.er CE.- ”.1
.3 v. p o I

K!’ ” T<

.6. Lama . in

5.34.» «ﬁrhhﬁwﬁ.
O

l... V. .. av}.
.. 3.06:“. pine...» is
V gin-«WWI; n. if... tut. 5L 3.
, . . . . . u .f z .fn§!hiﬁev:¢ir$ 7. «1!
V 1 V: .1 .6 “1.34.: and». «WAFEA .11..
l .. #7.. $3.63: .9 .9... Mum—nuvfﬁdmoqu
V ~W ..)rl.. 2! (2.7.2.. :3- .3327! :-
ﬂwwna‘i! 2...? . 2 * .. no
.5: l“ V?!
a rah... .
00“ Dr.”
3311». 18..

z ’si".n A
it

¢ , it“?
. v . . «
unnuurﬁnhugi v... (v...) 511%.... .3154

1:11, ﬂu". 14!..{...3. . . . . x
bit) In 1 15:90; ... 7.. ﬁn...
1 £" .
.l

,
11:1:
ilI‘...‘

. s. ,
. 2.1... 5
1 $162.53. . V
4: u:-
.tJtrnT.'k,a
19..
I5. .

II!!! V ~
1 .F’vam...y1nm :.
V. 1A...:0..rv.3

a V
:0;
:31!

I
s.
;
93.1.“... NF... - u}
1‘24}. I V9:
..

a it

£53.".

.3

. . .5 I!
21.... 33 5:
.£S!.III¢2;1!
I...

, L53?»

 

. .E<.u.t.a
.3.
Kalilbxt
1.. . 2:25:
.3.

a.
whine:
.iH...

) £31...
.1” at if}... z.
3......

«£7. 23.9:

. n hﬂﬁ.i¢l§ld.1§
v... P.xh..9...3.unhudnuuu
“3...“..“9 .... V
.

a. "2.5”“,
. . . .1
42...; 1.9
5.3.5,: .h I. :9
. is? .9".
I. ”V‘l‘ I.
‘ l I. I-IW .I‘ ‘h
.i: I: .35.
.33.?! t ‘5:
Au. Kafka...) V v.
y . . :1:
a 0.11.. 593.. . VA
a?xc§€.iﬁf. . n
2...... .52. :z

931...... {13:515.}:

K... ! Lr’..ﬁ«......l.v( “ﬁrtﬂﬂpug... xu.
i Q‘s-$5 .gtl‘vlr‘ v’li).’l d ‘-
0. ..I.V o

 

THESE

2000

names

ll\\\\\ \\\\\\l

\\\\\\\\\\\\\\\

mean»: srt\w£ WW5” "\

l\\\\\\\\\\\\\\\\\\\\ \l\\\\\

3 1293 02063 0509 r.

This is to certify that the

dissertation entitled

ABIOTIC AND BIOTIC INFLUENCES ON AQUATIC INVERTEBRATE
COMMUNITY STRUCTURE IN A LAKE HURON COASTAL WETLAND

presented by

Joseph Paul Gathman

has been accepted towards fulﬁllment
of the requirements for

PhoDo degree in Zoology

W

Major professor

Date ”1475 20%
I /

MS U is an Afﬁrmative Action/Equal Opportunity Institution 0-12771

 

 

LIBRARY

 

Unlvorslty

Michigan State

 

 

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

 

(Hm t W

 

 

 

 

 

 

 

 

 

 

 

 

 

1100 mm.“

 

ABIOTIC AND BIOTIC INFLUENCES ON AQUATIC INVERTEBRATE
COMIVIUNITY STRUCTURE IN A LAKE HURON COASTAL WETLAND

By

Joseph Paul Gathman

A DISSERTATION

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

DOCTOR OF PHILOSOPHY
Department of Zoology

2000

ABSTRACT

ABIOTIC AND BIOTIC INFLUENCES ON AQUATIC INVERTEBRATE
COMMUNITY STRUCTURE IN A LAKE HURON COASTAL WETLAND

By

Joseph Paul Gathman

Shoreline wetlands of large lakes are ecotones between terrestrial and aquatic
systems, and many ecological characteristics (e. g. water chemistry, vegetation) vary along
the lake-to-upland elevation gradient in these systems. In Great Lakes coastal wetlands,
this variation is enhanced by annual and multi-annual changes in water level. Aquatic
animal communities are likely to respond to this spatial variation in habitat conditions, but
few studies have sought such faunal variation or its causes. In this study I collected
animal community data along the elevation gradient in a northern Lake Huron coastal
wetland over a 3-year period during which the water level changed dramatically.

Replicate samples were collected along the slope at 10 cm elevation intervals,
covering a 60 cm elevation range in a ~120 m horizontal distance. In the ﬁrst year,
multivariate analysis revealed that animal community composition varied gradually with
elevation, and seasonal changes appeared to result from up-slope migrations of some
animal taxa. During the second year water levels were over 30 cm higher, and in the third
year water level dropped back toward initial levels. Most insect taxa were able to respond
rapidly to these changes, while many non-insects migrated more slowly. These diﬁ‘erences
led to increasing invertebrate densities and changes in community composition at diﬁerent
points on the gradient, and led to a loss of wetland diversity, as many rare taxa declined
and generalist animals thrived. Flooding duration was signﬁcantly correlated with

densities of the most common invertebrate taxa.

Because predator densities also varied over the gradient, I ran ﬁeld experiments to
determine whether they (small ﬁsh and larval dragonﬂies) partly caused the observed
faunal diﬂ‘erences over space and time. Fish reduced invertebrate densities, favoring
certain generalist taxa, while dragonﬂy eﬂ‘ects were less dramatic. Taxon-speciﬁc eﬂ‘ects
suggested that predation can prevent some prey taxa from occurring at gradient points
where predators are dense. Results led to a conceptual model of temporal and spatial
shifts in ﬂooding and predation controlling animal community composition in coastal

wetlands.

To my wife, Denise

iv

ACKNOWLEDGMENTS

I would like to thank my doctoral advisor, Dr. Thomas Burton, for his suggestions,
critiques, and encouragement, and for ensuring my continuous ﬁnancial support
throughout my doctoral program. I also thank the members of my advisory committee,
Dr. James Bence, Dr. Donald Hall, and Dr. Harold Prince, for their guidance and advice
on this research.

Financial support for this research was provided by The Nature Conservancy,
Michigan Chapter. Dr. David Ewert and other Nature Conservancy staff provided ﬁeld
logistical support and general enthusiasm. I also acknowledge the ﬁnancial and moral
support of the residents of the Les Cheneaux Islands area in Michigan's upper peninsula,
where this research was conducted. Additonal ﬁnancial support was provided by the
Michigan Department of Natural Resources, Non-game Wildlife Division, and I thank the
College of Natural Sciences for a Dissertation Completion Fellowship.

I am grateful to several colleagues for ﬁeld assistance and good-natured
comeraderie that made the many hours of ﬁeld work enjoyable. Sam Riﬂ‘ell and,
especially, Brian Keas were my constant ﬁeld companions and provided many insights and
reality checks. I also enjoyed the occasional company and assistance of Angie Conklin,
Donna Kashian, Tara Harmon, and Rebecca Serbin. Erica Butler and Sarah Harris
assisted with laboratory work. This work has been improved through discussions with
many people, particularly my lab colleagues and other graduate students.

The Zoology Department support staff were especially ﬁiendly and helpﬁil over
the years. Their jobs are thankless at times, so I would like to take time to thank Tracey
Barner, Lisa Craﬁ, Chris Keyes, Jan McGowan, and Judy Pardee for the countless "little"
things that needed to be done to support my work.

Finally, I thank my wife and my family, especially my parents, for support through

a period of time that has been challenging in many more ways than expected.

TABLE OF CONTENTS

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

FAUNAL VARIATION ALONG A FLOODIN G REGIME GRADIENT
IN A LAKE HURON COASTAL WETLAND: AQUATIC COMIVIUNITY

STRUCTURE AND SEASONAL CHANGES ................................................................ 4
Introduction ......................................................................................................... 4
Methods ............................................................................................................... 7

Study site ................................................................................................. 7
Sample collection and processing .............................................................. 8
Data collection and analysis ...................................................................... 9
Gradient analysis ..................................................................................... 10
Results ............................................................................................................... 12
Water level changes and wetland ﬂooding ............................................... 12
Multivariate analysis: Ordination by elevation ......................................... 13
Multivariate analysis: Ordination by depth .............................. ' ................. 15
Multivariate analysis: Monthly changes in ordination ............................... 16
Univariate analysis: Abundance differences among stations, plant zones
and depths .............................................................................................. 17
Univariate analysis: Temporal changes in gradient positions .................... 20
Univariate analysis: Microcrustaceans and ﬁsh ........................................ 21
Discussion .......................................................................................................... 22
Possible causal factors ............................................................................ 26

CHAPTER 2

IMPACTS OF HIGH WATER LEVELS ON THE BIOTA OF A LAKE HURON

COASTAL WETLAND: RESULTS OF A NATURAL EXPERIMENT ...................... 53
Introduction ....................................................................................................... 53
Methods ............................................................................................................. 55

Sampling procedures .............................................................................. 56
Data collection and analysis .................................................................... 58
Results ............................................................................................................... 60
Changes in total invertebrate densrty ........... 60
Community composition changes along the elevation gradient ................. 60
Diversity: Richness and equitability ........................................................ 62
Community composition variation: Individual taxa ................................. 63
Shifts in spatial position of peak taxa abundances among years ................ 64
Fish distributions .................................................................................... 65

Regression analyses of possible causal factors ......................................... 66

Plant community changes ........................................................................ 67
Discussion .......................................................................................................... 68
Possible causal mechanisms .................................................................... 70
Dispersal mechanisms and immigration rates ........................................... 75
Implications ............................................................................................ 77
CHAPTER 3

BIOTIC CONTROLS IN AN ABIOTICALLY-VARIABLE HABITAT:
EXPERIMENTS ON THE EFFECT OF PREDATORS ON A GREAT LAKES

COASTAL WETLAND AQUATIC INVERTEBRATE COMMUNITY ....................... 98
Introduction ....................................................................................................... 98
Methods ........................................................................................................... 100

Predator assemblage and ﬁsh density estimates ..................................... 101

Experiment 1 design and setup .............................................................. 102

Experiment 2 design and setup .............................................................. 104

Data collection and analysis .................................................................. 105

Results ............................................................................................................. 106
Experiment 1 ........................................................................................ 106

Experiment 2 ........................................................................................ 108

Discussion ........................................................................................................ 109
CONCLUSIONS ......................................................................................................... 132
LITERATURE CITED ................................................................................................ 135

vii

Table 1.1.

Table 1.2.

Table 1.3.

Table 2.1.

Table 2.2.

Table 2.3.

Table 2.4.

Table 2.5.

Table 3.1.

Table 3.2.

Table 3.3.

Table 3.4.

Table 3.5.

Table 3.6.

LIST OF TABLES

Signiﬁcant diﬁerences in abundances of taxa among sample stations,
plant zones, and depths; monthly and all months combined .................... 31

Mean gradient positions (weighted averages) of each taxon over all
months and for each month, and results from regressions of positions
against month. ........................................................................................ 35

Mean abundances of taxa by month and sampling station. ....................... 37

Mean invertebrate abundance (#/m2) over all months and transects
at each station during each year. ............................................................. 80

Shannon-Wiener diversity at each station during each year, based
on averages of taxa relative abundances over all months and transects ..... 80

Top ten most abundant taxa (ranked by relative abundance) at each
sampling station during each year ............................................................ 81

Relative abundance changes (increase or decrease) of individual taxa
between years at each station, and p-values from paired t—tests ................ 82

Taxa abundance changes (all stations) with paired t-test signiﬁcances,
and peak shifts among years .................................................................... 84

Ambient invertebrate community composition at experiment locations
(proportions of community comprised by individual taxa). .................... 117

Experiment 1 main eﬂ‘ects (% diﬁ‘erence between factor levels) and
signiﬁcance levels of model terms ﬁom ANOVAs based on log-
transformed abundances ........................................................................ 1 18

Signiﬁcant eﬂ‘ects of dragonﬂy larvae: percent changes in dependent
variables when dragonﬂy larvae were added to enclosures ..................... 119

Effects of ﬁsh addition on taxa abundances and signiﬁcance levels ﬁom
Experiment 2 t-tests on log-transformed abundances ............................. 120

Effects of mudminnows: diﬂ‘erences in number of signiﬁcant results
between prey of different sizes .............................................................. 121

Eﬂ'ects of mudminnows: % changes in some large-ﬁaction dependent
variables when intermediate size-class ﬁsh were added to enclosures. 122

Figure 1.1.

Figure 1.2.

Figure 1.3.

Figure 1.4.

Figure 1.5.

Figure 1.6.

Figure 1.7.

Figure 1.8.

Figure 1.9.

Figure 1.10.

Figure 2.1.

LIST OF FIGURES

Study site location. ................................................................................. 43

Schematic cross-section through a sample transect, showing vegetation
zonation and monthly water levels in 1996 at Mackinac Bay. .................. 44

Overhead view of Mackinac Bay and transect placement. ........................ 45

Correspondance analysis ordination (representing >69% of total inertia)
of invertebrate communities at six elevation stations during summer

(May-August), 1996, including positions of all taxa with combined

two-dimensional squared-cosine values greater than 0.5. ......................... 46

Correspondance analysis ordination (representing >71% of total inertia)
of invertebrate communities at six water depths during summer (May-
August), 1996 ......................................................................................... 47

Correspondance analysis ordinations of invertebrate communities at six
elevations during each of four months (May-August), 1996. ................... 48

Gradient distribution proﬁles of representative high-, mid-, and low-
gradient associated genera (Ilybius, Caenis, and T rianodes, respectively),
and a higher taxon (Chironomidae) with a more general distribution,

during summer (May through August), 1996 (means with standard

errors). ................................................................................................... 49

Comparison of gradient distributions of Isopoda: Caecidotea during each
month (May through August) of summer, 1996 (means with standard
errors). ................................................................................................... 50

Gradient distribution proﬁles of microcrustaceans during summer
(May through August), 1996 (means with standard errors). .................... 51

Variation in taxa richness (means with standard errors) across the
elevation gradient over all months (May through August) of
summer, 1996. ........................................................................................ 52

Schematic cross-section through representative sample transect in
study site. ............................................................................................... 86

ix

Figure 2.2.

Figure 2.3.

Figure 2.4.

Figure 2.5.

Figure 2.6.

Figure 2.7.

Figure 2.8.

Figure 2.9.

Figure 2.10.

Figure 2.11.

Figure 2.12.

Figure 3.1.

Figure 3.2.

Figure 3.3.

Lake Huron monthly average water levels in meters above International
Great Lakes Datum, measured at DeTour Village, MI, 1980 through

1998 (Data source: US Department of Commerce, NOAA/NOS,

Silver Spring, MD) ................................................................................. 87

Water level changes at Mackinac Bay over the study period. ................... 88

Average density of invertebrates among all sample stations, by
sampling date (means with standard errors) ............................................. 89

Invertebrate density differences among sampling stations, by year
(means with standard errors). .................................................................. 90

Correspondance analysis ordination of communities at each sampling
station during each year, with year-to-year shiﬁs indicated by arrows. ..... 91

Correspondance analysis ordination of invertebrate communities with
connecting lines representing community gradients during each year. ...... 92

Shannon diversity of invertebrate samples at each sampling station,
by year (means with standard errors) ....................................................... 93

Dominant taxa relative abundances, and changes in taxa richness by
year (means over all stations, with taxa richness standard errors) ............. 94

Mean ﬁsh catch per unit effort among stations and years (means with
standard errors). ..................................................................................... 95

Mean densities of mature dragonﬂy larvae among stations and years
(means with standard errors) ................................................................... 96

Densities of mature and early-instar dragonﬂy nymphs by year
(means with standard errors) ................................................................... 97

Wellbom et al. (1996) conceptual model of community structuring
forces in freshwater systems .................................................................. 123

Plan view of Experiment 1 in-site block layout ((1 = no dragonﬂy
larvae; D = dragonﬂy larvae added; f = no ﬁsh; F = ﬁsh added). ......... 124

Plan view of Experiment 2 in-site enclosure placement, indicating
ﬁsh addition and exclusion. ................................................................... 125

Figure 3.4.

Figure 3.5.

Figure 3.6.

Figure 3.7.

Figure 3.8.

Figure 3.9.

