape

and amphibian survey data allow.

Anuran presence was not more strongly related to factors at the 100 m buffer
scale directly adjacent to survey sites than the 1000 m buffer scale representing their
dispersal potential. Anuran abundance, on the other hand, was more strongly related to
land cover variables at the larger 1000 m buffer scale. Anurans use a variety of habitats
during their complex life cycles, and the overall species assemblage appears to benefit
from a mosaic of habitat types in the landscape surrounding breeding sites within the
distance most individuals are likely to disperse. Others have also indicated the
importance of the landscape surrounding anuran breeding habitats for the management
and conservation of these organisms (e.g., Hecnar and M’Closkey 1996, Semlitsch and

Bodie 1998).

The associations 1 found between anuran presence and abundance and land cover
variables provide valuable information for the management and conservation of these
species. Population trends determined from large-scale monitoring programs (e. g.,
MFTS) can be combined with such habitat and landscape evaluations to obtain a more
comprehensive understanding of the relationships of anurans to their potential habitats.
In order to fully understand amphibian population dynamics and properly manage and

conserve species and their habitats, we need to synthesize research evaluating influential

172

factors at multiple scales (e. g., within habitat, landscape) for a variety of life history

stages (e.g., larvae, juveniles, adults) over long time periods (at least 10 years).

173

APPENDIX

174

Appendix A

Survey questions presented to active volunteers on mail questionnaire. Responses were

edited and evaluated as described in the text of Chapter Two.

 

10.

11.

When did you become involved with the Frog and Toad Survey (1996, 1997,

1998, 1999, 2000, 2001)?

How many years have you submitted data (1, 2, 3, 4, 5, 6)?

How many years do you anticipate submitting data?

How many years have you been involved with frogs, herps, or wildlife in general
as an avid hobbyist or professional (0-5, 6— 10, 11-15, l6-20, >20)?

With respect to the Frog and Toad Survey, please rate your current level of
expertise at which you perceive yourself, on a scale from 1 (low) to 5 (high).

Please rate the level of expertise at which you perceived yourself before getting
involved with the Michigan Frog and Toad Survey, on a scale from 1 (low) to 5
(high)-

What forms of training did you participate in before beginning to survey your
routes (Please mark all that apply) (attended training workshop, listened to
training tape, practice in the field with a trained observer, other — specify)?

How many people participate in your survey runs each year?

Is this number consistent each year (yes, no)?

Do the same people participate in the survey each year (yes, no)?

Is there one person designated as the primary observer (yes, no)?

Are you the primary observer (yes, no)?

175

13.

14.

15.

16.

17.

18.

19.

Please estimate the amount of time you spend listening at each site along your
route, in minutes.

During the time of your involvement with this program, have there been
discrepancies among observers in species heard or call indices that should be
assigned (yes, no)?

If yes to previous question, please describe how these discrepancies are resolved
in the space that follows.

Are you also an avid birder (yes, no)?

If yes, do you most often identify birds by sight, song, or both?

Are you involved in any additional wildlife monitoring programs (yes, no)?

If yes, please indicate which monitoring programs in which you participate
(Marsh Monitoring Program, North American Breeding Bird Survey, Christmas

Bird Count, Other — specify).

 

176

 

 

LITERATURE CITED

177

Literature Cited

Abramovitz, J .N. 1996. Imperiled waters, impoverished future: the decline of freshwater
ecosystems. Paper 128. Worldwatch Institute, Washington, DC.

Adams, M.J., K.O. Richter, and WP. Leonard. 1997. Surveying and monitoring
amphibians using aquatic funnel traps. Northwest Fauna 4: 47-54

Alford RA. 1999. Ecology: resource use, competition, and predation. In R.W.
McDiarmid, R.A. Altig. (eds) Tadpoles: The Biology of Anuran Larvae, pp. 240-
278. Univ. of Chicago Press, Chicago, IL USA.

Alford, RA. and SJ. Richards. 1999. Global amphibian declines: a problem in applied
ecology. Annual Review of Ecology and Systematics 30: 133-165

Altig, R., R.W. McDiarmid, K.A. Nichols, and PC. Ustach. 1998. A key to the anuran
tadpoles of the United States and Canada. Contemporary Herpetology Information
Series. 1998 [Online] Available http://alpha.selu.edu/ch/chis/ l998/2/index.htm)

Aminot A., Kirkwood D.S. & Kérouel R. 1997. Determination of ammonia in seawater
by the indophenol-blue method: Evaluation of the ICES NUTS L’C 5
questionnaire. Mar. Chem. 56: 59-75

Ashley, ER and J .T. Robinson. 1996. Road mortality of amphibians, reptiles, and other
wildlife on the Long Point Causeway, Lake Erie, Ontario. Canadian Field
Naturalist 110: 403-412

Beebee, T.J.C. 1985. Discriminant analysis of amphibian habitat determinants in
southeast England. Amphibia-Reptilia 6: 35-43

Beebee, T.J.C. 1995. Amphibian breeding and climate. Nature 374: 219-220

Berger, L., R. Speare, P. Daszak, D.E. Green, and AA. Cunningham, C.L. Goggin, R.
Slocombe, M.A. Ragan, A.D. Hyatt, K.R. McDonald, H.B. Hines, K.R. Lips, G.
Marantelli, and H. Parkes. 1998. Chytriodiomycosis causes amphibian mortality
associated with population declines in the rain forests of Australia and Central
America. Proceedings of the National Academy of Sciences, USA. 95: 9031-9036

Bertram, S. and M. Berrill. 1997. Fluctuations in a northern population of gray treefrogs,
Hyla versicolor. Herpetological Conservation 1: 57-63

Berven, K.A. 1990. Factors affecting population fluctuations in larval and adult stages of
the wood frog (Rana sylvatica). Ecology 71(4): 1599-1608

178

w-‘n—‘Q It: .

