«gunman
A .

$5.
‘ .
41V... >Woud3rn
#34... A» ‘
. 5m. _
HR.)

1.,

...
.. a.
v

fiw

‘ u ”.5,

92h!

.313

. wmm‘mgm .

A

. .f:

3.:
.ufi...‘
".5. K

MF‘UM‘Y‘

w

. t
,.zi;. t.
fl... he Junta
fin .umqnu hfiuf. .
.Peu .0 ‘ A .

'

.917». r .

, ioiA ”0‘.ll
HI . 0 I‘m!

‘ :

 

 

t$~ ‘ 33...... 7

THFMK

1903

This is to certify that the

dissertation entitled

DETERMINANTS OF AQUATIC SNAIL COMMUNITY STRUCTURE
IN GREAT LAKES (LAKE HURON) COASTAL WET MEADOWS

presented by

Brian E. Keas

has been accepted towards fulfillment
of the requirements for

 

Ph . D . degree in Zoology

 

Major professor

Date (1%?QJT 313920007

MSUiJ an Affirmative Action/Equal Opportunin Institution 0-12771

 

LIBRARY
Michigan State
University

 

 

 

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

 

DATE DUE DATE DUE DATE DUE

 

fi‘Ec it 5064

 

 

 

 

 

 

 

 

 

 

 

 

 

6/01 cJCIRClDatoDuopes-p. 15

 

DETERMINANTS OF AQUATIC SNAIL COMMUNITY STRUCTURE
IN GREAT LAKES (LAKE HURON) COASTAL WET MEADOWS

By

Brian E. Keas

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

2002

ABSTRACT

DETERMINANTS OF AQUATIC SNAIL COMMUNITY STRUCTURE
IN GREAT LAKES (LAKE HURON) COASTAL WET MEADOWS

By

Brian E. Keas

Previous studies examining aquatic snail community structure have used a wide
variety of explanatory variables including water chemistry, area effects, habitat diversity,
and biotic interactions. Almost all have focused on within-patch characteristics and have
generally ignored the surrounding landscape context. The northern shoreline of Lake
Huron includes a dynamic wetland complex containing numerous Great Lakes coastal
wet meadows. I sampled aquatic snails from 26 coastal wet meadows in 1997 and 1998,
and measured 28 within-patch characteristics used to describe the water chemistry,
habitat structure, and plant community. A GIS database was used to calculate 27
landscape-level characteristics describing the adjacent and nearby patches and the
complexity of the shoreline. I used canonical correspondence analysis (CCA) to identify
patterns in the distribution and abundance of snails associated with the within-patch
characteristics. Landscape effects were examined by CCA afier accounting for within-
patch effects. Additional associations with water depth were examined using multiple
regression analyses and laboratory experiments.

Nineteen species of snails were collected with six species occurring abundantly in
most wet meadows. The occurrence of various plant species explained the greatest
amount of within-patch variation in the snail community, followed by characteristics of

habitat structure. Water chemistry explained the least amount, as expected. Significant

associations with multiple landscape context variables, especially bulrush, cattail, and
upland meadows, were observed after accounting for within-patch effects. In 1997 only,
there were significant associations with shoreline complexity as measured by sinuosity
and fractal dimension at multiple scales. A suite of variables that were related to urban
land use and habitat fragmentation was included in a 1998 model, suggesting some
concern for anthropogenic impacts on the snail community and the need for conservation
of these wetlands. In an analysis of water levels, mean depth was the most common
predictor, followed by spatial and temporal variation in water levels. Laboratory
experiments showed interspecific variation in snail motility and desiccation resistance
that were consistent in explaining snail community structure.

A conceptual model of the Great Lakes coastal wet meadow snail community was
developed based on a depth gradient from deeper wet meadows that remain permanently
flooded to shallow and hydrologically semi-isolated wet meadows in which periodic
drying occurred. The model included interactions at various scales, from within-patch to
shoreline complexity, and the proximate behavioral and physiological differences among
snail species. Further testing of the pathways in this model will help to clarify the

mechanisms behind the observed patterns.

ACKNOWLEDGMENTS

I am grateful for the support, advice, and patience of my major professor, Dr.
Tom Burton, throughout the entire research and writing process. Members of my
committee, Dr. Steve Hamilton, Dr. Rich Merritt, and Dr. Pat Muzzall, offered additional
insights and suggestions. Thank you.

Many enjoyable memories were made along the northern Lake Huron shoreline
and in the lab through the help and encouragement of Joe Gathman, Donna Kashian, and
Sam Riffell. Special thanks to Sam Riffell who helped with habitat design and sampling,
with construction of the GIS database, and with suggestions for data analyses.

Residents of Cedarville, DeTour Village, and Hessel provided access to coastal
wetlands on their properties. Computing resources were provided by the Department of
Geography and water analyses by the Department of Geological Sciences at Michigan
State. Support for my research was provided by grants from the Michigan Chapter of the
Nature Conservancy (to Tom Burton), the Department of Zoology and the Graduate
School at Michigan State University, and the John G. Shedd Aquarium.

Thanks to Angela Roberts for her loving encouragement and the inspiration to
finish writing, and to my parents for their support and patience.

Yes, my dissertation is finished!

iv

TABLE OF CONTENTS

LIST OF TABLES ................................................................................. vi
LIST OF FIGURES ................................................................................ ix
INTRODUCTION ................................................................................. 1
CHAPTER 1

RELATIVE IMPORTANCE OF WATER CHEMISTRY, HABITAT
STRUCTURE AND THE PLANT COMMUNITY TO COASTAL

WET MEADOW SNAIL DISTRIBUTION ................................................... 8
Introduction ................................................................................ 9
Methods .................................................................................... 12
Results ..................................................................................... 22
Discussion ................................................................................. 50

CHAPTER 2

SNAIL-LANDSCAPE ASSOCIATIONS IN COASTAL WET MEADOWS ........... 65
Introduction ............................................................................... 66
Methods .................................................................................... 69
Results ..................................................................................... 76
Discussion ................................................................................. 92

CHAPTER 3

INFLUENCE OF WATER LEVEL FLUCTUATIONS ON COASTAL WET

MEADOW SNAIL COMMUNITIES ........................................................ 105
Introduction .............................................................................. 106
Methods .................................................................................. 108
Results .................................................................................... 119
Discussion ............................................................................... 137

CONCLUSIONS ................................................................................. 1 55

LITERATURE CITED ......................................................................... 165

LIST OF TABLES

TABLE 1. Wet meadows of the northern Lake Huron shoreline sampled
for aquatic snails .................................................................................. 14

TABLE 2. List of species found in the study area and labels used in CCA
diagrams. Nomenclature follows Clarke (1981) and Burch and Jung (1992) ............ 23

TABLE 3. Mean abundance of snails from 26 wet meadows along the
northern Lake Huron shoreline in 1997 and 1998 ............................................ 24

TABLE 4. Summary statistics for the abundance and diversity of snails

measured in 26 wet meadows along the northern Lake Huron shoreline in

1997 and 1998. S = species richness, H ' = Shannon Index of diversity, and

J' = Shannon evenness statistic. See text for calculations .................................. 26

TABLE 5. Summary statistics for three groups of environmental variables

measured in 26 wet meadows along the northern Lake Huron shoreline in

1997 and 1998P values are results of t-tests (with Bonferroni correction)

examining differences between years ........................................................... 31

TABLE 6. Results of canonical correspondence analysis of aquatic snails
in northern Lake Huron coastal wet meadows in 1997 ....................................... 34

TABLE 7. Interset correlations of environmental variables with CCA

axes in 1997 for all variables included in significant model. Significant

variables (P<O.10) are indicated in bold type. Enviromnental variable

abbreviations are used in CCA diagrams ...................................................... 38

TABLE 8. Results of canonical correspondence analysis of aquatic snails
in northern Lake Huron coastal wet meadows in 1998 ....................................... 44

TABLE 9. Interset correlations of environmental variables with CCA

axes in 1998 for all variables included in significant model. Significant

variables (P<O.10) are indicated in bold type. Environmental variable

abbreviations are used in CCA diagrams ...................................................... 45

TABLE 10. Landuse-landcover categories used in the geographic
information system of the northern Lake Huron shoreline, Michigan ..................... 71

TABLE 11. Landscape variables calculated for each of 26 wet meadows
along the northern Lake Huron shoreline, Michigan ......................................... 72

vi

TABLE 12. Summary statistics for three groups of landscape variables.

*Complexity data are calculated for 24 wet meadows; sinuosity and fractal

dimension were unable to be calculated for two meadows located on

islands .............................................................................................. 77

TABLE 13. Results of canonical correspondence analysis of aquatic
snails in northern Lake Huron coastal wet meadows in 1997 .............................. 79

TABLE 14. Interset correlations of environmental variables with CCA
axes in 1997 for all variables included in significant model. Significant
variables (P<O.10) are indicated in bold type ................................................ 80

TABLE 15. Results of canonical correspondence analysis of aquatic
snails in northern Lake Huron coastal wet meadows in 1998 .............................. 88

TABLE 16. Interset correlations of environmental variables with CCA
axes in 1998 for all variables included in significant model. Significant
variables (P<O.10) are indicated in bold type ................................................. 89

TABLE 17. Comparison of the sampling effort in wet meadows of the
northern Lake Huron shoreline from 1997 to 1998, and the classification
of water level changes. See text for details .................................................. 1 13

TABLE 18. Criteria for selecting snail species used in laboratory
experiments ....................................................................................... 115

TABLE 19. Mean, standard deviation, and range of the water level
variables and snail species richness and abundance variables. N=59 in
1997, N=33 in 1998 ............................................................................. 121

TABLE 20. Results of stepwise multiple regression on species richness
and abundance, and presence/absence variables of snails in northern Lake
Huron coastal wet meadows ................................................................... 124

TABLE 21. Variables exhibiting negative and positive relationships to the
six predictors used in the regression modeling .............................................. 129

TABLE 22. Mean depth, range and standard deviation for three
classifications of coastal wet meadow sections in 1997 and 1998. N =
number of sections .............................................................................. 130

TABLE 23. List of species collected in cores from dry sections in 1998
and their mean abundance/section when flooded in 1997 ................................. 131

vii

TABLE 24. Results of statistical analyses comparing snail abundance and
presence/absence between 1997 and 1998 in three classes of wet

meadows. Bold P-values are significant at a<0.10 after correction for

multiple tests ..................................................................................... 133

TABLE 25. The number of snails in each tier at the end of the Snail

Movement experiment. x2 values are from the chi-square goodness of fit

test, DF =2. The total number of each species is also the number at the

beginning of the experiment ................................................................... 136

TABLE 26. Two-way AN OVA of snail survival in the desiccation
resistance experiment ........................................................................... 139

TABLE 27. Great Lakes coastal wet meadow snail community structure

in three categories of wet meadows. N = combined number of snails

collected in 1997 and 1998. ‘*’ = uncommon, ‘**’ = common, and

‘***’ = very common ............................................................................ 158

viii

LIST OF FIGURES

FIGURE 1. Schematic cross-section of a typical coastal wetland showing
the vegetation zones as they change with water depth. Approximate water
levels are indicated for 1997 and 1998. ......................................................... 5

FIGURE 2. Study area showing the locations of 26 wet meadows along
the northern Lake Huron shoreline of Michigan, USA.
Names of the wet meadows are in Table 1 .................................................... 13

FIGURE 3. Diagram of snail and habitat sampling procedures used in the

wet meadows during this study. A) Locations of timed aquatic dip net

samples for snail collections, of water chemistry sampling point, and

habitat sampling points along four radii. B) Detailed diagram of the three

methods used at each of the 20 habitat sampling points ..................................... 16

FIGURE 4. Shannon Index of diversity for wet meadows in the northern

Lake Huron Shoreline of Michigan. Data are arranged in order of the

Shannon Index value. Insets are probability plots for a normal

distribution. (A) 1997. (B) 1998. (C) 1997 and 1998 combined. Note

that wet meadow order is different between (A) and (B) and that the

number of wet meadows was 26 in 1997 and 19 in 1998 (7 sites were

dry in 1998) ......................................................................................... 27

FIGURE 5. Area relationships to abundance (A), species richness (B),

Shannon Index (C), and Shannon Evenness (D) of snail communities.

1997: solid symbols; 1998: open symbols. None of the comparisons

showed significant correlations ................................................................... 29

FIGURE 6. Canonical correspondence analysis of aquatic snails in

northern Lake Huron coastal wet meadows using the 1997 CHEMISTRY

variable group as explanatory variables: (A) species-environmental biplot

(B) sites-environmental biplot. All plotted environmental variables were

included in a significant model; those with solid arrows were significant

variables and dashed arrows indicate non-significant variables. Species

and sites scores are adjusted to scale = 0.3. Environmental variable

abbreviations as in Table 7, species labels as in Table 2, and sites labels as

in Table 1 ........................................................................................... 35

FIGURE 7. Canonical correspondence analysis of aquatic snails in
northern Lake Huron coastal wet meadows using the 1997 STRUCTURE
variable group as explanatory variables: (A) species-environmental biplot
(B) sites-enviromnental biplot. All plotted environmental variables were
included in a significant model; those with solid arrows were significant
variables and dashed arrows indicate non-significant variables. Species

ix

and sites scores are adjusted to scale = 0.3. Environmental variable
abbreviations as in Table 7, species labels as in Table 2, and sites labels as
in Table 1 ........................................................................................... 39

FIGURE 8. Canonical correspondence analysis of aquatic snails in

northern Lake Huron coastal wet meadows using the 1997 PLANT

variable group as explanatory variables: (A) species-environmental biplot

(B) sites-environmental biplot. All plotted environmental variables were

included in a significant model; those with solid arrows were significant

variables and dashed arrows indicate non-significant variables. Species

and sites scores are adjusted to scale = 0.3. Environmental variable

abbreviations as in Table 7, species labels as in Table 2, and sites labels as

in Table 1 ........................................................................................... 41

FIGURE 9. Canonical correspondence analysis of aquatic snails in

northern Lake Huron coastal wet meadows using the 1998 STRUCTURE

variable group as explanatory variables: (A) species-environmental biplot

(B) sites-environmental biplot. Species and sites scores are adjusted to

scale = 0.3. Environmental variable abbreviations as in Table 8, species

labels as in Table 2, and sites labels as in Table 1 ............................................ 46

FIGURE 10. Canonical correspondence analysis of aquatic snails in

northern Lake Huron coastal wet meadows using the 1998 PLANT

variable group as explanatory variables: (A) species-environmental biplot

(B) sites-environmental biplot. All plotted environmental variables were

included in a significant model; those with solid arrows were significant

variables and dashed arrows indicate non-significant variables. Species

and sites scores are adjusted to scale = 0.3. Environmental variable

abbreviations as in Table 8, species labels as in Table 2, and sites labels as

in Table 1 ............................................................................................ 48

FIGURE 11. Venn diagram of the CCA variance partitioning of 1997

species data among three variable groups. Total variance explained by a

variable group is enclosed in [ ] and represented by the size of the circle.

Variance shared by variable groups is indicated in the appropriate

intersections (although size is not proportional). Variance explained is

calculated as the sum of all canonical eigenvalues for a particular analysis

divided by the total inertia ....................................................................... 51

FIGURE 12. Venn diagram of the CCA variance partitioning of 1998
species data among two variable groups. There was not a significant
model for the CHEMISTRY variable group in 1998. Total variance
explained by a variable group is enclosed in [ ] and represented by the size
of the circle. Variance shared by variable groups is indicated in the
appropriate intersections (although size is not proportional). Variance

explained is calculated as the sum of all canonical eigenvalues for a
particular analysis divided by the total inertia ................................................ 52

FIGURE 13. Summary of the 1997 snail-environment relationships using

CCA for each of three environmental variable groups. Arrows represent

environmental gradients for canonical axes with percentage of the species

data explained above. Significant variables correlated with each axis are

listed below; correlated, nonsignificant variable are in ( ). Snail species

associated with each gradient are listed in the corresponding direction ................... 57

FIGURE 14. Summary of the 1998 snail-environment relationships using

CCA for each of three environmental variable groups. Arrows represent

environmental gradients for canonical axes with percentage of the species

data explained above. Significant variables correlated with each axis are

listed below. Snail species associated with each gradient are listed in the

corresponding direction. * No species were strongly associated with the
STRUCTURE gradient ........................................................................... 58

FIGURE 15. Canonical correspondence analysis of aquatic snails in

northern Lake Huron coastal wet meadows using the 1997 PERIMETER

variable group as explanatory variables: (A) species-environmental biplot

(B) sites-environmental biplot. All plotted landscape variables were

included in a significant model; those with solid arrows were significant

variables and dashed arrows indicate non-significant variables. Species

and sites scores are adjusted to scale = 0.3. Landscape variable

abbreviations as in Table 11, species labels as in Table 2, and sites labels

as in Table 1 ........................................................................................ 81

FIGURE 16. Canonical correspondence analysis of aquatic snails in

northern Lake Huron coastal wet meadows using the 1997 CONTEXT

variable group as explanatory variables: (A) species-environmental biplot

(B) sites-environmental biplot. All plotted landscape variables were

included in a significant model; those with solid arrows were significant

variables and dashed arrows indicate non-significant variables. Species

and sites scores are adjusted to scale = 0.3. Landscape variable

abbreviations as in Table 11, species labels as in Table 2, and sites labels

as in Table l ........................................................................................ 83

FIGURE 17. Canonical correspondence analysis of aquatic snails in
northern Lake Huron coastal wet meadows using the 1997
COMPLEXITY variable group as explanatory variables: (A) species-
environmental biplot (B) sites-environmental biplot. All plotted
landscape variables were included in a significant model; those with solid
arrows were significant variables and dashed arrows indicate non-
significant variables. Species and sites scores are adjusted to scale = 0.3.

xi

Landscape variable abbreviations as in Table 11, species labels as in Table
2, and sites labels as in Table 1 .................................................................. 86

FIGURE 18. Canonical correspondence analysis of aquatic snails in

northern Lake Huron coastal wet meadows using the 1998 PERIMETER

variable group as explanatory variables: (A) species-environmental biplot

(B) sites-environmental biplot. All plotted landscape variables were

included in a significant model; those with solid arrows were significant

variables and dashed arrows indicate non-significant variables. Species

and sites scores are adjusted to scale = 0.3. Landscape variable

abbreviations as in Table 11, species labels as in Table 2, and sites labels

as in Table l ........................................................................................ 90

FIGURE 19. Canonical correspondence analysis of aquatic snails in

northern Lake Huron coastal wet meadows using the 1998 CONTEXT

variable group as explanatory variables: (A) species—environmental biplot

(B) sites-environmental biplot. All plotted landscape variables were

included in a significant model; those with solid arrows were significant

variables and dashed arrows indicate non-significant variables. Species

and sites scores are adjusted to scale = 0.3. Landscape variable

abbreviations as in Table 11, species labels as in Table 2, and sites labels

as in Table l ........................................................................................ 93

FIGURE 20. Venn diagram of the CCA variance partitioning of 1997

species data among two variable groups. Total variance explained by a

variable group is enclosed in [ ] and represented by the size of the circle.

Variance shared by variable groups is indicated in the appropriate

intersections. Variance explained is calculated as the sum of all canonical

eigenvalues for a particular analysis divided by the total inertia ........................... 95

FIGURE 21. Venn diagram of the CCA variance partitioning of 1998

species data among two variable groups. Total variance explained by a

variable group is enclosed in [ ] and represented by the size of the circle.

Variance shared by variable groups is indicated in the appropriate

intersections. Variance explained is calculated as the sum of all canonical

eigenvalues for a particular analysis divided by the total inertia ........................... 96

FIGURE 22. Summary of the 1997 snail-environment relationships using

CCA for each of three environmental variable groups. Arrows represent

environmental gradients for canonical axes with percentage of the species

data explained above. Significant variables correlated with each axis are

listed below; correlated, nonsignificant variable are in ( ). Snail species

associated with each gradient are listed in the corresponding direction ................... 98

FIGURE 23. Summary of the 1998 snail-environment relationships using
CCA for each of three environmental variable groups. Arrows represent

xii

environmental gradients for canonical axes with percentage of the species

data explained above. Significant variables correlated with each axis are

listed below; correlated, nonsignificant variable are in ( ). Snail species

associated with each gradient are listed in the corresponding direction ................... 99

FIGURE 24. Diagram of the side view of one chamber in the Snail

Movement experiment showing the arrangement of sand, sediment and

water producing 3 tiers: shallow (A), medimn (B), and deep (C). Initial

water depth was adjusted so that water level was 4 cm above detritus. The

drains of all 6 chambers were connected to one common valve used to

maintain the same, constant reduction in water level. Width=42 cm .................... 116

FIGURE 25. Scatterplot matrix of the 6 water level explanatory variables

used in the regression analyses for 1997. Ellipses are 68% confidence

intervals (:1 SD. around x and y axes), N=59. Histograms are shown on

the diagonal for each variable. Values in each graph are r2 values that

includes all points, but note the presence of extreme values that would

influence this value in many variable pairs. Axes labels are explained in

the footnote for Table 19 ........................................................................ 122

FIGURE 26. Scatterplot matrix of the 6 water level explanatory variables

used in the regression analyses for 1998. Ellipses are 68% confidence

intervals (:1 SD. around x and y axes), N=33. Histograms are shown on

the diagonal for each variable. Values in each graph are r2 values that

includes all points, but note the presence of extreme values that would

influence this value in many variable pairs. Axes labels are explained in

the footnote for Table 19 ........................................................................ 123

FIGURE 27. Species richness (S) — Depth curve for snails in coastal wet
meadows of the northern Lake Huron shoreline during 1997. N = 59. The
line represents the significant (P<0.05) linear model, S = 0.027(Depth) +

4.642, R2 = 0.09 ................................................................................. 127

FIGURE 28. Abundance — Depth curve for Gyraulus deflectus in coastal
wet meadows of the northern Lake Huron shoreline during 1998. N = 33.
The line represents the significant (P<0.05) linear model, Abundance =

6.240(Depth) + 64.244, R2 = 0.20 ............................................................ 128

FIGURE 29. Number of sections (out of 15 that were dry in 1998) in

which living snails were present in 1997 (wet, sampled by dip nets) and

1998 (dry, sampled by cores). ‘*’ indicate significant differences between

years (P<O.10) .................................................................................... 134

FIGURE 30. Mean abundance of snails from three classes of coastal wet
meadows in 1997 and 1998. ‘Dry’ wet meadows were flooded in 1997
and dry in 1998 (results not shown here because of different sampling

xiii

methods), N=15; ‘Lower’ wet meadows were flooded in both years, but

shallower in 1998, N=24; and ‘Same’ wet meadows were the same depth

in both 1997 and 1998, N=5. Error bars are :t l S.E.M. ""’ indicates a

significant different between years (P<O.10). Note differences in scaling

of the vertical axes .............................................................................. 135

FIGURE 31. Desiccation survival curves of 7 species of snails. The

dashed lines in each graph indicate that all snails were assumed to be alive

at week 0. Species names not connected by a common line are

significantly different (Tukey HSD, P < 0.05). Error bars are i 1 SEM ................ 138

FIGURE 32. Correlation of the percentage of snails buried in experiment
I with the percentage surviving in experiment II (r = 0.91, N=5, P = 0.03) ............. 140

FIGURE 33. Conceptual model of the determinants of snail community

structure in a Great Lakes Coastal wet meadow complex (large box).

Block arrows indicate chemical, structural, and plant gradients within a

wet meadow gradient. Solid arrows are significant associations observed

in the present study whereas dotted arrows are weaker or implied

associations ....................................................................................... 156

FIGURE 34. Schematic cross-section of a coastal wetland along the

northern Lake Huron shoreline showing the three wet meadow snail

communities (semi-isolated, shallow and deep) described in a conceptual

model (Figure 33). Relative placement of snail species follows Table 27;

species in bold type are generalists. Species with ‘*’ are able to tolerate

drying as water levels decline (see Figures 29 and 31). Approximate water

levels are indicated for 1997 and 1998. ............................................... ' ........ 159

xiv

INTRODUCTION

Aquatic snails are usually an important component of freshwater invertebrate
communities (e.g. Weatherhead and James 2001). Many play key roles as consumers by
grazing upon algal and bacterial communities present in biofilms on the surfaces of
submersed objects and vegetation (Russel-Hunter 1978). Snails affect the microbial
biofilm by influencing the abundance and diversity of species within the biofilm itself
(Doremus and Harman 1977, Lowe and Hunter 1988), as well as invertebrate grazers
feeding upon it (Hawkins and Furnish 1987, Harvey and Hill 1991, Hill 1992). Other
snail species may consume living vegetation or act as detritivores by feeding on decaying
vegetation (Calow 1970, Reavell 1980). Snails, in turn, are often the prey of a variety of
predators (leeches: Townsend and McCarthy 1980, Bronmark and Malmqvist 1986;
crayfish: Crowl 1990, Weber and Lodge 1990; insects: Kesler and Munns 1989; and fish:
Brown and DeVries 1985, Turner 1996). Because of their importance in most aspects of
the aquatic food web, it is not surprising that many researchers have investigated factors
thought to affect snail distribution and abundance (e. g. Dussart 1976, Bronmark 1985,
Lodge et al. 1987).

Traditionally, explanations of the patterns seen in freshwater snail distributions
have focused on water chemistry, especially calcium concentration. Because calcium is a
major component of snail shells, the explanation seems obvious. However, many studies
have failed to find a significant correlation between environmental calcium and shell
calcium content (Hunter and Lull 1977, Mackie and Flippance 1983a,b). In some cases,
the abundance of snails was related to calcium (Dussart 1976, 1979), though diversity

declined only in lakes with very low calcium concentrations (<5 mg/l) in a study of 1500

Norwegian lakes (Okland 1983). Even in calcium-poor lakes, some species were found
in abundance (Rooke and Mackie 1984). The variability among previous studies
suggests a complex, inconsistent relationship between water calcium levels and snail
communities that is not well understood, and suggests that other factors should be
investigated. Lodge et al. (1987) summarized other examples and supported the
suggestion that, in most contexts, physicochemical factors (including calcium) were not
the primary determinants of snail distribution and abundance.

In addition to chemical factors, other studies have investigated how habitat
diversity and macrophytes (or productivity in general) affected snail communities. High
habitat diversity coupled with interspecific differences among snails in habitat and food
preferences may result in higher snail diversity (Ross and Ultsch 1980, Lodge et al.
1987). Numerous studies have shown either a general association between snails and
macrophytes (Brbnmark 1985) or associations between snails and specific macrophyte
species (Bickel 1965, Pip and Stewart 1976, Pip 1978, 1985, 1991, Lodge 1985, 1986,
Beckett et al. 1992). Costil and Clement (1996) examined snail and plant communities in
relation to trophic levels of the lakes, and underscored possible interactions between snail
distribution and physicochemical factors (especially nutrients), lake productivity, and
macrophyte abundance/diversity. Brown and Lodge (1993) cautioned that apparent snail-
macrophyte associations could be better explained by the greater surface area of these
habitats rather than a distinct preference per se.

The application of the theory of island biogeography (MacArthur and Wilson
1967) in which lakes are treated as islands within the surrounding land has been used

with some success in describing the mollusk species richness of lakes within a region

(Lassen 1975, Aho 1978a, Browne 1981, Jokinen 1987). In general, the number of
species was expected to increase with the surface area of the lake in a dynamic
equilibrium maintained by primarily biotic factors (e. g. dispersal rates, macrophyte
heterogeneity, etc.) rather than by abiotic chemical factors. Some studies have examined
the effects of variables above and beyond the species-area effect in an attempt to further
discern their role in determining snail distributions. For example, Aho (1978b) found
that oligotrophic lakes generally followed the predictions of the theory (i.e. biotic factors
were important), and that, only in lakes with low calcium concentrations, did the latter
variable better explain the variation in species distributions; likewise, the model did not
hold for larger lakes in the study. Lassen (1975) found differences in oligotrophic versus
eutrophic lakes, and Bro'nmark (1985) found that macrophyte diversity and the snail
species in nearby ponds added explanatory ability beyond area.

Clearly, a multitude of factors influence snail community structure, and the
interactions among them are not well understood. Lodge et al. (1987) attempted to
clarify the importance of physicochemical and biotic factors in the development of a
conceptual model that differentiated between three spatial scales. In this model,
chemistry (especially calcium) was the most important factor at the largest, regional
scale. If calcium levels were sufficient, then habitat diversity and food selectivity
became important among lakes and among habitats within lakes. Additional biotic
factors included disturbance, competition, and predation, but their effects were
constrained by a habitat continuum from temporary ponds to permanent lakes.

In order to further identify important determinants of freshwater snail community

structure, I conducted a multifaceted study in northern Lake Huron coastal wetlands of

Michigan’s Upper Peninsula. Coastal wetlands and their dominant vegetation types are
influenced by water depth and length of hydroperiod (e. g. permanent or seasonal
inundation). In most cases, there is a distinct zonation of vegetation types from the open
water of Lake Huron to the shoreline that includes bulrush, cattail, wet meadow, and
shrub zones (Mine and Albert 1998) (Figure 1). In the present study, snails were
collected from 26 wet meadows during 1997 and 1998 along with numerous
environmental characteristics based on field measurements and a geographic information
system (GIS). Additionally, laboratory experiments were conducted to help clarify the
effects of a drastic decrease in water levels that occurred during the study.

Three sets of analyses were performed to answer specific questions regarding the
aquatic snail community structure. Each of these analyses, which corresponds to a
separate chapter, is briefly described below.

Chapter One includes a description of the aquatic snail commmrity along the
northern Lake Huron shoreline and tests for associations between the snail community
and the more traditional determinants of structure: water chemistry, habitat structure
(heterogeneity), and the aquatic plant community. While there have been many similar
analyses (see above), my approach was unique because I combined them into one overall
analysis. If the patterns of previous studies in which no one factor strongly influenced
the snail community hold true, each of the three analyses alone would not provide strong
explanatory power, but by using a variance partitioning method, I hOped to describe the
relative importance of all three aspects.