Effects of Fish X Dragonﬂy-larvae interactions on abundances of
amphipods in experimental enclosures in Experiment 1 (means with
standard errors). ................................................................................... 126

Correspondance analysis plot of 1996 data including plotted positions

of predators used in experiments and their affected prey. Underlined

taxa are those that were signiﬁcantly affected by predators in

enclosures ............................................................................................. 127

Spatial distributions of mudminnows (Umbra limi) and all ﬁsh species
combined during each of 3 years ending with the experiment year
(means with standard errors) ................................................................. 128

Conceptual model of regulation of spatial community variation in
Mackinac Bay wetland. Solid arrows represent strong negative effects

and dashed arrows represent weaker effects (following diagram format

of Wellbom et al. 1996).; ...................................................................... 129

Observed spatial patterns in certain abiotic factors in Mackinac Bay

'wetland ................................................................................................. 130

Hypothesized chain of mechanisms describing how primary wetland
abiotic factors regulate biotic controls on prey invertebrates .................. 131

INTRODUCTION

Great Lakes coastal wetlands have received relatively little research attention,
despite their recognized importance as breeding and feeding habitats for ﬁsh (Brazner
1997, Jude and Pappas 1992, Liston and Chubb 1986), waterfowl (Prince et al. 1992), and
migrating songbirds (D. Ewert, pers._ com). This oversight may be because these
systems "fall through the cracks", not ﬁtting into the traditional deﬁnitions of wetland or
lake habitats. However, a special symposium (Prince and D'Itri 1985) and a 1992 special
issue of the Journal of Great Lakes Research each focused on Great Lakes coastal
wetlands, indicating a growing recognition of the need for more research on these
systems. While all components of these wetlands require much more work, Krieger
(1992) noted that invertebrate communities in particular had been almost completely
ignored. These communities are especially diverse and inhabit spatially-complex habitats
compared to non-wetland Great Lakes communities, and are likely aﬂ‘ected by a variety of
environmental factors such as water chemistry gradients, lake connectivity, wave
exposure, and predators (reviewed in Gathman et al. 1999). In the last two decades,
invertebrates in a few coastal wetlands have received attention (e. g. Duffy and Batterson
1987, McLaughlin and Harris 1990, Brady 1992, Brady and Burton 1995, Cardinale et al.
1997), but most other sites have remained unstudied.

Great Lakes coastal wetlands are like other wetlands in many respects, but their
position at the interface between uplands and large lakes subjects them to a unique suite of
water movements and water-level changes that appear to determine many biotic
characteristics. These hydrologic factors include waves from the open lakes and large
ships, storm surges, multi-annual and annual water level changes resulting from climatic
variation, and short-term seiche-induced changes. Coastal wetlands only develop where
shoreline morphology offers at least partial protection from the direct force of waves and

storm surges. Maynard and Wilcox (1997) classiﬁed Great Lakes coastal wetland types

based on shoreline form and, implicitly, exposure to waves. The turbulent mixing caused
by waves can aﬂ‘ect factors that inﬂuence benthic invertebrates such as vegetation,
substrate structure (Cole and Weigrnan 1983) and water chemistry (Suzuki et al. 1995,
Cardinale et al. 1997). Unlike wave-induced turbulence, water-level changes affect all
lake-connected coastal wetlands and are important in determining vegetation zonation
(Minc 1996) and wetland area (Burton 1985), though their effects on aquatic animals have
not been determined.

Over 50% of pre-settlement coastal wetlands of the Great Lakes have been lost to
human uses with losses of the more vulnerable habitats, such as wet meadows,
approaching 100% in certain areas (Brazner 1997, Comer et a1. 1995, Jaworski and
Raphael 1978). The Les Cheneaux Islands area in Northern Lake Huron (the location of
the present study) has largely escaped this fate so far, so it is one of the few remaining
Great Lakes shoreline regions with a wealth of coastal wetland acreage that has been left
mostly intact. However, other than fairly recent vegetation surveys (Albert et al. 1987 ,
Minc 1996), no formal studies including Les Cheneaux coastal wetlands have been
published. This is unfortunate because wetlands in this region are currently threatened by
increased commercial and residential shoreline development, higher human population
density, more inshore boat traﬁic, and invasions of several exotic plant and animal species.

In the studies presented here, I focused on the ecotonal nature of a northern Lake
Huron coastal wetland, emphasizing animal community variation along a continuum ﬁ'om
a relatively-deep emergent marsh "lake-like" habitat to a wet meadow "wetland-like"
habitat at the wetland-upland border. In Chapter 1, I test the hypothesis that animal
assemblages vary signiﬁcantly and gradually along this continuum, and argue that variation
in ﬂooding regimes is probably the most important inﬂuence on the gradient in community
composition. I also present evidence that animal movements along the gradient are
important, but that many animal taxa move very little, so community composition along

the gradient varies over the course of a summer season.

In Chapter 2, I focus on a natural "experiment" in ﬂood-regime alteration. A ,
substantial increase in Lake Huron water level in the second year of the study presented an
opportunity to study the effects of a change from temporary to year-round ﬂooding in the
wet meadow, while positions of vegetation zones changed little. Animal changes in the
wet meadow community were much greater than in lower-elevation wetland zones, so I
again argue for the importance of ﬂooding regime and animal movements. However, it
was also apparent that recent ﬂooding history was important, as invertebrate assemblages
varied depending on how long a given wetland area had been ﬂooded.

During the three years of water-level change, I also observed changes in
distributions of the major aquatic predators in the wetland. Because these changes co-
occurred with invertebrate community changes, I sought to determine whether variation in
predation pressure may have been an important factor contributing to the observed
community changes. Chapter 3 describes the results of ﬁeld experiments in which I
manipulated predators in mesocosms to observe their effects on their prey. I examine the
results in light of observations presented in previous chapters, and then suggest a
modiﬁcation of a literature-derived conceptual model of abiotic and biotic factors
determining aquatic invertebrate community structure. Finally, I conclude with a summary

of all results and their implications.

Chapter 1

FAUNAL VARIATION ALONG A FLOODING REGIME GRADIENT IN A LAKE
HURON COASTAL WETLAND: AQUATIC COMMUNITY STRUCTURE AND
SEASONAL CHANGES

INTRODUCTION

Coastal wetlands are important habitats for many game and forage ﬁsh (Jude and
Pappas 1992), waterfowl (Prince et al. 1992), and migrating songbirds (D. Ewert, The
Nature Conservancy, Michigan Chapter, pers. comm), but losses of coastal wetland
acreage have been large (Brazner 1997, Comer et al. 1995, Jaworski and Raphael 1978).
Despite the importance of these systems, advances in knowledge of Great Lakes coastal
Wetlands have not kept pace with studies of many other wetland types (Krieger 1992).
Similarly, work on the ecology of invertebrates and ﬁsh in wetlands has lagged behind
achievements in areas such as hydrology and vegetation dynamics (Batzer and Wissinger
1996), so knowledge of coastal wetland ﬁsh and invertebrate ecology is based on the work
of only a few authors. Krieger (1992) cited only two studies (Duﬂ‘y and Batterson 1987,
McLaughlin and Harris 1990) from the upper Great Lakes in his review of coastal wetland
invertebrate ecology , but at least one wetland in Saginaw Bay has received considerable
attention since that time (Brady and Burton 1995, Cardinale et a1. 1998).

Given the paucity of studies on coastal wetland fauna, we know very little about
the most basic environmental inﬂuences on these communities. However, it is becoming
apparent that some coastal wetland processes vary spatially along the substrate-elevation
gradient ﬁ'om the upland/wetland boundary down to the lower wetland interface with
deeper pelagic waters. Such factors as water quality (Cardinale et al. 1998) and
vegetation (Minc 1996) vary along this gradient, and may inﬂuence animal communities.

However, these factors are ultimately controlled by hydrologic factors (water depth, water

movements, and ﬂooding regime), so any faunal variation along the gradient is likely to be
caused by hydrologic variation, at least indirectly.

Hydrology in coastal wetlands is dynamic and very inﬂuential on wetland structure.
As in most wetlands, coastal vegetation occurs in distinct zones along elevation gradients
in response to spatial variation in ﬂooding regimes and soil moisture (Mitsch and
Gosselink 1993, Minc 1996). Many Great Lakes coastal ﬂooding regimes show a distinct
spatial pattern, progressing from permanently-ﬂooded lower elevations to temporarily-
ﬂooded high-elevation zones (reviewed in Gathman et al. 1999). The occurrence of
temporary ﬂooding regimes in these high zones is ensured by a fairly predictable seasonal
pattern in lake levels. The annual cycle begins with low water levels in mid-winter, which
rise through the spring as snow-melt water and spring rains run into the lakes. The highest
annual levels are observed in mid-summer, when reduced precipitation and high
evaporation and transpiration combine to cause a net water loss and water levels begin
their autumn decline. As the water rises, it ﬂoods progressively higher-elevation wetland
area, and as it falls, these areas are drained. The resulting wetland ﬂooding-regime
gradient was of particular interest to Krieger (1992), who called for researchers to look
for invertebrate community variation along the gradient, because he expected community
composition to vary in direct response to variation in ﬂood duration found there.

Many authors agree that water level ﬂuctuations (ﬂooding regime) are key
structuring features of wetlands (Wiggins et al. 1980, Williams 1987, Krieger 1992,
Mitsch and Gosselink 1993, Schneider and Frost 1996), and may even be the most
inﬂuential factor (Batzer and Wissinger 1996). Water depth, water and sediment
temperatures (Williams 1987), plant communities (V oights 1976, Zedler 1981), and
detritus quality (Nelson et al. 1990, Barlocher et al. 1978) are all strongly inﬂuenced by
ﬂooding regime, and may inﬂuence wetland fauna. Other potential inﬂuences on fauna,
such as turbidity and conductivity, are likely to be aﬂ‘ected by evaporation and, conversely,

by dilution from water inputs (Williams 1987). Furthermore, aquatic animals are directly

aﬂ‘ected by the habitat area increases and reductions resulting from water ﬂuctuations
(“frlliams 1987, Wiggins et al. 1980).

Whether through direct or indirect mechanisms, ﬂooding regime is important in
determining wetland animal communities. Wiggins et al. (1980) found that temporary
pools were inhabited by a distinct assemblage of macroinvertebrates which was less
diverse than that of permanent waters. They categorized the genera of this assemblage
according to life history adaptations to temporary aquatic environments, such as drought
resistance and active dispersal, which permanent-water animals apparently lacked.
Similarly, Williams (1987) found "relatively little overlap" in species between temporary
and permanent waters.

Collinson et al. (1995), Jeﬂiies (1994) and Hanson and Swanson (1989) found that
invertebrate species assemblages varied with ﬂooding regime, and these and other studies
(e. g. Bataille and Baldassarre 1993, Neckles et al. 1990) found community composition
diﬂ‘erences even at higher taxonomic levels. These apparent higher-taxa habitat
preferences were consistent across several studies, suggesting habitat constraints imposed
by body plan or other relatively-immutable similarities among related species. For
example, microcrustaceans, Hydracarina, Coleoptera, migrant Hemiptera, Gastropoda,
and certain families of Diptera were particularly abundant in temporary ponds, while
Porifera, most Megaloptera, and many families of Plecoptera, Odonata, and Trichoptera
were excluded ﬁom these systems.

While a distinction between fauna of isolated permanent and temporary waters is
well supported, such a distinction has not been sought in coastal wetlands which consist of
permanent and temporary zones that are directly connected. Ifﬂooding regime is an
important controlling factor in these systems, it is reasonable to expect that the fauna of
higher-elevation temporary zones differs from that found in permanent emergent marsh

zones. On the other hand, the relative ease of migration between these broadly-connected

zones may preclude strong faunal diﬁ‘erences, so investigation of this possible spatial
diﬂ‘erence in communities is needed.

In this study, I examined the invertebrate and ﬁsh community along a ﬂooding-
regime gradient in a single summer season. Rising water levels in the study site caused
progressively later inundation in successively higher-elevation wetland zones, in contrast
to low elevations, which were ﬂooded throughout the year. I hypothesized that the faunal
community would gradually and unidirectionally vary from low to high elevations as a
result of diﬂ‘erences in life-history and behavioral adaptations to temporary water
conditions. Further, I expected that animal taxa initially found at low elevations would
migrate up—slope as the advancing water line made available newly re-ﬂooded habitat.

I tested the following hypotheses: I. the community can be described as a
multivariate faunal " gradient" corresponding more to elevation (hence, ﬂooding regime)
than to speciﬁc water depths; 2. the gradient deteriorates over the season as immigrating
fauna mix into the temporary-zone community; 3. the community varies gradually from

low elevation to high, rather than in discrete steps corresponding to plant zones.

METHODS
Stun)» site

The study area was located on the northern shore of Lake Huron in Mackinac Bay,
which was immediately east of Hessel, Michigan (Figure 1.1). A two-lane, paved rural
highway formed the northern upland border of the site. The site was characteristic of
wetlands in the area, with a diverse plant community exhibiting typical wetland vegetation
zonation (Figure 1.2). Starting from the open water, this zonation began with a deep-
marsh community dominated by hardstem bulrush (Scirpus acutus) and interspersed with
yellow water lily (Nuphar sp.), pickerelweed (Pontedaria sp.), arrowhead (Sagittaria sp.),
and a variety of submergent plants. Moving up-slope the deep-marsh community gave

way to a transition zone, which consisted of a dense band of cattail (mostly T wha

angustzfolia) and spike rush (Eleocharis smallii) richly interspersed with species from
adjacent zones. This zone was relatively narrow and was succeeded by wet meadow at
higher elevations. The wet meadow was dominated by several sedge species of the genus
Carex (especially C. lasiocarpa and C. aquatilis), and included a diverse assemblage of
graminoid and broad-leaved facultative wetland plants. Finally, at the upper edge of the
wetland, several shrub and grass species were abundant among the sedges, which were

dominated by C. stricta, a hummock-forming sedge.

Sample collection and processing

In May, 1996, I established ﬁve replicate transects in the site running from high- '
elevation wet meadow down to permanently-ﬂooded emergent marsh (Figure 1.3). I
selected transect positions that were approximately evenly spaced, while avoiding either of
two small streams running through the site from upland to lake. Water depth was not a
reliable elevation measure because seiche activity caused constant water level ﬂuctuations
over a range of several cm, and because the wet meadow was not yet ﬂooded. Instead, I
used a laser level for elevation measurements, establishing six sampling stations on each
transect at 10 cm elevation intervals so that a range of six ﬂooding-regime zones would be
sampled (Figure 1.2). Stations were numbered 1 through 6, ﬁ'om highest-elevation to
lowest. The highest stations, 1 and 2, were in the wet meadow, while 3 and 4 were in the
cattail-dominated transition zone, and 5 and 6 were in the deep emergent marsh. Transect
4 was diﬂ‘erent ﬁ'om the others in that it ran through a break in the cattail stand, so the
plant community around Station 3 was sedge-dominated wet meadow, and that around
Station 4 was a bulrush-dominated deep marsh community.

Transect lengths, ﬁom Station 1 to Station 6, averaged 120 m, with inter-station
distances averaging 24 m. Because stations were selected based on elevation, horizontal
distances between them varied. In the transition/cattail zone, where substrate slopes were

relatively steep, sample stations were closer together than elsewhwere (average ~13 m

between Stations 3 and 4, all transects). I also installed a staﬂ‘ gauge at Station 6 of the
middle transect to monitor seasonal water-level changes, using the Station 6 substrate as a
benchmark of 0 cm depth.

To sample animals, I used activity traps made from 2 L plastic bottles placed
horizontally in the water on the substrate. These traps were constructed by cutting the
bottom oﬂ‘ one bottle, then cutting the top oﬂ‘ a second bottle and inserting it, narrow end
ﬁrst, into the open bottom of the other bottle and securing the two pieces with staples.
The result was a ﬁrnnel (created by the inserted, inverted top) leading into the bottle, with
a screw—on cap at the opposite end of the bottle for trap emptying. I placed traps ten
meters ﬁ'om each sample station, in a perpendicular direction from the transect and always
on the same side. Traps were held in place by ﬂagged dowels that passed through plastic
loops (made from cable ties attached to the trap sides) and into the ground. After 24
hours I emptied the traps to prevent specimen deterioration, and re-set them in the same
location. After a second 24-hour period I collected the traps, poured the contents through
a 250 um mesh sieve, euthanized ﬁsh in MS-222, and preserved all specimens in 95%
ethanol.

Data collection and analysis

_ To identify invertebrates, I used Merritt and Cummins (1996), Pennak (1989), and
taxon-speciﬁc keys for lower taxonomic levels, when appropriate. Fish identiﬁcation
followed Hubbs and Lagler (1967). I identiﬁed ﬁsh to species and invertebrates to the
lowest taxonomic level possible with a dissecting scope. I attempted to identify
invertebrates to the genus level, but this was not possible or feasible in many cases because
the majority of individuals were very immature, lacking some key characters. For each
sample, I counted all individuals of each taxon, then calculated the relative abundance of
each taxon as a proportion of all animals in the sample. Data were decidedly non-normal

and heteroscedastic because of a high ﬁequency of zero counts. This problem appeared to

 

be an artifact of the sampling method because many traps with ﬁsh and other predators in
them may have suﬂ°ered ﬁ'om high predation losses, and because activity traps are a
"passive" sampling method that depends on voluntary trap entry. The data could not be
transformed to meet the assumptions of parametric statistical methods, so I used methods

with less-restrictive assumptions, as described below.