Berven, K.A. and TA. Grudzien. 1990. Dispersal in the wood frog (Rana sylvatica):
implications for genetic population structure. Evolution 44: 2047-2056

Bishop, CA. 1992. The effects of pesticides on amphibians and the implications for
determining causes of declines in amphibian populations. Pp. 67-70 In: C.A.
Bishop and KB. Pettitt (eds.). Declines in Canadian Amphibian Populations:

Designing a National Monitoring Strategy. Occasional Paper No. 76. Canadian
Wildlife Service, Ottawa, ON, Canada

Bishop, C.A., K.E. Pettit, M.E. Gartshore, and DA. Macleod. 1997. Extensive
monitoring of anuran populations using call counts and road transects in Ontario
(1992—1993). Herpetological Conservation 1: 149-160

Blaustein, AR, and DB. Wake. 1990. Declining amphibian populations: a global
phenomenon? Trends in Ecology and Evolution 52203-204.

Blaustein, A.R., DB. Wake, and WP. Sousa. 1994a. Amphibian declines: judging
stability, persistence, and susceptibility of populations to local and global
extinctions. Conservation Biology 8: 60-71

Blaustein, AR, PD. Hoffman, D.G. Hokit, J .M. Kiesecker, S.D. Walls, and J .B. Hays.
1994b. UV repair and resistance to solar UV-B in amphibian eggs: a link to

population declines? Proceedings of the National Academy of Sciences, USA 91:
1791 ~1 795

Blaustein, AR, J.M. Kiesecker, D.P. Chivers, and R.G. Anthony. 1997. Ambient UV-B
radiation causes deformities in amphibian embryos. Proceedings of the National
Academy of Sciences, USA. 94: 13735-13737

Bonin, J ., J. DesGranges, J. Rodrigue, and M. Ouellet. 1997. Anuran species richness in
agricultural landscapes of Quebec: forseeing long-term results of road call
surveys. Herpetological Conservation 1: 141-149

Bowers , D.G., D.E. Andersen, and NH. Euliss, Jr. 1998. Anurans as indicators of
wetland condition in the prairie pothole region of North Dakota: an environmental
monitoring and assessment program pilot project. Pp. 369-378 in: M. J. Lannoo,
editor. Status and Conservation of Midwestern Amphibians. University of Iowa,
Iowa City, Iowa, USA

Bradford, DP. 1989. Allotopic distributions of native frogs and introduced fishes in high
Sierra Nevada lakes of California: implications of the negative effect of fish
introductions. COpeia 1989: 775-778

Bradford, D.F., F. Tabatabai, and D.M. Graber. 1993. Isolation of remaining populations

of the native frog, Rana muscosa, by introduced fishes in the Sequoia and Kings
Canyon National Parks, California. Conservation Biology 7: 882-888

179

Breden, F. 1988. The natural history and ecology of Fowler’s toad, Bufo woodhousei
fowleri (Amphibia: Bufonidae) in the Indiana Dunes National Lakeshore.
Fieldiana Zoology 49: 1-16

Brodman, R. and M. Kilmurry. 1998. Status of amphibians in Northwestern Indiana. Pp.
125-136 in: M. J. Lannoo, editor. Status and Conservation of Midwestern
Amphibians. University of Iowa, Iowa City, Iowa, USA

Brown, J .H., and A. Kodric-Brown. 1977. Turnover rates in insular biogeography: effect
of immigration on extinction. Ecology 58: 445-449

Bunnell, J .F . and RA. Zampella. 1999. Acid water anuran pond communities along a
regional forest to agro-urban ecotone. Copeia 1999: 614-662

Burton, TM. and GE. Likens. 1975. Salamander populations and biomass in the
Hubbard Brook Experimental Forest, New Hampshire. Copeia 1975: 541-546

Cantrell, K.J., S.M. Serkiz and EM. Perdue. 1990. Evaluation of acid neutralizing
capacity data for solutions containing natural organic acids. Geochim Cosmochim
Acta 54: 1247-1254.

Carey, C. and C .J . Bryant. 1995. Possible interrelationships among environmental
toxicants, amphibian development, and decline of amphibian populations.
Environmental Health Perspectives. 103 (Supp. 4): 13-17

Carey, C . and M.A. Alexander. 2003. Climate change and amphibian declines: is there a
link? Diversity and Distributions. 9: 111-121

Carr, I..W. and L. F arhig. 2001. Effect of road traffic on two amphibian species of
differing vagility. Conservation Biology 15: 1071-1078

Casper, GS. 1998. Review of the status of Wisconsin amphibians. Pp. 199-205 in: M. J.
Lannoo, editor. Status and Conservation of Midwestern Amphibians. University
of Iowa, Iowa City, Iowa, USA

Collins, JP. 1979. Intrapopulation variation in the body size at metamorphosis and timing
of metamorphosis in the bullfrog, Rana catesbeiana. Ecology 60: 738-749

Comer, PI. 1996. Wetland trends in Michigan since 1800: a preliminary assessment.
Report to the US. EPA. and Land and Water Division, Michigan Department of

Environmental Quality, Michigan Natural Features Inventory, Lansing, MI

Conant, R. and J .T. Collins. 1998. A field guide to reptiles and amphibians of eastern and
central North America. 3rd ed. (expanded). Houghton Mufflin co. New York, NY

180

Crouch, W.B. 1999. Evaluation of the North American Amphibian Monitoring Program’s

(NAAMP) calling anuran survey. MS. Thesis. University of Rhode Island,
Kingston, RI

Crouch, W.B. and P.W.C. Paton. 2002. Assessing the use of call surveys to monitor
breeding anurans in Rhode Island. Journal of Herpetology 36: 185-192.