Chapter Two expands upon the first by examining the less well-studied effects of

   

I r Open
Water

       

«'4‘: MIL

 

INN“ ‘

Shrub , ' """ ‘ """ 1998

Wet / [Vatelr
Meadow eve s

"' "‘ " “ """" 1997
Upland

Meadow

 

 

 

 

 

Cattail
Bulrush

Figure 1. Schematic cross-section of a typical coastal wetland showing the vegetation

zones as they change with water depth. Approximate water levels are indicated for 1997
and 1998.

the surrounding landscape on the snail community. While landscape ecology, which
deals with the effects of spatial patterns (e.g. patch size, shape, connectivity,
fragmentation and edge-effects) on organisms (see F orrnan 1995), has gained prominence
as a distinct field over the past two decades, only recently has it been applied to aquatic
systems (e.g. Palmer et al. 2000, Wiens 2002). I know of only two studies that
specifically examine the effects of landscape on snail communities. Magnin et al. (1995)
examined the effects of cultivation patterns and temporal changes on land snail
communities and Lewis and Magnuson (2000) found that differences in snail species
richness in lakes were dependent on the position of the lake in the catchment and their
stream connectivity. The success of landscape ecology in answering questions regarding
the distribution of a variety of animals suggested that it might be useful in explaining
wetland snail distributions. Because I wanted to know the effects of landscape patterns
on the snail community, I examined these relationships after factoring out the effects of
chemistry, structure and plant community. Thus, any significant effects were above and
beyond those of chemistry and habitat characteristics.

Between sampling periods from 1997 to 1998, the water levels of Lake Huron
dropped considerably (Figure l) leaving many sampling sites shallower or without
standing water — an obvious influence on aquatic snail communities. While water depth
was included directly in analyses of Chapter One and indirectly in Chapter Two, the
effects of water level changes were not specifically addressed. In Chapter Three, I
examined in more detail the effects of water level changes on the snail community
through regression modeling. The proximate causes of the observed changes were then

investigated using laboratory experiments of snail behavior and desiccation resistance.

Lastly, the results of my analyses were used in the construction of a conceptual
model describing the aquatic snail community structure in northern Lake Huron coastal
wetlands. My results add to the current understanding of determinants of snail
distribution as well as suggest further areas of research likely to increase knowledge of
these wetlands. In addition, the model can aid investigations of processes that affect the

biodiversity of this very important and declining habitat.

CHAPTER ONE

RELATIVE IMPORTANCE OF WATER CHEMISTRY, HABITAT
STRUCTURE AND THE PLANT COMMUNITY TO COASTAL WET

MEADOW SNAIL DISTRIBUTION

INTRODUCTION

Patterns in freshwater snail distributions have traditionally been explained on the
basis of water chemistry, especially calcium concentration. Because calcium is a major
component of snail shells, the explanation seems obvious. However, many studies have
failed to find a significant correlation between environmental calcium and shell calcium
content (Hunter and Lull 1977, Mackie and Flippance 1983a,b). In some cases, the
abundance of snails was related to calcium (Dussart 1976, 1979), though diversity
declined only in water with very low (<5 mg/l) calcium concentrations (Okland 1983).
Even in calcium-poor lakes, some mollusks were found in abundance (Rooke and Mackie
1984). The variability among previous studies suggests a complex, inconsistent
relationship between water calcium levels and snail communities that is not well
understood, and suggests that other factors should be investigated. Lodge et al. (1987)
summarized other examples and supported the suggestion that physicochemical factors
(including calcium) are not the primary determinants of snail distribution and abundance
in most contexts.

In addition to chemical factors, other studies have investigated how habitat
diversity and macrophytes (or productivity in general) affected snail communities.
Because each type of habitat or macrophyte assemblage (often a major component of
aquatic habitats) may support different species, higher species richness may be correlated
with increasing habitat diversity (Bronmark 1985). High habitat diversity coupled with
interspecific differences among snails in habitat and food preferences may result in
higher snail diversity (Ross and Ultsch 1980, Lodge et al. 1987). Numerous studies have

shown either a general association between snails and macrophytes (Bronmark 1985) or

associations between snails and specific macrophyte species (Bickel 1965, Pip and
Stewart 1976, Pip 1978, 1985, 1991, Lodge 1985, 1986, Beckett et al. 1992). Costil and
Clement (1996) examined snail and plant communities in relation to trophic levels of the
lakes, and underscored possible interactions between physicochemical (especially
nutrients), lake productivity, and macrophyte abundance/diversity. Brown and Lodge
(1993) cautioned that apparent snail-macrophyte associations could be better explained
by the greater surface area of these habitats rather than a distinct preference per se. Even
so, most studies agree that habitat diversity alone, and possibly the types of macrophytes
present affect the distribution of snails to some degree.

The application of the theory of island biogeography (MacArthur and Wilson
1967), in which lakes are treated as islands within the surrounding land, has been used
with some success in describing the mollusk species richness of lakes within a region
(Lassen 1975, Aho 1978a,b, Browne 1981, Jokinen 1987). In general, the number of
species was expected to increase with the surface area of the lake in a dynamic
equilibrium maintained by primarily biotic factors (e. g. dispersal rates, macrophyte
heterogeneity, etc.) rather than by abiotic chemical factors. Some researchers have used
this theory to examine the effects of variables above and beyond the species-area effect in
an attempt to further define their role in determining snail distributions. For example,
Aho (1978b) found that oligotrophic lakes generally followed the predictions of the
theory (i.e. biotic factors were important), and that only in lakes with low calcium
concentrations did the latter variable better explain the variation in species distributions;
likewise, the model did not hold for larger lakes in the study. Lassen (1975) found

differences in oligotrophic versus eutrophic lakes and Bronmark (198 5) found that

10

macrophyte diversity and the snail species in nearby ponds added explanatory power
beyond area.

Clearly, a multitude of factors influence snail community structure, and the
interactions among them are not well understood. Lodge et al. (1987) attempted to
clarify the importance of physicochemical and biotic factors by developing a conceptual
model that differentiated between three spatial scales. In this model, chemistry
(especially calcium) was the most important factor at the largest, regional scale. If
calcium levels were sufficient, then habitat diversity and food selectivity became
important among lakes and among habitats within lakes. Additional biotic factors
included disturbance, competition, and predation, but their effects were constrained by a
habitat continuum from temporary ponds to permanent lakes.

The objective of the present study was to describe the aquatic snail community of
a Great Lakes coastal wetland complex and evaluate the relative importance of water
chemistry, habitat structure, and the plant community in explaining snail distribution and
abundance. Because of the lack of research on these wetlands in general (Krieger 1992)
and specifically on the snail community, my approach was explorative in nature, with my
only hypothesis being that habitat structure and the plant community should be more
important than water chemistry in determining snail distributions. According to the
conceptual model of Lodge et al. (1987), this was to be expected because the wetlands

were located in the same biogeographic region.

11

METHODS
Study Area

I sampled 26 distinct coastal wet meadows along a 60 km length of the northern
Lake Huron shoreline in Michigan’s Upper Peninsula (Figure 2, Table 1). The wet
meadows all were part of a complex of Great Lakes marshes that were among some of
the most pristine coastal wetlands (Mine and Albert 1998). The area was part of the
Niagaran Escarpment composed of Silurian-age limestone and dolomite bedrock. The
vegetation was predominantly hummock-forming grasses (Calamagrostis canadensis)
and sedges (Carex stricta, C. aquatilis, and C. latifolia) with low densities of bulrush
(primarily Scirpus acutus) and cattail (primarily Typha angustifolia). The substrate,
composed of muck or mucky-peat sediments covered by dense layers of the previous
season’s grass and sedge leaves, provided a structurally complex habitat (along with the
hummock structure) in which most snails were found. Some sites also contained a shrub
zone including Salix spp., Myrica gale, and Alnus rugosa with Potentilla spp. scattered
throughout. Additional information was given by Burton et al. (1999).

Snail Collection

Transects running perpendicular to the shoreline of Lake Huron were established
in each wet meadow and divided into 50 m sections. The number of sections sampled in
each wet meadow ranged from 1 to 3 depending on the size of the wet meadow (Table 1)
and the presence of standing water deep enough to sample with a dip net (~10 cm).
Each wet meadow was assumed to by hydrologically connected with Lake Huron
although, in some, this was modified by natural barriers (sandy ridges creating a swale-

like effect) or roads (with culverts maintaining water flow) (Table 1). Snails were

12

._ 038. E 2m maoumoE 83 2: 90 onm2
.<mD .EmEumE we 2:227. :95: 8:5 Eocto: 2: was“ maouuoE 83 on .8 25:82 2: 9:39? was beam ._ Esmi

 

 

 

 

Exow

 

 

 

 

 

 

 

QQKEE EVNBxN

 

 

 

 

 

13

Table 1. Wet meadows of the northern Lake Huron shoreline sampled for aquatic snails.

 

Site
label

Wet meadow site

Area

Latitude (N) Longitude (W) (ha)

Sections

sampledl Barrier2

 

N~<><€<CHMWOWOZZWWHHIOWWUOW>

Pontchartrain 1
Pontchartrain 2

St. Martin’s
Carpenter 1
Carpenter 2
Search 1
Search 2
Mismer
Hessel 1
Hessel 2
Hessel 3
Mackinac
Rudd
Flowers
Hill Island
Lakeside
Port Dolomite
Prentiss
Scotty
Seymour 1
Seymour 2
Seymour 3
Sweets 1
Sweets 2
Sweets 3
Sweets 4

46°01 '45 "
46°01 '33 "
46°01 '59"
46°01 '1 5 "
46°00'49"
46°00'01 "
46°00'02"
46°00'37"
46°00'36"
46°00'37"
46°00'36"
46°00'24"
46°00'10"
45°59'28"
45°59'l 5 "
45°59’56"
45°59'43 "
45°59'36"
45°58'56"
45°58'29"
45°58'36"
45°58'44"
46°01 ’28"
46°01 '40"
46°01 '40"
46°01 '46"

84°35’14"
84°34'58"
84°34'48"
84°32'24"
84°32'16"
84°30'26"
84°30'13"
84°27’44"
84°27'09"
84°26’52"
84°26'13 "
84°24’46"
84°23'30"
84°19'25"
84°19’00"
84°18'02"
84°16'12"
84°13'41"
84°12'12"
83°55'54"
83°55'38"
83°55'30"
83°56'37"
83°56’22"
83°56'27"
83°56'29"

1.18
2.47
3.24
2.60
3.28
2.67
0.12
11.12
8.93
2.28
1.83
13.70
7.52
3.84
1.94
1.92
2.82
22.09
11.74
3.53
1.27
25.24
4.80
2.94
0.89
0.69

2/1
3/0
2/0
2/1
1/0
2/0
1/1
3/2
2/1
2/2
2/1
3/1
3/0
3/2
1/0
2/1
3/3
3/2
3/3
1/1
3/2
3/0
3/3
3/3
2/2
1/1

(59/33)

S
S
-R
-R
-R
S
S
R
-R

 

1. Number in 1997/ 1998

2. ‘-’ = no barrier; ‘S’= sandy ridge; ‘R’ = road between wet meadow and Lake Huron;

‘-R’ = road adjacent to wet meadow but not separating site from Lake Huron

l4

collected between 11-14 July 1997 and 7-10 July 1998 using timed D-frame aquatic dip
net samples (Brown 1979). Analyses were based on only these collections to minimize
variation in snail abundances caused by differences in life histories among species.
Sampling at this time minimized variation caused by both the late summer death of
annual, semelparous species and the early recruitment of juveniles into the population.

Two l-minute sweeps using a standard D-frame aquatic dip net were conducted in each
section covering an area of approximately 10 m2 (Figure 3). The contents were

combined and washed through a 1 mm sieve bucket, and then preserved in 95% ethanol.
In the laboratory, snails in the preserved samples were sorted to species and enumerated
(empty shells were not included). Taxonomic nomenclature followed that of Clarke
(1981) and Burch and Jung (1992).

Environmental Variables

Three variable groups were used to describe the wet meadow environment.

The first group contained seven water chemistry (CHEMISTRY) variables
measured in the center of each section (Figure 3) at the time of snail sampling (pH and
conductivity using an Altec monitor II meter and YSI model 31 conductivity bridge,
respectively) or in the laboratory using water samples collected during snail sampling.
Alkalinity was measured within 4 hours of collection by sulfuric acid titration to the
bromcresol green endpoint, and the cations (calcium, potassium, magnesium and sodium)
were measured using flame atomic absorption spectroscopy; samples were frozen until
analyzed. Only alkalinity, conductivity and pH were measured in 1998.

Nine variables describing the habitat structure (STRUCTURE) were measured for

each 50-meter section using two techniques at each of 20 habitat-sampling points located

15

 

 

 

 

 

 

I
}(-—— Sampling transect
l I.
: Dip net collection
I
l (\ Habitat sampling
' radius
Water chemistry\) '
sampling point I 5m 50 meters
l
I
1 15m
I
25m
\ 35m

\ “~\\
i ‘~‘ I
‘, ‘x\ } 45m VL
a “\‘ l
‘\ “\ I

\ ‘s '

I \s‘

B \ \\\

\ ‘s

\ ‘\

\ ‘s

| x

t

I

\

|

|

|

I

  
 
 
 

Habitat sampling
radius

    
 

Point-intercept
sampling point

1x1m sampling area

2.5 m radius
sampling area

Figure 3. Diagram of snail and habitat sampling procedures used in the wet meadows

during this study. A) Locations of timed aquatic dip net samples for snail collections, of
water chemistry sampling point, and habitat sampling points along four radii.

B) Detailed diagram of the three methods used at each of the 20 habitat sampling points.

16

on four habitat sampling radii originating at the center of each section (Figure 3). Five
sampling points were located at 10-m intervals along each of the four sampling radii.
Point-intercept sampling was conducted using a 2-m metal rod, stratified into
eight 25 cm sections (strata), that was passed through the vegetation (Rotenberry and
Wiens 1980). Using the rod, one of each of the following measurements was made at
each sampling point. Grass density was measured as the total number of grass and sedge
stems or leaves touching the rod (i.e. number of hits) in any of the eight strata; shrub
density was measured as the total number of hits by shrubs in all strata; and vegetation
density was measured as the total number of hits in all strata by any living vegetation
(grasses, other macrophytes, and shrubs). Foliage diversity, a measure of the shrub cover
and shading, was calculated as the number of 25-cm sections with at least one shrub hit.
Water depth, hummock height and grass height were also measured using the metal rod.
Water depth and grass height were standardized among sampling points by measuring
these from the bases of hummocks (i.e. positioning the rod in a valley between
hummocks) near the sampling point. This approach eliminated the possibility of
variation in grass height and water depth caused by random positioning of the rod on the

side or top of a hummock.

Hummock frequency and open water frequency were estimated by placing a l-m2
sampling quadrat (Figure 2) at each sampling point and recording the presence or absence
of hummocks and open water (areas >0.25 m2). The frequency of occurrence was

calculated as the proportion of sampling points at which each of these were present (out
of 20 sampling points per section).

The third group (PLANT) included 12 variables describing the frequency of

17

occurrence of common plant species or types. The frequency of occurrence for

macrophytes was measured as described above using a l-m2 sampling quadrat.

Macrophytes measured were grarninoids (including Calamagrostis canadensis, Carex
stricta, C. aquatilis, and C. latifolia), cattail (primarily T ypha angustifolia; also T.
latifolia and hybrids), bulrush (primarily Scirpus acutus; also S. americanus and S.
validus), duckweed (primarily Lemna minor; also L. trisulca), Nuphar (N. variegatum),
Polygonum (P. amphibium), Potamogeton (P. natans), and submersed vegetation
(Potamogeton spp., Myriophyllum spp., Utricularia spp.; also filamentous green algae).
Woody vegetation occurrence was sampled using a 2.5-m radius circle centered on each
habitat sampling point (Figure 2) and the presence/absence recorded. The variables
measured were alder (Alnus rugosa), willow (primarily Salix petiolaris), shrub (any other
shrub species <20 m in height, primarily Potentilla palustris and Myrica gale), and tree
(any >2.0 m in height, including Larix laricina and Populus spp.). Trees and shrubs were
measured in 1997 only because there was little or no change between years.
Statistical Analyses

Patterns of abundance and diversity. Abundance was estimated as the total
number of individuals collected in each 50-m section of wet meadow. The change in
abundance for each species from 1997 to 1998 was examined with t-tests for each
species; a Bonferroni technique was used to correct for multiple tests as part of the
SYSTAT software (SPSS 1999). Three measures of diversity were calculated for each
wet meadow using the average abundance in each section (in sites containing multiple
50-m sections, Table 1). Species richness (S) was measured as the total number of

species recorded at a site. Shannon Index (H ') (see Magurran 1988) was calculated as:

18

H' = ->3PI10gPI (1)
where p,- is the proportion of individuals of species 1'. This index takes into account both

the richness and the evenness of the community. Because each wet meadow was not

sampled with replicates, the variance of H ' was approximated by:
2 _ 2 2 2
S H’_[2fi108 fI -(2fI10gfi) Nil/'1 (2)
where f,~ is the number of snails of species 1', and n is the total number of snails (Zar 1999).

No statistical tests were used to compare H ' among wet meadows, however, confidence
intervals were constructed using the square root of the variance calculated in (2) as an
estimate of the standard deviation. The 95% confidence interval was then considered as
mean :t 2 s.d. J', a measure of evenness only, was calculated as:

J'=H’/H,,,ax' (3)
where:

Hmax' = log S (4)

The effect of wet meadow size on abundance and diversity was investigated
through regression analyses of each dependent variable separately with area (log
transformed). Species richness values were log transformed prior to analysis as is
customary (Rosenzweig 1995); other variables were untransformed. Regression analyses
were performed using SYSTAT 9 (SPSS 1999).

Multivariate analyses of environmental variables and species data. Canonical
correspondence analysis (CCA), using the statistical software CANOCO (ter Braak and
Smilauer 1999), was utilized to express major relationships between the species and

environmental variables. CCA is a technique that integrates ordination with regression

19

resulting in uncorrelated (i.e. canonical) linear combinations of environmental variables
(ter Braak and Smilauer 1998). CCA is especially appropriate when species show a
unimodal response to an environmental gradient as is often recognized in mollusks
(Bishop 1981). Over partial lengths of the gradient, the species response may appear
linear, but CCA is robust enough to meet these conditions (ter Braak and Verdonschot
1995, ter Braak and Smilauer 1998).

The three environmental variable groups were examined separately, limiting the
number of variables used in each analysis, as suggested by Borcard et al. (1992).
Abundance data were log (x+1) transformed to normalize the variance and rare species
were down-weighted using an algorithm available in CANOCO prior to analysis. Down-
weighting the rare species reduces the disproportionate effect that they may have on the
ordination (ter Braak and Smilauer 1998). Variables with high multicollinearity were
examined and those with a variable inflation factor greater than 20 were removed (ter
Braak and Smilauer 1998).

Two separate analyses were conducted. First, I wanted to determine which of the
environmental variables within each variable group were most significant in explaining
the variation in species data and second, to examine the relative importance of the three
variable groups. Data from individual sections were included in the data matrix rather
than each wet meadow as a whole, in order to increase the power of the analyses.
Although this did not account for effects of spatial autocorrelation, it was assumed that
snails would be influenced most strongly by habitat differences within each 50-m section,
and that any effects of spatial autocorrelation could be detected in the sites-environmental

biplots from the CCA analyses.

20

I used the forward selection procedure within CANOCO (ter Braak and Smilauer
1998) to select the largest number of variables from each environmental variable group
that resulted in a significant model (to determine overall variance explained by the
variable group). Significance was determined for both the lst axis and all axes combined
and a model was considered significant only if both of these met an a priori a < 0.10. 1
used a more liberal value because of the variability inherent in field observations, and to
maintain an acceptable Type 11 error rate. Significance testing was conducted using the
Monte Carlo permutation testing (199 permutations) procedure within CANOCO (ter
Braak and Smilauer 1998).

After defining a significant model, significant variables (i.e. those that explained
the greatest amount of species variation) were identified within each variable group. The
results were interpreted with the use of CCA biplots, in which the environmental
variables were represented as vectors among either site scores or species scores. Site or
species scores with positive relationships to the environmental variable were in the vector
direction whereas those with negative relationships were in the opposite direction. In
addition, the length of the vector represented the relative importance of that
environmental variable to the ordination (ter Braak and Verdonschot 1995).

In order to examine the overall variance explained and to partition the variance
among variable groups, I conducted partial CCA using the method of Borcard et al.
(1992), which removed the influence of covariables from the variable set in question.
This method was similar to a normal CCA but with the extra requirement that the
extracted gradients were uncorrelated with the covariables (ter Braak and Verdonschot

1995). The variation explained by a set of variables was calculated as the sum of all

21

constrained eigenvalues divided by the total inertia, which is analogous to the total
variance. In partial CCA, then, the variation explained by one variable group only was
based on the sum of all constrained eigenvalues after removing the effects of the
covariables (Borcard et al. 1992). In this way, the variance explained by one variable
group was separated from the variance shared by two or more variable groups (Okland

and Eilertsen 1994).

RESULTS
Patterns of abundance and diversity

A total of 20,410 individuals from 59 sections and 15,127 individuals from 33
sections were collected during 1997 and 1998, respectively. Fewer sections were
sampled in 1998 because Lake Huron water levels dropped 28 cm from the levels in 1997
(U SACE 2000) leaving sections or entire wet meadows without standing water. The
snails were distributed among 19 species in seven families (Table 2). The four most
abundant species were Physa gyrina, Gyraulus parvus, Planorbula armigera, and G.
deflectus (Table 3). These four species along with Planorbella trivolvis and F ossaria
parva were the most common, occurring in >40% of the sections sampled (Table 3).
Physa gyrina and F. parva decreased significantly (P<O.10), and G. deflectus and P.
armigera increased significantly (p<0.10) in abundance from 1997 to 1998; although
variable, all others did not change significantly from one year to the next (Table 3). Of
the 19 species collected, five were only collected in one year and were uncommon with
fewer than five individuals being collected for each species (abundance <0.10/section in

Table 3). Among wet meadows, the abundance varied greatly, ranging from only 27

Table 2. List of species found in the study area and labels used in CCA diagrams.
Nomenclature follows Clarke (1981) and Burch and Jung (1992).

 

 

Species Label
Pulmonata
Physidae
Physa gyrina Say, 1821 Pg
Aplexa elongata (Linnaeus, 175 8) Ac
Planorbidae
Gyraulus deflectus (Say, 1824) Gd
Gyraulus parvus (Say, 1817) Gp
Gyraulus cristus (Linnaeus, 175 8) Ge
Planorbula armigera (Say, 1821) Pa
Planorbella trivolvis (Say, 1817) Pt
Planorbella campanulata (Say, 1821) Pc
Promenetus exacuous (Say, 1821) Pe
Lymnaeidae
Stagnicola elodes (Say, 1821) Se
F ossaria obrussa (Say, 1825) F0
F ossaria parva (Lea, 1841) Pp
Acella haldemani (Binney, 1867) Ah
Ancylidae
F errissia parallela (Haldeman, 1841) F l
Prosobranchia
Hydrobiidae
Amnicola limosa (Say, 1817) Al
Amnicola walkeri Pilsbry, 1898 Aw
Pyrgulopsis Iustricus (Pilsbry, 1890) P1
Bithyniidae
Bithynia tentaculata (Linnaeus, 1758) Bt
Valvatidae
Valvata tricarinata (Say, 1817) Vt

 

23

can? J33 30% 923 can omumw

.2822 @8me Son M22222 Smovmv Eobbmu b2cmoEcw_mu** ”Eco combo .233 @223 mime 82 .ovmv Sesame 3258:222me u .2. .m
2628228 33 328% 22033 E 2202208 «0 0932288; .m
.32 222 28:03 mm .32 E 2226?... mm 28:03 2208.3 222 328:8 89222222 .2

 

 

- 8.22 0 Se 8.22 $2 N 8.22 8.22 assess seems
a 23 N2. 2.2 eNN $2 8 £2 £2 ss§2§ 3:52.222
- 8.22 o 8.22 8.22 a?” 2 No.22 8.22 acres E3322
- 8.22 o 8.22 8.22 a? N 8.22 8.22 2329: £8,225
a 2.2 2.2 3.22 E22 S2 2: 2.22 :22 use: 38.22%
a 2.2 2 2 one 8.22 New N 8.22 N222 3.225% SEEN
- SN 2 8.22 8.22 Se 22 8.22 8.22 22320352 Saw
.. 3.2 MN 32 SE 8.2 N2. $2 NAN ease nausea
8 SN 0 2.22 N22 3.2 2a 8.2 SN eases $238.2
a N. N N 8.22 Na New N 22.22 222.22 sees 58.53%
- N 2 .2 N 3.22 NN.2 8.22 o 8.22 8.22 asses erases
2 SN N Na 2N.o NE 2 8.22 222 2.3328258 323825
a N2..N2. c2 2 N3 0: 2.2. a N5 NNe eases 52335.:
.. 3.5 N2 SN NNNN 2.2.22 N 2 N 22222 222.8 cease eases:
- SN 2 8.22 8.22 8.22 o 8.22 8.22 .328 3:36
a 8.3 9% 2N.2.N N24N2 Sea 82. 22.2.2 No.8 stem 3.226
r. 3.8 22: SN 8.82 2.: 322. 2.2 2 one. seeps $238.6
8 Se . ow N2..N 2N 3.2 222 EN Now scare... Seat
1. 38 2A 3N2 8.2 223 NS 3.2 .2222 52.2% Sea
A >05:on 2223:2232 mm 82235:?V Nmocoscofi 22522222532 mm 85222

M

M32

32

_

.32 98 32 E 22:22? 22053 02.3 Eonto: 2: mac? mBoumoE 83 cm 802.“ madam .«o 3289:5222 532 .m 222d.

24

snails per section in Hessel 3 to 1035 snails in Port Dolomite during 1997. During 1998,
the range was similar, although values for particular wet meadows may have changed
(Table 3).

Species richness ranged from 3-11 with a mean of 6.3 and 6.5 in 1997 and 1998,
respectively (Table 4). Shannon Index diversity was relatively similar among the
majority of wet meadows (Figure 4, Table 4) in 1997 (mean = 0.52). In 1998, the mean
was the same (0.52) but Seymour 1 was unusually low (H ’ = 0.18) (Figure 4) because of
low richness (S = 3) and evenness (J' = 0.37); of the 19 individuals collected, 17 were G.
deflectus. In both years, values for the Shannon Index were normally distributed among
wet meadows as indicated by the generally straight, diagonal line in the probability plots
(Figure 4). Changes in H ' from 1997 to 1998 occurred primarily in wet meadows that
had lower than average H’ in 1997 (then increased in 1998) or higher than average H ' in
1997 (then decreased in 1998) (Figure 4).

Because the meadow Search 2 was much smaller (0.12 ha) than the other wet
meadows and exerted a significant effect on the regression analyses, it was not included
when examining the effects of area on abundance and diversity. Search 2 had the least
diverse snail community in 1997 but not 1998 (Table 4). Without Search 2 included in
the analyses, there were no significant area effects on abundance, species richness,
Shannon Index or evenness in either year (Figure 5).

C CA Analyses: Environmental Patterns in Species Distributions

Although selected for use in this study because of a general similarity, the wet

meadows did exhibit variation in the environmental variables selected for analyses (Table

5). A comparison between years was conducted using a t—test with Bonferroni correction.

25

Table 4. Summary statistics for the abundance and diversity of snails measured in 26 wet
meadows along the northern Lake Huron shoreline in 1997 and 1998. S = species
richness, H’ = Shannon Index of diversity, and J' = Shannon evenness statistic. See text
for calculations.

 

 

 

1_9_91 l9_9_8

Wet meadow AbundanceI S H' J' Abundance S H’ J'
A Pontchartrain] 232 5 0.48 0.68 294 4 0.47 0.78
B Pontchartrain2 538 6 0.36 0.46 — — — —
C St. Martin’s 370 7 0.51 0.61 — — — —
D Carpenter] 127 6 0.71 0.91 134 5 061 087
E Carpenter2 144 5 0.44 0.63 — — —-

F Search] 423 6 0.69 0.88 — — — —
G Search2 72 3 0.43 0.91 141 5 0.58 0.84
H Mismer 268 6 0.53 0.68 235 5 0.49 0.71
I Hessel] 533 7 0.61 0.72 906 1] 0.67 0.64
J Hesse12 562 8 0.47 0.52 1038 7 0.46 0.54
K Hessel3 27 4 0.51 0.84 268 7 0.45 0.53
L Mackinac 446 6 0.41 0.53 695 7 0.62 0.73
M Rudd 135 8 0.53 0.58 — — — —
N Flowers 570 7 0.47 0.56 1008 8 0.59 0.66
0 Hill Island 521 5 0.31 0.44 —— — — —
P Lakeside 278 7 0.58 0.68 55 4 0.59 0.97
Q PortDolomite 1035 7 0.63 0.75 596 7 0.52 0.62
R Prentiss 219 8 0.55 0.60 258 6 0.42 0.53
S Scotty 198 6 0.65 0.83 481 5 0.56 0.79
T Seymour] 94 5 0.48 0.68 19 3 0.18 0.37
U Seymour2 255 7 0.59 0.70 293 8 0.65 0.72
V Seymour3 471 6 0.47 0.60 — — — —
W Sweets] 360 7 0.52 0.6] 452 8 0.55 0.61
X SweetsZ 198 6 0.52 0.67 380 7 0.55 0.66
Y Sweets3 137 10 0.62 0.62 442 10 0.46 0.46
Z Sweets4 130 6 0.49 0.63 344 6 0.40 0.51
1. Number collected per 50-meter section.