Gradient analyses

To determine whether whole communities varied gradually over the elevation
gradient, I combined data ﬁom all months and from all ﬁve replicate samples at each
elevation (for a total of 6 multivariate observations, 1 per elevation), then conducted
correspondance analysis (CA) on the multivariate data set using SAS software, procedure
CORRESP. I chose this ordination method because: 1. it was designed for gradient
analysis of count data; 2. it assumes unimodal (rather than monotonic) distributions of
taxa along environmental gradients; 3. it is not constrained by assumptions of multivariate
normality; and 4. it allows simultaneous analyses of locations (elevations) and taxa. I
examined the resulting two-dimensional plot, looking for a numerical ordering of
elevations along the ﬁrst axis, which would indicate a community gradient corresponding
to the elevation gradient. Next, I plotted the positions of individual taxa in the two-
dirnensional display, and examined the diagnostic measures ﬁom the CA output. These
measures summarize the roles played by each taxon and each elevation in each CA
dimension, so each taxon and elevation are assigned a set of diagnostics for every
dimension. The number of dimensions is 1 less than the number of elevations or taxa,
whichever is lower. In this case, there were 6 elevations (much fewer than the number of
taxa), so there were 5 CA dimensions. The "Partial Contribution" measure in SAS
corresponds to Greenacre's (1984) "CRT", and represents how much a taxon or elevation
contributed to the total inertia of the dimension, i.e. the degree to which variation in that

taxon or elevation corresponded to variation in the dimension. The "Squared Cosine" in

10

SAS corresponds to Greenacre's "COR", and represents how much of the variation within
. a taxon or elevation was represented in the dimension, i.e. the degree to which a taxon's or
elevation's variation is accounted for in the dimension.

1 also ran correspondance analyses on two other aggregations of the data. First, I
separated all data by month, and aggregated the ﬁve replicate samples at each elevation
within each month (6 multivariate observations per analysis). Analyses of these individual
data sets resulted in separate ordinations for each month from May through August, which
allowed me to determine whether a faunal gradient occurred during individual months.
Second, I aggregated samples from the entire summer by water depth, regardless of
elevation, so May's Station 4 samples were grouped with June's Station 2 samples, and
Station 1 samples ﬁ'om July and August because all of these were taken in water
approximately 10 cm deep. The resulting ordination of depth groups allowed assessment
of whether these groups were numerically ordered along the ﬁrst axis, suggesting a strong
community response to water depth changes.

To explore distributions of individual taxa and their temporal changes, I examined
all data graphically, plotting taxa abundances and relative abundances against elevation,
depth, and plant zone for individual months and for all months combined. I then tested
whether taxa abundances signiﬁcantly varied by elevation (station), water depth, plant
zone, and month. Because the data could not be effectively transformed, as explained
above, I could not reliably use regression analysis to examine animal distributions along
the elevation gradient, as is common in direct gradient analysis. Given the data structure,
I usually used the non-parametric Knrskal-Wallis test followed by a graphical examination
of taxon abundances by each independent variable that produced a signiﬁcant result. Only
a few variables (e. g. total abundance) could be effectively transformed to meet
assumptions and allow the use of parametric methods. I used 90% or levels as the

standard for hypothesis tests because of the above-mentioned high within-group variances.

ll

To examine animal distribution changes over the sampling period, I calculated
mean elevation-gradient positions for all taxa during each month. This involved a
weighted averaging method in which I calculated the following for each taxon on each
transect:

6
ZixN,

GP = 1:1 .

6 ,
2 N.-
r=l
where GP=gradient position, i=sample station number, and Niﬁotal abundance of the
taxon at station i. The resulting " gradient position" was the abundance-weighted mean
elevation (using sample station number as an arbitrary elevation index) where the taxon in
question occurred. The integer values used for sample station identiﬁcations served as
eﬂ‘ective elevation indicators because the physical gradient that they depicted was
measured in equal intervals of elevation rather than horizontal distance. Performing GP
calculation for each transect resulted in ﬁve replicate mean gradient positions (1 per
transect) for each taxon. I then used simple linear regression of gradient positions vs.

month (thus 5 replicates per month for four months) to determine whether the gradient

position for each taxon changed unidirectionally over the course of the sampling season.

RESULTS
Water level changes and wetland ﬂooding

In early May, when I ﬁrst set transect positions and the depth gauge, the water line
was near the upper edge of the transition zone, where I placed Station 3 on each transect.
This water level corresponded to an initial reading of 30 cm on the Station 6 gauge, so
Station 5 was under 20 cm of water and Station 4 had a depth of around 10 cm, the
operational limit for the activity traps. These depths were approximate averages, but

continual seiche activity caused constant depth ﬂuctuations. Seiche amplitudes were

12

usually only a few cm except after storms, and depth changes were subtle and
undetectable to the casual observer.

The water level increased fairly rapidly through June, rising nearly 20 cm and
allowing trap placement at Stations 3 and 2 during the sampling period late that month
(Figure 1.2). By the same time in July, the water had risen an additional 10+ cm, ﬂooding
Station 1, and stayed at that level through the August sampling period. The August
ﬂooding of Station 1 was not predicted because, on average, Lake Huron levels peak in
late July and begin to drop shortly thereafter. The maintenance of high water ﬂooded the

wet meadow longer than expected, and presaged higher water levels in subsequent years.

Multivariate analysis: ordination by elevation

Correspondance analyis of the combined data ﬁ'om the six sampling stations
resulted in a numerical ordering of the stations in two-dimensional ordination space
(Figure 1.4). The combined inertias of the ﬁrst two axes exceeded 69% of total inertia,
indicating that the two-dimensional display represented 69% of the total variance in the ,
data set that was accounted for by the analysis. The relative plotted positions of
elevations were nearly identical whether family-level data were used or genera were -
substituted for family data when available. The numerical ordering of elevations indicated
a gradual shift in community composition from the upper wet meadow (Station 1) through
successively lower stations, to the deep emergent marsh at Station 6. Stations 3 and 4, in
the cattail/transition zone, were relatively close together, indicating relatively greater
similarity between samples collected at transition zone stations than between any other
two stations.

A subset of the animal taxa collected contributed the most inertia to the creation of
the ﬁrst two ordination axes, as indicated by relatively high "partial contribution"
parameters in the SAS CA output. The ﬁrst axis primarily represented a distinction

between the two wet meadow stations and Station 6 at the opposite end of the transects.

13

Three groups of taxa contributed most (contribution index > 0 in SAS output) to ﬁrst-axis
inertia. The ﬁrst group consisted of several taxa that were strongly associated with the wet
meadow stations, especially Ilybius (the dominant Dytiscidae beetle genus), early-instar
Libellulidae dragonﬂy nymphs, Isopoda, Sphaeriidae clams, Planorbidae snails, and
Nematoda. The second group included taxa that were strongly associated with the deep
marsh throughout the season: larvae of T rianodes (the dominant Leptoceridae caddisﬂy
genus), Hyalella (the dominant amphipod genus), young-of-year Amia calva (bowﬁn),
and Bufo tadpoles (which only occurred in the early summer). Finally, the
microcrustaceans and Chironomidae contributed to the axis because of their high
abundances. These groups were somewhat more widespread over the gradient, but still
showed associations with certain elevations. Chironomidae, Cladocera, and Copepoda
abundances peaked in the lower wet meadow/upper transition zone (Stations 2 and 3),
while ostracods were most abundant in the upper deep marsh (Station 5).

The second axis caused the "arch effect", a familiar phenomenon in
correspondance analysis that arose as a mathematical artifact of the occurrence of several
taxa that dominated the middle of the gradient but tended not to occur at either end. In
this study, these shared taxa were more abundant below the gradient middle, hence the
asymmetric arch evident in Figure 1.4. These taxa contributed strongly to the second axis
inertia and included the mayﬂy Caenis, the hydroptilid caddisﬂy Oxyethira, the corixid bug
Palmacorixa, and the Lepidoptera in general. In most cases, plotted positions of the
genera contributing strongly to these axes were very near the positions of their confarnilial
relatives, for example, the position of Palmacorixa was generally indicative of the position
of all Corixidae genera sampled, even though they were not identiﬁed as strong
contributors to the axis inertia (not having a contribution index > 0 in SAS output).
However, more variety in gradient position occurred at and above the family level. For
example, the Odonata families Libellulidae, Aeschnidae, and Coenagrionidae were

associated with progressively lower parts of the gradient.

14

Figure 1.4 includes the names of those taxa that were best represented in the ﬁrst
two ordination axes, as measured by the "squared cosine" parameters associated with each
taxon in the SAS CA output. I included all taxa with a summed two-dimension squared
cosine value greater than 0.5, indicating that over 50% of the variation in that taxon was
accounted for in the two-dimensional plot. Correspondance analysis plots commonly
include points for "sites" and " species", or stations and taxa in this case. I replaced taxa
points with taxa names because actual measured distances between taxa and elevation
positions cannot be interpreted directly in correspondance analysis. Consequently, the
proximity of taxa positions to elevation positions should be interpreted as general
correspondances between them, rather than quantitative associations. However, plotted
taxa positions relative to the "arch" can indicate habitat speciﬁcity vs. generality. For
example, the fact that the plotted position of Chirononridae was in the interior of the arch
suggested that it was more generally distributed on the gradient because it was relatively
close to several elevation points. Conversely, taxa that were found outside the arch or

past its ends usually corresponded strongly to the nearest elevation point.

Multivariate analysis: Ordination by depth

The ordination of all samples grouped by water depth did not result in a numerical
ordering of samples along the ﬁrst axis (Figure 1.5). Instead, the axis primarily
represented a distinction between two broods of tadpoles that appeared at different depths
during the summer. These groups may have been inﬂuential because they occurred at high
abundances, but these data were also used in the elevation ordination and did not have as
strong an inﬂuence. By contrast, the second dimension produced a near-complete
numerical ordering of communities at successively greater depths, although the 60-cm
depth occurred in the middle of the axis. This axis contrasted a group of shallow-water

taxa (e. g. Lirnnephilidae, Sphaeriidae, Hirudinea, Nematoda, Umbra Iimi, Agabus,

15

Isopoda, Ibeius, Chironomidae, Copepoda) with fewer deeper-water associates that

tended to occur at higher abundance (Ostracoda, Amia calva, Bufo tadpoles).

Multivariate analysis: Monthly changes in ordination

The analysis of the May data resulted in an ordination by elevation along the ﬁrst
axis (Figure 1.6a), though this is not remarkable because only three elevations were
included in the ordination (because low water precluded sampling higher elevations).
Nonetheless, the taxonomic groups that contributed most to the May ordination
comprised communities that were somewhat indicative of patterns seen later in the season.
The upper end of the reduced May gradient was characterized by Dytiscidae,
Chirononridae, Gastropoda, and Collembola, all of which became abundant in the wet
meadow later in the year. Meanwhile, the lower-station communities consisted largely of
Amphipoda, Corixidae, Hydracarina, and Ostracoda, all of which remained dominant at
low elevations through the season. Also, the May communities included several taxa that
apparently were early-season specialists, rarely or never appearing in later samples. These
included Iowa darters (Etheostoma exile), very large leeches, an unidentiﬁed
Lirnnephilidae caddisﬂy genus, and the Metropodidae mayﬂy genus Si'phloplecton, all of
which occurred in the highest-elevation (Station 4) May samples.

The June data included samples from ﬁve elevations because the water level had
risen almost 20 cm between the end of May and late June. The multivariate analysis of
these data also resulted in a numerical ordering by elevation along the ﬁrst dimension, but
Station 2, representing the newly ﬂooded wet meadow, appeared far to the right of the
other points, suggesting strong dissimilarity between the community at this elevation and
the others (Figure 1.6b). The ﬁrst axis, not surprisingly, was most strongly inﬂuenced by
a group of taxa that only appeared in the wet meadow: larvae of the small dragonﬂy
Sympetrum (Odonata: Libellulidae) and biting midges Serromyia and F orcipomyia

(Diptera: Ceratopogonidae), the exotic prosobranch snail Bithynia, and several

16

unidentiﬁed ﬂy larvae (Diptera: Brachycera). Also inﬂuential were several taxa associated
with the wet meadow in the overall ordination (e. g. Ilybius, Planorbidae).

The July and August analyses produced quite different results ﬁom the others. No
numerical ordination by elevation was apparent in the July plot's ﬁrst dimension, which
was primarily determined by a distinction between discontiguous elevations dominated by
Cladocera and those dominated by a combination of Chironomidae, Copepoda, and
Naididae worms. The second dimension, however, was largely determined by a distinction
between Station 1 and Stations 4 and 6 (based on partial contributions to inertia),
suggesting a partial ordination by elevation (Figure 1.6c). The taxa contributing most to
this axis were positioned at opposite ends of the plot, and were dominated by the wet
meadow (at one end) and deep marsh (at the other) associates identiﬁed in the overall
ordination.

The August plot included the elevation-based gradient, but in two separate parts
along the ﬁrst two axes (Figure 1.6d). The ﬁrst axis was determined by a distinction
between Stations 1 and 2 (wet meadow) and Station 3. The strongly contributing taxa
consisted of established wet meadow residents, and several taxa that apparently became
abundant there later in the season (e. g. Sphaeriidae, Copepoda, Isotomidae, Physa,
Umbra limi). Meanwhile, the second axis was driven by a distinction between Station 3
and Station 6, with the intervening elevations ordered numerically between them. As in
the overall ordination, Leptoceridae, Corixidae, and Amphipoda were inﬂuential deep
marsh taxa, and young-of-year Micropterus salmoides, the Branchobdellida (crayﬁsh
parasitoids), and Ephydridae ﬂy larvae were also inﬂuential at low elevations. Station 3

was strongly characterized by Cladocera, which inﬂuenced both axes.

Univariate analysis: Abundance diﬂ'erences among stations, plant zones, and depths
I followed the multivariate analyses with univariate analyses of individual taxa
(Table 1.1). Kruskal-Wallis comparisons of abundances of individal taxa at different

17

elevations tended to conﬁrm correspondance analysis results. The majority of analyses
performed on family-level (and higher) data indicated signiﬁcant inequality of abundances
along the gradient. However, signiﬁcant differences were not found for many individual
genera and less-common families, probably because these occurred at low abundances
(including many zero counts) so statistical tests could not detect diﬂ‘erences that may have
existed. I also used Kruskal-Wallis tests to compare taxa abundances among water depths
and vegetation zones, comparing p-values from these tests to each other and to those from
tests among elevations. While this is not a true statistical test of the relative importance of
environmental variables, lower p-values suggest lower probabilities that random chance
produced apparent animal associations with these variables. For example, Table 1.1
shows that the Anisoptera abundance difference among elevations produced a lower p-
value (p<0.01) than its diﬁ‘erence among vegetation zones or depths (both 0.05<p<0.001),
so the probabity that only random chance produced the elevation diﬂ‘erence was lower
than in the vegetation and depth tests. However, this method is qualitative and weak
compared to parametric variance-partitioning approaches, especially with such inter-
correlated environmental variables, so p-value differences among the three analyses for
any given taxon are only indicators of potential differences among these variables as
causative factors.

Only a few taxa showed strong associations with water depths but not with
elevation or vegetation zones (Table 1.1). These included Caenidae mayﬂies (and
Ephemeroptera in general), which occurred in low numbers at low elevations in May and
proceeded to increase in abundance at progressively higher elevations over time,
remaining consistently most abundant at 40 cm depth (p<0.01). Damselﬂy larva
(Odonata: Coenagrionidae) abundances were greatest at 40-60 cm depths (p<0.05) and
were not signiﬁcantly related to elevation (p>0.1), and only weakly related to the deep
marsh plant zone (0.1>p>0.5). Three caddisﬂy taxa were most associated with water

depth: Oxyethira (Hydroptilidae, 30-50 cm depths), T rianodes (Leptoceridae, 40-50 cm

18

depths), and Polycentropodidae (50 cm depth). The Collembola in general, and
Isotomidae in particular were most associated with the two lowest depths, which was
consistent with their importance in the wet meadow community. Oligochaetes tended to
concentrate in deeper water, with highest abundances at the 60 and 40 cm depths.

Qualitative distinctions between animal-vegetation and animal-elevation
associations were complicated by the fact that any taxon with a high association with a
single station would have a defacto association with a plant zone. Thus, I considered
vegetation to be a more important factor only in cases in which the p-value from the
vegetation test was lower than the value from the elevation test, though this decision rule
is qualitative and somewhat arbitrary. The majority of taxa that appeared to correspond
with wet meadow vegetation included several genera of Dytiscidae beetles, Haliplidae
beetles, various Diptera (e. g. three Ceratopogonidae taxa, Culicidae, unidentiﬁed
Brachycera), Isotomidae springtails, the Phryganeidae caddisﬂy Fabria, and Bithyniidae
snails. Ephemeroptera, some Hemiptera, and Hydroptilidae caddisﬂies were most
associated with the transition zone, and amphipods, damselﬂies (Coenagrionidae), and the
Leptoceridae caddisﬂy T rianodes were primarily found in the deep marsh zone.

Animal taxa could be generally classiﬁed according to four distribution types: wet
meadow specialists, mid-to-lower gradient generalists, deep marsh specialists, and
generalists. Representative gradient proﬁles of each type are shown in Figure 1.7. The
wet meadow specialists were those taxa that were almost exclusively found on the upper
halfof the gradient, especially Stations 1 and 2. Lower gradient generalists tended to
occur broadly among Stations 3 through 6, especially at Stations 4 and 5. Deep marsh
specialists were always concentrated at Station 5 or 6, or both, while generalists occurred
broadly over the gradient. This last group is of interest because it mostly consisted of
higher taxonomic levels. Had identiﬁcation been taken to lower levels, greater gradient

speciﬁcity would likely have been observed.