Crump, ML. 1989. Effect of habitat drying on developmental time and size at
metamorphosis in Hyla pseudopuma. Copeia 1989: 794-797

Crump, M. L., and N. J. Scott, Jr. 1994. Standard techniques for inventory and
monitoring-visual encounter surveys. Pages 84 - 92 in W. R. Heyer, M. A.
Dormelly, R. W. McDiarmid, L. C. Hayek, and M. S. Foster, editors. Measuring
and monitoring biological diversity: standard methods for amphibians.
Smithsonian Institution Press, Washington, DC.

Dahl, TE 2000. Status and trends of wetlands in the conterrninous United States 1986 to
1997. U .S. Department of the Interior, Fish and Wildlife Service. Washington,
D.G. 82 pp.

Daigle, C. 1997. Distribution and abundance of the chorus frog Pseudacris triseriata in
Quebec. Herpetological Conservation 1: 73-77

DcGraaf, RM. and DD. Rudis. 1990. Herpetofaunal species composition and relative
abundance among three New England forest types. Forest Ecology and
Management 32: 155-165

Delis, P.R., HR. M ushinsky, and ED. McCoy. 1996. Decline of some west-central
Florida anuran populations in response to habitat degradation. Biodiversity and
Conservation 5: 1579-1595

Dctenbeck, N.E., S.M. Galatowitsch, J. Atkinson, and H. Ball. 1999. Evaluating
perturbations and developing restoration strategies for inland wetlands in the
Great Lakes Basin. Wetlands 19: 789-820

Driscoll, D. 1998. Counts of calling males as estimates of population size in the
endangered frogs Geocrinia alba and G. vitellina. Journal of Herpetology 32:
475-481

Duellman, W.B. and L. Trueb. 1986. Biology of Amphibians. Johns Hopkins University
Press, Baltimore, MD

Dunson, W.A. and J. Connell. 1982. Specific inhibition of hatching in amphibian
embryos by low pH. J. Herp. 16: 314-316

181

Dunson, W.A. and J. Travis. 1991. The role of abiotic factors in community organization.
American Naturalist 138: 1067-1091

Erskine, A. J. 1978. The first ten years of the cooperative Breeding Bird Survey in
Canada. Canadian Wildlife Service Report Series No. 42, Hull, Quebec, Canada

F ahri g, L. 1997. Relative effects of habitat loss and fragmentation on population
extinction. Journal of Wildlife Management 61: 603-610

Fahrig, L., J .H. Pedlar, S.E. Pope, P.D. Taylor, and J .F . Wegner. 1995. Effect of road
traffic on amphibian density. Biological Conservation 74: 177-182

Findlay, CS. and J. Houlahan. 1997. Anthropogenic correlates of species richness in
southeastern Ontario wetlands. Conservation Biology 11: 1000-1009

Fisher, RN. and H.B. Shaffer. 1996. The decline of amphibians in Califomia’s Great
Central Valley. Conservation Biology 10: 1387-1397

Forrnan, R.T.T. and LE. Alexander. 1998. Roads and their major ecological effects.
Annual Review of Ecology and Systematics 29: 207-231

Freda, J. and W.A. Dunson. 1986. Effects of low pH and other chemical variables on the
local distribution of amphibians. Copeia 1986: 454-466

Freda, J ., W.J. Sadinski, and W.A. Dunson. 1991. Long term monitoring of amphibian
populations with respect to acidic deposition. Water, Air, and Soil Pollution 55:
445-462

Geissler, PH. and J .R. Sauer. 1990. Topics in route-regression analysis. Pp. 54-57 In:
J .R. Sauer and S. Droege (eds). Survey designs and statistical methods for the
estimation of avian population trends. US. Fish and Wildlife Service, Biological
Report 90(1)

Genet, KS. and LG. Sargent. 2003. Evaluation of methods and data quality from a
volunteer-based amphibian call survey. Wildlife Society Bulletin 31: in press

Gerhardt, H.C., M.B. Ptacek, L. Barnett, and K.G. Torke. 1994. Hybridization in the
diploid-tetraploid treefrogs Hyla chrysoscelis and Hylca versicolor. Copeia 1994:
5 1-59

Gibbs, J .P. 1993. Importance of small wetlands for the persistence of local populations of
wetland associated animals. Wetlands 13: 25-31

Gibbs, J .P. and AR. Breisch. 2001. Climate warming and calling phenology of frogs near
Ithaca, New York, 1900-1999. Conservation Biology 15: 1175-1178

182

 

Gosner, KL. 1960. A simplified table for staging anuran embryos and larvae with notes
on identification. Herpetologica 16: 183-190

Green, D.M. 1997a. Temporal variation in abundance and age structure in Fowler’s

Toads, Bufofowleri, at Long Point, Ontario. Herpetological Conservation 1: 45-
56.