26

 

 

 

 

 

 

 

 

 

 

0.7 . if
0.6, i i i f i
35 0.5 I f {{{{f
E i f § § 5 v t r r T
g 0.4 . { g ,_ ,
c i . '
2 O.3J{ E b
(D g 0_
0.2 . ‘25-.
>
g-Z- .. '1
0.1‘ §Q I Q 4 l
w 00 0.2 0.4 0.6 0.0 1.0
O 0 Shannon Index (H')
OBLGEVJNTAZKCWXMHRPUIYQSFD
WetMeadow
B 0.8-
0.7 r A
an 0.5. % ti
3 I 4 l} A . . , ,
c: 04 .5
2 . . % E1“ 0 a
C T 8 .
.2 0.3 . E 1i- /
(D 2 o-
3;
0.2 . 3 . -
A if ‘ :
0.1 , g2" '
‘30 01.2 01.4 E6 01.8 1.0
0. 0 Shannon Index (H')

 

 

 

TZRKJYAHQWXSGPNDLUI
WetMeadow

Figure 4. Shannon Index of diversity for wet meadows in the northern Lake Huron
shoreline of Michigan. Data are arranged in order of the Shannon Index value. Insets are
probability plots for a normal distribution. (A) 1997. (B) 1998. (C) 1997 and 1998
combined. Note that wet meadow order is different between (A) and (B) and that the
number of wet meadows was 26 in 1997 and 19 in 1998 (7 sites were dry in 1998).

27

d
0.6. 4% AI

aimlilliwl E

0.4 .
i I-
o... g

0.2 .

Shannon Index

0.1 -

 

 

 

0-0 .........................
OBLGEVJNTAZKCWXMHRPUIYQSFD

Wet Meadow

Figure 4. Continued.

28

1200 ..

 

 

 

 

1000 - 9'0
.5 0
‘3 800 .
8
i 600 0
~ O
E 0‘ . o
.g o . O O. .
c 400 - Ch
3 0 '
O
< 200- $ . § 8
00 .0 o
O
0 I T t T1] Q to T1111] I t Titrtl
O1 1 10 100
Area(ha)
B 1
¢ -
g A AAAAA ‘A A ‘
u A AA AA AA
in? A AAA A
.3 A A
g A
n.
(I)
1 l I rrfififi] r t I Irlrrr r I Irlrrrl
0.1 1 10 100
Area(ha)

Figure 5. Area relationships to abundance (A), species richness (B), Shannon Index (C),
and Shannon Evenness (D) of snail communities. 1997: solid symbols; 1998: open
symbols. None of these comparisons showed significant correlations.

29

.90 O

 

 

100

10

0.1

Area (ha)

IUUUI

ITIIIII

IIrlUll

 

D 1.0-

0.8 a

— q u
6. 4.
0 0

r: unoccgm

0.2 -

 

0.0

100

10

0.1

Area (ha)

Figure 5. Continued.

30

 

ma

8.8-8.8 w _ ._ 8+ 8.8-8.8 8.8 N: .x. 5885 as? :25

 

m: 8.8.8; R8 m2 868.8 38 V2 @8385 £225 .823...
8 8. a 28 8.8 2% 8.2-8.8 m3 :8 “50332 £85 85%;
8 8.78.8 8.8 28 8.78.8 8.8 8.8 86932 £85 £88
a 8.2-8.8 8.8 :2 8.8-8.8 28 8+ 88332 £85 W85
2 8.82-2.8 82.- NE : 8.3-2.: 8._ 3.2: :5 imam ago
8 8.8788 86 8.3 8,878.8 N? 8.3 .x. 5585 V38:8:
8 8.2.8.8 8; 8.2 8.2-8.8 m: 8.3 a... 2%: 85:5:
88v 2.8-8.8 m2 3.: 88.3.86 :4 8.8 :5 :88 533
mmapoampm
- I I I 8.8-8.8 8.8 a? is «z
- I I l was}? 8.8 8.2 fie w:
- I I I ohm-2.8 So 88 8.8 M
- I I NI 8. 5-8.: a: :8 8.5 8
2 5.8-8.8 88 Rs 2.33 8.8 8.5 I ma
88v 8.828s: 5 ~ .3 8.28 8.83-8.8 w? _ 8. 8m 62m: £23288
8 8.38-8.3 8.” NE: 8.808.? 8.” 8.82 606 38 288.2
555mg

1 89% mm :32 ”max mm 322 £5 UE§> 5%:

a a

 

.88» 59,33 Bounce-amt wfiEESB Eamon-coo Eobomcom 5:3 383 mo 3:88 2m mos—g m .waa 28 a
c053 8:3 F558: ofi mac? $5888 83 cm 5 3:588 “Wu—nets, 1358:8320 («o man—ohm 085 8m moumuflm bu

2 E 8:22:
Sam. .W oEmh

3]

 

-~u.4-:
~w-~fiu
. v - -

V Usnsc, -

.w00~ Gm Uohflmwoa “OZ .N

452-«Ema Hoar-ma moon-:0 .33» 525-45 ~83 mo 338m ._

 

 

 

m: $678.0 Rd Eum oo.mm-oo.o v0.0 mo;V °\° ooh-
? no. 569m 55 mmdm oodoTood ow.m made .x. 93.5

m: oodméod 3N 3.2 code-cod cm; 3.3 X. 32:3
m: mm.mm-oo.o mm; med mm. fl méod No; 86 ..\° 82¢.
m: oodwéod cod 3A: code-cod wed wag: o\o :o_§owo> 885.5%
3 8.3.86 omd Gd cod _ -oo.o mmd mwd o\o neomoESom
modv cad-cod £6 34 oo.mw-oo.o mm.m mcdfi o\o Ezaowboo‘
2 8.3.86 :3 mod oodméod ovd mad c\° x5332
- cod-cod cod cod $6 _ -oo.o 3.0 5 fl ._ X. 325—25
3 MOS-cod mmé NW: moan-cod nmd w _ .n X. 3330
we code-cod mod mm.mm oodoTood $.m 55.3 .x. 5815
2 862-8.? $6 3.3 code-MO? 84 360 X. 295580
H2483;

a 858 mm as: 85$ mm 502 85 ”88> “as“:
a N3

ll”

@8555 .m £85

32

Only three significant changes were observed. Conductivity was significantly higher in
1998 and water depth was significantly lower than in 1997 (Table 5). A decrease in the
frequency of occurrence of Polygonum amphibium from 1997 to 1998 was the only
significant change in the plant community.

Upon examination of the seven variables in the CHEMISTRY group (see Table 5)
in 1997, alkalinity and calcium were highly correlated; alkalinity explained more
variance and therefore calcium was removed from the analysis. The six remaining
variables explained a total of 18.0% of the variation in the species data (Table 6). Axes 1
and 2 explained 10.2% and 3.6% of the species data and 57.0% and 20.2% of the species
data when constrained by the environmental variables in the model (Table 6). Of the six
variables, alkalinity, conductivity and pH were significant (P<O.10). The interset
correlations (Table 7) were used to aid in the explanation of the CCA biplots with
particular weight given to the significant variables. Axis I was most closely associated
with pH and alkalinity. Examination of the sites plot (Figure 6) and the original data

indicate that this axis was most strongly influenced by alkalinity. Sites to the lower right

of the plot generally had much lower alkalinity (as low as 55 mg/l CaCO3) than the other

sites and only slightly higher pH. F our species, Amnicola Iimosa, Valvata tricarinata,
Stagnicola elodes, and Pyrgulopsis Iustricus, were associated with these low alkalinity
sites (Figure 6). Conductivity was somewhat negatively associated with this first axis but
also positively associated with the second axis. The tight clustering of the sites near the
Center of the plot indicated the overall similarity of the wet meadows in regards to water
Chemistr-y (Figure 6).

In 1997, total vegetation density and grass density were highly correlated in the

33

 

was 26 mém 5mm cows—oboe acogohéoéomoommmo

EN EN 3: as saussaflo

0285.» owficoeoa $323830

Rod awed Sod mid macaw—oboeEoficeficoéflooam

mmod mmod Sued owed mos—gcowfi

wood 33; omod $00 3% Need mmmd 32.9315

mow 3K ode w.mm coca—oboeEoficegcoéofiommmo
SN 3: 3; E 98863qu

35¢? owficoocoa gum—58:0
End wood 30.0 mood macaw—oboeEoficegcoéomoomm

 

 

 

Bod omod mvod mmod 8:325me

mood cow; god ommé m.mm Sod v2.0 32m~5HUDmHm
w.m3 fiww N: Q3 noun—oboe“cogegcoéomoommmo

M3: ”.2 a: 2: guacamfi

0289; away—cocoa gum—58:0

movd mwmd 39¢ 39o mcotfioboo Eoficegcoéflooam

83 mod v85 $3 8295me

Cod 3M: Cod w3.m QM: mood 32d 33~>~ElmHEmEU

m 228...“ Am 0:8-..“ 3:398 «:55 moflgaowmo v m m _
moxm =< 38 _ 8593» .x. :38. 32:88 N mox<

ml

 

.33 5 $5888 63 338 coat 8:3 F558: 5 282m 232% («o max—£8 85288on 32:88 no 338% .c 2an

34

+1.0

Cond Mg Na
. . ‘

Alk

 

:-1.o

 

Vt'Al

 

-1.0

Figure 6. Canonical correspondence analysis of aquatic snails in northern Lake Huron
coastal wet meadows using the 1997 CHEMISTRY variable group as explanatory
variables: (A) species-environmental biplot (B) sites-environmental biplot. All plotted
environmental variables were included in a significant model; those with solid arrows
were significant variables and dashed arrows indicate non-significant variables. Species
and sites scores are adjusted to scale = 0.3. Environmental variable abbreviations as in
Table 7, species labels as in Table 2, and site labels as in Table 1.

35

 

Figure 6. Continued.

H3 8301
° E1 U1L1P 90

l2 0 I13N1c2. 79‘

J19. 20.39 1
“5 “2 N2 92‘

+1.0

$M2

V

01

 

K1 . K313 .._:QZR92.F1

36

.. W1

 

L

STRUCTURE variable group (see Table 5), and vegetation density was removed leaving
eight variables in the analysis. These variables explained a total of 23.3% of the variance
in species data (Table 6) with axis 1 and axis 2 explaining 7.8% and 6.6% of the species
data and 33.8% and 28.2% of the species data when constrained by the environmental
variables (Table 6). Four variables: water depth, hummock frequency, shrub density and
frequency of open water, were significant within the model (Table 7). The first axis was
a gradient from deeper water to shallower water with a higher shrub density. The second
axis was a gradient from more dense hummocks and grass habitats to sparsely vegetated
habitat with more open water (Figure 7). Amnicola limosa, V. tricarinata, S. elodes and
F errissia parallela all tended to be associated with deeper habitats, whereas Aplexa
elongata was found in shallower, shrubby habitats with some open water pockets.
Bithynia tentaculata and Amnicola walkeri were found to have fairly strong association
with densely vegetated and hummock-filled habitats (Figure 7).

PLANT (see Table 5) accounted for the greatest proportion of variance explained
with 34.0% in 1997 (Table 6). All 12 variables were included in the final model and six
were significant (Table 7). Axis I and axis 2 accounted for 12.1% and 6.3% of the
variance in the species data and 35.7% and 18.5% of the species data when constrained
by the environmental variables (Table 6). A gradient from bulrush and cattails to shrub
habitats was associated with axis 1, and axis 2 was a gradient from alder and bulrush to
graminoid and Potamogeton habitats. F errissia parallela was strongly associated with
increasing bulrush frequency (Figure 8). Amnicola limosa, V. tricarinata, and S. elodes
were associated with axis 2, especially sites with lower frequency of grarninoids and

Potamogeton whereas Planorbella campanulata and B. tentaculata were associated with

37

Table 7. Interset correlations of environmental variables with CCA axes in 1997 for all
variables included in significant model. Significant variables (P<O.10) are indicated in
bold type. Environmental variable abbreviations are used in CCA diagrams.

 

Interset Correlations

 

 

Variable Abbrev. Axis 1 Axis 2 Axis 3 Axis 4
CHEMISTRY 1997
Alkalinity Alk ~0.388 0.101 0.142 -0.032
Conductivity Cond -0.200 0.362 0.084 -0.091
pH pH 0.408 -0.31 1 0.085 -0.125
Potassium K 0.197 0.245 -0.177 0.144
Magnesium Mg 0.055 0.359 0.039 0.016
Sodium Na 0.127 0.363 0.070 0.175
STRUCTURE 1997
Water depth Depth -0.390 -0.266 0.045 0.273
Hummock freq. Hummfrq 0.384 -0.391 -0.21 1 -0.045
Shrub density Shrubden 0.408 0.110 0.357 0.049
Open water freq. Waterfrq -0.051 0.286 —0.192 0.250
Foliage diversity Foldiv 0.380 0.060 0.270 0.071
Grass density Grassden 0.130 -0.331 -0.067 -0.335
Grass height Grassht -0.055 -0.307 -0.202 0.020
Hummock height Hummht 0.277 -0.156 -0.3 84 0.024
PLANT 1997
Alder Alderfrq -0.108 0.282 -0.314 0.007
Bulrush Bulfrq 0.507 0.223 -0.043 0.274
Cattail Catfrq 0.442 -0.070 -0.084 -0.147
Graminoid Gramfrq -0.1 15 -0.259 -0.306 -0.418
Potamogeton Potfrq -0. l 12 -0.293 -0. 103 0.397
Shrub Shrubfrq -0.221 -0.064 0.128 -0.249
Duckweed Duckfrq 0.137 -0.1 16 0.101 -0.245
Nuphar Nupfrq 0.186 -0.014 -0.106 0.290
Polygonum Polyfrq -0.031 —0.061 0.074 0.073
Submersed veg. Subfrq 0.148 -0.206 0.097 0.083
Tree Treefrq -0.177 0.273 -0.085 -0.157
Willow Wilfrq -0.1 10 -0.061 0.186 0.008

 

38

+1.0

Waterfrq

 
 
   
     

 

 

 

  

PC Shrubden
Se ______..-—§Foldiv
no: 7° ‘ e‘ W115
AI'Vt
‘\
Hummht
,.
I
De th "
p Grassht‘ §
Aw” Grassden
., .Bt Hummfrq
_ A
-1.0

 

Figure 7. Canonical correspondence analysis of aquatic snails in northern Lake Huron
coastal wet meadows using the 1997 STRUCTURE variable group as explanatory
variables: (A) species-environmental biplot (B) sites-environmental biplot. All plotted
environmental variables were included in a significant model; those with solid arrows
were significant variables and dashed arrows indicate non-significant variables. Species
and sites scores are adjusted to scale = 0.3. Environmental variable abbreviations as in
Table 7, species labels as in Table 2, and site labels as in Table 1.

39

 

 

+1.0 A2
Ii
M1 II
° 02
1'.
G1
9 'r
T1 ..
. 81
W1 .. P1 0
L3 -» A1
0
0
R1 01 OH D1
° ”3 M2 ’" E1 0
21 31° . 02o , Y1
Q '1
P2 . 02521303 N2 .
I I j I I I; I. H; O. I .I f I I I I I J
1" "F2'j" if
1.0 . x1. U2. M3 L1 V3 0
Y2 O 83 . Q 9
W2 F N1 0 9V1 O
. 12 . R3
wa’ R2 " . V2
00
L2 U1H1 . 33
° l 9 C1 0
X2 '
x3 . J1” J1 K1
.. o
o 11
01 N3
db
-1.0
.1.

Figure 7. Continued.

40

I

 

 

 

Bulfrq

Figure 8. Canonical correspondence analysis of aquatic snails in northern Lake Huron
coastal wet meadows using the 1997 PLANT variable group as explanatory variables: (A)
species-environmental biplot (B) sites-environmental biplot. All plotted environmental
variables were included in a significant model; those with solid arrows were significant
variables and dashed arrows indicate non-significant variables. Species and sites scores
are adjusted to scale = 0.3. Environmental variable abbreviations as in Table 7, species
labels as in Table 2, and site labels as in Table 1.

41

 

 

+1.0
A2
0 .-
._ P2
0
A1 "
. 02 $2
.. .
C1 B1
0K1 . R3 _ 03 ° T1
Y1. ° us. 31 Kg»
.. 33 x2 R1
0
°‘" D1 x3 G10 21 °
0 v3 , , .. .
E1: F2 “'1 . .
-_—‘fi 4 4? fi 1 ‘ -. ‘Xl' V I u v . u . u I
1~0 12 H1 . ’0 H3 F1 ”'0
H2 P10 1- 0..
té’JZ 033 W1 oz
WZOL‘l ’UquRZO” .
L3 01
NZ°""!I\/I°3bz°01v
lb .
N3
0
M1 ‘
O ..
v1 B
-1.0

Figure 8. Continued.

42

a higher frequency of these. Pyrgulopsis lustricus was associated with higher alder
frequency (Figure 8).

In 1998, there was no significant model for the CHEMISTRY group (P>0. 10) and
only hummock height was significant in STRUCTURE (Table 9). Thus, only one axis
was generated in the CCA and therefore explained 100% of the variance in species data
constrained by the environmental variable, but only 5.7% of the variance in species data
(Table 8). There were no strong species associations with this gradient (Figure 9).

Five variables were included in the significant model for PLANT in 1998 and
together they explained 24.2% of the species data. Bulrush, cattail and submersed
vegetation were significant (Table 9). Axes 1 and 2 explained 9.7% and 8.0% of the
variance in species data and 40.1% and 33.2% of the species data constrained by the
environmental variables (Table 8). Axis I was a gradient of cattail frequency and axis 2
was a gradient from high bulrush to high submersed vegetation frequency (Figure 10).
The majority of snail associations were with habitats containing low cattail frequency

(Figure 10). Bithym‘a tentaculata associated strongly with submersed vegetation.

C CA Analyses: Importance of CHEMISTRY, STRUCTURE, and PLANT
The total inertia (i.e. sum of all unconstrained eigenvalues) for the 1997 data was 0.662.
The proportion of variation explained by each of the three environmental variable groups
was calculated by dividing the sum of the canonical eigenvalues for each by total inertia.
The sums of unconstrained eigenvalues for CHEMISTRY, STRUCTURE, and PLANT
were 0.119, 0.154, and 0.225, resulting in 18.0%, 23.3%, and 34.0%, respectively, of the

variance explained by the environmental variables (Table 6). However, the correlation

43

.358 :m 3:205 .83 03:93 0:: 3:0 ._

 

Ndd
mdm

”3 Wm: mo:
NAN 5: 5d

nnvd mmmd Maud mend

:23:on ::o:::o:>:o-momoo:m do
8:: 86on do
859:» $8598: 03:25:50
30:22:00 ::o:::o:>:o-mo_oo:m

 

Edd add Edd omod mo:_m>:ow_m
odd an; odd mdwd New NGd mid $2923:
.I I I odd: 5:38.50 6085:2936on do
I I I Em 8:: 86on do
85:? owfifieom 03:23:30
I I: l. mmnd 30:38.56 6085:2936on
.l I :11 mmdd mos—35$“:
I _I. dmdd :2: 5m Sod mmdd onmMEUDMHm
6:08 H§Qm_:w_m :2
mad: >mHmHEm—EU
m 2:8-"— m 2:83.: 35298 35:: 829:qu v m m _
max: =< mi: _ oo§tm> .x. 30H 33:88 N mox<

F

 

d3: 5 958:2: 33 1:38 :83: SEA 505:0: E $.85. “5:26: we £335 oo:o::o%o.coo 30:88 .«o $33M d 033.

44

Table 9. Interset correlations of environmental variables with CCA axes in 1998 for all
variables included in significant model. Significant variables (P<O.10) are indicated in
bold type. Environmental variable abbreviations are used in CCA diagrams.

 

Interset Correlations
Variable Abbrev. Axis 1 Axis 2 Axis 3 Axis 4

 

 

CHEMISTRY 1998
No significant model

STRUCTURE 1998

Hummock height Hummht -0.722 — —- —
PLANT 1998

Bulrush Bulfrq 0.120 -0.449 0.367 -0.221

Cattail Catfrq 0.580 0.097 -0.144 0.095

Submersed veg. Subfrq 0.063 0.540 0.390 0.086

Alder Alderfrq 0.148 -0.254 -0.159 0.420

Shrub Shrubfrq 0.279 0.132 0.028 0.193

 

1. With one variable only one canonical axis is produced.

45

 

 

 

(+1.0
Ae
I
PC ,_ .Bt
' Pa 5;, AI
Hummht ' Fp '
I3.0"""ffis'e g'Fo""fi"1+1.0'
o Gd 0
G 0 Ah
Pe l'Pt
I
Fl

 

-1.0

Figure 9. Canonical correspondence analysis of aquatic snails in northern Lake Huron
coastal wet meadows using the 1998 STRUCTURE variable group as explanatory
variables: (A) species-environmental biplot (B) sites-environmental biplot. Species and
sites scores are adjusted to scale = 0.3. Environmental variable abbreviation as in Table
8, species labels as in Table 2, and site labels as in Table 1.

46

 

 

A1 +1.0
1.
Il-
J1 "
o
., N4
J2 111 N3. 0
’9 ., G1
x30$1 o
033 02 0L3 Y1
0
X1
‘44::44::;:1::::::::::
-1.0 X2 0W2 °H4 +1.0
.11- W4
U3 .9 O
, . W3 . T1
l1 ' Y2
o
21 R2."
0
03 K2. 1’
Q1
1- .
R1
lb.
P2
-1.0

Figure 9. Continued.

47

 

Catfrq

db
(b
h

 

+1.0'

 

 

-1.0

Figure 10. Canonical correspondence analysis of aquatic snails in northern Lake Huron
coastal wet meadows using the 1998 PLANT variable group as explanatory variables: (A)
species-environmental biplot (B) sites-environmental biplot. All plotted environmental
variables were included in a significant model; those with solid arrows were significant
variables and dashed arrows indicate non-significant variables. Species and sites scores
are adjusted to scale = 0.3. Environmental variable abbreviations as in Table 8, species
labels as in Table 2, and site labels as in Table 1.

48

 

 

 

+1.0 ”3
n-
'F
W24 ,N“
X3
. 1b
l1 W3
J1 O a
0 U1. Q81
" G1
X1 K2 R7- ‘33::
11.0fi""'J2#' "f' ""'+1.0'
0 U3
020 H3 0 . .01
-- ’32.02 oz
0
0x2 . Y1
«-
21. P2
0 ”
T1
-1.0

Figure 10. Continued.

49

that was likely to exist between environmental variables means that some of the variance
explained by each group was likely shared with one or both of the other groups. The
results of the partial CCA showed that less than half of the variance explained by
CHEMISTRY was due to CHEMISTRY alone (7.9% of the 18.0%) (Figure 11). The
remainder was shared with STRUCTURE and PLANT. Likewise, STRUCTURE alone
and PLANT alone accounted for 15.6% and 21.8%, respectively, of the explained
variance in the species data (Figure 11). A total of 58.2% of the variance in the species
data was explained by the CCA. Nearly 13% of the variance was shared among the three
variable groups, and 41.8% of the variance in species data was unexplained by the
environmental variables included in the three models (Figure 11).

In 1998, the CCA models were able to explain much less of the variation in the
species data (Table 8). There was no significant model for CHEMISTRY and only one
variable in the STRUCTURE group. The canonical eigenvalues for STRUCTURE
accounted for just 5.7% of the variance in species data (0.035 divided by the total inertia
of 0.612) and PLANT accounted for 24.2% (0.148/0.612). PLANT alone accounted for
22.1% of the variance and STRUCTURE alone for 3.6%; 2.1% was shared between the

two (Figure 12). The remaining 72.2% of the variance in species data was unexplained.

DISCUSSION
Snail Community Characteristics
Coastal wet meadows in the northern Lake Huron shoreline contained a very
diverse aquatic snail community composed of 19 species (Table 2). There have been no

systematic surveys of this habitat as far as this author can determine so comparisons are

50

 

  
 

UNEXPLAINED
[41.3]

   

CHEMISTRY
[18.0]

     
   
 

7.9

 
 

STRUCTURE
[23.3]

  

 

 

 

Figure 11. Venn diagram of the CCA variance partitioning of 1997 species data among
three variable groups. Total variance explained by a variable group is enclosed in [ ] and
represented by the size of the circle. Variance shared by variable groups is indicated in
the appropriate intersections (although size is not proportional). Variance explained is
calculated as the sum of all canonical eigenvalues for a particular analysis divided by the
total inertia.

51

 

UNEXPLAINED
[72.2]

 
 
  

STRUCTURE

[5.7] \‘

 

 

 

Figure 12. Venn diagram of the CCA variance partitioning of 1998 species data among
two variable groups. There were not a significant model for the CHEMISTRY variable
group in 1998. Total variance explained by a variable group is enclosed in [ ] and
represented by the size of the circle. Variance shared by variable groups is indicated in
the appropriate intersections (although size is not proportional). Variance explained is
calculated as the sum of all canonical eigenvalues for a particular analysis divided by the
total inertia.

 

difficult to make. Although not specifically listing wet meadows, Burch and Jung (1992)
list 31 species that might be found in habitats similar to those found in these wet
meadows (e. g. using categories such as quiet bay or pond, intermittent pool, or marsh
species, etc.) in the northern Michigan region. Bronmark (1985) found 18 species in
Swedish ponds, which were densely vegetated and provided similar habitats as those
found in wet meadows, and Pip (1987) identified 40 species in a variety of aquatic
habitats (41% were ponds, but also lakes and rivers) throughout central North America.
The northern Lake Huron shoreline is considered to contain some of the most pristine of
the remaining Great Lakes coastal wetlands (Minc and Albert 1998) and the snail
community reflects this.

Further investigations of diversity used the Shannon Index, a diversity index that
combined both species richness and a measure of the evenness of abundance among the
different species (Magurran 1988). The distribution of Shannon Index scores for the wet
meadows approximated a normal distribution (Figure 4), with the possible exception of
Seymour 1 in 1998, indicating an underlying relationship among wet meadows. The
normal distribution suggested that the snail communities within the wet meadows were
similar enough to be considered together in the current analyses. Others have also found
the distribution of Shannon Index scores to be normally distributed (see Magurran 1988).

In terms of the diversity represented by the Shannon Index, overall, the snail
communities showed little change from 1997 to 1998 (Figure 4). Those wet meadows
with low diversity in 1997 increased from 1997 to 1998 and those with relatively high
diversity decreased from 1997 to 1998. This suggests that a dynamic equilibrium process

may be important in structuring the snail communities.

53

Many studies of snails within lakes have found some evidence for an association
of diversity with lake area (Lassen 1975, Aho 1978a,b, Browne 1981, Jokinen 1987).
However, regressions of Shannon Index against area, and all other measures of diversity
used in the present study showed no evidence of an area effect (Figure 5). Similarly,
Bronmark (1985) found only a weak area effect in a study of aquatic snails in ponds.
One of the most likely reasons for the increase in snail diversity within lakes was because
lakes with greater area had higher habitat diversity that could support different snail
species (Harman 1972, Ross and Ultsch 1980). In those studies, however, the
measurement of habitat diversity was based primarily on substrate type, and macrophytes
were only considered in a general sense as one type of substrate (Harman 1972). Because
I sampled snails from sites that I identified as wet meadows, I automatically limited the
variability in sediment type and habitat in general (though not completely, see Table 5)
thereby reducing many effects of habitat type that might change as area increased. The
lack of an area effect at my scale of sampling suggested a dynamic equilibrium that is
more strongly influenced by other factors.

Habitat Characteristics of Coastal Wet Meadows

The 28 habitat characteristics used as explanatory variables provided a good
description of the environment in which the snails were located. Analysis of the original
data as well as CCA sites-environmental biplots suggested a relatively homogeneous
environment, especially in terms of water chemistry.

The sites-environmental biplot for CHEMISTRY in 1997 exhibited a clustering
effect of most points near the center of the two axes (Figure 6). This tight clustering was

indicative of low variation in the overall combination of water chemistry variables

54

examined (Jongman et al. 1995, ter Braak and Smilauer 1998). An exception was that
several wet meadows, Sweets 2-4, were separated and clustered in the lower right
quadrant of the plot. This clustering suggested that these sites were somewhat distinct in
their chemical characteristics, and direct analysis of the original data confirmed that these
wet meadows were much lower in alkalinity and slightly higher in pH. Examination of
the STRUCTURE and PLANT sites-environmental biplots (Figures 7-10) indicated
moderate variability in these characteristics (i.e. less clustering of points).

The majority of environmental variables also were consistent from 1997 to 1998
(Table 5). This was somewhat surprising given the dynamic nature of the water level
fluctuations in coastal wetlands. Not only do daily (seiches) and seasonal (due to
precipitation patterns) fluctuations occur (Keough et al. 1999), but also a substantial
decrease in water level occurred from 1997 to 1998. Lake Huron water level dropped by
28 cm (USACE 2000) during this time and the wet meadows showed a significant decline
in depth from 29 cm to 13 cm in the sampled sections (Table 5). The average decrease of
16 cm in the wet meadows was less than the lake level decline because of natural and
anthropomorphic barriers (Table 1) that affected water retention in the wet meadows.

The two other significant changes from 1997 to 1998 were most likely associated
with the lower water levels. For example, higher conductivity levels that were recorded
(Table 5) may have been an effect of concentrating solutes in the water column. The
only change in plant structure or frequency was a decrease in the occurrence of
Polygonum amphibium (Table 5). While the lower water levels will affect the plant
community over time, the lack of additional plant changes in the present study was

consistent with studies that show a lagged response to rapid water level changes (Keddy

55

and Reznicek 1985, 1986).
Snail — Environmental Variable Relationships

The significant snail-environment relationships detected in the CCA for 1997 and
1998 are summarized in Figures 13 and 14, respectively.