19

Univariate analysis: Temporal changes in gradient positions

Several taxa apparently migrated up-slope over the summer (e. g. Isopoda, Figure
1.8). Mean gradient positions (weighted averages) for each taxon during each month are
listed in Table 1.2, along with signiﬁcance levels of linear regressions of these positions on
the independent variable "month". Regressions were based on raw numbers per trap eﬁ‘ort
(averaged in Table 1.3), but taxa were not collected on every transect in every month, so
the number of data used in regressions varied widely from taxon to taxon. Table 1.2 only
includes those taxa that occurred during at least two months and on enough transects to
perform regression. Results of regressions of rare taxa should only be accepted
conditionally, and emphasis should be placed on regressions of the more common taxa and
higher taxonomic groupings. I

The regression results suggested three patterns: early wet meadow colonists
which only appeared at high elevations (shallow water), nrigrators which showed a
signiﬁcant trend of decreasing gradient position (increasing elevation) over time, and non-
nrigrators which remained at middle or low elevations (deeper water) through the summer.
The earliest wet meadow colonists were dominated by mites, copepods, certain biting
midge genera (Ceratopogonidae: Serromyia and F orcipomyia), various Cladocera,
Planorbidae snails, and Dytiscidae beetles (each taxon comprising at least 10% of
samples), all of which appeared to migrate ﬁ'om lower elevations. Later in the season, wet
meadow stations were increasingly dominated by Chironomidae, Libellulidae dragonﬂy
larvae, Isopoda, leeches, and Collembola. These groups apparently migrated more slowly
or, in the case of insects, up-slope dispersal was temporally limited to oviposition by ﬂying
adults during a certain period of the summer. Those groups that only appeared in the wet
meadow, showing no evidence of migrating from lower elevations, included Dytiscidae
larvae, Culicidae, Sphaeriidae, Bithyniidae, and Nematoda.

The majority of the remaining taxa were limited to lower elevations (deeper

water), showing little tendency to migrate to higher elevations. Among these were a few

20

taxa whose distributions widened, including higher elevations, but remained centered on
lower elevations (e. g. Caenidae, Leptoceridae, most Hemiptera, Ostracoda). All other
groups were more narrowly associated with mid-elevation stations (e. g. Mesoveliidae,
Aeschnidae, 0ecetis, Naididae) or the low end of the gradient (e. g. Corixidae,
Coenagrionidae, T rianodes, Amphipoda). One or more individuals of most taxa occurred
at most elevations (so presence-absence analysis would be ineffective), suggesting that the

spatial variation with elevation was a true gradient.

Univariate analyses: Microcrustaceans and ﬁsh

Although macrofauna were the primary focus of the study, microcrustaceans
collectively comprised the largest proportions of animals in most samples, so I included
them in the multivariate analyses described above and I also analyzed their distributions
with univariate tests. Certain spatial and temporal trends were apparent among these
animals (Figure 1.9), even though they were not identiﬁed ﬁrrther than the level of
Ostracoda, Copepoda (almost exclusively Cyclopoida), and Cladocera. Copepoda were
early wet meadow colonizers and remained most abundant there throughout the season.
They comprised a large proportion (17-28%) of Station 1 and 2 samples (p<0.05 relative
abundance), showing a strong association with wet meadow zone samples (p<0.05, raw
and relative abundances). They showed a signiﬁcant (p<0.001) up-slope migratory trend
because of their immediate strong presence at newly-ﬂooded wetland elevations. Overall
copepod abundance subtly rose until July, and then began to decline (p<0.05), possibly
because water level had stabilized by then.

Ostracoda were consistently most abundant at low elevations, peaking at Station 5
(p<0.001), though they also showed an up-slope migratory trend (p<0.05) as small
numbers appeared at higher elevations through the season. Because of their high
abundances at Stations 5 and 6, ostracods were strongly associated with the deep marsh

zone (p<0.01). Their signiﬁcant difference among depth classes (p<0.05) was an anomaly

21

resulting from a very high abundance in June at Station 6, and did not represent a
consistent association with the 50 cm depth. In contrast to ostracods, the Cladocera
gradient position was dynamic, showing a signiﬁcant (p<0.05) up-slope migratory trend as
they became more abundant in the wet meadow samples from June to July. A signiﬁcant
increase in abundance over the summer (p<0.01) led to high numbers at Station 3 in
August, which was an important inﬂuence in the August CA ordination, as mentioned
above.

The ﬁsh assemblage varied over the elevation gradient also (Table 1.1). Density
(p<0. 1) and taxa richness were highest at Station 4. Of 86 ﬁsh caught in all months, 38
were caught at Station 4, representing 7 of the 10 species. The most common species in
the wet meadow and the wetland as a whole was mudminnow (Umbra limi) comprising
34% (29 ﬁsh) of all ﬁsh caught, 15 of 18 (83%) caught at Station 2, and the only ﬁsh
caught at Station 1. Iowa darters (Etheostoma exile) were quite common in May,
presumably in the shallows to spawn, while few ﬁsh of any species were caught in June,
and total numbers caught increased in July and August as young-of-year mudminnows,
largemouth bass (Micropterus salmoides), bowﬁn (A mia calva), and yellow perch (Perca
ﬂavescens) became abundant. Brook stickleback were caught in all months, mostly at
Station 4. These ﬁgures do not include a brood of very young bowﬁn (Amia calva ~2-3
cm) that occurred in two June deep marsh traps at high densities (20 and 56 ﬁsh per trap),

or 7 unidentiﬁed larvae caught at Stations 2 and 3 in July.

DISCUSSION

My ﬁrst hypothesis was that ordination of the total data set would result in a
numerical ordering of sampling stations along the ﬁrst axis. Because there are 6! possible
sequences of the 6 stations, the probability of random chance producing the hypothesized
order is 1 in 720, or a probability of ~0.00139. Thus, the ordinations of the total data set

and of the May and June data are consistent with the hypothesis that a community gradient

22

occurs corresponding to the elevation gradient. The fact that ordinations produced similar
results regardless of taxonomic level suggested that the community gradient was a major
phenomenon relating to adaptations in body form found at high taxonomic levels, at least
in some of the taxa sampled. Had all animals been identiﬁed to genus, or at least family,
the gradient structure may have been ﬁrrther enhanced. However, the fact that I used
some relatively high taxonomic levels must be recognized as a limitation on ﬁrll
interpretation of the results.

My second hypothesis, that the community gradient would deteriorate over time,
was supported by the fact that the July and August ordinations did not result in a numeric
ordering of all stations in Dimension 1. Instead, the July ordination suggested a re-
shuﬂling of some members of communities at different elevations. One reason the six
stations were plotted in seemingly random positions is that high-level taxonomic groups
deﬁned communities in July while speciﬁc groups that I identiﬁed to lower levels were
migrating up-slope, homogenizing the communites at diﬂ‘erent elevations. Greater
taxonomic resolution may have revealed a numerical ordination in July, although it would
not change the fact that changes were occuning in the gradient structure as many taxa
shifted along the elevation gradient. Insects may have been particularly inﬂuential in
rearranging ordination plots because they could emerge as adults at low elevations and
oviposit at higher points, skipping the intervening elevations. This could create
community discontinuities as long as pre-emergent larvae and newly-hatched larvae are
present at discontiguous elevations at the same time.

The August ordination contained evidence of the community gradient re-asserting
itself, though in a somewhat different form. The wet meadow community still included
most early (June) constituents, but also included the newer immigrants (e. g. isopods,
Physidae snails) from lower elevations. I had anticipated that communities would
progressively homogenize, cauSing the gradient to break down by August, but this did not

occur, in part because certain taxa apparently did not move up gradient ﬁom their low and

23

mid-elevation positions. The distinction between taxa that migrated and those that
remained "true" to their position on the gradient may be an important piece of evidence in
any attempt to understand what factors caused the community gradient.

Combining the classiﬁcations by distribution type and by gradient position
regression results, I can assign most taxa collected to one of 4 categories: early wet
meadow colonists (dominating wet meadow elevations as soon as they are ﬂooded), wet
meadow immigrants (migrating up-slope from lower elevations), mid-gradient generalists
(occurring broadly, especially across Stations 3 through 5), and low-elevation specialists
(with peak abundances always at Stations 5 and 6). The early wet meadow colonists
included those taxa that only appeared in the wet meadow. Larvae of Dytiscidae beetles
and Culicidae (mosquitoes) appeared later in the season at high elevations, apparently as a
result of mid-season oviposition by adults in the newly-ﬂooded wet meadow. However,
the Sphaeriidae, Bithyniidae, and Nematoda are non-insects that were likely to have been
present in the soil/detritus matrix prior to ﬂooding. It is possible that even when surface
water is not present in the wet meadow, the underlying substrate remains wet enough to
allow these organisms to survive non-ﬂooded periods.

Those taxa that appeared to migrate to higher elevations over time probably used
different mechanisms to do so. Some probably were "true" migrators, in that individuals
crawled (e. g. Isopoda, Planorbidae, Physidae) or swam (e. g. Dytiscidae and Hydrophilidae
beetles, mudminnows, leeches) to higher elevations. Insects, as described above, are
capable of migrating through emergence as adults and subsequent oviposition at new
locations. This mechanism probably explains the gradient position shift of the Libellulidae
dragonﬂies, Chironomidae (midges), and Caenidae mayﬂies. It is also possible that some
animals may have been carried up-slope during the occasional storm surges that can
temporarily raise coastal water levels and batter shorelines with high-energy waves. Such
surges have been observed in the study site (S. Riﬂ‘ell, pers. comm), but whether they

move large numbers of invertebrates shoreward has not been studied. Another

24

complication in interpreting mechanisms of these apparent migrations is the lack of
taxonomic speciﬁcity in this study. Usually-speciose taxa (e.g. Chironomidae) may have
shown gradient position shiﬁs because diﬂ‘erent species occupied diﬂ‘erent positions at
diﬂ‘erent times, but these dynamics would be indistinguishable from true upslope
migrations without more detailed taxonomic work.

Meanwhile, many taxa that should be capable of relatively rapid migration did not
become abundant up-slope. These included fast swimmers such as Corixidae bugs,
Amphipoda, and most ﬁsh species. Also, many insects did not show position shifts,
despite the fact that they were obviously emerging and ovipositing during the summer, as
evidenced by the appearance of early instars in later months. These included most
Trichoptera, the Coenagrionidae, and some Hemiptera. Reasons for these failures to
migrate cannot be determined ﬁ'om these data, but there are several possibilities, including
"active" mechanisms of habitat selection and "passive" mechanisms of habitat exclusion, as
discussed below.

It may seem contradictory that the ordination including all months produced such
an obvious numeric ordering of elevations, while the July and August ordinations did not.
The explanation probably lies in the fact that community compositions at low elevations
remained relatively constant while higher elevations changed. Also, though high-elevation
communities changed, they still remained diﬁ‘erent ﬁ'om low-elevation communities. In
June the wet meadow community was strongly different ﬁom lower elevations, but as it
changed in later months, more taxa were added to the community rather than replacing
early colonists. Some of these appeared to migrate from lower elevations, but some (e. g.
Dytiscidae and Culicidae larvae) appeared for the ﬁrst time in June and August, and only
at high elevations. The result was an accumulation of taxa at higher elevations so overall
taxa richness was highest at high elevations (Figure 1.10).

In contrast to the ordination by elevation, the ordination by depth did not produce

a numerical ordering of communities at adjacent depths, suggesting that entire

25

assemblages did not move up-slope as water level rose even though many individual taxa
did. Water depth and substrate elevation are strongly correlated so distinguishing the
effects of these two factors on fauna is diﬂicult. Variance-partitioning methods could not
be used because the data could not be made to meet the assumptions of normality and
homoscedasticity. The method used here, comparing two ordinations, has the advantages
of being relatively assumption-ﬂee, and of comparing entire communities rather than

testing taxa individually.

Possible causal factors

. It is unlikely that a single causal factor was most responsible for all the animal
distributions observed in this study. Distinguishing among the various candidate causal
factors is further complicated by the inter-correlation of so many coastal wetland
environmental parameters. The data collected in this study did not allow for a robust
statistical distinction among causal factors, and I do not suggest that the analyses
presented here do any more than provide some clues to the relative inﬂuences of elevation,
vegetation, and depth on animal distributions.

My initial hypotheses were driven by an expectation that ﬂooding regime variation
would be directly responsible for most faunal variation, causing animals to sort themselves
into a temporary wetland community and a permanent wetland community at opposite
ends of the gradient. The gradation of community composition between these two
extremes was expected to comprise the community gradient that was revealed through
correspondance analysis. This expectation is supported by the fact that the majority of
taxa most associated with the wet meadow have been identiﬁed as common residents of
temporary waters in other studies (see Introduction for review): snails, beetles, mites,
midges (Chironomidae), mosquitoes, and copepods, Those groups generally considered
to be excluded from temporary waters were generally not common in the wet meadow in

this study: mayﬂies, caddisﬂies, amphipods, and most dragonﬂy/damselﬂy (Odonata) taxa.

26

Despite the above suggestion that drought-tolerance is an important determinant of
coastal wet meadow fauna, it is also possible that water quality factors were very
inﬂuential, as suggested by Cardinale et al. (1998). Dissolved oxygen is one such factor
that may have been important. The dominance of the wetland ﬁsh community by
mudminnows suggests that hypoxia is common in the shallows because these ﬁsh can
avoid oxygen stress by gulping air at the surface. Many invertebrates are affected
adversely by low dissolved oxygen levels (Merritt and Cummins 1996), and a large
proportion of the wet meadow inhabitants in this study were taxa equipped with air-
breathing mechanisms (pulrnonate snails, diving beetles, mosquitoes) or were taxa that are
known to have species with hypoxia tolerance (Chironomidae). It is also possible that the
variation in ﬁsh species over the gradient, possibly caused by dissolved oxygen variation,
caused a gradient of constantly-varying predation pressures. Invertebrate predators also
varied over the gradient, so the complex spatial and temporal variation in the assemblage
of predators could be important inﬂuences on the remainder of the invertebrate
community.

Plant community types are used in many wetland delineation and classiﬁcation
schemes (e. g. Cowardin et al. 1979), and have been the focus of several studies of wetland
invertebrates (reviewed in Batzer and Wissinger 1996). I initially hypothesized that
vegetation zonation would not be an important inﬂuence on invertebrate communities
because it may be more important as a human-perceived classiﬁcation than an
environmental factor inﬂuencing aquatic animals. This is not to say that plant community
composition is unimportant to animals, but that the division of the wetland continuum into
zones is not likely to be recognized by animal communities. In this case, I applied this
hypothesis speciﬁcally to my division of the wetland vegetation into three zones, based
primarily on the dominance of cattails in the middle of the transects, with very different
communities on either side. Transect 4 lacked cattails so Stations 3 and 4 were classiﬁed

as wet meadow and deep marsh respectively. If animals responded strongly to these plant

27

community diﬁ‘erences, or to their associated detritus diﬂ‘erences, I expected that sampling
stations in similar vegetation zones would occur relatively near each other in ordination
plots. This did not occur in any plots, except that Stations 3 and 4 were relatively close to
each other in the ordination of the entire data set. This may indicate high similarity among
animal communities in the transition zone, or it may simply reﬂect the fact that these two
stations were horizontally closer to each other than to other stations because of the
steeper substrate slope in the transition zone.

Despite the lack of strong evidence for animal community association with
vegetation zones, several individual taxa were strongly associated with plant zones. In at
least one case this may be explained by food preference: the Leptoceridae caddisﬂy
T rianodes feeds on soft live plant tissues, so this genus would be expected to remain in the
deep marsh zone where submergent and ﬂoating-leaved plants were abundant. In most
other cases, possible reasons are less apparent, yet vegetation zonation cannot be
dismissed as an important environmental factor for at least some invertebrates. The taxa
most likely to have been inﬂuenced by vegetation were probably those that were
signiﬁcantly associated with vegetation zones and also did not appear to migrate across
zone boundaries. Some of these (e. g. Haliplidae, Corixidae, Amphipoda, Sphaeriidae,
Gastropoda) may have been inﬂuenced by vegetation-associated food variables such as
detritus type and periphyton availability, but others (e. g. Dytiscidae, Ceratopogonidae,
Coenagrionidae) were predators, so the non-animal food base was not a likely direct
inﬂuence on them.

Some animals may have been inﬂuenced by plant structural complexity.
Complexity of submergent plant surfaces directly affects invertebrate colonists (Krecker
1939), but emergent plant stems rarely support high abundances of epiphytic animals
(Batzer and “Wissinger 1996). However, the degree of habitat complexity created by
emergent plants can aﬁ‘ect invertebrate abundances (Batzer and Resh 1992) and provide

refuge for ﬁshes (Werner and Hall 198 8). Further, invertebrates can beneﬁt from reduced

28

ﬁsh predation emciency in areas of high plant structural complexity (Gilinsky 1984). I did
not measure structural complexity, but related sampling (Gathman and Keas 1999)
indicated that emergent stem density decreased, and submergent plant coverage increased,
with lower elevations (greater depth), so variations in spatial structure occurred along the
gradient and may have inﬂuenced animals. Nonetheless, the ready migration and broad
distributions across zone boundaries observed in so many taxa in this study mitigate
against any suggestion of a strong role of vegetation zonation in deﬁning the community
gradient.