Green, D. M. 1997b. Perspectives on amphibian population declines: defining the
problem and searching for answers. Herpetological Conservation 1: 291-308

Griffiths, RA. 1985. A simple funnel trap for studying newt populations and an
evaluation of trap behavior in smooth and palmate newts T riturus vulgaris and T.
helviticus. Herpetological Journal 1: 5-9

Guerry, AD. and ML. Hunter, Jr. 2002. Amphibian distributions in a landscape of
forests and agriculture: an examination of landscape composition and
configuration. Conservation Biology 16: 745-754

Halley, J.M., R.S. Oldharn, and J .W. Amtzen. 1996. Predicting the persistence of
amphibian populations with the help of a spatial model. Journal of Applied
Ecology 33: 455-470

Harding, J .H. 1997. Amphibians and reptiles of the Great Lakes region. University of
Michigan Press. Ann Arbor, MI

Harkey, GA. and RD. Semlitsch. 1988. Effects of temperature on growth, development,

and color polymorphism in the ornate chorus frog, Pseudacris ornata. Copeia
1988: 1001-1007

Harris, RN. 1999. The anuran tadpole: evolution and maintenance. Pp. 279-294 In R.W.
McDiarmid, R.A. Altig. (eds) Tadpoles: The Biology of Anuran Larvae, Univ. of
Chicago Press, Chicago, IL USA.

Hay, R. 1998. Blanchard’s cricket frogs in Wisconsis: a status report. Pp. 79-82 in: M. J.
Lannoo, editor. Status and Conservation of Midwestern Amphibians. University
of Iowa, Iowa City, Iowa, USA

Hayes, MP. and MR. Jennings. 1986. Decline of frog species in western North America:
are bullfrogs (Rana catesbeiana) responsible? Journal of Herpetology 20: 490-
509

Hecnar, S. J. 1997. Amphibian pond communities in southwestern Ontario.
Herpetological Conservation. 1 : 1-15

Hecnar, SJ and RT. M’Closkey. 1996. Regional dynamics and the status of amphibians.
Ecology 77: 2091-2097

183

 

Hecnar, SJ. and RT. M’Closkey. 1997a. The effects of predatory fish on amphibian
species richness and distribution. Biological Conservation 79: 123-131

Hecnar, SJ. and RT. M’Closkey. 1997b. Spatial scale and determination of species
status of the green frog. Conservation Biology 11: 670-682

Hecnar, SJ. and RT. M’Closkey. 1998. Species richness patterns of amphibians in
southwestern Ontario ponds. Journal of Biogeography 25: 763-772

Hemesath, L. M. 1998. Iowa’s frog and toad survey, 1991-1994. Pages 206-216 in: M. J.
Lannoo, editor. Status and Conservation of Midwestern Amphibians. University
of Iowa, Iowa City, Iowa, USA

Hine , R.L., B.L. Les, and BF. Hellmich. 1981. Leopard frog populations and mortality
in Wisconsin, 1974-1976. Technical Bulletin No. 122, Wisconsin Dept. of Natural
Resources, Madison, WI

Hollands, G.G. 1987. Hydrogeologic classification of wetlands in glaciated regions.
National Wetlands Newsletter 9: 6-9

Home, M. T. and W. A. Dunson. 1995. Effects of low pH, metals, and water hardness on
larval amphibians. Archives of Environmental Contamination and Toxicology.
29:500-505.

Houlahan, J .E., C.S. Finlay, B.R. Schmidt, A.H. Meyer, and BL. Kuzmin. 2000.
Quantitative evidence for global amphibian population declines. Nature 404: 752-
755

Huff, ]. 1991. Frog and toad survey 1991. Madison: Wisconsin Department of Natural
Resources. Technical Report. pp. 114-123

Hunter, M.L., J r., A.J.K. Calhoun, and M. McCullough. 1999. Maine Amphibians and
Reptiles. University of Maine Press, Orono, ME USA

Johnson, B. 1992. Habitat loss and declining amphibian populations. Pp. 71-75 In: C.A.
Bishop and KB. Pettit (eds) Declines in Canadian Amphibian Populations:
Designing a National Monitoring Strategy. Occasional Paper No. 76. Canadian
Wildlife Service

Jongman, R.H.G., C.J.F. ter Braak, and O.F.R. van Tongeren (eds). 1995. Data Analysis
in Community and Landscape Ecology. Cambridge University Press, Cambridge,

England

Kats, L.B., J.W. Petranka, and A. Sih. 1988. Antipredator defenses and the persistence of
amphibian larvae with fishes. Ecology 69: 1865-1870

184

 

 

Kendall, W. L., B.G. Peterjohn, and J.R. Sauer. 1996. F irst-time observer effects in the
North American Breeding Bird Survey. The Auk 113: 823-829

Kiesecker, J .M. and AR. Blaustein. 1997. Population differences in responses to red-
legged frogs (Rana aurora) to introduced bullfrogs. Ecology 78: 1752-1760

Kiesecker, J .M, A.R. Blaustein, and L.K. Belden. 2001. Complex causes of amphibian
population declines. Nature 410: 681-684

Ki ffney, PM. and J .S. Richardson. 2001. Interactions among nutrients, periphyton, and
invertebrate and vertebrate (Ascaphus truei) grazers in experimental channels.
Copeia 2001: 422-429