As hypothesized, the water chemistry variables included in the CHEMISTRY
variable group explained the least amount of variation in the species data. In 1997,
18.0% was explained and since there was no model in 1998, none was explained (Tables
6 and 8). The most significant variables from the CHEMISTRY group in 1997 were
alkalinity, conductivity and pH (Figure 6). Calcium was not included in the analysis
because of the strong correlation with alkalinity, which was expected in this region
because of the dolomite and limestone bedrock. Whether the effect on snails was due to
alkalinity or calcium concentrations or an interaction between the two was beyond the
scope of this study. The first axis of this CCA ordination included both alkalinity and
pH, which were common factors in other studies as well (e.g. Aho 1966, Mackie and
Flippance 1983a, Okland 1983, McKillop 1985, Pip 1987). However, in the present
study, alkalinity and pH were unexpectedly negatively related. An examination of the
sites-environmental biplot (Figure 6) revealed that three sites were strongly related to this
axis. The three sites were all located on Sweets Point on the northern tip of a peninsula
(Figure 2). They were isolated from many of the other wet meadows and direct
examination of the data revealed much lower alkalinities at these sites relative to the rest
of the wet meadows. These sites also had slightly higher pH, leading to the negative
relationship in the CCA. Thus, the observed negative relationship between pH and

alkalinity should be regarded as a localized effect of a few influential wet meadows

56

.5886 wcmucoamotoo 05 5 Ba:

Ba :86me some £3, 38683 86on .mmcm A V E 8a 0335» EwocEwmmcoc €222.50 ”323 3%: 2a mes some .33 noun—2.50
$3352, “:35:me 65% 3:398 Sec 86on 2: mo owficoouoa 53> max“ 32:0ch Sm 356an 35:80:20 28352
maob< .358w @3352, 35:80:28 025 mo :03 no,“ <00 mama: 32328—8 Eoficozéoéacm 32 05 mo bmfigm .2 ouswi

32m @6385

 

98.282 £30393 A

 

 

 

 

 

gonzo?» mme< $00
2:5 95
=sz new 52.3 9.6385 .eSomoESoQ 665255 9.6385
wmbofi SooEmmum
«3:qu mummEmu A oxo r .9. V
._.z<.._n_
€92 59:52
.2902 van bacon $89
.3: $63 :80 96855 .8: x0955: @3355
A v $33323 $555
89:8 35me $0.0 team; $8EE<
399% 89.0:
35ch 925 @3385 Emma @5322: mumctmoE 59:9,
A v .386 SooEmmCm.
Someone 2%? Rows $23qu mummEmu
muons $8EE<
62 2a m2 washoamhm
A 33:03:00 9.3905
33.56 mtmmmou. oxamd
b_c__mx_m mEmmeoE EmctmoE m~w>um>
A 883 SooEmSm
Smmcsm mme< Rand? 36.:ng mumqoSuiu

mon: EooEE<

>m._.m=2m_._o

57

 

633% mME0D~Fm 05 .23 38888 bwgbm 80>» 86on oZ ...

.5526 w:€:o%otoo 05 :m 3%: 2a “:2:me :80 5:5 338%: $6on mam .323 3%: 8: fix: :80 5E, 363:8
83$?» BREEmfi .38: 3:398 Sm: 880% 0518 owfifieoq 53, max“ 32:88 :8 maomwfiw 3:250:35 E883:
@505 £38m 052:? 35:80:35 8:5 00 some :8 <00 wfim: 833322 ::o:::o:_>:o-=a:m M33 05 .00 $885 .3 oSwE

cozmuomg umflmEnam 95385

52.3 mEmeoE

 

smsosce 3:30am A $3

=sz 9.68.2:

V

 

A Qond

Ego; xooEEan mcfionE

 

A

Kid

£23qu £8.56”:
.EmEmEmc m§mo<

mCmuatqumo gmfioemi
8:966 88:?

._.z<n_n_

mmDPQDm—hm

6:02

>m._.m=2m_._0

58

 

among a generally homogeneous set, and therefore, the most significant variables were
alkalinity and conductivity (Figure 13).

Snail species that showed a strong relationship in the CCA included four species
that were more common than average in the Sweets point wet meadows (Figures 6 and
13). These included Amnicola limosa and Valvata tricarinata, which have been found in
low alkalinity waters in other studies (Mackie and Flippance 1983a, Rooke and Mackie
1984) although not limited to these (McKillop 1985). F ossaria obrussa was influenced
by environmental variables of the second axis, which is related to conductivity (as well as
magnesium and sodium). Aho (1966) also found correlations between conductivity and
the molluscan fauna in Finnish lakes and associated these with productivity.

The habitat structure or complexity as measured by the STRUCTURE variable
group accounted for 23.3% in 1997 and 5.3% in 1998 of the variation in snail abundance
and distribution (Tables 6 and 8). STRUCTURE was composed of primarily depth and
non-specific plant characteristics that provided complexity to the environment. Nearly
5% of the variation explained by STRUCTURE was shared by the CHEMISTRY group
in 1997, presumably as a result of interactions of the water chemistry variables on plant
growth and the changes in water chemistry from shallower to deeper water.

In 1997, CCA of the STRUCTURE variable group revealed the most significant
gradient as a change from deeper water to shallower water that contained a higher density
of shrubs (Figure 13). This gradient may have reflected the role of fish predation in
shaping the snail community. Snails in deeper water would likely be exposed to higher
levels of fish predation to which some snails may be better adapted (V ermeij and Covich

1978). Both A. limosa and V. tricarinata, which are positively related to depth (Figure

59

 

7), are prosobranchs and contain an operculum that may confer some resistance to
predators (Lodge et al. 1987). Furthermore the shell of V. tricarinata contains 2 or 3
raised carinae (keels) that help to strengthen the shell (V ermeij and Covich 1978).
Conversely, Aplexa elongata, is a thin-shelled physid, lacking an operculum. It is not
surprising that it was negatively related to depth and positively associated with shrub
density. Increasing plant density may afford protection from fish predators (Covich and
Knezevic 1978). A second gradient from higher occurrences of open water to increasing
frequency of hummocks was also identified in 1997 (Figure 7) and in 1998 hummock
height was the only significant variable (Figures 9 and 14). This gradient extended from
more open regions to more densely structured habitat (hummocks) within the wet
meadow, although it was independent of depth (as indicated by the right angle to the first
gradient). It may also indicate the influence of predation as discussed above.

In addition to the interaction of depth/open water and plant/hummock density
with predators, it is possible that these gradients affected the snail community in other
ways. Although not examined in the present study, this gradient may have reflected a
sun-shade gradient and/ or a temperature gradient. The open water areas were exposed to
more sun which may affect the periphyton community on which many snails feed
(Doremus and Harman ' 1 977, Russell-Hunter 1978, Lowe and Hunter 1988). In addition,
the shading by the shrubs and hummocks (including the grasses forming the hurnmocks)
may have maintained cooler temperatures in these regions that could have affected the
snails directly or indirectly through the periphyton community. These possibilities
deserve additional study.

The plant community explained the largest amount of variation in the species

60

 

data. The PLANT variable group used the frequency of occurrence of plant species in the
wet meadows as explanatory variables and explained 34.0% in 1997 and 24.2% in 1998
of the variation in snail distribution and abundance (Tables 6 and 8). In 1997, 9.4% of
this explained variation is shared with the CHEMISTRY group and 7% shared with
STRUCTURE, and 4.2% of this is shared among all three variable groups (Figure 11). In
both years, bulrush and cattail were significant variables (Figures 13 and 14). These
species may be found in large expanses just lakeward (Figure 1) of many of the wet
meadows and in other cases, individual cattail or bulrush plants may be found scattered
within the wet meadow (BEK: pers. obs.). In 1997, bulrush was included in a gradient in
which the opposite direction was increasing grarninoids, Potamogeton natans (a floating
leaved pondweed species) and shrubs. This gradient was probably influenced more by
the distribution of these plants within the wet meadow than by strong preferences for
these species per se as occur in some snail-plant relationships (Calow 1973, Pip and
Stewart 1976, Pip 1978). Bulrush was more common in deeper water whereas the others
increased in frequency towards shallower depths. F errissia parallela, an ancylid limpet,
showed a strong positive relationship with bulrush, as did Stagnicola elodes. Both of
these species do not ingest live bulrush stems, but rather, can be found strongly attached
to the stems where they are most likely feeding on the periphyton (BEK: pers. obs.,
Clarke 1981). Planorbella campanulata and Bithynia tentaculata showed a relationship
with grarninoids, Potamogeton, and shrub end of the gradient. In 1998, B. tentaculata
was also associated with submersed vegetation. In 1997, the frequency of alder was
significant and there was an association with A. elongata and Pyrgulopsis lustricus. It

was unclear what accounts for this association.

6]

 

Changes in the CCA from 1997 to 1998 (i.e. no significant model for
CHEMISTRY, and less variation explained overall) (see Figures 13 and 14) were most
likely due to the decrease in water levels that occurred. The elimination of a significant
water chemistry model in 1998 supported the generally weak association of the snails
with water chemistry observed in 1997. A decrease in water levels probably altered the
water chemistry by changing the importance of the sources. Lowered water levels
decreased the influence of Lake Huron and increased the importance of ground water and
rainfall. These changes most likely added enough variability to reduce the likelihood of
finding significant effects. In 1998 the only significant STRUCTURE gradient was
hummock height whereas in 1997 there was a depth-shrub density gradient as well. The
lower water levels in 1998 allowed only deeper sections from 1997 to be sampled in
1998. Thus, the depth gradient was reduced and the shallower sections that contain
higher shrub density did not contain standing water. The hummock height gradient
remained because hummock height was related to water depth; the taller hummocks were
found in sections still containing water in 1998. The significant PLANT gradients also
were plants generally associated with deeper portions of the coastal marshes: bulrush,
cattails, and submersed vegetation.

Patterns in common and uncommon species

A number of species were found near the center of the CCA ordination biplots for
the majority of the analyses, which indicated a limited association with the environmental
gradients. These included the six most abundant and widely distributed species: Physa
gyrina, Gyraulus parvus, G. deflectus, Planorbula armigera, Planorbella trivolvis and

Fossaria parva. In contrast, the species that were frequently identified with each

62

 

gradient were often less abundant and less widely distributed.

CCA is a technique widely used for the analysis of community composition data
because of its ability to handle large numbers of zeroes in the data. However, rare
species are known to have a disproportionately large influence on the analysis (ter Braak
and Smilauer 1998), and for this reason, they are given smaller weights in the analysis via
a downweighting procedure found in CANOCO (ter Braak and Smilauer 1998). I used
the downweighting procedure in the present study to minimize the effect of rare species,
thus reducing their influence in the analysis. The fact that they continued to appear in the
gradient detected by CCA suggested that these gradients are biologically meaningful in
comparison with the responses of the more abundant species, although caution should
continue to be used in their interpretation. Also, it is important to stress that, even though
the abundant species were not identified as having strong associations with the
environmental axes, because of their abundance, they were still influential in the
construction of the gradients.

The abundant species in the wet meadows were common throughout the region
(e.g. Pip 1985, 1987) and were described as generalists, being found in many different
environments (Burch and Jung 1992). The variation in their abundance and distribution
within coastal wet meadows of the northern Lake Huron shoreline was not consistently
associated with the environmental variables examined making it difficult to identify a
gradient applicable to these species. And, this was especially true in comparison to the
gradients detected with the less common species. Because these abundant species were
such generalists, it made it more difficult to identify factors that affected their

distribution. The environmental variables examined in the present study have shown

63

 

only a limited ability to explain the observed distribution patterns. Other factors should
be examined to further identify more important variables.
Conclusions

The coastal wet meadows of the northern Lake Huron shoreline contained a
diverse assemblage of aquatic snails that included 19 species from seven different
families. Snail abundance and diversity were not related to the size of the wet meadow
due to the relative homogeneity of the habitat. Environmental variables that may affect
the snail community composition were divided into three subsets, and the variance
explained by each was partitioned. The occurrence of various plant species, including
alder, bulrush, cattail, grasses and sedges, Potamogeton natans, various shrubs, and
submersed vegetation, explained the greatest proportion of variation in the snail
community. The structure of the habitat, as described by water depth, shrub density,
hummock height and frequency, and the frequency of open water, explained the second
greatest amount, and water chemistry (alkalinity and conductivity) explained the least.
Although there were only a few changes in the habitat variables from 1997 to 1998, fewer
significant snail-environment associations were observed in 1998, primarily because of
decreased water depth. The environmental gradients detected were most closely
associated with the less common snail species. The most common and abundant species
were less associated with the variables examined, suggesting that other factors (e.g.
surrounding landscape or fluctuating water levels) are more important determinants of

their distribution patterns in wet meadow habitats.

64

 

—4=.——‘- __...- ..__ ‘__ , _.

 

CHAPTER TWO

SNAIL—LANDSCAPE ASSOCIATIONS IN COASTAL WET MEADOWS

65

INTRODUCTION

Landscape ecology is the study of the interaction between spatial patterns and
ecological systems. F undamental to this is the idea that landscapes are composed of
relatively homogeneous, discrete elements called patches that when combined together
form a mosaic pattern (F orrnan 1995). Patch quality, boundaries with other patches (that
affect flow between them), patch context, and connectivity all may affect the ecological
processes within a landscape. Additionally, the scale at which the landscape is examined,
and the organisms involved are known to have an important influence (Wiens 2002).

As the name, “landscape”, suggests, the theories underlying landscape ecology
have been developed in terrestrial systems. It has been only relatively recently that many
of these concepts have been applied to aquatic, primarily lotic, systems (see Wiens 2002).
In addition to many other taxa and processes, these studies have shown that the stream
and surrounding landscape influence aquatic invertebrate distributions (e.g. Palmer et al.
2000, Malmqvist 2002). Comparatively few landscape level analyses have been
conducted that examine animal distributions in lentic systems. Lehtinen et al. (1999)
found landscape level effects to be important in structuring wetland amphibian
assemblages. In one of the earlier studies to include the importance of patch context,
Bronmark (1985) examined pond isolation in an analysis of aquatic snail communities.
In Minnesota lakes, Lewis and Magnuson (2000) found differences in species richness of
snails to be dependent on the position of the lakes in the catchment and their stream
connectivity.

The latter two studies demonstrate that the application of landscape level concepts

has the potential to add to our understanding of aquatic snail communities. This is

66

 

important since explaining the distribution and diversity of aquatic snails has proved
troublesome (see Lodge et al. 1987). For example, earlyresearch espoused the
importance of physicochemical factors, especially calcium content, of the water in
determining snail distribution (Macan 1950, Aho 1966, Dussart 1976). More recently,
some have concluded that physicochemical factors were important across
biogeographical regions, but that within regions, other factors had greater influence
(Lodge et al. 1987). The effect of lake size and area also received attention as the theory
of island biogeography (MacArthur and Wilson 1967) was developed and applied to snail
communities (Lassen 1975, Aho 1978a, Pip 1985, Jokinen 1987). The role that biotic
interactions play in structuring aquatic snail communities has been less well studied, in
part because of the difficulty in quantifying their importance. Although competition
among snails is possible, there are few studies examining it (but see Eisenberg 1966).
Lodge et al. (1987) suggested that other factors, such as predation by fish and crayfish
(Brown and DeVries 1985, Lodge et al. 1987, Turner 1996), might limit snail population
densities before competition became important.

The aquatic snail community found in 26 coastal wet meadows along the northern
Lake Huron shoreline was composed of 19 species (Chapter One). Of these, six were
relatively abundant and common; the others were found in relatively few of the wet
meadow sites. Because of the generalist nature of the common species, few within-wet
meadow characteristics were found to explain their distribution and abundance patterns,
whereas for the less common species, habitat structure and plant community composition
were important (Chapter One). The limited explanatory power of the within-patch

characteristics that were examined suggests that other, perhaps landscape-level, processes

67

(i.e. context or complexity) could help explain snail distributions.

The context of the wet meadow is likely important because of the potential
interactions with surrounding habitat types. Coastal marshes participate in the exchange
and transformation of inorganic and organic materials with adjacent waters and uplands
(Sager et al. 1985, Mitsch and Gosselink 1993). It is likely that these fluxes will vary
with the plant communities that are adjacent to the wet meadow; for example, whether
the adjacent patch is a bulrush (Scirpus spp.) marsh or forested wetland or upland.
Coastal marshes generally retain particulate matter rather than dissolved materials, which
are released (Sager et al. 1985). The particulate matter may add to the detritus base,
which is a food source for many species of snails (Russell-Hunter 1978, Reavell 1980,
Brown 1982) and for which there may be specific snail-detritus preferences (Sterry et al.
1983, Lodge 1986) that affect a species presence or abundance. Wet meadow context
may also affect the snail community composition by increasing or decreasing dispersal
rates of snails from adjacent habitats (Bronmark 1985) or possibly by influencing the
predator community.

Shoreline complexity may affect the import/export of materials and therefore the
distribution of snails in coastal wet meadows by altering hydrological fluxes with the
open water. Less complex shorelines should result in exposed wet meadows that are
more strongly influenced by the open water from Lake Huron (see Lougheed et al. 2001).
In addition, landscape theory suggests that complex landscapes are more stable (F orrnan
1995), which could allow for the development of more diverse snail communities. A
number of other physical, chemical, and biological processes also are affected by

shoreline complexity (Kent and Wong 1982).

68

The objective of the present study was to determine whether characteristics of the
landscape surrounding the wet meadows could explain any additional variation in the
snail communities. Specifically, I examined the context in which the wet meadows were
located by measuring habitat patches immediately adjacent to the wet meadows and those
present in the surrounding landscape (within 1 km of the wet meadow edge). I also
examined the relationship of shoreline complexity (i.e. convolutedness) with snail
community structure. These factors were examined after taking into account the
variation explained by within-patch factors (i.e. chemistry, habitat structure, and plant

community composition).

METHODS
Study Area
The northern Lake Huron shoreline in Michigan’s Upper Peninsula contains
numerous coastal wetlands (Figure 2). Twenty-six wet meadows among the extensive
marsh complex were selected for this study (Table 1). General characteristics of these
sites are described in Chapter One: Methods, Study Area.
Snail Collection
Snails were collected using timed dip net samples from 50-m sections in transects
through each wet meadow in 1997 and 1998. Since I defined an entire wet meadow as
one patch, I randomly selected a single section from sites in which multiple sections were
sampled in order to remove bias due to differential sampling effort. In 1998, low water
levels precluded sampling from seven wet meadows. Additional details of the snail

sampling procedures are described in Chapter One: Methods, Snail Collection.

69

Within-Patch Environmental Characteristics

To account for within-patch characteristics that might explain snail distributions, I
measured 28 environmental characteristics along each transect that were subdivided into
chemistry, habitat structure, and plant community groups. The methods for sampling
these characteristics, and descriptions of each of the variables are given in Chapter One:
Methods, Environmental Variables.

Landscape Characteristics

A geographic information system was developed using ArcView (ESRI, Inc.
1999) to examine landuse-landcover patterns of the northern Lake Huron shoreline.
Landuse-landcover patterns for the study area were interpreted based on color aerial
photography from 1992. Because multiple images were required to cover the entire area,
they were each georeferenced using existing digital data from the MIRIS (Michigan
Resource Information System) database, rectified and mosaiked into one image that was
used for digitizing landuse-landcover patterns. Landuse-landcover categories (Table 10)
were verified/updated after comprehensive visual ground-truthing in 1999.

The landscape patterns surrounding the wet meadows were examined at two
levels. First, patches that were immediately adjacent to the wet meadow perimeter were
included in a PERIMETER variable group. Second, the landuse-landcover within l-km
of the edge of each wet meadow was included in a CONTEXT variable group.

The PERIMETER variable group contained 11 variables. For each wet meadow,
the number of adjacent patches, the number of adjacent patch types, and the number of
interfaces (discrete segments of wet meadow perimeter) with other patches were counted.

Perimeter length of each wet meadow and proportion of perimeter length adjacent to

70

Table 10. Landuse-landcover categories used in the geographic information system of
the northern Lake Huron shoreline, Michigan.

 

Urban
Agricultural
Non-Forested Openings
Forests
Lakes, Ponds and Reservoirs
Forested Wetlands
Inland Emergent Wetlands
Coastal Wetlands:
Alvar Wetland
Bulrush Marsh
Cattail Marsh
Fen Wetland
Sand flats
Wet Meadow
Cobble Beaches
Sand Beaches or Sand Dunes

 

7l

 

72

 

.08: 0:23 0:0 00: .0000 0:23 n5:00 .258 59:: w880800 00:03: :00:0:o\° 0:005:
80:2: 000.: @0208: 8:955: 0:0 .0000: .5000 5:0 3.x. 5:0 3 5:0
.3 020:. 80:: 5:»: 0:00:03 :0 :0 005% 53:02:. 0000003 08:1
008:: 0002 0:0 0.000580: 0:03 0:803: 5000-0:23 0:00:52 :0 3:0 :8 000053 00:55:
.582 000.: 0: 00:00::00 0:035:80? :0: 50058 0:02: :0 32.x. 005003 :5w58m: 0:02:
05:03 8:005 00:08800 0w00m 0:0 000:0 000252,?O 300002 :0 3
0005:0820 00:05 :0 0808800 00:08 85:08.: _:0::00..\c 00:02 00:00
.0030: 0035.0 :0 0208800 00:08 85580 0:00:95 00:02 0:005:

5:085: 00:0: :0 80-: 8003

EXP—1200

00:00:08 0:0 00558800 020 2050000 0:0 5800: 3:50:00: 0008:: :00:D:..\O :00: D
00:05:05 0:0—:3 O:2:..\o 585:0 00:55:-52
.08: 0:23 0:0 00: .0000 0:23 80:00 .258 :53. 8:80:50 00:08: :00:0::..\.. 85:0:
2:08: 00:00 w80208: 0002000: 0:0 .005: .5001: 5:0 3:05 5:0 >7 5:0
.3 020:1 80:: 5:»: 0:0003 :0 :0 05:: 53:00:05 00:00:03 05:1
008:: 0002 0:0 0000580: 53 0:803: 5000-0:23 0:00:52 :0 3:0 ::.x. 0.000003 00:55:
.0::0\.::§m=0 00:0,: :0 0058800 00:08 85580 _:0:00:..\.. 00:02 00:00
.0550 000.00% :0 00:08800 00:08 8558”,: 000.0530: 00:02 08:05

”0: 8000.200

558:5: 00:0: :0 505:8:
.300008 :03 0: :500_.00 0:585: 00:0: 05500 0:05:00 :0 508:2 0000:5800 500:0: 50:0 003 500:5:8 :0 0
5:085: 300008 :03 0: 8000.200 00:»: 00:0: ::0:0:::0 :0 508:2 00:31:: 5%: 00:0: :500 .200 :0 0
5:085: 300008 :03 0: :500 _.00 500:0: 05500 :0 50:82 00080:»: 500:0: :500 .200 :0 0

mafia—20:0:
50:00:00 500.: 0.005> 0:080:00

 

8520:): 6:00:00: 5.82 0000 050:5: 00: w:0_0 0300008 :03 cm :0 0000 :0: 00:25:00 00205:: 0:080:01: .2 020:.

 

0800000 :0 55582 _0:00::

50.00::
5:30:08:
580.52:

50.0085
3700850

8890:8050

55585 850::

 

0:80:05 503:00 55:80 08308000 00: :0 000000 085: 080500 3:00:82
808-5000
800-5000
8209:0000
0:00:05 5:02 000‘: :0 0&5: 008502 0:80.: 080500
80 N 0:0 .80: _ .8 com 5:000 m :0 00880.00 83000: 00: :0 000m
:Ehxm—aE—ZOU
00:0 :5800 80:3 0000: 00>0: 0:0 _0>0:w :0 0&5: 000M 80:0Q 000m
00:0 05800 8003 085:0 :0 0:80.: 800::m 80:09 800::m
00:00:08 0:0 00558800 020 505000 0:0 5800: 8:50:00: 8:080: 50:30:. 50::
58:5: 0:0 0:0—:0:”: 0:585 88:88:90
005505 0:083 0:280 585:0 00:5:0:..:02
508553 020::0> 0:080:00

 

08880 .: 030:

73

eight landuse-landcover categories were calculated using ArcView (Table 11).

The CONTEXT variable group consisted of 13 variables describing the landscape
surrounding the wet meadows for a distance of l-km from the edge of each. The
proportion of this buffer area assigned to 11 landuse-landcover categories was calculated
as well as the length of streams and roads (Table 11).

The complexity or convolutedness of the shoreline in which the wet meadows
were located was measured using vector shoreline data (Great Lakes Environmental
Laboratory, Ann Arbor, MI). Buffer operations in Arc/Info were used to create shoreline
vectors at three different scales (radii) from each wet meadow: SOO-m, l-km, and Z-km.
Three separate measures of COMPLEXITY were calculated for each shoreline segment
at each scale (Table 11); larger values of each of these metrics indicate greater
complexity in the shoreline shape. Coastline length was the overall length of the
shoreline in the buffer radius and was calculated directly from Arc/Info attribute tables.
Sinuosity was a measure of the coastline length divided by the straight-line (Euclidean)
distance between the endpoints. The fractal dimension was calculated using the walking
divider method in the F ortran program, Fract_Line (Lam and de Cola 1993).

Statistical Analyses

Environmental characteristics. Canonical correspondence analysis (CCA), using
CANOCO (ter Braak and Smilauer 1999), was used to identify environmental variables
from within each wet meadow (i.e. within-patch) that were related to snail distribution
and abundance. Three separate analyses using water chemistry, habitat structure, and
plant community composition were conducted using the forward selection procedure in

CANOCO to select variables explaining the most variance within the species data.

74

Variables that were selected using this procedure were tested for significance ((1 <0.10)
using Monte Carlo permutation testing (ter Braak and Smilauer 1998). All significant
variables from each analysis were combined for each year and used as covariables in
further analyses. Detailed descriptions of these analyses are provided in Chapter One:
Methods, Statistical Analyses.

Multivariate analyses of landscape variables and species data. Canonical
correspondance analysis (CCA), using the statistical software CAN OCO (ter Braak and
Smilauer 1999), was also utilized to express the major relationships between the species
and landscape variables. The three landscape variable groups were examined separately,
limiting the number of variables used in each analysis, as suggested by Borcard et al.
(1992). Snail abundance data were log (x+1) transformed and rare species were down-
weighted to reduce their influence using an algorithm available in CANOCO prior to
analysis. Variables with high multicollinearity were examined and those with a variable
inflation factor greater than 20 were removed (ter Braak and Smilauer 1998).

I used the forward selection procedure within CAN OCO (ter Braak and Smilauer
1998) to select the largest number of variables from each landscape variable group that
resulted in a significant model (to determine overall variance explained by the variable
group). Because I was interested in the variation in snail data that could be accounted for
by landscape variables above and beyond within-patch environmental characteristics,
significant environmental variables were entered as covariables during the CCA (Borcard
et al. 1992). In essence, this procedure removes the influence of covariables from the
variable set in question, producing extracted gradients that are uncorrelated with the

covariables (ter Braak and Verdonschot 1995). Significance was determined for both the

75

lst axis and all axes combined and a model was considered significant only if both of
these met an a priori a < 0.10. I used a more liberal value because of the variability
inherent in field observations, and to maintain an acceptable Type 11 error rate.
Significance testing was conducted using the Monte Carlo permutation testing (199
permutations) procedure within CANOCO (ter Braak and Smilauer 1998).

After defining a significant model, significant variables (i.e. those that explained
the greatest amount of species variation) were identified within each variable group. The
results were interpreted with the use of CCA biplots, in which the landscape variables are
represented as vectors among either site scores or species scores.

I conducted partial CCA using the method of Borcard et al. (1992) to determine
the amount of variability shared between the PERIMETER and CONTEXT variable
groups. Two of the wet meadows, Flowers and Hill Island, were located on islands;
therefore, sinuosity and fractal dimension could not be calculated because no endpoints
were present to measure straight-line distance. This prevented the inclusion of the
COMPLEXITY variable group in the partial CCA (Okland and Eilertsen 1994).
Additional details of this method are described in Chapter One: Methods, Statistical

Analyses.

RESULTS
Wet meadows were located among a variety of surrounding habitats. All wet
meadows were adjacent to at least two different patches and up to 17 interfaces with
adjacent patches (Table 12), representing a very simple to very complex landscape

configuration. The most common landuse-landcover category immediately adjacent to

76

Table 12. Summary statistics for three groups of landscape variables. *Complexity data
are calculated for 24 wet meadows; sinuosity and fractal dimension were unable to be

calculated for two meadows located on islands.

 

 

Landscape Variable Units Mean SE Range
PERIMETER
# of adjacent patches #/patch 4.46 0.49 2.00-11.00
# of adjacent patch types #/patch 3.38 0.25 2.00-6.00
# of interfaces with other patches #/patch 5.46 0.74 2.00-17.00
Proportion of patch perimeter adjacent to:
Bulrush Marsh (Scirpus spp.) % 19.26 3.66 0.00-58.29
Cattail Marsh (Typha spp.) % 3.65 1.48 0.00-25.25
Forested Wetlands % 5.37 2.82 0.00-68.05
Total Wetlands % 28.28 4.50 000-7539
Open water % 8.61 3.65 0.00-76.50
Forests % 53.00 4.51 13.27-93.10
Non-Forested Openings % 0.90 0.54 0.00-12.13
Urban % 7.52 2.74 0.00-62.40
CONTEXT
Within l-km of patch perimeter:
Bulrush Marsh (Scirpus spp.) % 4.13 0.84 0.00-16.45
Cattail Marsh (T ypha spp.) % 0.16 0.05 0.00-0.91
Wet meadow % 4.24 0.68 0.23-11.01
Inland Emergent Wetlands % 0.40 0.24 0.00-5.19
Forested Wetlands % 2.14 0.58 0.12-12.34
Total Wetlands % 11.13 1.29 1.48-26.55
Open water % 37.29 3.56 1 1.48-71.01
Forests °/o 70.50 3.47 23.14-94.32
Non-Forested Openings % 3.83 0.84 0.00-17.96
Agricultural % 0.18 0.10 0.00-1.96
Urban % 14.09 2.80 0.00-55.4
Stream Density m/ha 5.44 0.42 2.52-10.37
Road Density m/ha 11.92 1.03 0.10-24.49
COMPLEXITY*
Coastline Length
500 m m/ha 6.32 0.46 2.52-11.47
1 km m/ha 12.46 0.64 685-1869
2 km m/ha 27.89 1.76 11.85-44.84
Sinuosity
500 m — 3.29 0.37 1.18-8.35
1 km — 2.73 0.26 1.24-6.76
2 km — 2.65 0.20 1.25-5.67
Fractal Dimension
500 m — 0.02 1.03-1.42
1 km — 0.02 1.03-1.29
2 km —— 0.01 1.04-1.28

 

77

and within the surrounding (1-km) context of the wet meadows was forests, accounting
for 53.0% and 70.5%, respectively (Table 12). In terms of the shoreline complexity, the
shorelines ranged from nearly straight-line (i.e. sinuosity and fractal dimension near 1.0)
to very convoluted (e.g. sinuosity 44.84 and fractal dimension 1.42) (Table 12).