I used multivariate analysis to deterrrrine whether entire communities were strongly
associated with depth ranges. The fact that such associations were not evident was not
because animals did not follow rising water up-slope, but because various groups
responded in different ways to the rising water. As a result, the situation was more
complicated than my initial hypotheses anticipated. Communities changed over time as
some taxa migrated and others expanded their ranges, all doing-so at diﬁ‘erent rates against
the backdrop of the more spatially-stable groups. Depth was not a good indicator of
community composition, but the increasing area ﬂooded over the season that resulted
ﬁom water level rise allowed many animals to move to new habitats. The failure of some
taxa to move up-slope may have resulted from any of the habitat variables suggested
above, or may have resulted from taxonomically-diﬂ‘erential mortality along the elevation
gradient. The fact that abundances of predators (e. g. ﬁshes, Odonata) varied spatially
suggested that predation pressures may have varied over the gradient, possibly limiting the
abilities of some taxa to occupy some habitats.

This movement, which appeared to diﬂ‘er among taxa, was probably a very
important factor determining community composition along the elevation gradient. For
example, some taxa may have been able to survive the de-watered period in the substrate,
while others were able to migrate into the wet meadow when it became ﬂooded, though

some did so more quickly than others. Still other taxa did not appear to migrate,

29

remaining at the low-elevation end of the gradient. However, factors other than direct
migration and dispersal movements may have produced the appearance of movement, or
lack thereof. For example, some low-elevation taxa may have engaged in up-slope range-
expansion movements, but encountered high mortality in higher-elevation habitats,
precluding their successful establishment there. Whatever the causes, this difference in
apparent migration and dispersal responses to changing water level was an important
determinant of community composition at successively higher points on the gradient.
Despite the seasonal changes resulting from migrations, the correspondance
analysis results clearly indicated a gradual variation in community composition along the
wetland elevation gradient in the study site. The numerical ordination of sampling stations
along the ﬁrst axis indicated that the main component of community variation in this
wetland site was strongly related to substrate elevation, either directly or indirectly. Thus,
the diﬂ‘erent environmental conditions at different elevations appeared to create a spatial
distinction in habitats that was important to animals in the wetland. What could not be
clearly determined from this study is what proximate factors were responsible for the
variation observed. However, the observation of the pattern is a necessary ﬁrst step

toward understanding coastal wetland animal community structure.

30

Table 1.1. Signiﬁcant diﬁ’erences in abundances of taxa among sample stations, plant

zones, and depths; monthly and all months combined (Kruskal-Wallis test;
* 0.1>p>0.05, ** 0.05>p>0.01, *** p<0.01).

 

TAXON \

Independent variable:
Month: ALL May Jun

Station
Jul Aug

Plant zone

ALL May Jun Jul

Aug

Lem

ALL

 

Coleoptera
Dytiscidae
Dytiscidae adult
Laccophilus
Ilybius
Rhantus
Hygrotus
Agabus
Uvarus
Hydroporus
Dytiscus
Coptotomus
unident. Dytisc.
Dytiscidae larvae
Dytiscus
H ydroporus
Agabus
Gyrinidae -
Haliplidae -
Hydrophilidae
T ropisternus
Laccobius
Helochares
Curculionidae -
Scirtidae
Diptera
Ceratopogonidae
Serromyia
F orcipomyia
Atrichopogon

Gyrinus
Haliplus

Lixus

Bezzia/Palpomyia

Culicoides
unident Cerat
Chironomidae
Ephydridae
Sciomyzidae
Culicidae
Culex
Anopheles
Dixidae
Brachycera
Ephemeroptera
Baetidae
Metropodidae -

Siphloplecton

***
***
***

***

***

***

**

***
***

***

***
*** *
***

**

**

**

***

***

***

***

***

***
***

***
***

***

**
**

***

**

**

***
** ***
** ***

***

**
**

**

**

**

**
**
*

**

***
***
***

**

***
**

 

31

Table 1.1 (cont'd).

 

Independent variable:
Month: ALL May Jun Jul

TAXON \

Ephemeroptera - cont'd
Caenidae - Caenis **
unident Ephem.

Hemiptera
Belostomatidae
(knbddae ***
Corix. adult ***

Hesperocorixa
Sigara
Palmacorixa ***
unident. Corix.
Corix. larvae
Saldidae
Mesoveliidae
Nepidae
Veliidae
unident Hemipt.

Lepidovtera

Pyralidae
Parapoynx
other Pyral.

other Lepidopt.

Ckknuna ***
AUﬂSOpuna ***
Aeshnidae
Libellulidae ***

Leucorrhinia **
Sympetrum
immature ***

Libel.

Coenagrionidae
T elebasis

Homoptera *

Trichoptera
Hydroptilidae

Oxyethira
immature Hydropt.

Leptoceridae
Trianodes *
0ecetis
Mystacides

Limnephilidae

Phryganeidae

Polycentropodidae

Collembola **
Sminthuridae

Station

**

***

**

**

***

***

***

**

Plasma;

*
**

*** ***
*** * *

**

** **

**

***

*** ** *

*** *
** ***
***
***

**
** **

**

** ** *

Depth

Aug ALL May Jun Jul AIL ALL

***

**

**

**

**

**

**

**

**
**

Table 1.1 (cont'd).

 

Independent variable:
Month: ALL May Jun Jul

TAXON \

m

Plant zone

Alli ALL May Jun Jul Aug

 

Collembola - cont'd

Isotomidae
Poduridae
other Collembola
Hymenoptera
Thysanoptera
Amphipoda
Gammaridae
Hyalellidae
Crangonyctidae
Asellidae
Cambaridae
Argulus
Hirudinea
Oligochaeta
Lumbriculidae
Naididae
Tubiﬁcidae
other Oligo.
Branchobdellida
Mollusca
Sphaeriidae
Gastropoda
Bithyniidae
Lymnaeidae
Physidae
Planorbidae
Suocineidae
ACARI
NEMATODA
TURBELLARIA
CLADOCERA
OSTRACODA
COPEPODA
HYDRIDAE
FISH
Etheostoma exile
Ameiurus nebulosus
Umbra liml'
Culaea inconstans
Amia calva
Pungiti us pungiti us
Gasterosteus aculeatus
Cwﬁnidac
Micropterus salmoides
Percaﬂavescens

Orconectes

**

*III

**

#331!

***
**
*1!

**

1H“?

***

**

*1!

t

**

**

*Ihk

**

**

**

13*

*4:

*

**

**

In"!

*Ihk
***
**
***

**

***

***
*3?

**

***

**

***

*1!

in?
** It
**

*** *

*Il‘
**

**

***

**

**

 

33

Table 1.1 (cont'd).

 

 

Independent variable: Station Plant zone Depth
TAXON \ Month: ALL May Jun Jul Aug ALL May Jun Jul Aug ALL
ﬁsh larvae * *
tadpole
Bufo tadpole
Rana tadpole **
unknown tadpole * **
# taxa per mp 11!!ka IN: *1: *1“!
Total abundance * **

 

34

Table 1.2. Mean gradient positions (weighted averages) of each taxon over all months
and for each month, and results from regressions of positions against month.

 

 

Mean Gradient Position Remsion
Taxon All Months May Jun Jul Aug slope p—value n
Coleoptera 2.32 4.33 2.98 1.94 2.11 -0.717 0.001 18
Dytiscidae 2.21 4.00 2.90 1.96 1.98 -0.625 0.004 17
Dytiscidae adult 2.22 4.00 2.90 1.99 2.05 0598 0.006 17
Ilybr‘us 2.19 n.a. 3.00 2.23 2.21 -0.360 0.297 13
Rhantus 2.19 n.a. 2.67 1.67 n.a. -1.000 0.101 6
Hygrotus 2.25 n.a. 3.50 1.50 1.50 -1.000 0.076 6
Agabus 1.41 n.a. 3.00 1.33 1.00 -l.000 0.111 5
Dytiscidae larvae 2.08 n.a. n.a. 1.80 2.30 0.500 0.300 4
Haliplidae 2.88 n.a. 3.50 1.00 3.00 -0.250 0.879 3
Hydrophilidae 3.00 5.00 n.a. 3.00 1.75 -1.083 0.001 5
Diptera 3.22 4.58 3.66 2.67 3.32 -0.456 0.013 19
Ceratopogonidae 3.19 4.50 3.07 3.49 2.96 -0.364 0.212 13
Atrichopogon 1.81 n.a. 3.00 1.50 n.a. -1.500 0.330 3
Chironomidae 3.26 4.33 3.83 2.70 3.37 -0.373 0.053 19
Culicidae 1.88 n.a. n.a. 2.17 2.00 -0. 167 0.868 4
Brachycera 2.81 n.a. 3.89 3.13 1.83 -1.028 0.124 9
Ephemeroptera 3.73 5.17 5.00 4.50 3.52 -0.552 <0.001 13
Caenidae 3.66 5.00 5.00 4.83 3.52 -0.616 0.008 11
Hemiptera 4.43 5.56 4.17 4.32 3.83 -0.479 0.002 19
Belostomatidae 4.33 n.a. n.a. 6.00 3.50 -2-.500 0.812 3
Corixidae 4.76 5.56 4.32 4.97 4.63 -0. 191 0.287 18
Corixidae adult 5.08 5.56 4.38 5.00 6.00 0.151 0.493 15
Corixidae larvae 4.38 5.00 4.08 4.58 4.26 -0.074 0.760 12
Mesoveliidae 3.46 n.a. 3.75 2.30 n.a. -1.540 0.154 6
Nepidae 3.69 n.a. 4.00 n.a. 3.50 -O.250 0.667 3
Veliidae 2.33 n.a. n.a. 3.00 1.67 -1.330 0.219 5
Odonata 3.26 4.83 4.25 3.33 2.95 -0.654 0.005 17
Anisoptera 3.02 4.83 4.00 3.05 2.49 -0.790 0.001 16
Aeshnidae 4.22 4.50 n.a. 3.33 n.a. -0.583 0.133 5
Libellulidae 2.88 5.00 3.50 2.89 2.49 -0.765 0.005 14
Coenagrionidae 4.41 n.a. 5.00 4.83 4.54 -0.250 0.567 9
Trichoptera 4.46 4.17 4.80 4.33 3.98 -0. 198 0.326 15
Leptoceridae 5.01 5.00 5.03 4.00 3.87 -0.500 0.079 10
Trianodes 5.32 5.00 5.42 4.00 6.00 0.086 0.796 8
0ecetis 4.60 n.a. 5.00 n.a. 3.50 -0.750 0.312 4
Limnephilidae 3.48 4.00 2.50 3.00 6.00 0.731 0.337 5
Collembola 2.93 4.25 3.39 2.08 1.25 -1.054 0.001 12
Sm'mthuridae 3.55 4.25 4.00 2.50 n.a. -0.897 0.118 6
Isotomidae 2.76 4.00 3.00 3.50 1.25 -0.798 0.042 7
Hymenoptera 3.80 n.a. n.a. 4.50 2.00 -2.500 0.312 3
Amphipoda 4.47 4.63 4.32 4.21 4.59 -0.025 0.844 19
Hyalellidae 4.48 4.63 4.32 4.17 4.73 0.010 0.943 19
Asellidae 2.41 5.00 4.00 2.88 2.07 -0.972 0.096 9
_C_ambaridae 4.76 n.a. 4.50 n.a. 6.00 0.750 0.333 3

 

35

 

Table 1.2 (cont'd).

 

 

Mean Gradient Position Regression

Taxon All Months May Jun Jul Aug slope p-value n
Naididae 4.08 4.83 3.50 3.87 3.70 -0.278 0.275 16
Mollusca 3.10 4.67 4.07 3.05 2.64 -0.712 <0.001 18
Sphaeriidae 1.63 na na 1.37 2.35 0.483 0.266 7
Gastropoda 3.35 4.67 4.07 3.31 2.73 -0.660 <0.001 18
Bithyniidae 1.81 n.a. 2.00 n.a. 1.67 -0.167 0.667 4
Lymnaeidae 3.18 5.00 3.30 2.98 3.33 -0.255 0.587 14
Physidae 3.95 4.50 4.73 3.68 3.16 -0.592 0.040 16
Planorbidae 3.02 5.00 4.03 3.00 2.89 -0.647 0.003 16
Suocineidae 3.71 n.a. n.a. 3.67 4.00 0.330 0.800 4
Acari 3.81 4.77 3.93 2.88 3.94 -0.356 0.066 20
Nematoda 2.75 4.00 2.58 1.40 2.36 -0.438 0.225 11
Turbellaria 3.50 n.a. n.a. 2.00 3.11 1.111 0.652 4
Claodooera 3.72 5.00 4.20 3.62 3.58 -0.456 0.025 18
Ostracoda 4.71 4.88 4.68 4.70 3.81 0.321 0.021 20
Copepoda 3.54 4.99 3.75 3.42 2.97 -0.640 <0.001 20
Hydridae 2.75 n.a. n.a. 3.81 2.00 -l.806 0.206 7
All Fish 4.39 4.13 4.55 3.98 3.25 -0.326 0.075 18
Etheostoma exile 4.23 4.14 4.00 n.a. n.a. -0. 143 0.667 3
Umbra limi 2.90 4.00 n.a. 3.92 2.54 -0.592 0.117 9
Culaea inconstans 3.67 4.00 3.00 4.50 3.00 0.008 0.984 8
Amia calva 5.16 n.a. 4.90 5.00 5.00 0.062 0.922 6
Micropterus salmoides 4.33 n.a. n.a. 4.00 4.25 0.250 0.904 5

 

36

 

 

ad ad ad ad ad ad ad ad ad ad ad ad ad ad ad ad Nd ad ad ad 6.526th
3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 MN 3 3 N._ 333228
3 3 3 3_ 3 N.“ 3 3 3 3 3 N.: 2 3 3 3 3 3 N._ 3 seen
3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 338m
3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3: $33325
3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 sets»:
3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 333883
3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 sesame:
3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 gaineam
ed od od ad ad ad o.o ad ad ad ad Nd cd od No No ed o.o ad ed 3358: - 323:5.—
3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 22.2.6 -3255
ad ad No No ed Nd ed :6 ad ad ad ad ad ad ad od ad ad od od 25%:
od od od No ed 06 ad ad ad ad o.o od od ed o.o od o.o ad ad o.c 223933
ed ad ad ad ad ad o.o o.o o.c No .5 9o ed o.c ad ad ad c.o ed :5 363:3
od Qo Nd ed to Nd o.o o.o od No .5 9o 3. ed ad ad od o.o od od 3.3. 3283.5
cd od ad 3. od o.o od o.o od od Nd ad ad ed do od No ed o.o to 8:5 352::
ad ad ad ad ad Nd ad ad od ad ad ad o.o od od ad ad ad ad ad 352230
od od od ad ad ad ad od ad ad ed .3 cd ed ed ed ed ed od ed ESSAQ
od ad ad od ed ad ad o.o ad ad Nd ed o.o od od ad ad ad od o.o 453835
ad ad ad ad ad ad ad ad No ed No ad ad ad ad ad ad ad o.o ad .2295
ed od o.o ad ad Nd o.o ad ad o.o No 3 ed od od do o.o ad ad ad azcemw
3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 323:
ad ad od ad ad ed o.o od o.o ed to Nd od od ed mac Yo 06 ed ed 3.523:
Nd Na «d to 3 ad ad No to v._ 3. w.~ ad ad No 2. ad ad ad ad 3.33:
ad ad ad od o6 ad ad ad od o6 od o6 ed od No ad ad Qo ed to 3.52383
3 3 3 3 N._ I 3 3 3 E 3 3 3 3 3 N._ 3 3 3 3 338335
3 3 3 3 S E 3 3 3 3 3 N.“ 3 3 S N._ E 3 3 3 33.5
3 3 3 2 N.N M: 3 3 3 3 3 3 3 3 N._ I Z 3 3 3 2238.8
8 3n 2. 8 3N 2 8 an 2. 3m 3N 2 3n 2. 9.. 3N e 8 3N 2 ”Eamo / 205:.
e n 4 m N _ e n 4 N N _ a n a. N. N e n .. ”29.2.5
elamli 3|. mag :12. ”5202

 

douse 3:953 28 5:08 .3 83 be 828233 :32 .m._ 033.

37

 

 

 

3 3 3 3 3 3 3 3 3 3 3 3 3 .3 3 3 3 3 3 3 23.3
3 3 N._ 3 3 3 3 3 3 3 3 3 3 3 Na 3 3 3 3 3 853:8
3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 230.283
3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 33833
3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 N._ 3 3 83m
3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 N._ 3 3 3:823:
3 3 3 3 3 3 3 3 3 3 3 3 2 3 3 3 3 3 3 3 =§.xnoo
N._ 3 N._ 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 08380
3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 833283
N._ 3 3 3 3 3 3 3 f 3 3 3 3.. E 3 3 3 3 3 3 822:5:
3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3.239323
3 N._ 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 2&5 -335
3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 228333 $328352
3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 823m
3 N._ 3 3 3 3 3.. S 3 3 3 3 3 3 3 3 3 3 N._ 3 $2328.23
3 3 3 3 3 3 3 3 3 3 3 N._ 3 3 3 3 3 3 3 3 88.285
3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 82x5
3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 $333
3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 ES
3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 36:5
3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 33:53
3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3323
3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 2. 3 3 3 3 332.820
3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 N._ 38352::
3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 838:5
3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 .aamENsm
3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 83333.
3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3538a
8 on 8 em 8 2 8 cm 3. 33 GN 2 cm 8 cm 3N 2 8 ON 2 ”Emma , 292...
om43N_ememN_ememNemvnzoFEu
3 3 34 3 “5.202

 

 

.6383 3 23.