Kline, J. 1998. Monitoring amphibians in created and restored wetlands. Pages 360-368
in: M. J. Lannoo, editor. Status and Conservation of Midwestern Amphibians.
University of Iowa, Iowa City, Iowa, USA

Knutson, M.G., J.R. Sauer, DA. Olsen, M.J. Mossman, L.M. Hemesath, and M.J.
Lannoo. 1999. Effects of landscape composition and wetland fragmentation on
frog and toad abundance and species richness in Iowa and Wisconsin, USA.
Conservation Biology 13: 1437-1446

Knutson, M .G., J .R. Sauer, D.A. Olsen, M.J. Mossman, L.M. Hemesath, and M.J.
Lannoo. 2000. Landscape associations of frog and toad species in Iowa and
Wisconsin, USA. Journal of the Iowa Academy of Science 107: 134-145

Koloszvary, M .B., and R.K. Swihart. 1999. Habitat fragmentation and the distribution of

amphibians: patch and landscape correlates in farmland. Canadian Journal of
Zoology 77: 1288-1299

Kupferberg, SJ. 1997. The role of larval diet in anuran metamorphosis. American
Zoologist 37: 146-159

Kupferberg, S.J., .I.C. Marks, and ME. Power. 1994. Effects of variation in natural algal
and detrital diets on larval anuran (Hyla regilla) life history traits. Copeia 1994:
446-457

Laan, R. and B. Verboom. 1990. Effects of pool size and isolation on amphibian
communities. Biological Conservation 54: 251-262

Lannoo, M.J., K. Lang, T. Waltz, and GS. Phillips. 1994. An altered amphibian
assemblage: Dickinson County, Iowa, 70 years after Frank Blanchard’s survey.
American Midland Naturalist 13l(2):311-319

Lannoo, M.J. 1998. Status and Conservation of Midwestern Amphibians. University of
Iowa, Iowa City, Iowa, USA

185

 

Lee, Y. M. 1998. Final report on Blanchard’s cricket frog (Acris crepitans blanchardi)
surveys in southern Michigan. A report to the US. Environmental Protection
Agency, EPA wetland enhancement grant and critical small non-contiguous
wetlands project and Michigan Dept. of Natural Resources, Parks and Recreation
Division. Michigan Natural Features Inventory report no. 1998-03. 13 pp. +
appendices.

Lehtinen, R.M., S.M. Galatowitsch, and J .R. Tester. 1999. Consequences of habitat loss
and fragmentation for wetland amphibian assemblages. Wetlands 19: 1-12

Lehtinen, R. M. 2001. Surveys for three rare amphibians in southeastern Michigan:
Blanchard’s cricket frog (Acris crepitans lanchardi), the four toed salamander
(Hemidactylium scutatum), and the smallmouth salamander (Ambystoma
texanum). Final report to the Michigan Dept. of Natural Resources, August 2001.
18 pp. + appendices.

Lepage, M., R. Courtois, and C. Daigle. 1997. Surveying calling anurans in Quebec using
volunteers. Herpetological Conservation 1: 128-140

Lepczyk, CA. 2002. Effects of human activities on birds across landscapes in the
Midwest. Ph.D. Dissertation. Dept. of Fisheries and Wildlife, Michigan State
University

Licht, LE. and KP. Grant. 1997. The effects of ultraviolet radiation on the biology of
amphibians. American Zoologist 37: 137-145

Link, W.A. and J.R. Sauer. 1994. Estimating equations estimates of trends. Bird
Populations 2:23-32

Link, W.A., and Sauer, J .R. 1997. Estimation of population trajectories from count data.
Biometrics 53(2), 488-497.

Lips, KR. 1998. Decline of a tropical montane amphibian fauna. Conservation Biology
12: 106-117

Loman, J. 2002. Rana temporaria metamorph production and population dynamics in the
field: effects of tadpole density, predation, and pond drying. Journal for Nature
Conservation. 10: 95-107

Lucas, EA. and W.A. Reynolds. 1967. Temperature selection by amphibian larvae.
Physiol. 2001. 40: 159-171

Lynch, J .A., G.B. Rishel, and ES. Corbett. 1984. Thermal alteration of streams draining

clearcut watersheds: quantification and biological implications. Hydrobiologia
111: 161-169

186

Mader, H.J. 1984. Animal habitat isolation by roads and agricultural fields. Biological
Conservation 29: 81-96

Maxwell, SE. and H.D. Delaney. 1990. Designing experiments and analyzing data: a
model comparison perspective. Wadsworth, Belmont, CA

Merrell, DJ. 1977. Life history of the leopard frog, Rana pipiens, in Minnesota.
Occasional Paper Number 15. Bell Museum of Natural History. University of
Minnesota, Minneapolis, MN, USA

Mierzwa, KS. 1998. Status of Northeastern Illinois amphibians. Pp. 115-124 in: M. J.
Lannoo, editor. Status and Conservation of Midwestern Amphibians. University
of Iowa, Iowa City, Iowa, USA

Miller, R. G. 1981. Simultaneous statistical inference, 2nd ed. Springer Verlag, Berlin

Mitchell, J.C., S.C. Rinehart, J.F. Pagels, K.A. Buhlmann, and CA. Pague. 1997. Factors
influencing amphibian and small mammal assemblages in central Appalachian
forests. Forest Ecology and Management 96: 65-76