Three PERIMETER variables were included in the significant model in 1997,
explaining 18.1 % of the variation in snail distributions after accounting for within-patch
habitat characteristics (Table 13). Non-forested openings was the only significant
variable although the number of interfaces with other patches and adjacency to cattail
marsh were included in the significant model (Table 14). The first canonical axis
represented the increasing proportion of adjacent non-forested openings and the second
axis was correlated with the number of interfaces and increasing cattail (Figure 15, Table
14). Aplexa elongata was highly correlated with the increasing proportion of non-
forested openings and all other species were only moderately associated with these axes
(Figure 15).

In the 1997 CONTEXT variable group, the proportion of forests was highly
intercorrelated with the other variables and was removed from the analyses. The full
model included three variables explaining 18.9% of the species data. The proportion of
non-forested openings and bulrush marsh in the l-km surrounding the wet meadow were
significant variables (Tables 13 and 14). The first axis of the CCA corresponded to an
increasing proportion of non-forested openings whereas the second axis was a gradient
from bulrush marsh to inland emergent wetlands (Figure 16, Table 14). Amnicola
walkeri and A. elongata were associated with wet meadows situated in a context with

increasing non-forested openings whereas Bithynia tentaculata was associated with

78

.358 E cops—2: 803 353:2» 085 3:0 .—

 

I odo— odn on
I CNN 0: W:

I wand oood vad
I wmod mmod mood
Sod mid mood mood CNN oomd ammo

I 3:: wdw Q?
I $3 52 2:

I oond nowd and
I owed mmod wood
Sod mmvd mnod omod QM: omod aZd

I ode Q; 0.0m
I 2: 5.3 wro—

I 036 vowd 033
I mmod wmod mood

Sod ENN cmod Sad 22 omwd v2.0

5:30:00 Bogobéoéofiomm Co
8% 86on mo
beefing owficoeom 3:23:50
80:28.50 Eoficob>coa2ooqm
mos—miswfi

32 >H~qumEOU

sown—2.50 EoEco:>:o-mo_oomm («0
Saw 86on mo
85:? owficooaom 0233830
28:38.80 EoE:8_>:o-mo_ooqm
moflgcowmm

32 HXmHZOU

20:28.50 “cogegcoéfloomm mo
8% 36on mo
853:; owficoeom 02:23:50
macaw—oboe “58:83:98?on
mos—gamma

32 duhmwzamm

 

a 285* m 0:8-..“ 35398 «:85 mos—«>520 v m m _
moxm =< 399% 3599 o\° 30H 33:88 w 35..

 

 

flag 5 95888 63 13930 .825 8:3 Boston 5 mafia osmium mo marge monogamous E08058 mo 938m .2 2an

79

Table 14. Interset correlations of environmental variables with CCA axes in 1997 for all
variables included in significant model. Significant variables (P<O.10) are indicated by

 

 

 

bold type.
Interset Correlations
Variable Axis 1 Axis 2 Axis 3 Axis 4
PERIMETER 1997
% Non-Forested Openings 0.716 -0.009 0.446 —‘
# of interfaces with other patches -0.044 0.738 -0.329 —
% Cattail Marsh -0.211 0.672 0.410 —
CONTEXT 1997
% Non-Forested Openings 0.745 0.276 0.308 ~—
% Bulrush Marsh 0.253 -0.736 -0.311 —
% Inland Emergent Wetlands -0.032 0.684 -0.430 —
COMPLEXITY 1997
Sinuosity - 500 m -0.660 0.565 -0.011 —
Sinuosity - 2 km -0.527 -0.096 -0.615 —
Fractal Dimension - 1 km -0.009 0.457 -0.687 —-—

 

1. With three variables only three canonical axes are produced.

80

1'+1.o

 

#Interfaces

%PCattail I.
h» u
\
Fl
Pc AW
\ o 9
g‘ Gd %PNFO
as = - - 8.3%.,- _ - 4, --...o—
Pathp Ae

 

db
v
I

 

Figure 15. Canonical correspondence analysis of aquatic snails in northern Lake Huron
coastal wet meadows using the 1997 PERIMETER variable group as explanatory
variables: (A) species-environmental biplot (B) sites-environmental biplot. All plotted
landscape variables were included in a significant model; those with solid arrows were
significant variables and dashed arrows indicate non-significant variables. Species and
sites scores are adjusted to scale = 0.3. Landscape variable abbreviations as in Table 11,
species labels as in Table 2, and sites labels as in Table l.

81

Figure 15. Continued.

 

82

l

I

+1.0

°/oanet 1'

 

I
‘I
I

l
I...
I

I

I

   

 

 

Aw
Pc 0
Fl
0
Foo %NFO
Ae
Pt Fp -
-1.0' ' ' ‘ ' T ‘T 11.6
.89
l.
%Bu|rush
.. A
-1.0

 

Figure 16. Canonical correspondence analysis of aquatic snails in northern Lake Huron
coastal wet meadows using the 1997 CONTEXT variable group as explanatory variables:
(A) species-environmental biplot (B) sites-environmental biplot. All plotted landscape
variables were included in a significant model; those with solid arrows were significant
variables and dashed arrows indicate non-significant variables. Species and sites scores

are adjusted to scale = 0.3. Landscape variable abbreviations as in Table 11, species
labels as in Table 2, and sites labels as in Table 1.

83

31.0

Figure 16. Continued.

84

 

+1.0

ID

increasing bulrush marsh and Stagnicola elodes was intermediate with both bulrush
marshes and non-forested Openings (Figure 16).

Significant variables at all three scales were included in the 1997 COMPLEXITY
model and explained 22.0% of the variation in species patterns (Table 13). Axis l was
correlated with the sinuosity of the shoreline at the 500-m and 2-km scales and axis 2 was
correlated with the 500-m sinuosity and l-km fractal dimension (Figure 17, Table 14).
Aplexa elongata and A. walkeri were clearly associated with less complex, straighter
shorelines, whereas F errissia parallela, Planorbella trivolvis, and F ossaria obrussa were
associated with more complex, convoluted shorelines (Figure 17).

In 1998, seven variables were included in the PERIMETER model, combining to
explain 43.1% of the variation in species data (Table 15). The variable, number of
patches, was removed due to high multicollinearity. As in 1997, the proportion of
adjacent non-forested openings was significant and included in a gradient with increasing
bulrush on the opposite end (Figure 18, Table 16). Cattail marsh was the other
significant variable and was highly correlated with the second axis (Table 16). Aplexa
elongata was associated with increasing non-forested openings whereas F. parallela and
Promenetus exacuous were correlated with increasing bulrush marsh (Figure 18).
Bithynia tentaculata and Planorbella campanulata were negatively associated with
cattail marsh whereas F. obrussa, P. trivolvis, and Gyraulus crista were all positively
associated with cattail and several other variables, including % urban (Figure 18).

The proportion of bulrush marsh was the only significant variable in the 1998
CONTEXT group along with three additional variables in the model. Together they

explained 30.2% of the variation in snail distributions (Tables 15 and 16). The first axis

85

"+1.0

Sinuosity-500m

   

 

.L Fractal-1k Ae

 

 

l L A A A le l l l A j
-1.0' ' i t j . ' . ' t +10
Sinuosity-2k P P9 GP .Aw
a
II A
-1.0

 

Figure 17. Canonical correspondence analysis of aquatic snails in northern Lake Huron
coastal wet meadows using the 1997 COMPLEXITY variable group as explanatory
variables: (A) species-environmental biplot (B) sites-environmental biplot. All plotted
landscape variables were included in a significant model; those with solid arrows were
significant variables and dashed arrows indicate non-significant variables. Species and
sites scores are adjusted to scale = 0.3. Landscape variable abbreviations as in Table 11,
species labels as in Table 2, and sites labels as in Table 1.

86

 

 

 

Figure 17. Continued.

87

II
II

 

ESE 83$ch 02
woo_ >EXm—QAEOU

Odm: m.m© OWN. m6? Gomwm—OCOO uGOECOfi>CDImomoomm MO

 

 

Sm 2a SN 3; $3208.“?

859?. 098523 3:23:50

oood Cad wood wood mucus—oboe EoEconmEoéomooam

Sod omod oood vofio won—gcowmm

omod moo; oo_d SON Ndm mid vmmd woflthHZOU

#3 9mm 5:. Y? 8:22.80Bogota/5-360%?

mam Em 3m 2: smuwaoaflo

85E.» owficoonoo 3325830

ind omwd nmod Rod mcouflotoo“55835-86on

Rod mmod wood ofld mos—gcowfi

mmod vmo; mmod oovd d? mid ommd woo_MmHm22mo
m own: m 28H-..— ooEmExo «FEE mos—gauge v m m _

move =< 38 mm 8593 .x. 30H 13:8me N %

 

.woo_ E m3oomoE $3 338 :05: 834 Boston 5 28cm cannon oo £9298 cosmos—comets 1889830 338% .2 033.

88

Table 16. Interset correlations of environmental variables with CCA axes in 1998 for all
variables included in significant model. Significant variables (P<O.10) are indicated by

 

 

 

bold type.
Interset Correlations
Variable Axis 1 Axis 2 Axis 3 Axis 4
PERIMETER 1998
% Non-Forested Openings 0.845 0.078 0.069 -0.028
% Bulrush Marsh -0.374 0.127 -0.087 -0.049
% Cattail Marsh 0.261 -0.708 -0.310 -0.311
# of adjacent patch types 0.082 -0.479 0.287 0.247
# of interfaces with other patches -0.027 -0.495 0.269 -0.l38
% Urban -0.065 -0.330 0.292 0.450
% Forests -0.118 0.204 -0.298 -0.544
CONTEXT 1998
% Bulrush Marsh 0.597 0.591 -0.044 -0.239
% Non-Forested Openings -0.3 l 8 0.463 0.271 0.488
% Forested Wetlands 0.566 0.099 0.599 -0.164
% Urban 0.241 -0.037 -0.476 0.532

COMPLEXITY 1998
No significant model

 

89

+1.0

 

 

 

#lnterfaces 0

1|

 

Figure 18. Canonical correspondence analysis of aquatic snails in northern Lake Huron
coastal wet meadows using the 1998 PERIMETER variable group as explanatory
variables: (A) species-environmental biplot (B) sites-environmental biplot. All plotted
landscape variables were included in a significant model; those with solid arrows were
significant variables and dashed arrows indicate non-significant variables. Species and
sites scores are adjusted to scale = 0.3. Landscape variable abbreviations as in Table 11,
species labels as in Table 2, and sites labels as in Table 1.

90

I
T

Figure 18. Continued.

II

91

+1.0

db
II
II
II
J.
II

+1.0

 

was positively correlated with bulrush and forested wetlands and the second with bulrush
and non-forested openings (Figure 19, Table 16). Snails positively associated with the
first axis were F. parallela and B. tentaculata whereas A. elongata and P. trivolvis were
negatively associated. Aplexa elongata was associated with non-forested openings
(Figure 19).

There was no significant model for shoreline COMPLEXITY in 1998. Two wet
meadows were not included in the COMPLEXITY analyses because of their location on
islands, which prevented the calculation of sinuosity fractal dimension. Thus, in the
partial CCA, only the PERIMETER and CONTEXT variable groups were included.

In 1997, partial CCA indicated that both PERIMETER and CONTEXT explained
an approximately equal percentage of the variation in the snail data and also shared a
large proportion of variance explained. Combined, they explained 28.9% of the variation
in the snail data above and beyond that which was explained by within-patch habitat
characteristics (Figure 20). In 1998, a greater overall proportion (62.7%) of the variance
was explained by PERIMETER and CONTEXT (Figure 21) beyond that of the within-
patch characteristics. The PERIMETER variable group explained the majority of the
variation, although 10.6% of the variation explained was shared by both groups (Figure

21).

DISCUSSION
Snail distribution and abundance patterns are influenced by many different factors
that vary with the scale being examined (Lodge et al. 1987). Among these are water

chemistry, habitat diversity, and plant community composition (e. g. Aho 1966, Harman

92

+1.0

 

%Bulrush

   

%ForWet

 

 

0
Pt

   

Pa P9 PI %Urban ”'0
.I C
Gd .
.. ' Fl
.I
«r
up A
-1.0

 

Figure 19. Canonical correspondence analysis of aquatic snails in northern Lake Huron
coastal wet meadows using the 1998 CONTEXT variable group as explanatory variables:
(A) species-environmental biplot (B) sites-environmental biplot. All plotted landscape
variables were included in a significant model; those with solid arrows were significant
variables and dashed arrows indicate non-significant variables. Species and sites scores
are adjusted to scale = 0.3. Landscape variable abbreviations as in Table 11, species

labels as in Table 2, and sites labels as in Table 1.

93

q
1!
in

Figure 19. Continued.

1b

94

+1.0

 

'11 .d

 

UNEXPLAINED
[71.1]

   
    
  

PERIMETER
H81]

    
 
   
   
 

10.0 10.8

CONTEXT
[18.9]

 

 

 

Figure 20. Venn diagram of the CCA variance partitioning of 1997 species data among
two variable groups. Total variance explained by a variable group is enclosed in [ ] and
represented by the size of the circle. Variance shared by variable groups is indicated in
the appropriate intersections. Variance explained is calculated as the sum of all canonical
eigenvalues for a particular analysis divided by the total inertia.

95

 

UNEXPLAINED
[37.3]

PERIMETER
[43.1]

CONTEXT
[30.2]

 

 

 

 

Figure 21. Venn diagram of the CCA variance partitioning of 1998 species data among
two variable groups. Total variance explained by a variable group is enclosed in [ ] and
represented by the size of the circle. Variance shared by variable groups is indicated in
the appropriate intersections. Variance explained is calculated as the sum of all canonical
eigenvalues for a particular analysis divided by the total inertia.

96

1972, Pip 1978, Brénmark 1985). These characteristics of a water body are in turn
affected by their position in the surrounding landscape due to regional and local
differences in topography, groundwater inputs, and soil conditions (Kratz et al. 1997).
Lewis and Magnuson (2000) demonstrated that lake snail assemblages were influenced
by the spatial pattern of lakes in Wisconsin. The present study provides evidence that
snail communities of northern Lake Huron coastal wet meadows also respond to
characteristics at the landscape scale.
Landscape Context and Snail Associations

Because I examined the landscape variables and snail distributions after
accounting for variations in the snail community due to within-wet meadow (i.e. within-
patch) characteristics, the associations identified were responses at the landscape scale.
Snail communities in wet meadows were influenced by adjacent patch characteristics
(PERIMETER) as well as the characteristics of a broader area (CONTEXT) surrounding
the wet meadow (Figures 22 and 23). It was not surprising that a substantial percentage
of the variation in the snail communities was shared by both the immediate as well as the
broader context of the wet meadows (Figures 20 and 21). The majority of snail
associations with landscape features involved the less common species and therefore
these associations must be interpreted with caution (see Chapter One: Discussion,
Patterns in Common and Uncommon Species). However, the patterns observed are
consistent and interpretable allowing conclusions to be made (ter Braak and Verdonschot
1995). Because this is one of the first studies to examine snail-landscape associations,
these conclusions should not imply causal mechanisms, but should be regarded as a

starting point to future investigations and to experiments that involve more rigorous

97

.:o50::: 56:35:50 2: :_ 5%:

0:: E255 55 5:5 55585 55on mam A o E 0:: 532:5, :50E:w_m:o: .5385”: ”Bo—op 5%: 0:: 3:8 55 55 5:22:00
55:5.» Boomimmm .265 5:535 85 55on 0:: mo 58505: 5:3 5:: 73:85.0 :8 35:55 35:80:35 :55:5:
@505 .855 255:, 55:80:35 5:5 mo 55 8.: <00 mam: 83505—5: :5:::o:m>5-_:w:m 32 on: do 55m .9 BawE

Eden 5 >585 n5

 

 

 

 

 

 

5350 mtmmmou Ex-.. 5 55556 =59: 0.55905
$29.5 m5€o5i A
5:88 £3.58”: $5
59.50 mtmwmom Ex-u :5 Eden 5 >535 9555:.
$29.5 mooofiocmi A .2952 38::de
$2:qu mommEon: com: 8.8.5:: 98:?
>._._xm._as_oo
3:553 :cooEEm
55E 52.3: xx“ 95555 95.5 .x. 9555::
$5585: NEASE Y
883 aoofimam $3
5555
5592-5: .x. 95555

559: $8EE< A
83:05 mon< #09

3.05:: 5:5 .xo UAW—.200

A585: 585 :o a:
32:98 :85”: $3.
355::
5598-5: .x. 95555

33:03 8:me A $03

”GEN—2.1mm

98

80980 883:

88:08:08 8:85
80:80: 800:8“:

0502:: 80088::
80.:0 0:38am
000380 0.20000”:

80950 8:83:

3:588 8:00.020: oi
:88: 52.2: ..\o 05:00:05

80200:: 85800280 05 :_ :80:
0:: 805:8 8:0 5:5 :0::8000: 00:00am =::m A v :m 0:: 058:? 885868: 50:80:80 ”3200 :05: 0:: 8x: :0:0 5:5 “00:80:80

838:8, 8:00:85 .080: “005898 8:: 00:00am 05 .8 88:00:00 5:5 000:: 308080 8.: 885:8 8808:0880 800080:
030:5. 098% 0E8::> 8:088:38 00:5 .8 8:0 8.: <00 wEm: 0020:0580: 8088580885 :00. 05 .8 9:885 .mm 08me

 

A

3:50:03 :8008: $0
:88: :023: .x. 0:600:05

£3:

 

A

$0.3.

A009: :08: 3: 000885 a . :05: ch

:88: .8300 o\o 90:08:.

 

A

098000
8:008:78: .x. 0500805

$5~F

:88:
52.0: .x. 93885

 

A

$03

V

0:02

>._._xm._n=200

8202:: 8:050:81
80980 0x0E<

UAW—.200

88:89:00 8:050:8Q
88:08:08 8.35.5

0000088 0808808
80:80: 800E0m

mmhmEP—mn—

99

testing of specific hypotheses.

The most common landscape characteristic and the one that explained the most
variance in the CCA models was non—forested openings, either immediately adjacent to
the wet meadow or within a l-km area around the wet meadow. In the GIS classification
scheme used (Table 10), non-forest openings were primarily upland grassy and/or
shrubby areas. Thus, wet meadows adjacent to non-forest openings were essentially wet
meadows that graded into grasslands as the elevation increased and the water saturation
levels decreased. The grass uplands may act as a buffer between the terrestrial and
aquatic habitats, limiting nutrient or other inputs that may affect the wet meadows and the
snail communities (Karr and Schlosser 1978, Krieger and Klarer 1991 , Crosbie and
Chow-Fraser 1999). Snails found in wet meadows associated with increasing non-
forested openings included Aplexa elongata and Amnicola walkeri (Figures 22 and 23).

Second in frequency of occurrence in the CCA models was the proportion of
bulrush marsh. Similar to the buffering capabilities of adjacent non-forested openings,
bulrush marsh, which was commonly adjacent but in deeper water than the wet meadows
(Figure 1), probably acts as a buffer to the open waters of Lake Huron. Marsh vegetation
reduces the velocity of incoming waves (Leonard and Luther 1995) and this may have
significant effects on the communities of some organisms (Cardinale et al. 1997). The
reduction in water flow into the wet meadow may settle out suspended solids creating
clearer water conditions in wet meadows adjacent to extensive bulrush marshes or may
allow for greater development of the wet meadow region. Differences in water quality
may directly affect the snail communities or indirectly affect them through some related

mechanism such as changes in the periphyton used for food (Russel-Hunter 1978).

100

Snails showed varying associations with the proportion of bulrush marsh.
Stagnicola elodes, F errissia parallela, Bithynia tentaculata and Promenetus exacuous
were positively associated with bulrush whereas A. elongata and Planorbella trivolvis
were negatively associated. Bithynia tentaculata and P. exacuous, were most commonly
linked with more densely vegetated habitats and therefore the positive association with
bulrush was likely associated with the greater development of the wet meadow region
allowed by the presence of the bulrush marsh. On the other hand, S. elodes and
especially F. parallela were often found on bulrush stems (Clarke 1981). The bulrush
marsh may act as a source for the dispersal of these snails into the adjacent wet meadows.
Aplexa elongata was generally found in habitats with high vegetation density (Clarke
1981, Chapter One) so the negative relationship with bulrush marsh (which would favor
greater wave forces reaching the wet meadow allowing less vegetation development) was
unexpected. However, the sites at which A. elongata was common and abundant
included Pontchartrain 1 and 2, both of which have wet meadows protected from the
open waters of Lake Huron by a sandy ridge (see Table l).

The proportion of cattail marsh was the final significant variable in the CCA
models. F ossaria obrussa, Gyraulus crista, and P. trivolvis were positively associated
with wet meadows adjacent to cattail marshes. The presence of cattail marshes is
generally a sign of disturbance and/or increased nutrient loadings (Galatowitsch et al.
2000) and these snail species are usually associated with heavily vegetated, eutrophic
wetlands (Clarke 1981, Burch and Jung 1992). Thus, this association appears to be
another example of the adjacency pattern reflecting a source for the dispersal of these

snails into the wet meadow. Additionally, cattail marshes are very productive (Mitsch

lOl

and Gosselink 1993) and there may be a transfer of detritus and other nutrients to the wet
meadow that influences these snail species.

A number of other landscape variables were included in significant models
although they were not significant by themselves. These generally included factors
related to number of patch types or interfaces with other patches, and suggested that the
interconnectivity of wetlands with the surrounding environment was a factor in
structuring the snail community. A loss of the diversity of surrounding patches may,
therefore, lead to changes in the wet meadow snail community, and this suggests that
caution is needed to prevent additional anthropogenic changes to the northern Lake
Huron shoreline. Although the proportion of urban patches was included in one model,
surprisingly the density of roads in the area, which are known to affect many aquatic
communities (Trombulak and Frissell 2000), did not seem to strongly affect the snail
community.

Shoreline Complexity

The complexity or convolutedness of the shoreline in 1997 was significantly
associated with many of the same species as above. This was likely due to effects of
shoreline shape on the development of the different patch types (Kent and Wong 1982).
For example, more complex shorelines may permit greater wetland development because
of protection from Lake Huron waves and storm surges.

Significant COMPLEXITY variables included sinuosity and fractal dimension at
all three of the measured scales. Although fractal dimension is theoretically independent
(i.e. self-similar) of the measurement scale, it was only significantly associated with the

snail community at the l-km scale. This was likely an artifact of the sampling scale and

102

may reflect differences in the processes related to the convolutedness (Forrnan 1995).
Sinuosity was significant at the SOO-m and 2-km scales.

Three species of snails were positively associated with sinuosity. F ossaria
obrussa and P. trivolvis were characteristic of densely vegetated wet meadows and F.
parallela with bulrush marshes that would be expected to develop in a convoluted bay
protected from intense wave action and storm surges coming off of Lake Huron. Aplexa
elongata and Amnicola walkeri were associated with “straighter” shorelines that were
directly impacted by the open waters of Lake Huron. The wet meadows in which they
were found were created and protected by sandy ridges produced by the wave action
present on these stretches of shore and that are rare in the protected bays.

In 1998, there was no significant model for COMPLEXITY. The major
environmental change from 1997 to 1998 was a significant decline in water levels
(U SACE 2000) suggesting that it was a factor in the annual change. It is possible that a
drawdown event may affect the wetlands in complex shorelines differently than those in
straighter shorelines and thereby affect the snail community. These changes may have
added additional variation to the data that made detection of a significant model in the
already noisy data impossible.

Conclusions

In conclusion, this study has shown that patterns of landscape context
(associations with non-forested openings, bulrush and cattail marshes) and shoreline
complexity exist in coastal wet meadow snail communities. However, I must reiterate
that the determination of cause and effect was beyond the scope of this study and

explicitly designed studies are needed to establish the mechanisms behind the observed

103

patterns. Future efforts should be targeted towards identifying other landscape effects

and/or confirming those found, and towards the determination of causal mechanisms.

104

CHAPTER THREE

INFLUENCE OF WATER LEVEL FLUCTUATIONS ON COASTAL WET MEADOW

SNAIL COMMUNITIES

105

INTRODUCTION

In coastal wetlands of the Laurentian Great Lakes, water level fluctuations occur
over short-term, seasonal, and multi-annual time scales (Keough et al. 1999). Short-term
fluctuations include seiches and storm surges that may vary from less than an hour to
several hours with amplitudes of tens of centimeters. Seasonal fluctuations often include
mid-summer peaks that are about 30 cm higher than winter lows (Keough et al. 1999).
Multi-annual water level fluctuations vary in periodicity and amplitude from year to year
depending on regional climatic factors (Bedford 1992). Multi-annual water levels may
vary over a range of almost 2 m (Kelley et al. 1984, Keough et al. 1999).

Water level fluctuations strongly affect structure and function of Great Lakes
wetlands (Burton 1985, Keough et al. 1999). Previous studies have documented changes
in community characteristics and species occurrences for wetland plants (Keddy and
Reznicek 1985, 1986, Edsall et al. 1988), fish (Liston and Chubb 1985), and birds
(McNicholl 1985). Invertebrates, including aquatic snails, are understudied (Krieger
1992). It has been suggested that they show a lagged response to the changing plant
communities (Gathman et al. 1999).

Patterns in freshwater snail distributions have been explained on the basis of water
chemistry (Dussart 1976, 1979, McKillop 1985, Okland 1983), and especially the
concentration of calcium (Macan 1950, Aho 1966, Okland 1983, Rooke and Mackie
1984, McKillop 1985). Others have used the theory of island biogeography to explain
snail distributions in lakes by treating lakes as islands within the mainland (Lassen 1975,
Aho 1978a,b, Browne 1981). These studies have focused on variables such as area and

isolation (Jokinen 1983, Bronmark 1985), position within the landscape (Lewis and

106

Magnuson 2000), and the role of habitat variability and disturbance on the snail
community (Bronmark 1985, Lodge and Kelly 1985). Lodge et a1. (1987) synthesized
these ideas into a conceptual model to explain the distribution of freshwater snails over
various scales. Wellbom et a1. (1996) proposed a similar model for freshwater
communities in general. Both models stressed the importance of hydroperiods along a
gradient from temporary ponds to permanent lakes. Lodge et al. (1987) suggested that in
temporary habitats with similar chemistry, habitat availability and disturbance (e.g.
periodic drying) were the most important structuring forces for snail communities.

Great Lakes coastal wetland habitats do not fall neatly into the gradient fi'om
temporary ponds to permanent lakes described in the conceptual models of Lodge et al.
(1987) and Wellbom et al. (1996). Coastal wetlands often occur along a continuum of
habitats, from higher elevation shrub and wet meadow zones (Figure 1), which often lack
standing water, to deeper, permanently flooded emergent marshes (Mine 1996).
Connectivity of wetlands along this continuum allows mobile aquatic animals to seek
refuge in the emergent marsh during low water levels. During high water levels, mobile
aquatic animals may quickly move back into the newly flooded habitat in the wet
meadow. Behavioral responses (e. g. movement towards and dispersal from refuge) may
be more important to mobile species whereas physiological (e.g. desiccation resistance)
adaptations are more likely to be of importance for more sessile organisms found in wet
meadows.

Movement patterns and desiccation resistance are well known for many marine
intertidal snails subjected to daily tidal fluctuations (Levings and Garrity 1983, Marchetti

and Geller 1987, Chapman and Underwood 1996). Freshwater snails in non-coastal

107

temporary habitats exhibit behaviors and resistance to desiccation that enable them to
survive seasonal dry periods (Jokinen 1978, Brown 1979, Cantrell 1981, Betterton et al.
1988, Vera et al. 1995) with desiccation resistance probably being the most important
determinant of species occurrence (Lodge et al. 1987). In coastal wetlands, however, the
continuum from permanent to temporary marsh suggests that species that are mobile may
be able to survive and compete with species that are poor dispersers but are highly
resistant to desiccation.

The water levels in coastal wetlands of northern Lake Huron declined 28 cm from
1997 to 1998 (U SACE 2000) providing a natural experiment to examine the responses of
the snail community. In this study, I used multiple regression analyses to examine the
role of spatial and short-term variation in water levels in structuring the aquatic snail
communities. Specifically, I tested six different explanatory water level variables to
determine how well they predicted the occurrence of snails in these wetlands. I also
contrasted changes in coastal wet meadow snail communities where water level declines
left the sites dry, to sites where water levels decreased but the sites remained flooded, and
to sites where the water levels remained constant between years. Finally, I used two
laboratory experiments to examine the proximate behavioral response of snails to water
level decreases and their desiccation resistance as two possible mechanisms influencing

the observed patterns in community structure.