38

 

 

3 3 3 3 3 3 3 3 3 3 3 3 3 S 3 3 3 3 3 3 .0882:
3 3 3 3 3 3 3 3 3 3 3 3 3 S 3 3 3 3 3 3 085883
3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3382:3825
3 3.. 3 3 3 3 3 3 3 3. 3 3 3 3 3 3 3 3 3 3 E2536
3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 8258.3:
3 M: 3 3 3 3 3 3 3 3 3 3 3 2 3 3 3 3 3 3 8232.05.
3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 .2233:
3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 38802233
3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 2883
3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 82:88:80
3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 2823.52.83
3 3 3 Z 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 2.832382%
3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 53835,
3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 253.2883
3 3 3 Z 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 83:23:
3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 82532
3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 2282.2
3 3 Z 3 3 3 3 f 3 3 3 3 3 3 3 3 3 3 3 3 2825
3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3231285
3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 88.328580
3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 Erase
3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 888.5
3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 825$
3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 223233
3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3228.882:
3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 82=o>
3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 2282
3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 Z 3 3 3 3 082382
8 cm 8 on on 2 8 cm 8 cm 8 2 on 8 on 8 2 cm 33 2 ”Emma / 2000:.
onemn_envmﬁ_enemaoneuzoﬁﬁm
3 3|. 33. 8.2 ”5.202

 

 

.6880 S 282

39

 

ad ed 04 ad ad od od Nd ed ad ad ad ad ad ad od od od 06 ad mzmuwm§Wuuxmm

 

3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 382323322:
3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 263.5
3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3838:3838
3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 Samziuszmai
3 3 3 3 3 3 3 3 3 3 3 3 3 w: 3 3 3 3 3 3 383:3
3 3 3 3 3 3_ 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3385836
3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 £335
3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3.232332%
3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 32332833
3 3 3 2 3 3 3 3 2 3 S 3 3. E. 3 3.. 3 3 3 3 Him
3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 26:3:
3 3 3 3 3. 3.2 3. 3;: 3 3 w: 3_ 3 3 3 3 3 3 3 3 38325
3 3 3_ 3 3 3 3.2 3 3 3 3 3 3.: .5. 33 3.: 3 3 3_ 3 «88:6
3: 3 3.: 3m 3 3 3.: 3.3 3. 33 3.: N: 3. 2 3 3 3. 3 3 3 80836
3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 532.3
3 3 3 3 3 3 3 3 3 3 3 2 3 3 3 Z 3 3 3 3 €25.52
2 3 3 3 M: S 3 3 3 3 3 3 3 3 3 3 3 3 3 3 E3
3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 820583
3 3 Z 3 3 3 3 3 3 3 E S 3 3 3 3 3 3 3 3 $2885
3 3 3 3 3 3.. Z 3 3 3 3 3 3 Z 3 3 3 3 3 3 8233
3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 I S 3 3 3 332:3
3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 oéziém
3 3 3 3 3 3 3 3 3 M: 3 3 .2 3. 3 3 S 3 3 3 882.3
3 3 3 3 3 3 3 3 3 3 2 3 3 3 3 3 3 3 3 3 825.2%
3 3 3 3 3 3 3 3 3 3 3 3 S 3 3 3 3 3 3 3 83o:
3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 §__oBoESEm
8 an S cm on 2 8 on 8 on 8 2 on 3 an 3.... 2 on 8 2 “Emma / 2929
cn¢m~_onvm~_eWVMNoneuzoﬁﬁm
3:4: 3 3 3 ”5.202

 

 

.3253 2 03.;

4O

. . 14
.- -h-«-‘

- I‘

 

 

3 3 3 3 3 3 3 3 9.. 3 3 N. 3 3 3 3 3 3 3 3 3.0.2....
3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 8202...:
3 3 3 3 N. N. 3 3 3 N. 3.. 2. 3 3 3 3 3 3 3 3 08.25,.
3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 32.835...
3 3 3 3 N. a. 3 3 3 3.. 2. 3 3 3 3 3 3 3 3 3 502.835
3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 822...:
3 3 3 3 N. 3 3 3 3.. 3 3 3 3 3 3 3 3 3 3 3 332
3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 8.8285 082258
3 3 3 3 Z 3.. 3 3 3 3 3 3 3 3 3 3 3 3 3 3 2.2.32
3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 82.0.8320
3, 3 3 o. 3 3 3.. N. Z 3 o. 3 3 3.. I 3 3 3 3 3... 82:29...
3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 22.2.58
3 3 3 .... 3 3 m. 2. 3 3 3.. 3 3 o. E 3 3 3 3 3... 2.2.23...
3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 823...». 2339.5
3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 82852:
3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 2385......
3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 2822.828
3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 2.2.2.9.
3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 2.2228.
3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 2.2.2.225
3 3 3 3 3 3 3 3 3 3 N. 3 3 3 3 3 3 3 3 N. 2822.8
3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 ...2...ea§e.
3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 022.338.5059.
3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 22053.5.
3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 32.283:
3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 $3253....
3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 32923..
3 3 3 3 3 3 3 3 3 3 3 3 3.. 3 3 3 3 3 3 3 2.88
8 cm 8 on 8 c. 8 an 3. cm 8 c. an 2. cm 8 c. on 8 3. ”Emma , 20x3
emvm~.cn¢m~.enemmeneuzo...‘<..m
.223. 31.4 2... 3:2 H5.202

 

......=8. m. 2......

41

 

 

 

o5. N.mm vdw wdw Nam 93 New dawn 04m adv v.5 ado adﬂ mew OS. t: v.2 eda 3:. adv conga—.5353.
92:2. 88;
N.: N.N_ ad— aé. ed. N.m_ N.N_ a.N_ ww— v.N_ N.m_ N.m_ ad ad N.N_ Nd ed a.m ad ad
ad ad ad ad ad vd ad ad ad ad ad ad ad ad ad ad ad ad ad ad :36 «52
ad ad ad ad ad ad ad ad ad ad ad Nd ad ad ad ad ad ad ad ad .16“ oazm
ad ad ad ad ad ad ad ad ad ad ad ad 3% ad ad ad ad ad ad ad 0.038 555?:
ad ad ad ad ad ad ad ad Nd ad ad ad ad ad md ed ad ad ad ad 2268 «:5.
ad ad ad ad ad ad ad ad ad ad ad ad ad nd .2 ad ad adw ad ad Bananaoaam
ad ad ad ad ad ad ad ad Nd ad ad ad 3% md a._ vd ad adw ad ad 2268
ad ad ad ad ad ad ad Nd ad ad ad ad ad ad ad ad ad ad ad ad 82.:
ac am av an aN 2 do an ac am aN a_ an ac am aN a_ an aN a_ ”Ema / 299E.
a n v n N _ a n v m N _ a n v m N a n v H29.55%
la: .2: 5|. 0.3 32 ”5.202

 

.5253 2 033

42

 

   

Mackinac Bay

 

 

 

 

 

Figure 1.1. Study site location

43

 

Water

 

  
 

 

 

 

 

 

Levels
_____ _ ‘_. July-August
----- __ June
Sample 1 2 —— May
Station: 3
\
Vege-
tation Wet Cattail Bulrush-
Z0“: Meadow Water Lily

 

 

Figure 1.2. Schematic cross-section through a sample transect, showing vegetation
zonation and monthly water levels in 1996 at Mackinac Bay.

 

 

 

2Km

 

 

 

< »
M-1 34
\ample
transects
N Mackinac
Bay

w+e %
S E;
m

Figure 1.3. Overhead view of Mackinac Bay and transect placement.

 

 

 

45

 

 

 

 

 

Sphaeriidae
1 _.
Hydridae 1
Brachycem
Ascllidae Otlicidae
N Corixidae . .
z 6 Naididae Dynw‘i“
Copepoda .
Q Isotomidae
U) - Chironomidae
Z polymmmm'd“ 2 Planorbidae
E 0 — d“ Ostracoda Nematoda
Hyalellidae 4 3 Cladocera
5 Turbellaria
Coenagrionidae Mesoveliidae,
Succincidac
Aeschnidae
-1 I T
-1 DINIENSION l 1

Figure 1.4. Correspondance analysis ordination (representing >69% of total inertia) of
invertebrate communities at six elevation stations during summer (May-August), 1996,
including positions of all taxa with combined two-dimensional squared-cosine values
greater than 0.5.

46

 

 

 

 

 

N
z
O
:75
z 20
E 0 “ 60
a
30 4o
50
'1 l T l
-1 DIMENSION 1 1

Figure 1.5. Correspondance analysis ordination (representing >71% of total inertia) of
invertebrate communities at six water depths during summer (May-August), 1996.

47

 

 

 

 

 

 

 

 

 

 

I I
h mMay : 6 . b.June :
I I
= 3.
I
: 5
4 3 i
I I
IHIIII.IIIIIIIII?IIIIIIIIIIIIIIIII IHIIIIIIIIIIII:IIIIIIIIIIIIIIIIIII‘
: 3
5 5 6
i .
I I
N P : I : 2
I I
E 4 l l + I j l
'53
-1 0 l -l 0 l 2
t% _ a
z - -
C3 .. c.July : . d.August 6

A

6

IIIIII
m IIIIII

S

‘HIIIIIIIIIIIIIII‘III 3 IIIIIIIIIIII HIIIIIIIIIII 4 IIII 2 tIIIIIIIIId

 

 

 

 

 

 

I I
I I
2 : 3 :
: 1 : 1
I I
I I
_ i h- 3
I I
J I J 1 j l
-l O l -l 0 l
DIMENSION 1

Figure 1.6. Correspondance analysis ordinations of invertebrate communities at six
elevations during each of four months (May-August), 1996.

48

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Dytiscidae: Ilybius 5 ‘ Caenidae: Caenis
4 "' 4 -
3 -
2 -
2 ..
a.
i3 1 '
m
E
:9 0 I I I I ' 0 I I r I I r
.2.
'2 2 Le . . . 10 . .
.... ptocendae. T nanodes Chironomidae
:4:
ii
0
E
1 - 5 -
0‘ I Iﬁ I 0 I I I I I I
1 2 3 4 5 6 1 2 3 4 5 6
Wet ' Deep Wet Deep
<——> <———>
Meadow Marsh Meadow Marsh
Sampling Station

Figure 1.7. Gradient distribution proﬁles of representative high-, mid-, and low-
gradient associated genera (l/ybius, Caenis, and Trianodes, respectively), and a
higher taxon (Chironomidae) with a more general distribution, during summer
(May through August), 1996 (means with standard errors).

49

 

1.5

0.5

1.5

Mean number of isopods/trapt

0.5

 

 

 

 

 

 

 

 

 

 

June

 

 

 

 

 

May
I I I I I f
July
I l l I" I
l 2 3 4 5 6
Wet Deep
<-————>
Meadow Marsh

Sample Station

 

 

 

 

 

 

 

 

 

I f
1 2 3 4 5 6
Wet Deep

Figure 1.8. Comparison of gradient distributions of Isopoda: Caecidotea during
each month (May through August) of summer, 1996 (means with standard

errors).

50

 

 

 

0.3

 

 

 

 

 

 

 

 

 

 

 

 

g. Copepoda
; 0.2 L ---------
E
'8 Cladocera
D
.2.
E
g Ostracoda
g 0.1 h" ______
2
0-0 I I I I I I
l 2 3 4 5 6
Sample Station
Wet ‘ > Deep
Meadow Marsh

Figure 1.9. Gradientdistribution profiles of microcrustaceans during summer
(May through August), 1996 (means with standard errors).

51

 

15

 

 

 

Mean taxa richness/trap
'5
I

 

 

 

 

 

 

10 '-
9 I T I I I
l 2 3 4 5 6
W Sample Station Dee
et I p
Meadow Marsh

Figure 1.10. Variation in taxa richness (means and standard errors) across the elevation
gradient over all months (May through August) of summer, 1996.

52

 

Chapter 2

IMPACTS OF HIGH WATER LEVELS ON THE BIOTA OF A LAKE HURON
COASTAL WETLAND: RESULTS OF A NATURAL EXPERIMENT

INTRODUCTION

Great Lakes water levels ﬂuctuate on several time scales, including fairly
predictable seasonal cycles, and erratic long-term cycles that are determined by regional '
climatic variation. On average, the level of Lake Huron rises approximately 25 cm from
midwinter to midsummer each year, falling again through autumn and early winter (Prince
and Burton 1996). This cycle, combined with gentle substrate slopes, creates a gradient of
hydrologic conditions in coastal wetlands: from permanently-ﬂooded deep emergent
marsh to progressively higher zones that are ﬂooded for shorter durations as the water line
moves slowly up, and then down, the slope (Krieger 1992, also see Figure 2.1). Animal
communities gradually vary in composition along this gradient (Chapter 1), possibly
because of diﬁ‘erences among taxonomic groups' adaptations to temporary ﬂooding.
Drought-resistant taxa appear to comprise the majority of the communities at higher
elevations, while those groups without such adaptations are progressively more abundant
at lower elevations. Further, some taxa migrate up the gradient as the duration of
seasonal ﬂooding lengthens, though some conspicuously do not.

Determining the causes of animal community variation along the elevation gradient
is diﬁcult because most environmental factors that may aﬂect aquatic animals co-vary
with the gradient. While variation in ﬂooding regime is one likely cause of fauna]
variation, correlated variables such as plant community composition (Voights 1976,
Zedler 1981), detritus composition and quality (Barlocher et al. 1978, Nelson et al.
1990), water quality (Cardinale et al. 1997), and predator assemblages (W ellbom et al.

1996) may also affect ﬁsh and aquatic invertebrates. The inﬂuences of some of these

53

factors may be distinguishable through controlled experimentation, but the effects of other
factors are more difﬁcult to separate from each other. .

The above description of ﬂooding regime gradients in coastal wetlands is overly
simple because it does not incorporate the effects of year-to-year lake level changes.
Amplitudes, durations, and directions of these long-term changes are unpredictable, but
they can be equal to or greater than seasonal changes (Figure 2.2). Rising water levels
cause permanent ﬂooding in previously temporary zones, and falling water levels reverse
this effect. Even relatively small changes can ﬂood or drain large areas of the shallow-
sloped coastal wetlands, and these ﬂooding changes are known to cause plant community
changes over time (Burton 1985). However, eﬂ‘ects of such hydrologic change on
wetland ﬁsh and invertebrates have not been well studied.

Aquatic fauna diﬂ‘er among isolated wetlands with different ﬂooding regimes (e. g.
permanent, semi-permanent, seasonal, temporary, ephemeral; Williams 1987 , Batzer and
Wissinger 1996), so changes in coastal wetland ﬂooding may be expected to change
animal communities. The apparently-distinct communities at high elevations (Chapter 1)
should change when these areas are subjected to year-round ﬂooding, rather than the usual
seasonal inundation pattern. Some members of these distinct communities may even be
temporarily lost where up-slope migration is impeded by human shoreline development.
Also, hydrologic variability may be an important determinant of long-term survival of
animal species, especially if it increases in the ﬁiture, as predicted by some global climate
change models (Mortsch 1998). Animals that cannot adjust to high-amplitude and/or
short-period water level change will probably give way increasingly to a lower-diversity
community of generalists, as is typical of temporary waters (Wiggins et al. 1980).

From late 1996 to early 1998 Lake Huron's water level rose abruptly and then fell,
creating a high-water year (1997) between two lower-water years and dramatically
altering coastal wetland hydrologic conditions. During this period, high-elevation wet

meadow communities switched from temporary ﬂooding (4 months or less) to year-round

54

ﬂooding and back again, so faunal changes in response to a ﬂooding-regime change could
be studied. Also, plant communities did not change dramatically during this period, and
the vegetation zonation pattern along the elevation gradient remained the same, offering
an opportunity to determine whether animal distributions changed horizontally along the
gradient or whether they remained in the elevation/plant zones where they had previously
been most abundant. Thus, at least temporarily, many variables directly associated with
vegetation zonation and those associated with water level were de-coupled.

In this study I hypothesized that if ﬂooding regime differences among wetland
zones are very important faunal structuring forces, as is argued in the previous chapter,
dramatic alteration in ﬂooding from year to year should alter the animal communities most
in coastal wetland zones that experienced the greatest change in ﬂooding regime,
particuarly the high-elevation wet meadow. Meanwhile, deep marsh communities that
were ﬂooded continuously throughout the study period would change little. On the other
hand, if vegetation zonation and associated variables (e.g. detritus characteristics, shading,
etc.), which respond more slowly to hydrologic changes, are of key importance to animal
communities, then these communities should show relatively little change in spatial

distribution among zones when water levels change.