Moriarty, J]. 1998. Status of amphibians in Minnesota. Pp. 166-168 in: M. J.
Lannoo, editor. Status and Conservation of Midwestern Amphibians. University
of Iowa, Iowa City, Iowa, USA

Morin, P. J. 1983. Predation, competition, and the composition of larval anuran guilds.
Ecological Monographs 53: 119-138

Morrison, DE. 1990. Multivariate statistical methods (3rd ed.). McGraw Hill. New York,
NY USA

Mossman, M.J. and R.L. Hine. 1984. The Wisconsin frog and toad survey: establishing a
long-term monitoring program. Endangered Resources Report No. 9. Wisconsin
Dept. of Natural Resources, Madison, WI

Mossman, M. J ., L. M. Hartman, R. Hay, J. R. Sauer, and B. J. Dhuey. 1998. Monitoring
long-term trends in Wisconsin frog and toad populations. Pages 169-198 in: M.J.
Lannoo, editor. Status and Conservation of Midwestern Amphibians. University
of Iowa, Iowa City, Iowa, USA

Moyle, PB. and RA. Leidy. 1992. Loss of biodiversity in ecosystems: evidence from
fish faunas. Pp. 127-169 in P.L. F iedler and SK. Jain (eds). Conservation
biology: the theory and practice of nature conservation, preservation, and
management. Chapman and Hall. New York, NY

Muths, E., P.S. Corn, A.P. Pessier, and DE. Green. Evidence for disease-related
amphibian declines in Colorado. Biological Conservation 110: 357-365

187

O’Brien, R.G. and MK. Kaiser. 1985. MANOVA method for analyzing repeated
measures designs: an extensive primer. Psychological Bulletin 97: 316-333

Pais, R.C., S.A. Bonney, and WC. McComb. 1988. Herpetofaunal species richness and
habitat associations in an eastern Kentucky forest. Proceedings of the Annual
Conference of Southeastern Associations of Fish and Wildlife Agencies 42: 448-
455

Pandian, T.J. and MP. Marian. 1985. Predicting anuran metamorphosis and energetics.
Physiol. Zool. 58: 538-552

Patterson, J .W. and A]. McLachlan. 1989. Larval habitat duration and size at
metamorphosis in frogs. Hydrobiologia 171: 121-126

Pechmann, J .H.K., D.E. Scott, R.D. Semlitsch, J .P. Caldwell, L.J. Vitt, and J .W. Gibbons.
1991. Declining amphibian populations: the problem of separating human impacts
from natural fluctuations. Science 253: 892-895

Pechmann, J .H.K. and HM. Wilbur. 1994. Putting declining amphibian populations in
perspective: natural fluctuations and human impacts. Herpetologica 50: 65-84

Pentecost, ED. and RC. Vogt. 1976. Amphibians and reptiles of the Lake Michigan
drainage basin. Environmental Status of the Lake Michigan Region, Vol. 16.
Argonne National Laboratory, Argonne, IL

Peterjohn , B.G., J .R. Sauer, and W.A. Link. 1994. The 1992 and 1993 summary of the
North American Breeding Bird Survey. Bird Populations 2: 46-51

Pope, S.E., L. F ahrig, and HG. Merriam. 2000. Landscape complementation and
metapopulation effects on leopard frog populations. Ecology 81: 2498-2508

Pounds, J .A. and ML. Crump. 1994. Amphibian declines and climate disturbance: the
case of the golden toad and the harlequin frog. Conservation Biology 8: 72-85

Power, T., K.L. Clark, A. Hamfeist, and DB. Peakall. 1989. A review and evaluation of
the amphibian toxicological literature. Technical Report No. 61. Canadian
Wildlife Service, Ottawa, ON, Canada

Reh, W. and A. Seitz. 1990. The influence of land use on the genetic structure of

populations of the common frog, Rana temporaria. Biological Conservation 54:
239-249

Richards, S.J., K.R. McDonald, and RA. Alford. 1993. Declines in populations of

Australia’s endemic tropical rainforest frogs. Pacific Conservation Biology 1: 66-
77

188

 

Richter, KC. 1995. A simple aquatic funnel trap and its application to wetland
amphibian monitoring. Herpetological Review 26: 90-91

Robbins, C.S., D. Bystrak, and PH. Geissler. 1986. The Breeding Bird Survey: its first
fifteen years, 1965-1979. US. Fish and Wildlife Service, Resource Publication
157

Rosenberg, EA. and BA. Pierce. 1995. Effect of initial mass on growth and mortality at
low pH in tadpoles of Pseudacris clarkii and Bufo valliceps. J. Herp. 29: 181-185

Rowe, CL. and W.A. Dunson. 1995. Impacts of hydroperiod on growth and survival of

larval amphibians in temporary ponds of central Pennsylvania. Oecologia 102:
397-403

Saber, PA. and W.A. Dunson. 1978. Toxicity of bog water to embryonic and larval
anuran amphibians. J. Exp. Biol. 204: 33-42

Sargent, L. G- 2000. Frog and toad population monitoring in Michigan. .Ioumal of the
Iowa Academy of Sciences 107: 195-199

Sarkar, S. 1996. Ecological theory and anuran declines. BioScience 46: 199-207

Sauer, J. R., B. G. Peterjohn, and W. A. Link. 1994. Observer differences in the North
American Breeding Bird Survey. Auk 111:50-62.