METHODS
Study Area

I sampled 26 distinct coastal wet meadows along a 60 km length of the northern

108

Lake Huron shoreline in Michigan’s Upper Peninsula (Mackinac and Chippewa counties)
(Figure 2). The wet meadows all were located within 500 m of the shoreline and water
levels were influenced by Lake Huron water levels (Table 1). The vegetation was
predominantly hummock-forming grasses (Calamagrostis canadensis) and sedges (Carex
stricta, C. aquatilis, and C. latifolia) with low densities of bulrush (primarily Scirpus
aeutus) and cattail (primarily Typha angustifolia). The substrate was muck or mucky-
peat, covered by dense layers of the previous season’s grass and sedge leaves, and
provided a structurally complex habitat (along with the hummock structure) in which
most snails were found. Some sites also contained a shrub zone including Salix spp.,
Myrica gale, and Alnus rugosa. A more detailed description is in Chapter One: Methods,
Study Area.
Snail Sampling

Transects running perpendicular to the Lake Huron shoreline were established in
each wet meadow and divided into 50 m sections. The number of sections sampled in
each wet meadow ranged from 1 to 3 depending on the size of the wet meadow and
whether the sections contained standing water deep enough to sample with a dip net (~10
cm) (Table l). Snails were collected using timed D-frame aquatic dip net samples
(Brown 1979). A more detailed description of this sampling protocol is provided in
Chapter One: Methods, Snail Collection.

Sections that were dry (lacked standing water over >50% of the area) in 1998
(Table 17) were sampled by removing five round cores (20 cm diameter, 15 cm deep)
from each dry section using a plastic bucket with the bottom removed and the edge

serrated. Samples were collected from depressions between hummocks making sure to

109

maintain the sediments and detritus layer intact. After collection, cores were returned to
the lab, placed separately in buckets and 5 liters of non-chlorinated well water were added
to each. After being flooded for 2 (1, water and detritus were poured off and rinsed twice
through a SOO-um sieve. The contents of the sieve and any snails observed were
preserved in 95% ethanol to which Rose Bengal stain had been added. In the laboratory,
snails in the preserved samples were sorted from the detritus, identified to species, and
enumerated (empty shells were not included).
Water Level Measurements

Four SO-m radii were established at right angles to one another from the center of
each section. Every 10 m along these radii, the water depth between hummocks was
measured using a marked, metal rod (nearest 1 cm), for a total of 20 measurements per
section (see Figure 3). Mean, maximum, minimum and range of depth were calculated
for each section from these 20 measurements. Variation, in addition to the range, was
quantified as the coefficient of variation (hereafter referred to as the spatial CV), which is
a relative standard deviation based on the mean.

Because water levels in these coastal wet meadows vary not only spatially, but
also temporally because of fluctuations in Lake Huron levels and rainfall events, a
temporal CV was also calculated. Temporal CV was calculated using depth
measurements taken approximately 1 week apart, 4 times during June of each year. Each
of these measurements was the mean of 2 depths measured at the start and end of the

transect in each section.

110

Multiple Regression Analyses

All six water level explanatory variables (mean, maximum, minimum, range,
spatial CV, and temporal CV) were used in regression models as predictors of the snail
community. Because lake levels declined in 1998, analyses were conducted separately
for data from 1997 and 1998. All statistical analyses were conducted using SYSTAT
software (SPSS 1999).

Snail variables. Species richness was calculated as the number of different
species in each section. I calculated total abundance as the number of all individuals of
all species collected in a section. I also calculated the abundance of each of the common
species (i.e. found in at least 2/3 of all sections sampled). Presence/absence was
determined for less common species (found in >10%, but <2/3 of the sections). Species
found in fewer than 10% of the sections were not included in analyses.

Regression models. Stepwise multiple linear regression was used to examine
significant water level predictors for total abundance, richness, and individual species
abundances. An F-to-stay=0.15 was used to establish an initial set of water level
predictors and models containing each possible subset of this initial set were then
assessed. Criteria included overall significance of the model (a<0.10), significant t-
values for each predictor variable (on<0.10), and all assumptions of regression (e.g.
normality of residuals and constant variance) had to be satisfied. Outliers (>3 SD.) and
points with high leverage (>0.5) were removed and the model reassessed. Models that
explained the highest amount of variance (R2) and met these criteria were kept.

Stepwise multiple logistic regression was used to test presence/absence of species.

111

Again, an F-to-stay=0.15 was used and models containing each subset were assessed.
Criteria included overall significance of the model and significant t—ratios for each
predictor variable (a<0.10). Additionally, each model passed the Hosmer-Lemeshow
goodness-of-fit test (Hosmer and Lemeshow 1989). Models meeting these criteria that
explained the highest amount of variance (McFadden’s p2) were kept.
Annual Comparisons

Mean depths were calculated from the water level measurements for each section
and sections divided into three classes: (1) ‘Dry’ sections were those that lacked standing
water in 1998; (2) ‘Lower’ sections contained standing water in both years but were
shallower in 1998; and (3) ‘Same’ sections in which the water levels were approximately
equal in 1997 and 1998 (Table 17). Each section was treated as an independent
observation since snails were assumed to respond independently to the differences found
over the 50 m distance.

Because Dry sections could not be sampled using the same method in 1997 and
1998, the abundance of snails could not be compared directly. For these sections, the
presence/absence of each species was calculated and the proportions between years
compared using the Fisher exact test. This test has no statistic, and instead directly
calculates a P-value using a 2x2 contingency table (year and presence/absence) (Zar
1999). The abundance of snails in the Lower and Same sections were compared using the
Wilcoxon signed ranks test. This non-parametric test was used because the data did not
meet the assumption of normality. Statistical tests were computed using SYSTAT

software (SPSS 1999) using an a priori 0t<0.10. This more liberal value was used

112

Table 17. Comparison of the sampling effort in wet meadows of the northern Lake
Huron shoreline from 1997 to 1998, and the classification of water level changes. See
text for details.

 

Number of sections sampled
Site 1 997 1 998 Classification

label Wet meadow NetI Core2 Net Core Dry Lower Same

 

Pontchartrain 1
Pontchartrain 2
St. Martin’s
Carpenter 1
Carpenter 2
Search 1
Search 2
Mismer
Hessel 1
Hessel 2
Hessel 3
Mackinac
Rudd

Flowers

Hill Island
Lakeside

Port Dolomite
Prentiss
Scotty
Seymour 1
Seymour 2
Seymour 3
Sweets 1
Sweets 2
Sweets 3
Sweets 4

|"""‘|

IINN"|"‘N|

|"‘|

~—Nwmww—wwww~wwwwwww~t\)v—NNwN
lflwl

N<><€<CHMWOWOZZPWHHEQWWUOW>
l
'—"I\)Li~’l~&JOI\)'-‘b~)l\)b-"—‘OI\)O'—‘'—“l\)'—“l'\)""OO"-‘OO'—‘h
I

U}
\0

0

DJ
DJ
p—d
Kl!
p—a
LII
N
A
{It

Totals

 

1. Number of sections sampled with a D-frame aquatic dip net as described in Methods.
2. Number of sections lacking standing water that were sampled with a coring device as
described in Methods.

113

because of the variability inherent in field observations, and to maintain an acceptable

Type 11 error rate. To protect against inflated Type I error rates, I corrected for multiple

comparisons within test families (the 3 tests for each species) (Hochberg 1988).
Laboratory Experiments

Species Selection. The species used in the experiments were selected based on 4
criteria: (1) familial relationships, (2) frequency of occurrence in the wet meadows, (3)
presence or absence of an operculum (i.e. Prosobranchia or Pulmonata), and (4) primary
habitat preferences (Table 18). Representatives of four families were used to examine
whether responses were affected or constrained by phylogenetic relationships. Because I
was attempting to explain species occurrence patterns, I also selected species that were
widespread in the coastal wet meadows as well as those restricted to only a few sites in
the area (see Table 3). The presence of an operculum that covers the apertural opening
when the snail retracts into its shell was used because it may confer additional resistance
to desiccation (Gibson 1970). I also selected one species (Amnicola limosa) because it
was primarily considered a permanent water inhabitant rather than a resident of the more
temporary wet meadows (Clarke 1973, 1981). The other species were either shallow
water inhabitants or generalists across a wide variety of habitats (see references in Burch
and Jung 1992).

Experiment I: Snail Movement. Six experimental chambers were prepared by
arranging sand in plastic containers to provide a 3-tiered system (Figure 24). The sand
provided an even drainage of water through chambers to the drain spouts. A simulated
natural substrate, consisting of sieved (1 mm) organic sediment from a nearby wetland,

was placed on top of the sand to a regular depth of approximately 3 cm. The sieving of

114

“customs E08968 :95 05 5 wow: 8: 203 86on .m
b.8358 u H Java—«Egon u m dove u G .3235 n m .N
baa no.8 ~83». 2: E 859800 mo >283on A

 

 

| m d 8% 5888.5 BEBBPAE C88 ESE: SSEEV

m I m d .m mo> :oEEoQCD omEEEHmm ad 8539822 S§§m

ovm m d .m 02 58:80 bo> onEfiocmE 9mg union. mSEBAC

ofi m d .m 02 8888 Eo> 035855 039 wSomRmh wEEfiAU

cm A .D .m 02 :oEEoU 033853 9m@ $303.: SBESSE

c2 m .m o Z 5888 bo> entries—L Axmmv ogmmmzto 5‘58ng

o3 H .m 0 Z :oEEchb omEmbE enemy SEES» 883V

o2 H d d .m oZ :oEEoo bo> 823.5 Axumv 32AM 33.3

3650 ”8:20on $055.80 mo

82830 8:an 823825 55:60am 38mm 36on

225.62

 

.macoEtomxo b88022 5 com: $6on :98 mason—om com mtoEU .M: 033.

115

—-ll-

 

96 cm

 

——

 

 

 

 

 

 

 

 

 

 

 

 

Water -
A Coarse Detritus
WKAKNKKNqANAWA’A/A B
3 cm \ WKKKKNNKNNKKKKM C
l \ Organic Sediment \wmmmcmmmcmc’ .
Sand \
To common
drain valve

15cm

Figure 24. Diagram of the side view of one chamber in the Snail Movement experiment
showing the arrangement of sand, sediment and water producing 3 tiers: shallow (A),
medium (B), and deep (C). Initial water depth was adjusted so that water level was 4 cm
above detritus. The drains of all 6 chambers were connected to one common valve used
to maintain the same, constant reduction in water level. Width=42 cm.

116

the substrate ensured that no snails were present except the experimental animals. On top
of this was placed equal amounts of coarse detritus (primarily dead Carex spp. leaves
with snails removed) to provide a complex structure representative of the natural habitat
and to provide a microbially-rich biofilm food source. Water from Lake Huron was
added to each chamber to cover the shallowest tier to a depth of 4 cm. Water and
substrate in the chambers were allowed to settle for 24 h before snails were added to
them.

Each focal snail species was added to individual chambers (equal numbers on
each tier) on 4 June 1999, in numbers corresponding to the size of each species and in the
approximate densities found in the wet meadows from which they were collected (Table
18). Snails were allowed to acclimate to the experimental chambers for one week.
Starting 11 June 1999, the water was slowly and continuously drained from the chambers
at a constant rate resulting in a lowering of water levels by approximately 5 cm/wk. The
experimental chambers were maintained outdoors in semi-shade to provide the natural
die] temperature fluctuations (max 21 .9i4.1 C, min 12.5i3.7 C, meanil SD.) and
light:dark photoperiod (~1529). By day 24, the experimental chambers had been
completely drained, and the experiment was concluded.

The detritus in each tier was carefully searched and the number of snails in each
tier was recorded and all snails were removed by hand or by sieving the substrate. The
chi-square goodness of fit test was used to test whether the final distribution of snails was
different from random (i.e. different than the original, equal numbers of snails in each

tier) using an a priori a < 0.05.

117

Experiment 11: Desiccation Resistance. The desiccation resistance of seven
species of snails (unfortunately enough Gyraulus deflectus adults were not able to be
collected at the beginning of the experiment) was determined by exposing them to
extended lengths of time without water. I used a 7x3 factorial design (N=3), sampling
(without replacement) the seven species at three different time periods (2, 7 and 12 wks).
Each experimental chamber was a 500 ml cylindrical container (12 cm height) filled with
5 cm sieved (1 mm) organic sediment covered by 5 cm of water. Only 1 species was
added to a chamber so a total of 63 chambers was required (7 species x 3 wks x 3
replicates). Snails were added to the chambers as follows [species and (#/container)]: P.
gyrina (5), A. elongata (8), G. parvus (12), P. armigera (8), P. trivolvis (2), B.
tentaculata (5), and Alimosa (12). These numbers were based on snail size (for example,
two P. trivolvis at 20mm each versus 12 G. parvus at 4 mm each) to maximize sample
size without overcrowding the chambers. The snails were acclimated to the chambers for
3 days after which small pin holes were drilled in the bottom of each chamber to drain the
water slowly, allowing the snails a short time to react (possibly bury themselves, over-
hydrate, etc.) before experiencing the dry conditions. Thirty-six hours after the draining
began, all containers were void of standing water. All snails were assumed to still be
alive at this point in time (week 0). Experimental containers were maintained at room
temperature and humidity.

Survival was measured afier 2, 7 and 12 weeks by removing the snails from the
experimental containers and placing them in aquaria filled with water. Most surviving

snails began moving in <1 hr. Snails were assumed to be dead if they did not move by

118

the end of 24 hr; no additional snails began moving after more than 24 hr. SYSTAT
(SPSS 1999) statistical software was used to perform a two-way AN OVA to test for the
main effects of species and time and their interaction (wk 0 was not included in analyses
since the snails were all assumed to be alive, but were not actually sampled). Data were
arcsine transformed prior to analyses in order to meet normality assumptions (Zar 1999).

Tukey HSD tests were used for post hoc analyses of significant effects.

RESULTS

A total of 35,537 individuals of 19 species of aquatic snails were collected in dip
net samples 1997 and 1998 (Table 2). Physa gyrina and Gyraulus parvus were the most
common species (frequency of occurrence) in both 1997 and 1998 along with G. deflectus
in 1998. Likewise, P. gyrina and G. parvus were the most abundant snails in 1997
whereas in 1998, G. parvus and G. deflectus were most abundant (Table 3).

Physa gyrina and F ossaria parva declined significantly from 1997 to 1998
(paired t=2.28, df=29, P<0.05 and paired t=2.26, df=29, P<0.05, respectively).
Planorbula armigera and G. deflectus both increased in abundance from 1997 to 1998
(paired t=2.92, df=29, P<0.01 and paired t=3.71, df=29, P<0.01). Species richness per
section was similar in both years, ranging from 3-8 in 1997 and 3-11 in 1998. However,
abundance of all snails collected per section varied dramatically (Table 3).

The dominant species in these coastal wet meadows, based on frequency of
occurrence and abundance, were P. gyrina, G. parvus, P. armigera, F. parva, G.
deflectus, and Planorbella trivolvis. The remaining species (see Table 3) were

encountered infrequently and in low abundance in the wet meadows.

119

Multiple Regression Analyses
Mean water levels dropped approximately 15.8 cm from 1997 to 1998 resulting in
26 sections that were sampled in 1997 but were dry or too shallow to sample with D-
frame dip nets in 1998 (Table 19). Thus, in 1998 only 33 sections were analyzed,
compared to 59 in 1997.
Because the six variables all were based on the water level variation in the wet

meadows, they were interrelated (Figures 25 and 26). The majority of pairs had an r2 of

less than 50%, and many were influenced by extreme values. The most highly correlated
were mean and maximum depth, mean and minimum depth, and maximum and range of
depth. Maximum and range of depth were especially correlated in 1998 because the low
water levels caused many section to have dry areas (minimum depth = O), which made the
maximum depth equal to the range of depth. When these variables were included in the
regression models, models including related variables were also examined to ensure that
the best predictor was being entered.

Regression Models. Species richness, total abundance, and a total of 11 species
met the criteria for inclusion in the regression modeling (see pages 108-109). At least
one water level variable was a significant predictor in 7 of 11 species as well as richness
and total abundance during at least one year (Table 20). The data were variable (see
Figures 27 and 28 for examples of the variability), and the variance explained by the
models (R2) ranged from 6 to 32%.

Mean depth was the most commonly selected variable in the models. As mean

depth increased, the abundance of F. parva and G. deflectus increased, as well as species

120

Table 19. Mean, standard deviation, and range of the water level variables and snail
species richness and abundance variables. N=59 in 1997, N=33 in 1998.

 

 

 

 

Mean :8 SD. Range for all sections
Variable‘ 1997 1998 1997 1998
Depth 28.9 a 13.1 13.1 a 7.7 5.1 — 57.8 0.0 — 26.1
Min 10.4 a 12.1 1.2 a: 2.8 0.0 — 40.0 0.0 — 9.0
Max 52.1 a 27.1 38.3 a 28.2 11.0 — 151.0 10.0 — 120.0
Range 41.6 a 24.1 37.1 a 28.1 11.0 — 126.0 9.0 — 120.0
sCV 48.7 a 36.2 106.4 a 72.5 8.4 — 236.0 27.1 — 292.6
tCV 31.6 a 26.6 21.9 a 20.0 5.3 — 125.4 4.1 - 94.2
Richness 5.4 a 1.2 5.9 a 1.5 3 — 8 3 — 11
Abundance 345.9 a 260.4 472.5 a 341.9 5 — 1107 19 — 1257

 

1. Depth = mean depth (cm), Min = minimum depth (cm), Max = maximum water depth
(cm), Range = range of water depth (cm), sCV = spatial coefficient of variation, tCV =
temporal coefficient of variation, Richness = number of snail species, Abundance = total
abundance/section of snails collected.

121

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

E
8
E
r, nr:
32’ .
é :fl [MED
5 , g . .~ .
'....... 21 ..
g 1,, ;..
ft" :OQn; .. .l .f‘ lTLn n

 

 

 

 

 

 

 

 

 

 

 

 

DEPTH MIN MAX RANGE SCV TCV

Figure 25. Scatterplot matrix of the 6 water level explanatory variables used in the
regression analyses for 1997. Ellipses are 68% confidence intervals (:1 SD. around x
and y axes), N=59. Histograms are shown on the diagonal for each variable. Values in
each graph are r2 values that includes all points, but note the presence of extreme values
that would influence this value in many variable pairs. Axes labels are explained in the
footnote for Table 19.

122

 

DEPTH

 

MIN

 

 

 

 

MAX
5;...
3.:
.: .
:1
El

 

 

 

n ITI' -
.00 . 0.00

2” ° . .
.8.\.“.’.t 053°
. 0 0 .

0.26 0.01 .07 ° . 0.07 0.19

 

RANGE
a. . '
‘i‘é .

SCV

 

 

 

 

 

 

 

 

 

TCV

. "Erie.

DEPTH MIN MAX

 

 

 

 

 

 

 

Figure 26. Scatterplot matrix of the 6 water level explanatory variables used in the
regression analyses for 1998. Ellipses are 68% confidence intervals (:1 SD. around x
and y axes), N=33. Histograms are shown on the diagonal for each variable. Values in
each graph are r2 values that includes all points, but note the presence of extreme values
that would influence this value in many variable pairs. Axes labels are explained in the
footnote for Table 19.

123

Table 20. Results of stepwise multiple regression on species richness and abundance, and
presence/absence variables of snails in northern Lake Huron coastal wet meadows.

 

Variable Predictors in model R2
Richness

1997 (+) Mean depth 009‘

1998 -2 -
Total Abundance

1997 — _

1998 (16g)3 (—) Spatial cv 0.32

Species Abundance

F ossaria parva

1997 (+) Range of depth, (+) Temporal CV 0.16

1998 (log) (+) Mean depth, (+) Temporal CV 0.20
Gyraulus deflectus

1997 (log) (+) Mean depth, (+) Spatial CV, 0.22

(-) Range of depth

1998 (+) Mean depth 0.20
Gyraulus parvus

1997 — —

1998 — —
Planorbula armigera

1997 — —

1998 — —
Physa gyrina

1997 (log) (-) Mean depth, (+) Minimum depth 0.22

1998 — —

Presence/A bsence

Aplexa elongata

1997 (+) Temporal CV 0.09
Amnicola limosa

1998 (+) Minimum depth 0.11

 

124

Table 20. Continued.

 

Variable Predictors in model

R2

 

F ossaria obrussa
1997 (+) Maximum depth, (+) Temporal CV

F errissia parallela
1 998 —

Planorbella trivolvis
1997 (+) Minimum depth
1998 (-) Spatial CV

Stagnicola elodes
1 998 _

0.21

0.06
0.16

 

1. Values are the coefficient of determination for richness and abundance variables and

McFadden’s p2 for presence/absence variables.

2. A ‘—’ indicates there was no significant predictors in the model (P>0.10).
3. Indicates that variables were log transformed to meet regression assumptions.

125

richness (Figures 27 and 28). Physa gyrina decreased in response to increasing mean
depth (Table 21).

Minimum depth, spatial CV and temporal CV were significant predictors for 3
species each (Table 21). As minimum depth increased, P. gyrina, A. limosa, and P.
trivolvis became more common. Gyraulua deflectus increased in abundance in sections
with higher spatial CV whereas P. trivolvis was less common, and total abundance of all
species was lower. F ossaria parva and F. obrussa as well as A. elongata were more
common in sections with high values for temporal CV (Table 21).

F ossaria obrussa was the only species positively associated with maximum depth
and no species were negatively associated. F ossaria obrussa increased as the range of
depth increased, whereas G. deflectus was negatively associated with range of depth
(Table 21).

Annual Comparisons

A subset of 44 sections was analyzed for additional annual comparisons of the
snail community. The omitted 15 sections were primarily dry sections for which cores
were not accessible. Fifteen of these 44 sections were dry in 1998 and 24 of the 44
sections had significantly lower water levels in 1998 (paired t=3.24, df=23, P<0.01) with
a decline in these sections of approximately 27 cm (Table 22). The depths in 5 sections
were found to remain constant from 1997 to 1998 (paired F047, df=4, P=0.665) (Table
22).

Nineteen species of aquatic snails were collected from the flooded wet meadows
in 1997 and 1998 (see Table 2), and 9 species were collected in cores from dry sections in

1998 (Table 23). The 9 species collected in the dry sections were the most frequently

126

 

Species Richness

 

 

0 IfirllIIII'ITIIIIIVIIIIUI1IIII|

0 10 20 30 40 50 60
Mean Depth (cm)

Figure 27. Species richness (S) — Depth curve for snails in coastal wet meadows of the
northern Lake Huron shoreline during 1997. N = 59. The line represents the significant

(P<0.05) linear model, 3 = 0.027(Depth) + 4.642, R2 = 0.09.

127

400.
. O

Abundance

 

 

 

0 5 10 ' I 15 ' ' 20 25 30
Mean Depth (cm)

Figure 28. Abundance - Depth curve for Gyraulus deflectus in coastal wet meadows of
the northern Lake Huron shoreline during 1998. N = 33. The line represents the
significant (P<0.05) linear model, Abundance = 6.240(Depth) + 64.244, R2 = 0.20.

128

Table 21. Variables exhibiting negative and positive relationships to the six predictors

used in the regression modeling.

 

 

Negative (-) Predictor Positive (+)
Physa gyrina (1997) Mean depth F ossaria parva (1998)
Gyraulus deflectus (1997, 1998)
Richness (1997)
—- Minimum depth Physa gyrina (1997)

— Maximum depth

Gyraulus deflectus (1997) Range of depth
Planorbella trivolvis (1998) Spatial CV
Total Abundance (1998)

-— Temporal CV

Amnicola limosa (1998)l
Planorbella trivolvis (1997)

F ossaria obrussa (1 997)1

F ossaria parva (1 997)

Gyraulus deflectus (1997)

Fossaria parva (1997, 1998)
Aplexa elongata (1997)1
F ossaria obrussa (1997)l

 

1. Only examined in one year.

129

Table 22. Mean depth, range and standard deviation for three classifications of coastal
wet meadow sections in 1997 and 1998. N = number of sections.

 

 

 

 

Mean :1: SD. Range
Class N 1997 1998 1997 1998
Dry 15 21.9 a 8.3 —‘ 10.2 — 37.4 ——
Lower 24 40.2 a 9.8 13.6 :1: 6.9 14.4 — 57.8 1.1 — 26.1
Same 5 10.3 a 6.8 9.6 a 6.8 1.4 — 17.1 1.5 — 17.1

 

1. Sections did not have standing water.

130

Table 23. List of species collected in cores from dry sections in 1998 and their mean
abundance/section when flooded in 1997.

 

 

Species 1997 Abundance
Physa gyrina 130.6
Aplexa elongata 5.0
Gyraulus deflectus 40.3
Gyraulus parvus 90.7
Planorbula armigera 60.3
Planorbella trivolvis 4.2
Planorbella campanulata 0.7
F ossaria obrussa 2.7

F ossaria parva 9.6

 

131

occurring species in 1997, and were used in the remaining analyses. Sample sizes for the
other species were too small to be reliably tested.

‘Dry’ Sections. The frequency of occurrence of live snails in Dry sections was
significantly less in 1998 than it had been in 1997 only for Physa gyrina (Figure 29). All
other species showed no significant change (P>0.10 corrected) in frequency from 1997 to
1998 (Figure 29, Table 24).

‘Lower’ Sections. Planorbula armigera and Gyraulus deflectus both increased in
abundance in wet meadow sections where water levels declined from 1997 to 1998, but
did not dry out, whereas F ossaria parva declined (Figure 30 and Table 24).

‘Same ’ Sections. Physa gyrina was the only species to show a response in wet
meadows that remained constant in depth from 1997 to 1998; it decreased in abundance
(Figure 30 and Table 24). No other species showed significant differences between years
(Figure 30 and Table 24).

Laboratory Experiments

Experiment I: Snail Movement. Snail movement patterns of individual species
varied widely in response to the reduction in water levels. Planorbula armigera and
Planorbella trivolvis were found more often in the shallow tier, Physa gyrina was found
significantly more often in the deep tier, and the remaining species showed random
distributions among the tiers (Table 25). The percentage of snails that were completely
buried at the end of the experiment ranged from 2.5% of Gyraulus deflectus to 43.3% of
P. trivolvis (Table 25).

Experiment 11: Desiccation Resistance. Some individuals of all species tested,

except for Amnicola limosa, were able to survive for the 12 week length of the

132

 

.mcouoom omofi 5 950m 8: 83 83256858 3:38:63 .m
.waa 8 33 Set 09:23 mo .8585 .—

 

 

 

 

 

888 8.8- cs... 88. 3 38.8 stem outback
8 _ .o X. 3- 89° :8 3 83 8458 sense

a- N1 s _ 2 8. 7 3 88 863888 88883
:8 8. 3- :8 8. _- 3 83 $68ch 883..
88 8.8- 88 8.8- 3 M88 .823 3:38
888 8.8 _8.=v 3m 3 83 88:8 38:6
83 8.8 83 N3- 3 32 36.38 6:88:62
83 8.8 8.8 wow 35 83 E888 5885
88... SN. mm 3 .o 8. _- _$ 38.? been 63:

o:_m>um N Case—m3 o=~w>lm N Genoa—$5 03—57; 5on “onmmm
25m .8304 ha
820 3882 63

 

.33“ 23:2: Sm :ouooboo Home Edva 8 Ewe—mama 2a mos—STA 2.5 832038 83

mo momma—o 025 E moi van 32 50253 oocomnfloocomua 28 85.93% :85. 9:89:00 mama—«5 1833830 338m .vm 2an

133

.Aofiovnc 989m 5253 meanest? “:85:me 88:8: 2.: .828 3 @2958 53 moi cam
30: fit 3 338mm J25 32 E 880E v.83 manna wE>= :03? E Goa E S 203 35 2 mo 88 28:03. «o “3:52 .am 233m

.86me

  

 

 

 

 

 

 

 

 

 

 

 

 

a.
y 0%.
% % @ ox,
9 8o. em. a? O
C _I_ -

I...

I la

8

1 .D

- w

Im u

I O

1 A

1 mm

1 m.

1 9

D

88! H m.
82D .. M

.. I?

134

P. gyrina

 

 

‘2—
3
31::—

 

G. deflectus

5

P. campanulata

gm

Meadow

LowarSame
Class

100-

 

. P. armigera

 

roa-

 

('3
v v v

l

Lower

, G. parvus

 

1997

DryLoworSm

F. obrussa

 

 

El 1998

1%..

> P. trivolvis

DtyLowar

 

I'D

1oq-
» A. elongata

50 [hi]
Sm

DtyLowor

 

1'3

F. parva

Figure 30. Mean abundance of snails from three classes of coastal wet meadows in 1997
and 1998. ‘Dry’ wet meadows were flooded in 1997 and dry in 1998 (results not shown
here because of different sampling methods), N=15; ‘Lower’ wet meadows were flooded
in both years, but shallower in 1998, N=24; and ‘Same’ wet meadows were the same
depth in both 1997 and 1998, N=5. Error bars are i 1 S.E.M. ‘*’ indicates a significant
different between years (P<O. 10). Note differences in scaling of the vertical axes.

135

Table 25. The number of snails in each tier at the end of the Snail Movement experiment.
)8 values are from the chi-square goodness of fit test, DF=2. The total number of each
species is also the number at the beginning of the experiment.

 

 

 

Species Total Shallow MTelgium Deep x2 P % Buried
P. gyrina 120 9 34 77 59.15 <0.001 27.5
A. elongata 120 36 48 36 2.40 0.301 10.8
P. armigera 120 68 33 19 31.85 <0.001 35.8
P. trivolvis 3O 25 3 2 33.80 <0.001 43.3
G. deflectus 120 36 47 37 1.85 0.397 2.5
G. parvus 240 83 64 93 5.43 0.066 12.9

 

136

experiment, and this percentage ranged from 3 to 93% (Figure 31). There was a
significant species effect (Table 26) and post hoc tests revealed three significantly
different survival patterns. (1) Aplexa elongata and Gyraulus parvus had a large decrease
in survivorship at 2 weeks and then a slow decline through week 12, whereas (2) Physa
gyrina, Planorbula armigera, Planorbella trivolvis, and Bithynia tentaculata maintained
a relatively high survival rate throughout the experiment; (3) Amnicola limosa did not
survive for even 2 weeks (Figure 31). There was also a significant time effect (Table 26)
and post hoc tests indicated an overall significant decline in survivorship between weeks
2 and 12.

Individuals in the desiccation experiment were not identified as buried or visible
on the surface when determining if they were alive at the end of each time period.
However, a comparison of the percentage surviving at week 12 (Figure 31) in the
desiccation experiment with the percentage of snails buried (Table 25) in the snail
movement experiment shows a positive correlation (Figure 32). Species with higher
burial rates in Experiment I tended to be the same species that exhibited better survival in
Experiment 11.