METHODS

The study site was the same northern Lake Huron coastal wetland, with the same
sampling transects used in the 1996 analysis described in Chapter 1. In 1997, I observed a
substantial water level rise in the wetland, which was the impetus for this study. Water
levels changed over the three years (1996-1998) of the study as follows (also see Figure
2.3). In Year 1 (1996) the water line was at the upper edge of the cattail/transition zone
in early May. I considered this an "average" year, because cattail plants are generally
considered an indicator of the long-term average water level, and later examination of

NCAA data indicated that the rise from January to May was just over half the total rise for

55

the year. Year 2 (1997) was the high-water year. By May of that year, the water level
was at least 30 cm higher than it had been one year earlier, resulting in year-round
ﬂooding of the wet meadow zone. Because the high water reached much farther into the
wet meadow than in Year 1, I added an additional high-elevation sample station (Station ‘
0) to each transect. The water peaked in late July and then began a multi-year general
decline. In May of Year 3 (1998) the water line was approximately intermediate between
the May levels of the previous two years. Aﬁer a relatively brief rise, the level peaked in
June and declined slowly for the rest of the year. Under these conditions the upper wet
meadow only remained ﬂooded until late June so Station 0 samples could not be taken
aﬁer that month. By the August sampling period there was just enough water at Station 2

sites to allow sampling.

Sampling procedures

To estimate relative ﬁsh abundances at diﬁ‘erent elevations among years, I
continued the activity trap sampling described in Chapter 1. However, for invertebrate
samples I used a Wilding-type stovepipe sampler to collect quantitative substrate/water-
column samples. This method was a measure of a standardized subset of invertebrates
found on and above a known surface area of wetland substrate, so it could be used to
estimate invertebrate density diﬂ‘erences among sample stations and time periods. I
collected this type of sample from even-numbered sampling stations resulting in one
sample ﬁ'om each vegetation zone (wet meadow, transition, bulrush) on each transect
(thus ﬁve replicates per zone), during each sampling period (once monthly during June,
July, and August of 1996 through 1998). Because the water level was high in the second
year of the study, I established, and sampled, an additional station (Station 0) at an
elevation 10 cm higher than the previously-highest station (Station 1) of each transect.

The sampling method consisted of forcing one end of a metal stovepipe-like tube

(14.5-cm diam.) into the substrate to enclose 165.1 cm2 of substrate with its overlying

56

water. Large detritus and any attached animals were scooped out of the cylinder by hand
and placed in a sieve-bottom bucket (250 pm mesh). The water in the cylinder was then
stirred to suspend ﬁne particulate matter and small animals, and the cylinder contents were
pumped into the bucket using a plastic manual transfer pump. The pump was very
inexpensive and commonly available in stores, and consisted of a hand-squeezed bulb atop
a one-way valve assembly with incurrent and excurrent tubes. I standardized the volume
of sampled material by squeezing the bulb 100 times per sampler placement.

Sample sites were selected at a random distance from 5 to 15 meters from each
sampling station, perpendicular to the transect and always on the same side. From the
randomly-selected location, I collected three stovepipe samples at one-meter intervals,
combining the three in the sieve bucket to reduce the eﬁ‘ects of small-scale spatial
variations on the data. I rinsed all material in the sieve bucket to remove small particles,
then transferred the bucket contents to a one-quart canning jar. Upon returning ﬁom the
ﬁeld site each evening, I drained excess water from each jar and preserved the contents in
95% ethanol with Rose Bengal dye to stain animal tissues for easier lab sorting.

Each sample was processed as follows. First, all sample material was rinsed
through a l-cm mesh to remove large detritus that would interfere with the subsampling
procedure. The sample was washed through a 500 um mesh sieve and two random one-
sixth subsarnples were collected and combined, so a total of one-third of each sample was
selected for ﬁrrther processing. The subsampling device was a ray-ﬁnned divider that ﬁt
into a standard sieve and consisted of six plastic dividers radially arranged and glued
together at the center. The device "cut" the sample into six equal-sized "pie" pieces in the
sieve. To do this, I placed the sieve in a tub of water deep enough to submerge the sample
but shallow enough that the top edge of the sieve was above the water surface. I then
agitated the sample with a stirring rod for two minutes, and placed the subsampler in the
sieve, twisting the device as I slowly lowered it to keep sample material suspended. Once

the divider was lowered to the mesh surface, I removed the sieve ﬁ'om the water and the

57

 

 

randomly-selected "wedges" were easily removed from the sieve with a rinse bottle. All
invertebrates were picked from each subsample under 10X magniﬁcation and retained for
identiﬁcation.

1 also collected plant community data from each sampling station, though I did not
begin this sampling until the second year. For the ﬁrst year I was able to use data ﬁom a
single transect that were collected as part of a related study, but for the next two years I
collected data ﬁom all stations on all transects. Plant sampling was performed as follows.
Three 0.5m X 0.5m quadrats were selected at random angles and distances (1 to 20 m)
ﬁ'om each sampling station. Stems of each emergent species in the quadrat were counted

and percent areal coverage of each submergent species was visually estimated.

Data collection and analysis

Invertebrates were identiﬁed according to Merritt and Cummins (1996), Pennak
(1989), and taxon-speciﬁc keys for lower taxonomic levels, when appropriate. Fish
identiﬁcation followed Hubbs and Lagler (1967). I identiﬁed ﬁsh to species and
invertebrates to the lowest taxonomic level possible using a dissecting microscope, and
counted individuals in each taxon. This resulted in diﬁ‘erent levels of taxonomic
resolution, so subsequent comparisons of taxa richness and Shannon-Weiner taxa diversity
may not have fully reﬂected differences at the genus level. After data entry, I used
correspondance analysis to examine and display among-year changes in invertebrate
communities at diﬂ‘erent elevations. First I aggregated the monthly data from each station
during each year resulting in 10 data records (3 sampling stations during each of three
years, plus the additional high wet meadow station in Year 2). I used the CORRESP
procedure in SAS to simultaneously ordinate the ten samples, and plotted the samples in
two-dimensional ordination space to examine changes in community similarities among

years.

58

 

To explore changes in abundances of individual invertebrate taxa, I ﬁrst examined
data distributions and decided to use a ln(x+1) transformation to normalize the data and
equalize variances. I initially used a two-way AN OVA approach (with Transect as a
"blocking" factor, and Year and Sampling Station as main effects) to analyze differences in
abundances and relative abundances of all taxa, total abundance, and taxa richness.
However, the Year X Station interaction term was signiﬁcant (p<0.05) in almost all cases,
so I analyzed data from each station separately. I used individual paired t-tests to analyze
diﬁ‘erences in observations matched by transect and month from Year 1 to Year 2, and
then from Year 2 to Year 3. This approach accounted for the effects of repeated
measures from the same transects, and accounted for differences among months without
including them in the analysis. The three-year sequence of events, then, was treated as
two separate natural experiments.

After the "experiment" analysis, I analyzed data records from all stations and
sampling periods with multiple regression to explore likely causes of the changes observed
in response variables. The response variables tested included total invertebrate abundance,
total insect abundance, and abundances of the most abundant taxa. Independent variables
used in the regressions (both complete and stepwise) included sampling station, water
depth, month, time since ﬁrst re—ﬂooded, plant stem density, % cover of submergent
plants, dragonﬂy larva density, and ﬁsh catch per unit eﬂ‘ort. The "time since re-ﬂooded"
variable was a sum of all non-winter months of continuous ﬂooding since the last time the
sample location had been "dry" (not ﬂooded). For low-elevation sampling stations, the
last "dry" time was inferred from NOAA water-level records. I ran regressions with and
without the vegetation variables because I lacked vegetation data for all records, so the
with-vegetation regressions had to be based on fewer data records. I examined the
residuals from regressions on untransformed data and decided to re-run the analyses on

ln(x+l) transformations of the dependent variables.

59

RESULTS
Changes in total invertebrate density

The absolute density of invertebrates almost tripled over the three-year period
(Figure 2.4), ﬁom nearly 15,000 animals/m2 (~253/sample for all 1996 samples) to over
41,500 (~705/sample for all 1998 samples), averaged over all stations (p<0.001, ANOVA
on log-transformed abundances). Density was highest in the deep marsh initially (Table
2.1), but rates of increase were higher at higher points on the gradient, so that by 1998
densities in the wet meadow and transition zone equalled or exceeded that in the deep
marsh (Figure 2.5). The greatest increase (~3 75%) occurred in the wet meadow, with a
higher rate of increase in the ﬁrst period (~181%, 1996-97) than in the second (~69%,
1997-98). On the other hand, density increase in the transition zone and deep marsh were
higher in the second period. The transition zone density increased ~3 0% and ~220% in
the ﬁrst and second periods, respectively; deep marsh density rose ~20% and ~73% in
these periods. To test whether invertebrate density growth approached exponential rates
at each station, I log-transformed the total abundance data and ran simple linear
regressions of log(density+1) against sampling time (June through August of each year, or
nine sample times). The data ﬁt a linear ﬁniction well at each station (p<0.001) with an r2
as high as 0.64 at Station 4, indicating that the invertebrate density increase approximated

an exponential growth curve.

Community composition changes along the elevation gradient

The correspondance analysis (CA) plot (2-dimensional inertia 56.74% of total)
illustrated changes in communities among years and stations as distances between their
plotted positions (Figure 2.6). The positions of the wet meadow, transition zone, and
deep marsh communities in 1996 indicated the community gradient ordered along the ﬁrst
ordination axis. The 1997 communities at these stations, plus the added "hummock"

station in the high wet meadow, also were ordered along the ﬁrst axis but two diﬁ‘erences

6O

were evident. First, the total 1997 gradient, even with the additional station, was shorter
than the 1996 gradient, indicating less variability from end to end in 1997 than in 1996.
Second, the 1997 gradient was oriented diﬁ‘erently along the second axis than the 1996
gradient, indicating partial changes in community composition among years. The 1998
gradient was the shortest of the three, indicating a substantial homogenization of
communities from deep marsh to wet meadow. Yet, the orientation of the gradient was
roughly parallel to the 1996 gradient, indicating a partial return to community
compositions characteristic of the ﬁrst year of the study (Figure 2.7). The greatest
magnitudes of community change (represented by arrows in Figure 2.6) occurred in the
wet meadow community during both periods.

Examination of the partial contributions to inertia for each axis (from the SAS
output, as explained in Chapter 1) revealed that the taxa contributing most to the ﬁrst axis
(contribution index > 1) were from two groups: those that were most associated with the
highest points on the gradient during the ﬁrst two years (and subsequently decreased in
relative abundances), as opposed to those that were more generally distributed and
increased in relative abundance at most stations over time (discussed in detail below). The
ﬁrst group could be split into two subgroups that contributed to the second axis, which
represented among-year differences at the highest-elevation stations. These sub-groups
were those that were most characteristic of the wet meadow community in 1996 (and
1998 to a lesser extent) vs. those that became more abundant in the wet meadow and
hummock communities in 1997, at the expense of the ﬁrst sub-group.

Overall, the greatest spatial heterogeneity in community composition (ordination
axis 1) occurred in 1996, indicated by the gradient length in the CA plot (Figures 2.6 and
2.7). Over the three-year period, wet meadow communities became more like lower-
elevation communities, whose ordination positions changed little. This loss of gradient
variation was a result of lowered relative abundances of many taxa that characterized the

wet meadow community initially, and the increasing dominance of a few groups that had

61

been initially abundant only at lower elevations, but became more generally distributed (as
described below). Secondary variation (represented by the second ordination axis) .
apparently consisted of diﬁerences between wet meadow communities under high water

and low water conditions.

Diversity: Richness and equitability

Shannon-Wiener taxa diversity differed both spatially and temporally (Table 2.2).
These diﬁ‘erences indicated variation in taxa identiﬁed to various taxonomic levels: genus
in most cases, but also higher taxa that were diﬁicult to identify ﬁirther at low
magniﬁcation. I present these data as indicators of diversity differences, though they may
not fully reﬂect differences in species or genus diversity. Initially, diversity was greatest in
the wet meadow, but after a decline at all stations in 1997, diversity in the deep marsh
recovered while it continued to decrease at other stations. This resulted in a reversal of
the diversity spatial trend: from highest diversity in the wet meadow in 1996 to highest
diversity in the deep marsh in 1998 (Figure 2.8). The greatest diversity loss (ﬁ'om 3.02in
1996 to 2.20 in 1998) occurred in the wet meadow, while deep marsh diversity changed
much less. Transition zone diversity decreased most from 1996 to 1997 (2.71 to 2.17)
and very little after that. The hummock zone had the highest diversity of all stations
(2.63) in 1997, and dropped to 2.45 in 1998, though this number only represents a single
month's data.

These diversity differences reﬂected spatial and temporal differences in the two
components of diversity: richness and equitability. Taxa richness decreased among the
wet meadow, transition zone, and deep marsh communities from 1996 to 1997 partly
because several taxa associated with the higher elevations in 1996 did not occur in 1997
samples, or were only found at the newly-established "hummock" stations. Richness
rebounded partially in 1998 as some of these taxa returned to lower elevations (Figure

2.9). On the other hand, equitability decreased throughout the study period. This was

62

primarily the result of increases in community dominance by a few taxa, especially isopods
(mostly Caecidotea), midge larvae (Chironomidae), Caenidae mayﬂies (all Caenis) and, to
a lesser extent, amphipods (mostly Hyallela). Increasing dominance by these groups was
typical of communities at all stations, with Chironomidae reaching their highest relative
abundances in 1997 and the other three groups generally increasing in abundance and

relative abundance at all stations over the whole study period (Figure 2.10).

Community composition variation: Individual taxa

Community compositions varied among stations and years (Table 2.3). In general,
Chironomidae larvae numerically dominated all stations in 1996, but Caenidae mayﬂy
larvae and Isopoda grew in relative abundance all along the gradient over time (Table 2.4).
The increase of these two groups proceeded from lower to higher stations, though
Caenidae increased at higher elevations sooner than isopods. Amphipods showed a similar
trend to isopods, though they increased less and remained most abundant at Station 6
(deep marsh) at all times.

The wet meadow community changed most dramatically and somewhat diﬁ‘erently
during both periods (1996-1997 and 1997-1998). While isopods and amphipods increased
in importance over the whole period, other changes occurred that made the 1997 high-
water wet meadow community relatively distinct (as was indicated by the ordination plot,
discussed above). Several taxa "invaded" the wet meadow in 1997, only to decrease in
abundance again as water fell in 1998. These changes were accompanied by decreases in
abundance of several taxa that were characteristic of the 1996 wet meadow community.
For example, Nematoda worms, Oribatei soil mites, Harpacticoida copepods,
Limnephilidae caddisﬂy larvae, Ceratopogonidae ﬂy larvae, Sphaeriidae clams, and
Dytiscidae diving beetles all decreased in wet meadow relative abundance in 1997, and
most of them returned to higher abundances the following year (Table 2.4). Taken

collectively, all taxa that were most abundant at Station 2 in 1996 were signiﬁcantly less

63

 

abundant (p=0.002) among all stations in 1997, though many of these groups occurred at
the newly-ﬂooded Station 0. Conversely, these taxa collectively increased (p<0.001) over

all stations (but predominantly at Station 2) from 1997 to 1998.

Shifts in spatial position of peak taxa abundances among years

Abundances of taxa generally peaked at single sampling stations, but these peaks
changed position on the gradient ﬁ'om year to year for most taxa (Table 2.5). Four peak- r
shiﬁ categories could be distinguished, and they corresponded to the spatiotemporal

relative abundance changes summarized in Table 2.4. The most common pattern was seen

 

in about one-third of the taxa and consisted of up-slope peak shifts in 1997 when water ‘
was high, followed by down-slope shifts during the declining water levels of 1998. ,
Groups showing this pattern included many insect taxa, most microcrustaceans, some
snails, and water mites. Some of these taxa shifted ﬁom low parts of the gradient to
higher elevations, and back again, while certain others (e. g. Dytiscidae, most
Ceratopogonidae genera) were wet meadow associated taxa that shifted to the hummock
stations when water was high and returned to the lower wet meadow the following year.
Other peak shiﬁ patterns included taxa that progressively moved to higher
elevations through the entire study period, particularly those that shifted ﬁom the wet
meadow to the hummock zone and stayed there (e. g. Libellula, Planorbula, Haliplus, an
unidentiﬁed Limnephilidae genus, Harpacticoida, Isotomidae, Oribatei, Nematoda) or
disappeared completely from the samples over time (e. g. Ceratopogonidae: Forcipomyia
and Atrichopogon, Sminthuridae). A different, and very important, pattern was shown by
the Pericarida crustaceans (Isopoda and Amphipoda): they increased in relative
abundance at all stations, as mentioned above, but nonetheless remained abundant at the
lowest elevations during all years. A few other taxa showed this pattern of stable »

abundance peaks in the transition zone and deep marsh (though not expanding up-gradient

64

as did the Peracarida): limpets (Ancylidae), Libellulidae: Leucorrhinia dragonﬂy nymphs,
Chrysomelidae beetles, Polycentropodidae caddisﬂies, and non-Stylaria Naididae worms.
A few taxa showed anomalous patterns. Brachycera initially (1996) peaked in the
wet meadow, then in the deep marsh in 1997 and 1998; Leptoceridae caddisﬂies showed
two disconnected abundance peaks in 1997 (in wet meadow and deep marsh);
Sphaeriidae were always most abundant at high elevations, but consistently showed a
secondary peak in the deep marsh. In each case it is likely that these discontinuities
represented diﬂ‘erent distributions of distinct, but unidentiﬁed, genera within the higher
taxonomic category. Bithyniidae snails and Oribatei soil mites remained most abundant in
the lower wet meadow (Station 2) communities during all years, but the Bithyniidae were
only common on the westem-most transects, suggesting a localized population within the
site. Hydrobiidae snails and Corduliidae dragonﬂy nymphs, on the other hand, were not
present in 1996 samples, then peaked in the deep marsh and transition zone during 1997
and 1998, respectively, suggesting that they were migrating up-slope ﬁ'om elevatons that

were lower than the deep marsh stations.