Saunders, D.L., J.J. Weeuwi g, and A.C.J. Vincent. 2002. Freshwater protected areas:
strategies for conservation. Conservation Biology 16: 30-41

Scott, N. J ., Jr. and B. D. Woodward. 1994. Surveys at breeding sites. Pages 118-125 in:
W.R. Heyer, M.A. Donnelly, R.W. McDiarmid, L.C. Hayek, and MS Foster,
editors. Measuring and monitoring biological diversity: standard methods for
amphibians. Smithsonian. Washington, D.G., USA

Semlitsch, RD. 2000. Principles for management of aquatic-breeding amphibians.
Journal of Wildlife Management 64: 615-631

Semlitsch, RD. 2002. Critical elements for biologically-based recovery plans of aquatic-
breeding amphibians. Conservation Biology 16: 619-629

Semlitsch, R.D., D.E. Scott, J .H.K. Pechmann, and J .W. Gibbons. 1996. Structure and
dynamics of an amphibian community: evidence from a sixteen-year study of a
natural pond. Pp. 217-248 In: M.L. Cody and J. Smallwood (eds). Long-tenn
Studies of Vertebrate Communities. Academic Press, New York, NY

Semlitsch, RD. and J .R. Bodie. 1998. Are small, isolated wetlands expendable?
Conservation Biology 12: 1129-1133

189

Shirose, L.J and RI. Brooks. 1997. Fluctuations in abundance and age structure in three
species of frogs (Anura: Ranidae) in Algonquin Park, Canada, from 1985 to 1993.
Herpetological Conservation 1: 16-26

Shirose, L.J., C.A. Bishop, D.M. Green, C]. MacDonald, R]. Brooks, and NJ.
Helferty. 1997. Validation tests of an amphibian call count survey technique in
Ontario, Canada. Herpetologica 53: 312-320

Sinsch, U. 1990. Migration and orientation in anuran amphibians. Ethology, Ecology, and
Evolution 2: 65-79

Sjogren, P. 1991. Extinction and isolation gradients in metapopulations: the case of the
pool frog (Rana lessonae) Biological Journal of the Linnean Society 42: 135-147

S jogren-Gulve, P. 1994. Distribution and extinction patterns within a northern
metapopulation of the pool frog, Rana lessonae. Ecology 75: 1357-1367

Skelly, D.K. 1995a. A behavioral tradeoff and its consequences for the distribution of
Pseudacris treefrog larvae. Ecology 76 (1): 150-164

Skelly, D.K. 1995b. Competition and the distribution of spring peeper larvae. Oecologia
103: 203-207

Skelly, D.K., 1996. Pond drying, predators, and the distribution of Pseudacris tadpoles.
Copeia1996: 599-605

Skelly, D.K. 1997. Tadpole communities. American Scientist 85: 36-45

Skelly, D.K., E.E. Werner, and SA. Cortwright. 1999. Long-term distributional
dynamics of a Michigan amphibian assemblage. Ecology 80(7): 2326-2337

Skelly, D.K., L.K. Friedenburg, and J .M. Kiesecker. 2002. Forest canopy and the
performance of larval amphibians. Ecology 83: 983-992

Smith, DC. 1983. Factors controlling tadpole populations of the chorus frog (Pseudacris
triseriata) on Isle Royale, Michigan. Ecology 64(3): 501-510

Smith, DC. and J. Van Buskirk. 1995. Phenotypic design, plasticity, and ecological
performance in two tadpole species. Am. Nat. 145: 211-233

Smith-Gill, SJ. and K. Berven. 1979. Predicting amphibian metamorphosis. Am. Nat.
113: 563-585

Snodgrass, J .W., M.J. Komoroski, A. L. Bryan, Jr., and J. Burger. 2000. Relationships

among isolated wetland size, hydroperiod, and amphibian species richness:
implications for wetland regulations. Conservation Biology 14: 414-419

190

Staniszewski, MS. 1995. Amphibians in Captivity. T.F.H. Publications, Neptune, NJ,
USA. 544 pp.

Stebbins, RC. and NW. Cohen. 1995. A Natural History of Amphibians. Princeton
University Press, Princeton, NJ

Steinwascher, K. and J. Travis. 1983. Influence of food quality and quantity on early
larval grth of two anurans. Copeia 1983: 238-242

Stevens, J. 1992. Applied multivariate statistics for the social sciences, 2"d edition.
Lawrence Erlbaum, Princeton, NJ

Strijbosch, H. 1980. Habitat selection by amphibians during their terrestrial phase. British
Journal of Herpetology 6: 93-98

Tej edo, M. and R. Reques. 1994. Does larval grth history determine timing of
metamorphosis in anurans? A field experiment. Herpetologica 50: 113-118

Thomas, L. and K. Martin. 1996.The importance of analysis method for Breeding Bird
Survey population trend estimates. Conservation Biology 10:479-490

Travis, J. 1994. Calibrating our expectations in studying amphibian populations
Herpetologica 50(1): 104-108

Trombulak, SC. and CA. Frissell. 2000. Review of the ecological effects of roads on
terrestrial and aquatic communities. Conservation Biology 14: 18-30

Ultsch, G.R., D.F. Bradford, and J. Freda. 1999. Physiology: Coping with the
environment. Pp. 189-214 In R.W. McDiarmid, R.A. Altig. (eds) Tadpoles: The
Biology of Anuran Larvae, Univ. Chicago Press, Chicago, IL