DISCUSSION

With an average decline of >15 cm in 1 year, extensive changes occurred in the
coastal wet meadows of northern Lake Huron from 1997 to 1998. Significant results in
all 3 approaches to examining water level effects on snails in coastal wet meadows
suggested water levels were important. Here, I first discuss the water level fluctuations
used in the regression analyses and annual comparison, and then examine the proximate

responses of the snails to water level declines shown in the laboratory experiments.

137

 

.Emm _ H
08 33 8me .God v m .Qmm 883.5 “:08”.wa raga—mama 08 0:: 5888 a .3 88288 8: mafia: momooam .o x83 “a 033 on
9 @2558 203 £85 :8 85 28:8: gnaw some 5 3:: 3:36 2:. .mmmcm m0 860% N. go 823 8233 838333 ._m onE

 

 

 

 

 

 

 

 

 

 

 

 

 

 

v.00; xfl0>> x00;
Npiow n b O V W O Nw OP 0 O v N O NF or o O v N O
1 u . t . q A O . O . . 1 3 3 11 O
» 8
ON ON ..ON
m. ov 3. cw .9. Mn.
.... 8 ..... 8 u... .8 W
..uOO .... OO LOO
OOP CO? “00—.
meE. .< 338 .0 Emacsm .<
«.025 #003 x0; x8;
NF O.. O 0 v N 0 NF Ow O O v N O N.. or O 0 v N O Nw Or a O v N O
. 1 q . . . O . 3 q q q - O a . q - 4 . O . . q q 7 J O
k A
18 ...ON ...ON ON
4 l A
row nov 40? O? ”my
row row row on W
.4 A A
row row ..OO on
> ._ ...... . 4 +1/
of cocoa.
..... IOOP bOO-V ...OOF .3... OF
Egaoficfi .m w_>_o>E .n. 969.com .n. 2:3 .n.

138

Table 26. Two-way ANOVA of snail survival in the desiccation resistance experiment.

 

 

Source of Variation ss df ms F P
Species 20.677 6 3 .446 70.515 <0.001
Time 0.353 2 0.176 3.609 0.036
Species * Time 0.812 12 0.068 1.385 0.211
Error 2.053 42 0.049

 

139

100—

   

 

 

80—
O)
E 60?
a _
3
”3, 40—
o\
20—
0 ' 1 L J . 1 r 1 1 I
0 20 40 60 80 100
°/o Buried

Figure 32. Correlation of the percentage of snails buried in experiment I with the
percentage surviving in experiment II (r = 0.91, N=5, P = 0.03).

140

Finally, I combine the results of all 3 approaches in examining responses of individual
species to water level fluctuations in the coastal wet meadows and the overall effect it has
on the snail community.
Water Level Fluctuations

I used 6 explanatory water level variables in the regression analyses. Mean,
minimum, and maximum depths all related directly to the depth of water in the wet
meadow although each represented a slightly different aspect of the habitat. Minimum
depth reflected the shallowest parts of the wet meadow. Sections with shallow minimum
depths were those that were likely to dry out completely when Lake Huron water levels
were low (e.g. in 1998 - classified as ‘dry’ in the annual comparison), but would be wet
during high water years (e. g. 1997). In 1998, these dry sections were moist or saturated
(presumably from rainfall and/or groundwater inputs), and afier periods of heavy rain,
some depressions among the hummocks contained very shallow puddles. These sections
were probably similar conceptually to traditional temporary ponds in general models
(Lodge et al. 1987, Wellbom et al. 1996). Sections with deeper minimum depths were
likely to remain permanently flooded in all but the lowest water levels, and were
significantly shallower in 1998 (classified as ‘lower’ in the annual comparison). The
rapid decline in water levels observed during this study was likely too rapid for major
changes to occur in the plant community, except for some opportunistic annuals (Keddy
and Reznicek 1985). Thus, the changes observed in the wet meadow snail community
were probably direct responses to the drawdown event, rather than indirect effects from
plant community shifts.

The significance of maximum depth was more difficult to assess. In this study,

141

deeper maximum depths represented sections that were more lakeward and bordered the
more sparsely vegetated, emergent marsh (Mine and Albert 1998). Mean depth was
conceptually a combination of minimum and maximum depths and reflected the overall
water level in the section.

Range of depth and spatial CV were both measures of spatial variability in water
levels within a section. Range of depth was a good indicator of the extremes in depths
within a section but gave little indication of the evenness of the depth and could be
influenced by a small number of extreme depths. Spatial CV incorporated the variation
in depth relative to the mean and was less influenced by extreme values. Low spatial
CVs indicated a more uniform depth throughout the section.

Temporal CV was a measure of the variation in water levels over time. Temporal
CVs in this study were not indicators of seasonal and multi-annual fluctuations in water
levels that occur in Great Lakes coastal wetlands (Bedford 1992, Keough et al. 1999), but
rather represented the magnitude of water fluctuations in the 1-month immediately prior
to sampling. Low temporal CVs indicated a tighter connection to the hydrology of Lake
Huron, and higher CVs indicated sections that were less connected to Lake Huron, which
includes those isolated by naturally formed dune ridges as well as anthropogenic
impoundment (e.g. road building). The water levels fluctuated more over time because
these semi—isolated wetlands were more strongly influenced by rainfall pattems/droughts.

In comparison to the short-term water level changes above, in 5 sections, water
levels did not change significantly from 1997 to 1998 (classified as ‘same’ in the annual
comparison). Two of these sections occurred in wet meadows partially isolated

hydrologically from Lake Huron by small dune ridges, and were probably influenced by

142

rainfall. The other 3 sections were in a wet meadow separated from Lake Huron by a
road, but connected via a culvert and with a stream running through it.
Experiment 1: Movement Patterns

The behavioral responses of freshwater snails to declining water levels have been
examined in several studies of temporary pond habitats. In these populations, the snails
must be able to resist desiccation to survive from one season to the next. The movement
patterns that were often exhibited have been interpreted as an adaptive mechanism to
increase the probability of successfully resisting desiccation by finding more suitable
microhabitats, either in shallower or in deeper habitats. Jokinen (1978) and Brown
(1979) found that adult Stagnicola elodes did not react to changing water levels, but that
juveniles did, by moving out of the drying pond to the surrounding mud (i.e. toward
shallow areas). Betterton et a1. (1988) suggested that the opposite pattern (i.e. moving
toward deeper water) may also be advantageous in two other species, Bulinus rohlfsi and
B. globosus. They were both observed to move to the bottom of the drying ponds before
aestivation, and there was no apparent difference in the movement of adults and
juveniles, although older snails did survive better after reviving (Betterton et a1. 1988).

Overall, my results mirror these studies; some species moved to shallow areas and
others followed the receding water (Table 25). In environments with fluctuating water
levels, moving away from the receding water to shallow, dry areas as Planorbella
trivolvis and Planorbula armigera did in the current study may have several benefits.
First, premature revival from aestivation, e.g. after a single, early rainfall (Betterton et al.
1988), may be prevented. A second possible advantage is escaping predation. In

temporary ponds, the falling water levels would decrease the suitable habitat available,

143

and confine the snails into small areas of high densities, making them vulnerable to both
aquatic and terrestrial predators. Supporting this, both P. trivolvis and P. armigera
exhibited the greatest percentage of burying behavior (Table 25), a response that may
decrease predation rates.

These explanations for moving to shallow areas may be important for temporary
ponds, but do not entirely explain the behavior in coastal wet meadows where a
permanent refuge is available if the snails would follow the receding water. It is possible
that these behaviors are advantageous, even in coastal wet meadows, because of the
heterogeneity within the wetlands. Isolated pools may act like non-coastal temporary
ponds even though at a larger scale, the wetland is contiguous with a permanent water
environment. Thus, the behaviors that are advantageous to snail populations in
temporary ponds may also be advantageous to those in coastal wet meadows.

Physa gyrina was the only species to move with the water as it declined and most
were found in the deepest tier (Table 25). Clampitt (1974) found a maximum dispersal
rate of approximately 7 m in 24 h for the closely related congener P. integra. This rate by
itself would be sufficient to move to refuge in the coastal wetlands. In addition, as the
water flows out of the wetlands during frequent seiches, snails may float with the current,
either attached to the surface film or buoyed by air bubbles in their shells (BEK: pers.
obs.). The combination of these behaviors suggested that directed movement following
the water is a viable, adaptive response by P. gyrina to falling water levels.

The remaining species tested (Aplexa elongata, Gyraulus parvus, Gyraulus
deflectus) did not show any significant direction to their movement, indicating that they

may not respond behaviorally to declining water levels. It was possible that rather than

144

moving in preparation for aestivation (as done by P. trivolvis and P. armigera), they
were preparing physiologically by increasing the hydration of their tissues (Arad 1990).
They apparently did not use burying as a behavior to increase survival after aestivation as
they showed the lowest burial rate (Table 25).

Experiment [1: Desiccation Resistance

Freshwater snails from non-coastal temporary habitats were frequently observed
to resist desiccation for various lengths of time until more favorable conditions retumed
(Jokinen 1978, Brown 1979, Betterton et al. 1988). In the present study, six of seven
species were able to survive at least 12 weeks of desiccation.

The four species that showed the highest survival rate (Figure 31) were a diverse
group phylogenetically. Physa gyrina, Planorbula armigera, and Planorbella trivolvis
were pulmonates from two different families whereas Bithynia tentaculata was a
prosobranch. Desiccation resistance in B. tentaculata was probably enhanced by the
presence of the thick, calcareous operculum characteristic of the family, and a tendency to
bury in the substrate. The presence of an operculum did not guarantee resistance, as the
only other operculate snail in the study, Amnicola limosa, did not even survive two weeks
(Figure 31).

It was not clear why the three pulmonates should show such high survival rates.

It may be a function of physiological tolerance of high hemolymph osmolality resulting
from water loss, or of other behavioral adaptations for reducing water loss. Adaptations
to resist desiccation may include retaining water in the mantle cavity (Blinn 1964),

creating a soil and mucus plug covering the apertural opening (Parashar and Rao 1998),

or burying in the sediment or other moist habitat (Jokinen 1978, Thomas and McClintock

145

1996). They were observed to bury quite frequently in the present study (Table 25).

Gallo et al. (1984) found that Planorbella trivolvis survived <5 (I at low relative
humidity (12-l4%) and <40 d at high humidity (96-98%). Clampitt (1970) found 51% of
Physa gyrina surviving after 12 days of desiccation (humidity not indicated). In contrast,
I found a much longer survival time (at least 84 d). The difference in results may be
because Gallo et al. (1984) maintained the snails in moist paper towel (they were
simulating shipping conditions) and Clampitt (1970) in dry dishes, rather than in a
substrate in which the snails could burrow, as was the case in the present study. This
emphasizes the importance of behavioral adaptations to increase desiccation resistance,
and should be taken into consideration when making comparisons of desiccation
resistance among studies with different methods. Additionally, for some species, there
was limited evidence that populations from temporary habitats were able to survive
longer periods of desiccation than populations from permanent habitats (Olivier 1956).

Some individuals of Aplexa elongata and Gyraulus parvus were able to survive
12 weeks of desiccation but with a much lower rate (Figure 31). Why there was such a
difference between these species (both pulmonates) and the other three pulmonates was
unclear, but it may be due to behavioral adaptations (i.e. reduced burying rates, Table 25)
rather than major physiological differences.

Snail Responses

Species richness was significantly related to increasing mean depths in 1997 but
there were no predictors in 1998 (Table 20). The deeper water levels recorded in 1997
allowed snails that were normally associated with the deeper, more permanent habitat of

the emergent marsh (i.e. bulrush marsh) to occur in the wet meadow regions. Thus, in

146

1998 when the water levels receded, and these species were not found in as many wet
meadows, on average, there was a decline in species richness in relationship to depth.
Total abundance of snails was related to decreasing water level spatial variability
in the wet meadows in 1998 (Table 20). Sections with low spatial variation (i.e. more
uniform depth) may have had higher abundances because of the more stable water levels.

Physa gyrina. The negative association in 1997 with mean depth, and a positive

 

association with minimum depth, suggested that P. gyrina was most abundant in
shallower sections, but only those that were likely to remain flooded. Thus, in the annual
comparison, P. gyrina was found less frequently in the Dry sections (Table 20). This
pattern was consistent with behavioral and physiological results that showed high
desiccation resistance and active movement to deeper water (Figure 31, Table 25). In
some sections, the snails may have been able to move out of shallow sections with the
receding water because of sparser vegetation or shorter distances to permanent water
(Clampitt [1974] observed a very similar congener to move up to 7 m in 24 hr). This
directed movement would result in relatively few individuals aestivating in the shallow
sections (<8% remained in the dry sections during the experiment, Table 25). Because of
the decreased density, few snails would be collected and a significant decline (Figure 29)
would be expected. Instead, they could be found in permanently flooded, shallow
sections as in this study. They were apparently not as common in deeper water, perhaps
because of increased predation risk from fish (Turner 1996). Physa gyrina were thin-
shelled which made them especially vulnerable to fish predation (Sadzikowski and
Wallace 1976, Mittelbach 1984).

Abundance of P. gyrina was not significantly predicted by any of the water level

147

 

variables in 1998. Because of the lower water levels in 1998 (Table 19), the sampled
sections were those that were deeper in 1997. Since P. gyrina was found in shallower
sections in 1997, it was not surprising that they were less common in 1998. Life history
characteristics of P. gyrina suggest that they have a high reproductive rate (Brown 1979)
that may allow them to recover after one or more generations, and become the most
abundant species once again, as it was in 1997.

I found P. gyrina to be the most common and abundant species overall (see Table
2) which was consistent with other studies which have found that where P. gyrina occurs,
it was often the most abundant (Clarke 1981). For example, McKillop (1985) found it at
all sites sampled and in the highest abundances in southeastern Manitoba, Canada, and
Pip (1985, 1987) found it to be the most common species in her sites in central Canada.

Fossaria parva. F ossaria parva, in both 1997 and 1998, was significantly predicted by

 

higher temporal CV (Table 20). It was also predicted by increasing range of depth (1997)
and mean depth (1998). These results suggested that F. parva was more commonly
found in sections that were less connected to Lake Huron (positive temporal CV) and that
were more likely to remain permanently flooded (positive range and mean depths). The
behavioral response and desiccation resistance of F. parva were not studied, but these
results suggested that it probably would not move in response to changing water levels,
and it probably has a low desiccation resistance. These attributes may explain the
significant decrease in abundance from 1997 to 1998 as water levels dropped and F.
parva was not able to respond (Table 2).

Previous researchers have reported F parva to be amphibious, occurring above

the water line or at the edges of large, permanent water bodies (Baker 1928, Laursen et al.

148

 

1992). My results agreed with this; that F. parva was found in more permanent sections
of coastal wet meadows. However, I could not distinguish where in the water column the
individuals were when collected. It was possible that even in these deeper sections, they
were amphibious and occupied the water’s edge, occurring at the upper edges of

hummocks.

 

Gyraulus deflectus. In both 1997 and 1998, G. deflectus was associated with deeper 9
mean depths. In those deeper sections in 1997, it was also predicted by a narrow range of
depths and increased spatial CV (Table 20). In the behavioral experiments, G. deflectus

did not react to changing water levels (Table 25), a result consistent with a species found

 

2'"
in deeper water where fluctuating water levels would not be as important. These results

supported those of Pip (1985) who also reported that G. deflectus was found in habitats
where seasonal variation was reduced, and of Clarke (1981) who reported that it was
found in permanent habitats.

Gyraulus deflectus had significantly higher abundances per section in 1998
compared to 1997 (Table 2). This result was primarily due to the increase observed in
sections that had lower water levels in 1998 (Figure 30). This may have been a delayed
response to the high water levels in 1997. Since G. deflectus was found in deeper water,
the lower water levels in 1997 may have been conducive to high recruitment which
manifested itself in the 1998 population. Supporting this conclusion that the population
increase was a lagged response, in sections that remained the Same in 1998, abundance
remained at high levels and did not significantly change from 1997 (Figure 30).

I followed the nomenclature of Burch and Jung (1992) and counted both hirsute

(G. hirsutus) and non-hirsute forms as one species, G. deflectus. Both forms were present

149

in the sampled coastal wet meadows.

Gyraulus parvus. Gyraulus parvus did not move in response to changing water levels

 

and had a low tolerance for desiccation (Table 25, Figure 31). The abundance of G.
parvus remained the same in 1997 and 1998 (no significant difference, Table 2), there
was no difference in Lower and Same sections (Figure 30), G. parvus was common in
cores from Dry sections (Figure 29), and there were no predictors in either year (Table
20). This was also consistent with the behavioral response, but its low desiccation
resistance suggested that G. parvus should be more common in deeper, permanent water.
It was very clear that G. parvus populations were not affected by water level changes in
the coastal wet meadows. G. parvus was a very small species (3-5 mm) and may be able
to remain in shallower sections, despite the possibility of desiccation, because the
periodic pooling in shallow depressions afier rainfall was enough for these species to
survive dry periods during drawdown events (i.e. because they are not actually exposed to
a desiccating environment).

Planorbula armigera. There were no significant water level predictors of P. armigera

 

abundance in the current study. In contrast, in laboratory experiments, P. armigera were
highly desiccation resistant and moved to shallower regions in response to falling water
levels (Table 25, Figure 31). Those results suggested that water levels may be important
for this species and that it should be found in shallower, more variable sections. In the
current study, P. armigera did increase in abundance from 1997 to 1998 (Table 2), was
more abundant in Lower sections (Figure 30), and was found in cores in all Dry sections
(Figure 29). This pattern was similar to that for G. deflectus and a lagged response in

population increase from 1997 to 1998 may also be responsible for the observed changes.

150

Previous reports for the habitat of P. armigera varied widely. Pip (1985) found it
more cormnonly in rivers (>2 m deep) whereas Clarke (1981) reported it from stagnant
and heavily vegetated habitats. Coastal wet meadows, although more dynamic than
stagnant, are similar to Clarke’s description of the habitat requirements.

Planorbella trivolvis. The occurrence of P. trivolvis was explained by increasing
minimum depth in 1997 and decreasing spatial CV in 1998 (Table 20). These both
reflected deeper, more stable water levels. In contrast, the behavior experiments
suggested that perhaps P. trivolvis, was adapted to more variable, shallower environments
since they moved towards shallower water (Table 25), could resist desiccation for
extended periods of time (Figure 31), and were more likely to bury themselves (Table
25). Because P. trivolvis was more likely to remain in deeper sections of the wet
meadows, these responses may instead serve to decrease predation risk.

Planorbella trivolvis was often found on the bottom substrate, which was usually
mucky, rather than on the vegetation as were the majority of other species (Boerger 1975,
Burch and Jung 1992). In the coastal wet meadows such habitat was more common in
the deeper sections where vegetation was less dense and the bottom was less covered by
decaying vegetation (BEK: pers. obs.). Clarke (1981) reported that it was usually found
in permanent water habitats.

Aplexa elormta. Aplexa elongata occurrence was predicted by high temporal CVs

 

(Table 20) which indicated that populations of A. elongata were found in sections that
were variable over time and that may have hydrology separated from Lake Huron.
Hydrology that is less dependent on Lake Huron may show increased temporal variability

due to variability in rainfall. Although not quantified, I observed water level changes that

151

corresponded to heavy rainfall and short drought periods during regular visits to wet
meadows where A. elongata was common. Desiccation studies showed that it was not
very resistant to drying out and moved little in response to decreasing water levels (Table
25, Figure 31). Other researchers have found A. elongata in temporary habitats that dry
seasonally (Baker 1928). The positive association with temporal CV suggested it could
be found in temporary habitats, but the low desiccation resistance of the individuals in the
current study required more permanent water levels. It was possible that the coastal wet
meadows in which A. elongata were found generally had permanently standing water,
from rainfall, regardless of the water level fluctuations of Lake Huron, and it was possible
that in years with drought conditions and low water levels of Lake Huron, that these
sections would dry out completely causing populations of A. elongata to become locally
extirpated.

F ossaria obrussa. F ossaria obrussa, similar to A. elongata, was found in sections with a

 

positive temporal CV (maybe influenced by rainfall) that remained permanently flooded
(positive maximum depth) (Table 20). Previous studies suggested that F. obrussa was
found in similar habitats and locations (at water’s edge) to F. parva (Baker 1928, Laursen
et al. 1992).

Amnicola limosa. Behavioral responses were not examined for A. limosa but it was

 

found to be extremely susceptible to desiccation with no individuals surviving over 2
weeks (Figure 31). Those results suggested that A. limosa should be found in deeper,
permanently flooded wet meadows. In the current study, it was associated with
increasing minimum depths (Table 20) which supports the laboratory observations.

These observations are also consistent with other researchers that reported A. limosa was

152

a resident of permanent lakes and rivers (Clarke 1973, Pip 1985).

F errissia parallela. There were no significant water level predictors for F. parallela, the
only ancylid (freshwater limpet) found in the current study. There are few reports of
habitat preferences for F. parallela but Burch and Jung (1992) reported it from quiet
waters (shallow lakes, ponds and ditches) in northern Michigan and Clarke (1981) stated
that it was often found attached to cattail (Typha sp.) and bulrush (Scirpus sp.) stems,
which was consistent with personal observations in the coastal marshes. The lack of
response to water levels and its rarity in the wet meadows supported its primary habitat as
a deeper water species (bulrush and cattail are found in deeper water).

Stagnicola elodes. Stagnicola elodes had no significant water level predictors in the

 

current study. It was known to be very adaptable species, found in a range of habitats
from temporary woodland ponds to more permanent water bodies (Clarke 1981, Laursen
et al. 1992). The low frequency and abundance in the current study could reflect the
temporary nature of this species in the coastal wet meadows and it may be more common
in the deeper, emergent marsh. Because of its known adaptability, it was unclear why it
should not also be able to flourish in the more variable and temporary wet meadows.
Conclusions

The objective of this study was to examine the role of water level, independent of
other environmental characteristics, in structuring the snail communities in coastal wet
meadows. As suggested by their responses in laboratory experiments, the occurrence of
snails in coastal wet meadows can be partially explained by the six variables examined.
Mean depth was the most common predictor, explaining in part, four snail variables.

Variation in the water levels spatially, as well as through time, were significant predictors

153

for three snail variables each. Increasing minimum depth also explained three snail
variables. Finally, range of depth was a significant predictor for two variables and
increasing maximum depth for one snail variable. Although significant, the variables
examined in this study only explained a small amount of the variance in the snail data.
However, these results suggested that in part, the response to water levels by coastal wet
meadow snails was based on relatively simple and immediate responses made by the

snails.

154

 

CONCLUSIONS

My research has confirmed the complex nature of factors affecting the
distribution and abundance of aquatic snails, and suggested that many factors operating at
different scales (see Lodge et al. 1987) combined to determine snail community structure.
As the first comprehensive study of aquatic snails in Great Lakes coastal wet meadows, I
have identified several consistent variables that were associated with one or more snail
species, although the causal mechanisms of many remain unconfirmed. Here I synthesize
the results of research examining within-patch and surrounding landscape characteristics
and the effects of water level changes, by presenting a conceptual model describing the
identified determinants of snail community structure. It is hoped that this model will
provide a guide for further research aimed at investigating the mechanisms behind the
observed patterns.

The Model

The conceptual model (Figure 33) was based on water depth as the single most
important factor affecting snail distributions. Many of the factors in the model were
directly or indirectly related to water depth. For example, bulrush marsh was associated
with deeper water, whereas shrubs and hummocks were more common in shallower
depths. In many cases, snail community associations with water depth were an indirect
result of the water depth affecting the physical habitat, and a presumed differential
response by snail species to the physical habitat. In addition to the indirect effects, some
snails responded directly to changing water levels in laboratory experiments.

In order to describe the factors affecting the snail community structure, I

separated the apparent depth gradient into three relative categories: deep, shallow and

155

 

 

 

\ Shallow.

Semi-isolated Temporarily
Wet Meadow Flooded
Snail Community Meadow

Alkalinity Shrubs\ A

 

    
 
  

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Conductivity Hummocks
Grasses
A Sub. Veg.
Desiccation
Resistance
/ l 8
Shoreline é
Complexity 1 9
(Sinuosity 8’ Shallow 3‘
Fractal Dim.) Wet Meadow it:
Snail Community 5
.C
‘6.
d)
o
/ r-z
ID
3
Motility
Cattail /
v Bulrush

 

 

 

.... v

 

 

 

 

 

Wet Meadow
Snail Community Deep
Usually
K 1| / Flooded
‘\\ . Meadow
‘ Frsh Predation

 

 

 

Figure 33. Conceptual model of the determinants of snail community structure in a Great
Lakes coastal wet meadow complex (large box). Block arrows indicate chemical,
structural, and plant gradients within a wet meadow gradient. Solid arrows are
significant associations observed in the present study whereas dotted arrows are weaker
or implied associations.

156

semi-isolated wet meadow snail communities (Figure 33). However, it is important to
realize that these categories represent a continuum, rather than discrete changes in the
snail community. Based on my results, I assigned the snails and their relative abundance
to each wet meadow category (Table 27, Figure 34).

Deep Wet Meadow Snail Community

The deep wet meadow snail community was influenced by its proximity to the
open water of Lake Huron and the deeper emergent marsh vegetation, especially bulrush
(Scirpus spp.) (Figure 33). In all but the most extreme cases, it would remain
permanently flooded. Snails common to this area were those that were intolerant of
desiccation in experiments and in cores examined from dry sections of wet meadows
(Figure 34).

Within-patch characteristics of the deep wet meadow included lower alkalinity
and conductivity when compared to shallower depths. These values of the deep wet
meadow were expected to be closer to the values of the open water in Lake Huron
because of increased mixing, and previous studies suggested that this may affect the
invertebrate community (Cardinale et al. 1997, 1998). The deep wet meadow also
included a higher frequency of bulrush relative to shallower depths. Ferrissia parallela
and Stagnicola elodes showed a significant association with the bulrush, most likely as a
preferred substrate for attachment (Clarke 1981). It is possible that decaying bulrush
enriched the detrital base and could have affected other snail species (Brown 1982).

Bulrush and cattail marshes adjacent to and nearby the wet meadows were
associated with the snail community structure (Figure 33). Testing the mechanism for

this association was beyond the scope of the present study, but three hypothesized

157

 

 

 

00:20:08 088.00 00.: 800 30.: 00 P c N 805.805 89:08
000003.;
0300008 003 8 8088005 .30:0:m 2. e 02 . 80:88:00 850:5
08880
00080008 38:00 08 800 30.“ 00 a. a N §0§§~ anmeimim
000000.808 08800 08 800 30.: 00 H a 0 2039: 300850.
0300008 003 8 0088008. £000 ... mm 0.88.: 300.85“
58er
03052080...—
0300008 03 8 8088008. £009 0 ha . 30:83 8.3...0k
8855
00080008 0.8—.000 00.: 800 30.: 00 H ... _ 805030: 0:00!
0050850 32 8 805008 40:80:00 ... 3 ... n2. SEE eteuuek
3:80:00 a a a E.— 08300 8880K
0300008 «03 8 0088008: doom— ... m.— 3080 £00.8M8m.
000088»:
0300008 003 8 0088008. doom . NV 303030 308.58%
«300008 .03 00308-88 0:0 30:05 a ... av 8:388:00 0:030:03
0300008 .03 30:05 «a 00m 0.3095 0:003:03
0300008 08208-88 0:0 30:80 40:80:00 2. :3 c: 800 808.50 03000.83
00:80:08 38:00 00.“ 800 30.: 00 a. c _ .3800 .33an
00:80.88 0: 40:80:00 I. .3 2. 03.0 eaten 33055
0300008 003 0080800 0008 40:80:00 2: «a... .5 3%. 30380 338.6
00058007:
0300008 «03 08208480m a 2:. film 8088? 303V
8900008 .03 0000.08-80». 0:0 30:85 a 2;. 2. 83 053M 3.3%
2.0005
3808....—
..00:8000.& “8303: @000 30:25 080—00388 Z 0.080%

 

#008800 300002 «03

 

808800 80> n 3:. 0:0 £08800 n .3: 6088008. n 1: .maa— 0:0 33 8 0080:00 0:05
:0 00088 0080800 M Z 0300008 003 :0 00803800 008: 8 0.80080 3808800 :08 300008 03 .8800 00003 8000 SN 030.:

158

Semi-isolated Shallow Deep
7 3 ‘ Wet Meadow

VL‘XL 'N l/ ’ , ‘ ——>
4! ' " ' \v 7 ‘ // """"" 1997
Aplexa elongata‘ !‘ fl‘ ‘ ‘f‘ ..... -l ........ 1 998

 

 

Physa gyn‘naA Water
Planorbula armigera" G "I“ when”. Levels
Planorbella campanulata Plano rbe Ila WW7:
Gyraulus parvus‘ Stagnicola elodes
Bithynia tentaculata"Fossm1a parva' Ferrissia parallela
Fossaria obrussa Amnicola limosa
Promenetus exacuous

Figure 34. Schematic cross-section of a coastal wetland along the northern Lake Huron
shoreline showing the three wet meadow snail communities (semi-isolated, shallow and
deep) described in a conceptual model (Figure 33). Relative placement of snail species
follows Table 27; species in bold type are generalists. Species with “" are able to
tolerate drying as water levels decline (see Figures 29 and 31). Approximate water levels
are indicated for 1997 and 1998.

159

 

mechanisms may be postulated. Bulrush and cattail marshes may have provided a source
of snail species (e. g. F. parallela) that were not as commonly found on wet meadow
vegetation. The bulrush and cattail marshes may have limited the wave action from Lake
Huron, preventing thorough mixing and altering water chemistry (see Cardinale et al.
1997) and/or the wet meadow vegetation characteristics. Finally, larger fish from the
deeper emergent marsh may have selectively preyed on susceptible snail species that
were found in the deep wet meadow.

The location of the wet meadow within coves or sheltered bays (i.e. high sinuosity
or fractal dimension) decreases the hydraulic effects of Lake Huron and generally allows
for greater wetland development (Keough et al. 1999). Therefore shoreline complexity
may have affected snail community structure both directly by affecting the development
of wet meadow characteristics, and indirectly by the development of bulrush marshes and
its effects as described above.

When Lake Huron water levels decline, as was the case from 1997 to 1998, snail
motility may play a role in structuring the snail community (Figure 33). Species such as
Physa gyrina were able to move with the declining water levels from the shallower wet
meadows. Snails that may have initially been more common in shallow depths could
become more abundant in the deeper wet meadow as the lake levels decline.

Shallow Wet Meadow Snail Community

The shallow wet meadow region lies between the terrestrial environment and the
permanently flooded deep wet meadow. In some years all or part of the shallow wet
meadow may dry out from seasonal or longer-term water level declines.

Higher alkalinity and conductivity levels in the shallow wet meadows probably

160

 

resulted fiom the decreased volume of water, lower dilution of salts fi'om the
limestone/dolomite bedrock and higher evapotranspiration rates, combined with limited
mixing with Lake Huron water. Aplexa elongata and F ossaria obrussa were the only
two species closely associated with increasing alkalinity (and the correlated calcium
levels) and conductivity in 1997 and there were no associations in 1998. As Lodge et al.
(1987) points out, water chemistry was not likely to influence snail distributions within
regions. The alkalinity (calcium) levels present in the shallow wet meadow were well
above the threshold level for limiting snail presence/absence. Therefore the higher levels
in the shallow wet meadow more likely affected snail abundance rather than distribution.

There was an association between several snail species and the higher density and
frequency of grasses, hummocks, submersed vegetation and shrubs that were found in the
shallow wet meadows. Hypothesized mechanisms for these observations included the
increased diversity of specific plant-snail associations or an increased surface area for
growth of the microbially-rich biofilm (see Brown and Lodge 1993). The structurally
complex habitat coupled with the shallower water levels may also have reduced the rates
of predation from larger predators such as fish. Additionally, the structural complexity
and the resulting shading and temperature differences may have affected the biofilm
community, oxygen availability and snail growth rates.

Adjacent landscape features associated with the snail community were cattail
marshes and non-forested openings (e. g. dry, upland meadows). The mechanisms behind
the cattail marsh association were probably similar to those discussed above for the deep
wet meadows. The positive association of dry meadows with the snail community may

be affected by the buffering effects fiom nearby terrestrial systems (Karr and Schlosser

161

 

1978). The exact nature of the interface between the terrestrial systems and the shallow
wet meadows deserves more research.

Because the shallow wet meadow snail community may be exposed to short- or
long-term desiccation events caused by seasonal or annual changes in water level,
interspecific differences in motility and behavioral or physiological mechanisms for
resisting desiccation may have affected snail community structure (F igurc 34). Some
species (e. g. P. gyrina) may move to deeper water as the water levels decrease, whereas
others may have been able to resist desiccation for long periods of time (at least 12 weeks
in laboratory experiments and longer from cores taken in dry wet meadows). When
water levels rise again, the snail community will be dominated by desiccation-resistant
species (Figure 34) until others can re-colonize from adjacent habitats. Nearby inland
wetlands (forested and emergent marshes) were included in two models, possibly
reflecting their role as sources for re-colonization. Investigations of these source
communities and their spatial relationship with specific wet meadows may further help to
elucidate their role in the re-colonization of coastal wet meadows after desiccation
events.

Semi-Isolated Wet Meadow Snail Community

The semi-isolated wet meadow category includes wet meadows that were partially
separated from adjacent wetlands and Lake Huron. The separation may have been from
natural causes such as sand ridge formation in exposed, simple shorelines (Figure 33) or
fi'om anthropogenic influences such as roads (see Table 1). My results and observations
suggested that because of their decreased connectivity with Lake Huron, these wet

meadows experienced greater water level fluctuations due to rainfall and a potentially

162

increased likelihood of drying out during drought periods. However, I did not design my
study to address these mechanisms directly and specific investigations into the
hydrological link between rainfall and water levels in many of these wet meadows may
provide new insights.

The snail community in these semi-isolated wet meadows was influenced by
many of the same factors as the shallow wet meadow community (Figure 33). However,
because of their limited connectivity to the deeper wet meadow and Lake Huron, they
were probably more affected by their interface with the terrestrial environment, and the
effects of periodic drawdowns (Figure 34). The snail community was composed of snails
with high desiccation resistance or with the ability to disperse rapidly into the wet
meadows following a desiccation event.

Human Influences on Great Lakes Coastal Wet Meadows

Because of increased use and development along the northern Lake Huron
shoreline and the number of roads that were located adjacent to or through coastal
wetlands, their impact on the wetland flora and fauna was of special concern. My results
suggested that at the current time, these impacts were having limited or no effect on the
aquatic snail community. Two areas of potential concern for future investigations were
highlighted by my research: the hydrologic isolation of wet meadows by road
construction and the impacts of fiagmentation of wetlands by continued development.

Approximately half of the wet meadows examined in the present study were
situated adjacent to or were bisected by a road. Wet meadows bisected by roads may
have experienced altered hydrological regimes as they became isolated fi'om Lake Huron.

As discussed above, isolation influenced the snail community structure by increasing the

163

 

abundance of species that could tolerate desiccation and changes in the water chemistry
and vegetation. However, a direct examination of roads in the surrounding landscape
context showed no negative (or positive) associations with roads. Because of the change
in hydrology, roads have been inferred to affect many aquatic communities (see
Trombulak and F rissell 2000). Therefore continued monitoring of wet meadows adjacent
to roads is important to document any impacts that do occur.

A suite of variables that indicated human disturbance was included in a
statistically significant model in 1998 (see Chapter Two). The variables included
increasing cattail marsh, urban contexts, number of patch interfaces and patch types.
Development along the northern Lake Huron shoreline consisted largely of the building
of new homes and vacation properties. The development of these urban patches within
the more contiguous shoreline increased the fiagrnentation (F orrnan 1995) of both aquatic
and terrestrial systems (indicated by the increased number of patch interfaces). Also,
cattail marshes may increase afler human disturbances and introduction of nutrients such
as those produced by improperly located septic systems (Mitsch and Gosselink 1993).

Overall, my results indicated the importance of both the within-patch
characteristics of coastal wet meadows as well as their surrounding context. The
structure and vegetation within a wet meadow had the greatest effect on the snail
community structure with a minor role for water chemistry. Wet meadows within larger
wetland complexes that included adjacent bulrush and cattail marshes and that were
buffered on the terrestrial side by meadows also had more diverse snail communities.
Maintaining this context will be important as the northern Lake Huron shoreline

continues to be developed for human use.

164

 

LITERATURE CITED

Aho, J. 1966. Ecological basis of the distribution of littoral freshwater molluscs in the
vicinity of Tarnpere, South Finland. Annales Zoologoci Fennici 3:287-322.

Aho, J. 1978a. Freshwater snail populations and the equilibrium theory of island
biogeography. I. a case study in southern Finland. Annales Zoologoci Fennici
15: 146-154.

Aho, J. 1978b. Freshwater snail populations and the equilibrium theory of island
biogeography. 11. relative importance of chemical and spatial variables. Annales
Zoologoci F enrrici 15:155-164.

Arad, Z. 1990. Resistance to desiccation and distribution patterns in bush-dwelling land
snails. Journal of Zoology, London 221 :1 13-124.

Baker, RC. 1928. The freshwater Mollusca of Wisconsin. Wisconsin Geological and
Natural History Survey, Bulletin 70:1-507+.

Beckett, D.C., T.P. Aartila, and AC. Miller. 1992. Contrasts in density of benthic
invertebrates between macrophyte beds and open littoral patches in Eau Galle
Lake, Wisconsin. The American Midland Naturalist 127: 77-90.

Bedford, K.W. 1992. The physical effects of the Great Lakes on tributaries and
wetlands. Journal of Great Lakes Research 18:571-589.

Betterton, C., G.T. Ndifon, and RM. Tan. 1988. Schistosomiasis in Kano State, Nigeria
11. field studies on aestivation in Bulinus rohlfsi (Clessin) and B. globosus

(Morelet) and their susceptibility to local strains of Schistosoma haematobium
(Bilharz). Annals of Tropical Medicine and Parasitology 82:571-579.

Bickel, D. 1965. The role of aquatic plants and submerged structures in the ecology of a
freshwater pulmonate snail, Physa integra Hald. Sterkiana 18: 17-20.

Bishop, M.J. 1981. Quantitative studies on some living British wetland mollusc faunas.
Biological Journal of the Linnean Society 15:299-326.

Blinn, W.C. 1964. Water in the mantle cavity of land snails. Physiological Zoology
37:329-337.

Boerger, H. 1975. Movement and burrowing of Helisoma trivolvis (Say) (Gastropoda,
Planorbidae) in a small pond. Canadian Journal of Zoology 53:456-464.

Borcard, D., P. Legendre, and P. Drapeau. 1992. Partialling out the spatial component of
ecological variation. Ecology 73: 1045-1055.

165

 

Bronmark, C. 1985. Freshwater snail diversity: effects of pond area, habitat
heterogeneity and isolation. Oecologia 67:127-131.

Bronmark, C. and B. Malmqvist. 1986. Interactions between the leech Glossiphonia
complanata and its gastropod prey. Oecologia 69:268-276.

Brown, K.M. 1979. The adaptive demography of four freshwater pulmonate snails.
Evolution 33:417-432.

Brown, K.M. 1982. Resource overlap and competition in pond snails: an experimental
analysis. Ecology 63:412-422.

Brown, K.M. and D. DeVries. 1985. Predation and the distribution and abundance of a
pond snail. Oecologia 66:93-99.

Brown, K.M. and D.M. Lodge. 1993. Gastropod abundance in vegetated habitats: the
importance of specifying null models. Limnology and Oceanography 38:217-
225.

Browne, R.A. 1981. Lakes as islands: biogeographic distribution, turnover rates, and
species composition in the lakes of central New York. Journal of Biogeography
8:75-83.

Burch, J .B and Y. Jung. 1992. Freshwater snails of the University of Michigan
Biological Station area. Walkerana 6:1-218.

Burton, TM. 1985. The effects of water level fluctuations on Great Lakes coastal
marshes. Pages 3-13 in Prince, H.H. and FM. D'Itri, eds., Coastal wetlands.
Lewis Publishers, Inc., Chelsea, Michigan.

Burton, T.M., D.G. Uzarski, J .P. Gathman, J .A. Genet, B.E. Keas, and CA. Stricker.
1999. Development of a preliminary invertebrate index of biotic integrity for
Lake Huron coastal wetlands. Wetlands 19:869-882.

Calow, P. 1970. Studies on the natural diet of Lymnaea pereger obtuse (Kobelt) and its
possible ecological implications. Proceedings of the Malacological Society of
London 39:203-215.

Calow, P. 1973. Gastropod associations within the Malham Tarn, Yorkshire.
Freshwater Biology 3:521-534.

Cantrell, MA. 1981. Bilharzia snails and water level fluctuations in a tropical swamp.
Oikos 36:226-232.

Cardinale, B.J., T.M. Burton, and V.J. Brady. 1997. The community dynamics of
epiphytic midge larvae across the pelagic-littoral interface: do animals respond to

166

changes in the abiotic enviromnent? Canadian Journal of Fisheries and Aquatic
Sciences 54:2314-2322.

Cardinale, B.J., V.J. Brady, and TM. Burton. 1998. Changes in the abundance and
diversity of coastal wetland fauna fiom the open water/macrophyte edge towards
shore. Wetlands Ecology and Management 6:59-68.

Chapman, MG. and A.J. Underwood. 1996. Influences of tidal conditions, temperature
and desiccation on patterns of aggregation of the high-shore periwinkle, Littorina
unifasciata, in New South Wales, Australia. Journal of Experimental Marine
Biology and Ecology 196:213-237.

Clampitt, RT. 1970. Comparative ecology of the snails Physa gyrina and Physa integra.
Malacologia 10:113-151.

Clampitt, RT. 1974. Seasonal migratory cycle and related movements of the fresh-water
pulmonate snail, Physa integra. American Midland Naturalist 92:275-300.

Clarke, A.H. 1973. Canadian freshwater molluscs (Lymnaeidae). Malacologia 13-14:1-
509.

Clarke, A.H. 1981. The fieshwater molluscs of Canada. National Museums of Canada,
Ottawa, Canada.

Costil, K. and B. Clement. 1996. Relationship between freshwater gastropods and plant
communities reflecting various trophic levels. Hydrobiologia 321 :7-16.

Covich, A. P. and B. Knezevic. 1978. Size-selective predation by fish on thin-shelled
gastropods (Lymnaea): the significance of floating vegetation (Trapa) as a
physical refuge. Verhandlungen der Internationale Vereinigung fur Theoretische
und Augewandte Limnologie 20: 2172-2177.

Crosbie, B. and P. Chow-Fraser. 1999. Percentage land use in the watershed determines
the water and sediment quality of 22 marshes in the Great Lakes basin. Canadian
Journal of Fisheries and Aquatic Sciences 56:1781-1791.

Crowl, TA. 1990. Life-history strategies of a freshwater snail in response to stream
permanence and predation: balancing conflicting demands. Oecologia 84:23 8-
243.

Doremus, CM. and W.N. Harman. 1977. The effects of grazing by physid and
planorbid freshwater snails on periphyton. The Nautilus 91 :92-96.

Dussart, G. 1976. The ecology of freshwater molluscs in north west England in relation
to water chemistry. Journal of Molluscan Studies 42:181-198.

167

 

Dussart, G. 1979. Life cycles and distributions of the aquatic gastropod molluscs
Bithynia tentaculata (L.), Gyraulus albus (Muller), Planorbis planorbis (L.), and
Lymnaea peregra (Muller) in relation to water chemistry. Hydrobiologia 67:223-
239.

Edsall, T.A., B.A. Manny, and ON. Raphael. 1988. The St. Clair River and Lake St.
Clair, Michigan: an ecological profile. US. Fish and Wildlife Service,
Washington, DC, Biological Report 85(7.3).

Eisenberg, RM. 1966. The regulation of density in a natural population of the pond
snail, Lymnaea elodes. Ecology 47:889-906.

ESRI, Inc. 1999. ArcView 3.2. Redmond, California.

F orrnan, R.T.T. 1995. Land mosaics: the ecology of landscapes and regions. Cambridge
University Press, New York, New York.

Galatowitsch, S.M., D.C. Whited, R. Lehtinen, J. Husveth, and K. Schik. 2000. The
vegetation of wet meadows in relation to their land use. Environmental
Monitoring and Assessment 60: 121-144.

Gallo, G.J., B. Fried, and CW. Holliday. 1984. Effects of desiccation on survival and
hemolymph osmolality of the freshwater snail, Helisoma trivolvis. Comparative
Biochemistry and Physiology 78A:295-296.

Gathman, J .P., TM. Burton, and B.J. Arrnitage. 1999. Coastal wetlands of the upper
Great Lakes: distribution of invertebrate communities in response to
environmental variation. Pages 949-994 in Batzer, D.P., R.B. Rader, and SA.
Wissinger, eds., Invertebrates in freshwater wetlands of North America: ecology
and management. John Wiley & Sons, Inc., New York, New York.

Gibson, J .S. 1970. The function of the operculum of Thais lapillus (L.) in resisting
desiccation and predation. Journal of Animal Ecology 39:159-168.

Harman, W.N. 1972. Benthic substrates: their effect on freshwater mollusca. Ecology
53:271-277.

Harvey, BC. and W.R. Hill. 1991. Effects of snails and fish on benthic invertebrate
assemblages in a headwater stream. Journal of the North American Benthological
Society 10:263-270.

Hawkins, GP. and J .K. Furnish. 1987. Are snails important competitors in stream
ecosystems? Oikos 49:209-220.

Hill, W.R. 1992. Food limitation and interspecific competition in snail-dominated
streams. Canadian Journal of Fisheries and Aquatic Sciences 49:1257-1267.

168

Hochberg, Y. 1988. A sharper Bonferroni procedure for multiple tests of significance.
Bionmetrika 75:800-802.

Hosmer, D.W. and Lemeshow, S. 1989. Applied logistic regression. John Wiley &
Sons, Inc., New York, New York.

Hunter, RD. and W.W. Lull. 1977. Physiologic and environmental factors influencing
the calcium-to-tissue ratio in populations of three species of freshwater pulmonate
snails. Oecologia 29:205-218.

Jokinen, EH. 1978. The aestivation pattern of a population of Lymnaea elodes (Say)
(GastropodazLymnaeidae). American Midland Naturalist 100:43-53.

Jokinen, EH. 1983. The freshwater snails of Connecticut. State Geological and Natural
History Survey of Connecticut, Bulletin 109.

Jokinen, EH. 1987. Structure of freshwater snail communities: species-area
relationships and incidence categories. American Malacological Bulletin 5:9-19.

Karr, J .R. and LI. Schlosser. 1978. Water resources and the land-water interface.
Science 201 :229-234.

Keddy, RA. and A.A. Reznicek. 1985. Vegetation dynamics, buried seeds, and water
level fluctuations on the shorelines of the Great Lakes. Pages 33-50 in Prince,
H.H. and FM. D'Itri, eds., Coastal wetlands. Lewis Publishers, Inc., Chelsea,
Michigan.

Keddy, RA. and A.A. Reznicek. 1986. Great Lakes vegetation dynamics: the role of
fluctuating water-levels and buried seeds. Journal of Great Lakes Research
12:25-36.

Kelley, J .C., T.M. Burton, and W.R. Enslin. 1984. The effects of natural water-level
fluctuation on N and P cycling in a Great Lakes marsh. Wetlands 4: 159-175.

Kent, C. and J. Wong. 1982. An index of littoral zone complexity and its measurement.
Canadian Journal of Fisheries and Aquatic Sciences 39:847-853.

Keough, J .R., T.A. Thompson, G.R. Guntenspergen, and DA. Wilcox. 1999.
Hydrogeomorphic factors and ecosystem responses in coastal wetlands of the
Great Lakes. Wetlands 19:821-834.

Kesler, DH. and W.R. Munns, Jr. 1989. Predation by Belostomaflumineum

(Hemiptera): an important cause of mortality in freshwater snails. Journal of the
North American Benthological Society 8:342-3 50.

169

Kratz, T.K., K.E. Webster, C.J. Bowser, J .J . Magnuson, and BJ. Benson. 1997. The
influence of landscape position on lakes in northern Wisconsin. Freshwater
Biology 37:209-217.

Krieger, K.A. 1992. The ecology of invertebrates in Great Lakes coastal wetlands:
current research knowledge and research needs. Journal of Great Lakes Research
18:634-650.

Krieger, K.A. and D.M. Klarer. 1991. Zooplankton dynamics in a Great Lakes coastal
marsh. Journal of Great Lakes Research 17:255-269.

Lam, NS. and L. de Cola. 1993. Fractals in geometry. Prentice Hall, Englewood Cliffs,
California.

Lassen, H.H. 1975. The diversity of freshwater snails in view of the equilibrium theory
of island biogeography. Oecologia 19:1-8.

Laursen, J .R., G.A. Averback, G.A. Conboy, and BE. Stromberg. 1992. Survey of
pulmonate snails of central Minnesota. I. Lymnaeidae. Journal of Freshwater
Ecology 7:25-33.

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.

Leonard, LA. and ME. Luther. 1995. Flow hydrodynamics in tidal marsh canopies.
Limnology and Oceanography 40: 1474-148.

Levings, SC. and SD. Garrity. 1983. Diel and tidal movement of two co-occurring
neritid snails; differences in grazing patterns on a tropical rocky shore. Journal of
Experimental Marine Biology and Ecology 67:261-278.

Lewis, DB. and J .J . Magnuson. 2000. Landscape spatial patterns in freshwater snail
assemblages across northern highland catchments. Freshwater Biology 43:409-
420.

Liston, CR. and S. Chubb. 1985. Relationships of water level fluctuations and fish.
Pages 121-134 in Prince, H.H. and F .M. D'Itri, eds., Coastal wetlands. Lewis
Publishers, Inc., Chelsea, Michigan.

Lodge, D.M. and P. Kelly. 1985. Habitat disturbance and the stability of freshwater
gastropod populations. Oecologia 68:111-1 17.

Lodge, D.M. 1986. Selective grazing on periphyton: a determinant of freshwater
gastropod microdistributions. Freshwater Biology 16:831-841.

170

 

Lodge, D.M., K.M. Brown, S.P. Klosiewski, R.A. Stein, A.P. Covich, B.K. Leathers, and
C. Bronmark. 1987. Distribution of freshwater snails: spatial scale and the
relative importance of physicochemical and biotic factors. American
Malacological Bulletin 5:73-84.

Lougheed, V.L., B. Crosbie, and P. Chow-Fraser. 2001. Primary determinants of
macrophyte community structure in 62 marshes across the Great Lakes basin:
latitude, land use, and water quality effects. Canadian Journal of Fisheries and
Aquatic Sciences 58:1603-1612.

Lowe, R.L. and RD. Hunter. 1988. Effect of grazing by Physa integra on periphyton
community structure. Journal of the North American Benthological Society 7:29-
36.

Macan, T. 1950. Ecology of freshwater mollusca in the English Lake District. Journal
of Animal Ecology 19:124-146.

MacArthur, RH. and E0. Wilson. 1967. The theory of island biogeography. Princeton
University Press, Princeton, New Jersey.

Mackie, G.L. and LA. Flippance. 1983a. Intra-and interspecific variations in calcium
content of freshwater mollusca in relation to calcium content of the water.
Journal of Molluscan Studies 49:204-212.

Mackie, G.L. and LA. Flippance. 1983b. Relationship between buffering capacity of
water and the size and calcium content of freshwater mollusks. Freshwater
Invertebrate Biology 2:48-55.

Magnin, F ., T. Tatoni, P. Roche, and J. Baudry. 1995. Gastropod communities,
begetation dynamics and landscape changes along an old-field succession in
Provence, France. Landscape and Urban Planning 31:249-257.

Magurran, A.E. 1988. Ecological diversity and its measurement. Princeton University
Press, Princeton, New Jersey.

Malmqvist, B. 2002. Aquatic invertebrates in riverine landscapes. Freshwater Biology
47:679-694.

Marchetti, KB. and J .B. Geller. 1987. The effects of aggregation and microhabitat on

desiccation and body temperature of the Black Turban Snail, Tegulafimebralis
(A. Adams, 1855). Veliger 30:127-133.

McKillop, W.B. 1985. Distribution of aquatic gastropods across the Ordovician

dolomite - Precambrian granite contact in southeastern Manitoba, Canada.
Canadian Journal of Zoology 63:278-288.

171

 

McNicholl, MK. 1985. Avian wetland habitat functions affected by water level
fluctuations. Pages 87-92 in Prince, H.H. and FM. D'Itri, eds., Coastal wetlands.
Lewis Publishers, Inc., Chelsea, Michigan.

Minc, L. 1996. Michigan's Great Lakes coastal wetlands: definition, variability and
classification. Report to Michigan Natural Features Inventory, East Lansing, MI.

Minc, L. and DA. Albert. 1998. Great Lakes coastal wetlands: abiotic and floristic
characterization. Report to the Michigan Natural Features Inventory, East
Lansing, Michigan.

Mitsch, W.J. and J .G. Gosselink. 1993. Wetlands, 2nd ed. Van Nostrand Reinhold,
New York, New York.

Mittelbach, G.G. 1984. Predation and resource partitioning in two sunfishes
(Centrarchidae). Ecology 652499-513.

Okland, J. 1983. Factors regulating the distribution of fresh-water snails (Gastropoda) in
Norway. Malacologia 24:277-288. '

Okland, RH. and O. Eilertsen. 1994. Canonical correspondence analysis with variation
partitioning: some comments and an application. Journal of Vegetation Science
5:117-126.

Olivier, L. 1956. Observations on vectors of Schistosoma mansoni kept out of water in
the laboratory. 1. Journal of Parasitology 42:137-146.

Palmer, M.A., C.M. Swan, K. Nelson, P. Silver, and R. Alvestad. 2000. Strearnbed
landscapes: evidence that stream invertebrates respond to the type and spatial
arrangement of patches. Landscape Ecology 15:563-576.

Parashar, ED. and K.M. Rao. 1998. A novel mechanism for protection against
desiccation in a freshwater planorbid snail, Indoplanorbis exustus, vector of

animal schistosomiasis in India. South African Journal of Science 94:567-568.

Pip, E. 1978. A survey of the ecology and composition of submerged aquatic snail-plant
communities. Canadian Journal of Zoology 56:2263-2279.

Pip, E. 1985. The ecology of freshwater gastropods on the southwestern edge of the
Precambrian Shield. Canadian Field Naturalist 99:76-85.

Pip, E. 1987. Species richness of freshwater gastropod communities in central North
America. Journal of Molluscan Studies 53:163-170.

172

 

Pip, E. 1991. Macrophyte and associated mollusk communities in a meteor crater lake
on the Precambrian Shield in Manitoba. Canadian Field-Naturalist 105:483-487.

Pip, E. and J .M. Stewart. 1976. The dynamics of two aquatic plant-snail associations.
Canadian Journal of Zoology 54:1192-1205.

Reavell, RE. A study of the diets of some British freshwater gastropods. Journal of
Conchology 30:253-271.

Rooke, J. and G. Mackie. 1984. Mollusca of six low-alkalinity lakes in Ontario.
Canadian Journal of Fisheries and Aquatic Science 41:777-782.

Ross, M.J. and GR. Ultsch. 1980. Temperature and substrate influences on habitat
selection in two Pleurocerid snails (Goniobasis). American Midland Naturalist
103:209-217.

Rosenzweig, M.L. Species diversity in space and time. Cambridge University Press,
New York, New York.

Rotenberry, J .T. and J .A. Wiens. 1980. Habitat structure, patchiness, and avian
communities in North American steppe vegetation: a multivariate analysis.
Ecology 61:1228-1250.

Russell-Hunter, W.D. 1978. Ecology of freshwater pulmonates. Pages 335-383 in V.
Fretter and J. Peake, eds., Pulmonates (vol. 2A): systematics, evolution and
ecology. Academic Press Inc., London, England.

Sadzikowski, MR. and DC. Wallace. 1976. A comparison of food habits of size classes
of three sunfishes (Lepomis macrochirus Rafinescue, L. gibbosus Linneaus, and
L. cyanellus Rafinescue). American Midland Naturalist 95:220-225.

Sager, P.E., S. Richman, H.J. Harris, and G. F ewless. 1985. Preliminary observations on
the seiche-induced flux of carbon, nitrogen and phosphorus in a Great Lakes
coastal marsh. Pages 59-68 in Prince, H.H. and F .M. D'Itri, eds., Coastal

wetlands. Lewis Publishers, Inc., Chelsea, Michigan.
SPSS. 1999. SYSTAT version 9. SPSS, Inc., Chicago, Illinois.

Sterry, P.R., J .D. Thomas, and R.L. Patience. 1983. Behavioural responses of
Biomphalaria glabrata (Say) to chemical factors from aquatic macrophytes

including decaying Lemna paucicostata (Hegelm ex Engelm). Freshwater
Biology 13:465-476.

ter Braak, C.J.F. and P. Smilauer. 1998. CANOCO Reference manual and user’s guide

to Canoco for Windows: canonical community ordination (version 4).
Microcomputer power. Ithaca, New York.

173

 

ter Braak, C.J.F. and P. Smilauer. 1999. CANOCO for Windows (version 4.0). Centre
for biometry Wageningen. Wageningen, Netherlands.

ter Braak, C.J.F. and P.F.M. Verdonschot. 1995. Canonical correspondence analysis and
related multivariate methods in aquatic ecology. Aquatic Sciences 57:255-289.

Thomas, D.L. and J .B. McClintock. 1996. Aspects of the population dynamics and
physiological ecology of the gastropod Physella cubensis (Pulmonata: Physidae)
living in a wann-temperate stream and ephemeral pond habitat. Malacologia
37:333-348.

Townsend, CR. and T.K. McCarthy. 1980. On the defense strategy of Physafontinalis
(L.), a freshwater pulmonate snail. Oecologia 46:75-79.

Trombulak, SC. and CA. F rissell. 2000. Review of the ecological effects of roads on
terrestrial and aquatic communities. Conservation Biology 14:18-30.

Turner, AM. 1996. Freshwater snails habitat use in response to predation. Animal
Behaviour 51:747-756.

USACE. 2000. Water levels continue to decline. Great Lakes Update 139:1-7.

Vera, C., P. Bremond, R. Labbo, F. Mouchet, E. Sellin, D. Boulanger, J .P. Pointier, B.
Delay, and B. Sellin. 1995. Seasonal fluctuations in population densities of
Bulinus snegalensis and B. truncatus (Planorbidae) in temporary pools in a focus
of Schistosoma haematobium in Niger: implications for control. Journal of

Molluscan Studies 61:79-88.

Vermeij, G.J. and AP. Covich. 1978. Coevolution of freshwater gastropods and their
predators. American Naturalist 112:833-843.

Weatherhead, MA. and MR. James. 2001. Distribution of macroinvertebrates in
relation to physical and biological variables in the littoral zone of nine New
Zealand lakes. Hydrobiologia 462:1 15-1 19.

Weber, L.M. and D.M. Lodge. 1990. Periphytic food and predatory crayfish: relative
roles in determining snail distribution. Oecologia 82:33-39.

Wiens, J .A. 2002. Riverine landscapes: taking landscape ecology into the water.
Freshwater Biology 47:501-515.

Wellbom, G.A., D.K. Skelly, and BE. Werner. 1996. Mechanisms creating community
structure across a freshwater habitat gradient. Annual Review of Ecology and
Systematics 27:337-363.

Zar, J .H. 1999. Biostatistical analysis, 4th ed. Prentice-Hall, Inc. Upper Saddle River,
New Jersey.

174