US. EPA. 2002. Methods for evaluating wetland condition: using amphibians in
bioassessments of wetlands. Office of Water, US. Environmental Protection
Agency, Washington, DC. EPA .822-R-02-022

Valderrama, J .C . 1981. The simultaneous analysis of total nitrogen and total phosphorus
in natural waters. Mar. Chem. 10: 109-122.

van Buskirk, J. 2000. The costs of an inducible defense in anuran larvae. Ecology
81(10): 2813-2821

van Dorp, D. and P.F.M. Opdam. 1987. Effects of patch size, isolation, and regional
abundance on forest bird communities. Landscape Ecology 1: 59-73

191

Varhegyi, G., S.M. Mavroidis, B.M. Walton, CA. Conaway, and AR. Gibson. 1998.
Amphibian surveys in the Cuyahoga Valley National Recreation Area. Pp. 137-
154 in: M. J. Lannoo, editor. Status and Conservation of Midwestern Amphibians.
University of Iowa, Iowa City, Iowa, USA

Vogt, RC. 1981. Natural History of Amphibians and Reptiles of Wisconsin. Milwaukee
Public Museum, Milwaukee, WI

von Ende, C.N. 1993. Repeated-measures analysis: growth and other time-dependent
measures. Pp. 113-137 In: S.M. Scheiner and J. Gurevitch (eds). Design and
Analysis of Ecological Experiments. Chapman and Hall, New York, NY

Vos, CC. and J .P. Chardon. 1998. Effects of habitat fragmentation and road density on
the distribution pattern or the moor frog, Rana arvalis. Journal of Applied
Ecology 35: 44-56

Vos, CC. and HP. Stumpel. 1995. Comparison of habitat-isolation parameters in relation
to fragmented distribution patterns in the tree frog (Hyla arborea). Landscape
Ecology 11: 203-214

Waldick, R.C., B. Freedman, and R.J. Wassersug. 1999. The consequences for
amphibians of the conversion of natural, mixed-species forests to conifer

plantations in southern New Brunswick. The Canadian Field Naturalist 113: 408-
418

Wassersug, R.J. and EA. Seibert. 1975. Behavioral responses of amphibian larvae to
variation in dissolved oxygen. Copeia 1975: 86-103

Watermolen, D.J. andH. Gilbertson. 1996. Keys for the identification of Wisconsin’s
larval amphibians. Wisconsin Endangered Resources Report #109, Bureau of

Endangered Resources, Wisconsin Department of Natural Resources. Madison,
WI

Weir, L. A. and M. J. Mossman. In press. The protocol and history of the amphibian
calling survey of the North American Amphibian Monitoring Program (NAAMP).
Chapter 39 in: M. J. Lannoo, editor. Declining amphibians: a United States’
response to the global phenomenon. University of California, Berkeley,
California, USA

Wellbom, G.A., D.K. Skelley, and BE. Werner. 1996. Mechanisms creating community

structure across a freshwater habitat gradient. Annual Review of Ecology and
Systematics 27: 337-363

Weller, W.F. and D.M. Green. 1997. Checklist and current status of Canadian
amphibians. Herpetological Conservation 1: 309-328

192

Welsh, H.W., Jr. 1990. Relictual amphibians and old-growth forests. Conservation
Biology 4: 309-319

Welshmeyer, NA. 1994. Fluorometric analysis of chlorophyll a in the presence of
chlorophyll b and pheopigrnents. Limnol. Oceanogr. 39: 1985-1992

Werner, BE. 1986. Amphibian metamorphosis: growth rate, predation rate, and optimal
size at transformation. American Naturalist 128: 319-341

Werner, BE. and KS. Glennemeier. 1999. Influence of forest canopy cover on the
breeding pond distributions of several amphibian species. Copeia 1999(1): 1-12

Wetzel, R.G. and GE. Likens. 2000. Lirnnological analyses, 3rd ed. Springer-Verlag.
429 pp.

Wilbur, HM. 1980. Complex life cycles. Annual Review of Ecology and Systematics
11:67-93

Wilbur, HM. 1984. Complex life cycles and community organization in amphibians. In:
P.W. Price, C.N. Slobodchikoff and W.S. Gaud (eds). A New Ecology: Novel
Approaches to Interactive Systems. John Wiley and Sons, New York, NY USA

Wilbur, H. M. 1987. Regulation of structure in complex systems: experimental temporary
pond systems. Ecology 68: 1437-1452

Wilbur, HM. and J .P. Collins. 1973. Ecological aspects of amphibian metamorphosis.
Science 182: 1305-1314

Winter, TC. 1988. Conceptual framework for assessment of cumulative impacts on the
hydrology of non-tidal wetlands. Environmental Management 12: 605-620

Wyman, R.L. 1988. Soil acidity and moisture and the distribution of amphibians in five
forests of south central New York. Copeia 1988: 394-399

Zar, J. R. 1998. Biostatistical Analysis. Fourth edition. Prentice Hall. Englewood Cliffs,
New Jersey, USA

Zimmerman, B. L. 1994. Audio strip transects. Pages 92-97 in: W.R. Heyer, M.A.
Donnelly, R.W. McDiarmid, L.C. Hayek, and MS. Foster (eds). Measuring and

monitoring biological diversity: standard methods for amphibians. Smithsonian
Press. Washington, DC, USA

Zug, G.R., L.J. Vitt, and and J .P Caldwell. 2001. Herpetology: an introductory biology of
amphibians and reptiles (2"d ed.). Academic Press.

193