Fish distributions

In 1996 most ﬁsh caught were mudminnows (Umbra Iimi), especially in the wet
meadow. Juvenile largemouth bass (Micropterus salmoides), bowﬁn (Amia calva), and
yellow perch (Percaﬂavescens) occasionally occurred there, but were more common in
the transition zone. Brown bullhead (Ictalurus nebulosus) were only caught in the deep
marsh. I observed common carp (Cyprinus carpio) breeding activity in the deep marsh
and into the transition zone, but dense vegetation and shallow water (<20 cm) prevented
their invading the wet meadow that year.

Wet meadow and transition zone ﬁsh densities were apparently lower in 1997 than
in 1996 (Figure 2.10), though this is somewhat misleading. Because the activity traps
used for ﬁsh sampling were size limited, any ﬁsh that could not ﬁt through the opening

65

(~23 mm inner diameter) were not sampled. Other ﬁsh sampling with large fyke nets
(conducted as part of a related study) indicated that larger ﬁsh (e. g. yearling bowﬁn,
pumpkinseed, bullhead) were present in the wet meadow in 1997. However,
mudminnows (the most common species) of all sizes were small enough to ﬁt into the
traps, and they were less abundant in 1997 than in 1996, but increased again in 1998
(p=0.014, Kruskal-Wallis test). Juvenile bowﬁn, bullhead, and bass extended their range
far into the wet meadow in 1997 where they had been rare in 1996.

In 1998, the initial water level was about as high as the 1996 peak. The ﬁsh
assemblage resembled the 1996 assemblage in composition and spatial distribution, except
that juvenile pumpkinseed were much more abundant (p=0.009 Wilcoxon rank-sum test),
and mudminnows were concentrated in the transition zone rather than the wet meadow.
As water level fell, ﬁsh were apparently forced to move ﬁirther down-gradient until only
small mudminnows were present in the wet meadow. Densities, however, remained lower

than in 1996.

Regression analyses of possible causal factors

Multiple regression analysis revealed that total invertebrate density was
signiﬁcantly associated with duration of ﬂooding (p<0.001) largely because of increased
densities of the most abundant taxa: isopods (p<0.001), Caenidae mayﬂy larvae
(p<0.001), Chironomidae (p=0.016), amphipods (p=0.011), and snails (p=0.01).
However, ﬂooding duration was not a signiﬁcant inﬂuence on the abundances of
Dytiscidae beetle larvae, Ceratopogonidae larvae, or microcrustaceans, all of which were
inversely associated with water depth (p<0.05), which was consistent with their tendency
to peak at high elevations. Trichoptera as a group were not associated with ﬂooding time,
but were inversely associated with water depth. A small minority of taxa were weakly
(0. l>p>0.05) associated with emergent stem density, submergent plant coverage, and

densities of the dominant predators (ﬁsh and dragonﬂy larvae). However, none of these

66

 

factors consistently inﬂuenced dependent variables, with both positive and negative
relationships occurring, and some factors being included or excluded (during stepwise

regression) depending on which data transformations and aggregations were used.

Plant community changes

This study was partly based on the idea that plant communities changed little while
water levels were changing. Data collected on Mackinac Bay plant communities in this I"
study, and on several nearby wetlands in a related study (Gathman and Keas 1999),

revealed that vegetation zones did not change position on the elevation gradient during the

 

study period. I deﬁned wet meadow communities as those that were numerically- if .
dominated by sedges (Carex) and which included grasses, shrubs, and a characteristic i
assemblage of facultative wetland plants (e. g. Campanula aparanoides, Lysimachia
thyrszﬂora, Iris versicolor, etc.) Deep marsh communities were characterized by a very
diﬂ'erent assemblage of emergent plants (Sciipus acutus, Pontedaria cordata, Sagittaria
sp., etc.) as well as ﬂoating-leaved and submergent species. The transition zone
community was characterized by dense growths of cattails (T who angustifolia and T.
latifolia). Over all 3 years, plant communities surrounding my sampling stations remained
in the same categories, although some changes occurred among individual species.
These relatively few vegetation changes could be summarized as follows. First, in
the transition zone and wet meadow, emergent stem density decreased progressively over
the 3-year period, and submergent plant coverage increased under high water (decreasing
again when water-level was falling). As described above, I included stem density and
submergent coverage in regression analyses and found that they were not signiﬁcantly
associated with most invertebrate abundances. Second, many rarer wet meadow plant
species disappeared or declined under high water, and most did not return immediately the
next year. Third, plant community composition shifts under high water were the opposite

of hypothesized shifts (i.e. the transition zone became more similar to the wet meadow

67

rather than the deep marsh, etc.) This occurred in part because distributions of certain
ﬂooding-tolerant sedge species (Carex aquatilis and C. lasiocarpa) spread down-slope

from the lower wet meadow into the transition zone and upper deep marsh.

DISCUSSION

I used correspondance analysis to determine whether community composition at
each station shifted from year to year according to the following expectations. Under high
water conditions, the wet meadow community should become more like transition zone
and deep marsh communities, and the transition zone community should become more like
the deep marsh community. Ifﬂooding regime differences are very important structuring
factors, then the most change should be seen in the wet meadow community, where
ﬂooding regime changed the most from 1996 to 1997. The magnitude of change should
be less in transition zone communities, while deep marsh corrununities should not change.

These hypotheses were generally supported by the data. The plotted position of
the wet meadow community in 1997 was far to the left of the 1996 wet meadow position,
and very near most of the other plotted points. This indicated a relatively large change in
community composition that strongly increased the similarity between the wet meadow
community and communities at lower elevations. Also, the loss of diversity was greatest
in the wet meadow because many taxa that were dominant there in 1996 decreased in
abundance in 1997, while a smaller number of taxa originally ﬁ'om lower elevations
became more abundant. Shifts along the second CA axis reﬂected this turnover in
dominant taxa. The greatest difference between communities in this second dimension
was between the 1996 wet meadow and the hummock zone in 1997, indicating that the
wet meadow changes reached far up the gradient because the water level was high long
enough to allow many taxa to migrate many tens of meters across the wetland.

In general, the 1996-1997 changes can be described as up-slope distribution shifts

or expansions of most taxa, though these shifts occurred at diﬂ‘erent rates among taxa.

68

The 1997-1998 changes, on the other hand, were more varied. While many taxa reversed
their previous responses by shifting back down-slope, many others apparently did not.
Some of the initially characteristic taxa of the wet meadow that appeared to ﬁnd a high-
water reﬁrge in the hummock zone failed to return to the lower wet meadow stations
when the water level was dropping. These tended to be rare taxa and non-insects, while
those that readily reversed their shifts were insects. The taxa that most inﬂuenced
community composition changes (isopods, mayﬂies, midges, and amphipods) showed
different patterns ﬁom those described above. Midge larvae numerically dominated
communities at all stations in 1996, but gave way to mayﬂies and isopods as their numbers
increased dramatically. By 1998, when many taxa were returning to lower elevations,
isopods, mayﬂies and amphipods continued to increase at all stations, including the wet
meadow.

Vegetation zonation must not have played a strong direct role in determining
invertebrate community composition in this study. Invertebrate distributions shifted along
the elevation gradient despite the lack of similar shifts in plant zones. In fact, the down:
slope spread of two very abundant sedge species was an unexpected shift in the opposite
direction to shifts observed in invertebrates. Nonetheless, the observed changes in plant
community structure cannot be fully dismissed as inﬂuences on invertebrates. The
literature does not suggest that changes in emergent species composition are likely to
strongly aﬂ‘ect invertebrates, although detritus produced by diﬂ‘erent species may have
some affects because it can vary in food quality (Batzer and Wissinger 1996). Also, the
observed wet meadow reduction in emergent stem density and increase in submergent
plant coverage probably had some impact on some invertebrates: submergent plants
provide colonizable surfaces and complex structure for invertebrates, and lower emergent
stem density allows greater light penetration, which may have increased water and

substrate temperatures and enhanced algal growth. Ifthese factors aﬂ‘ected invertebrates,

69

 

the eﬂ‘ects were too weak to be detected by most of the regression analyses described
above.

The water level changes observed in Mackinac Bay from 1996 through 1998, and
resulting biotic changes, are best understood when placed in their historical context. The
lake level had been quite low the year before I began sampling (Figure 2.2), so in 1996 the
wet meadow was ﬂooded for the ﬁrst time in two years. The transition zone had been
only seasonally ﬂooded the previous year and even the deep marsh had experienced very
low winter water levels, so it was probably subjected to deep substrate freezing and ice
scour. As a result, initial invertebrate densities were very low, probably because all zones
were recovering ﬁom the effects of low water conditions. The wet meadow community,
however, was probably the most adversely aﬂ‘ected by the previous year because only a
minority of invertebrate taxa can survive dormancy in non-ﬂooded soil for over a year
(\Vrlliams 1987).

The 1996 dominance of midges was probably a result of their aerial dispersal and
short generation time which make them rapid colonizers of newly-ﬂooded areas. The
other taxa dominating the 1996 wet meadow (e. g. pulmonate snails, Sphaeriidae clams,
beetles, etc., copepods) were also either rapid re-colonizers, or were groups with
adaptations for tolerating de-watered periods by estivating in the detritus/ soil substrate.
Subsequent increases in abundances of other taxa resulted from slower re-colonization of
the wetland ﬁ’om lower elevations. These other taxa, including isopods, amphipods,
mayﬂies, most caddisﬂies, and others, were all groups typically restricted to permanent

waters.

Possible causal mechanisms
While the gradient distributions of most animal taxa were clearly affected by water
level changes, the actual mechanisms driving these changes were less clear. The rapid

changes in animal distributions among years suggested that plant zone characteristics were

70

not very important determinants of gradient positions, but certain other questions remain
to be answered: Why did many deep marsh taxa fail to occupy the wet meadow until the
water level rose in Year 2? Why did the initial wet meadow taxa decline in 1997? Why
did isopods, mayﬂies, and amphipods remain abundant in the wet meadow in 1998 even
though they had not been there in 1996 under similar ﬂooding conditions?

It seems likely that the initial distinctions (in 1996) between deep marsh and wet
meadow communities were partly caused by differences in life history adaptations to
temporary ﬂooding because the wet meadow community consisted of common temporary-
water inhabitants (as identiﬁed by Williams 1987). This hypothesis is ﬁrrther supported by
the fact that the wet meadow had not been ﬂooded the previous year, so the "drought"
period in the wet meadow had lasted through an entire summer season. Had the water
level followed the average annual pattern, falling back to the transition zone in late 1996
and rising the next spring, a similar pattern would likely have been observed in 1997.
Presumably, any animals without drought-tolerance or avoidance mechanisms that
migrated into the wet meadow would not survive the de-watered period. However, the
water did not drop in late 1996 and many taxa did appear to migrate into the wet meadow
(Chapter 1) that year. They may have simply been continuing an up-slope migration that
had begun a year earlier, although quite a few taxa did not show this pattern.

Perhaps these non-migrating taxa were responding to water chemistry gradients, as
Cardinale et al. (1997) observed in a Saginaw Bay coastal wetland. Gradients in dissolved
oxygen, conductivity, and pH were detected in 1998 in Mackinac Bay in a related study
(Gathman and Keas 1999) and showed evidence of changing with water level changes. In
particular, dissolved oxygen decreased from supersaturation in the deep marsh to hypoxic
conditions in the wet meadow, but this gradient moved up-slope and down-slope as water
levels rose and fell with oscillating seiches. There was a direct correlation between water

level changes of ~10 cm in the wet meadow and dissolved oxygen levels, so it is likely that

71

 

 

under the very high water- conditions of 1997, wet meadow dissolved oxygen levels were
considerably higher than in 1996 or 1998.

Dissolved oxygen or other water chemistry variables may have been important to
some invertebrate taxa. The 1996 wet meadow community. consisted of taxa that were
not only drought-tolerant, but also had low-oxygen adaptations such as the ability to
breathe atmospheric oxygen (pulrnonate snails, diving beetles), high-eﬁciency gas
exchange mechanisms (cases in Limnephilidae caddisﬂy larvae), or high surface-area to
volume ratios (Chironomidae, Nematoda, Copepoda). However, similar statements could
be made about most lower-elevation taxa as well. Using "Tolerance" values ﬁ'om the
Hilsenhoﬁ‘Biotic Index (Hilsenhoﬂ‘ 1988) as a measure of low dissolved-oxygen
adaptation, none of the invertebrate taxa collected were strongly intolerant (Tolerance
below 3), and the few that could be considered moderately intolerant were not restricted
to the deep marsh. Aeschnidae dragonﬂy larvae (Tolerance value 3) were common in the
wet meadow in 1998, and Limnephilidae caddisﬂy larvae were characteristic of the wet
meadow despite having the same tolerance value as the other caddisﬂy families sampled
(Tolerance value 4). On the other hand, the damselﬂy larvae (Coenagrionidae) and
amphipods (mostly Talitridae) had high tolerance (Tolerance value 8) but nonetheless
were rare at high elevations in 1996. The dominant taxa showed even less evidence of
oxygen limitation. As the wet meadow water levels fell to <10cm in 1998, isopods,
amphipods, and Caenidae mayﬂies remained very active and abundant, even in small
isolated depressions high in the wet meadow where dissolved oxygen levels were below
the reliable detection limits of the instrument used. However, many other taxa did not
remain or survive in these environments. Overall, then, it seems unlikely that dissolved
oxygen was a strong limiting factor, except at extremely low levels for some taxa.

It is possible that dissolved oxygen played a different role in regulating invertebrate
communities. Ifit varied spatially and temporally as argued above, it may have

determined ﬁsh community compositions, which could, in turn, affect invertebrate

72

communities through predation. The dominance of mudminnows at higher elevations
indicated a probable dissolved oxygen limitation on most ﬁsh because mudminnows can
occupy low-oxygen habitats by "breathing" air at the surface. This spatial variation in ﬁsh
species may have imposed spatially-variable predation pressures on invertebrates, though
such differences in feeding preferences among these ﬁsh species are little known. Fish
density, on the other hand, is considered to be a strong inﬂuence on wetland invertebrate

communities (Batzer and Wissinger 1996, Wellbom et al. 1996). Wet meadow small ﬁsh

. Iftw
I

densities decreased over the study period, coinciding with increased invertebrate density.

These reduced ﬁsh densities may have resulted ﬁom the invasion of the wet meadow by

 

piscivorous ﬁsh under high-water conditions, as suggested by data on distributions of

ﬁ .
I“

larger ﬁsh (Gathman and Keas 1999). Based on the multiple regression results, ﬁsh
density did not appear to be an important inﬂuence on invertebrate abundances, but the
obvious variations in ﬁsh communities in this study cannot be dismissed as partial
determinants of invertebrate faunal composition.

Dragonﬂy larvae may have imposed important predation pressures also, though
attempts to experimentally determine whether invertebrate predators can structure
invertebrate prey communities have produced mixed results (W ellbom et al. 1996).
Although the increase in mean wet meadow dragonﬂy larva density was non-signiﬁcant
from 1996 to 1997, all of these larvae collected in 1996 were late-season new hatchlings
which were probably too small eat larger macroinvertebrates such as isopods. Excluding
immature larvae, dragonﬂy larva density increased in the wet meadow from O in 1996 to
1.23 per sample in 1997 (p=0.014), and up to 1.71 per sample in 1998 (p=0.034).
However, while wet meadow dragonﬂy larval density was highest in 1998, these were all
mature larvae that showed evidence of reduced development rates, which suggests
reduced feeding. In previous years dragonﬂy nymphs emerged as adults in July, but in
1998, 40 dragonﬂy nymphs in wet meadow enclosures (in a related experiment) failed to

develop to the emergent adult stage by late July, and even in early September mature

73

larvae were very common in the wet meadow. Relatively few dragonﬂy larvae collected in
1998 were early-instar hatchlings (Figure 2.12), and none of these were collected from the
wet meadow, suggesting that dragonﬂy development was stunted. Thus, it appeared that
the effective dragonﬂy larva density (in terms of predation pressure) was highest in the
wet meadow in 1997, which may have contributed to the decreased abundances of many
taxa characteristic of the low-water wet meadow.

It may seem contradictory to discuss possible predation eﬂ‘ects after regression In
analyses suggested little association between invertebrate prey and the most abundant

predators. However, the regression analysis was a weak test of predation effects, in part

 

because the ﬁsh abundance data used were based on a passive trapping method which did -
not measure actual ﬁsh density. These data were probably reliable measures of relative w
ﬁsh abundance differences among sample stations on average, but they had high within-
group variability with many zero counts, even at elevations with relatively high total ﬁsh
catches over all transects. Because ﬁsh predation is generally considered to be an
important community structuring force (Wellbom et al. 1996), and because large
invertebrate density increases coincided with reduced ﬁsh density in most cases in this
study, further study involving controlled experimentation is called for.
The large invertebrate density increases that I observed were not always associated
with reduced ﬁsh abundance. The transition zone and deep marsh density increases from
1997 to 1998 coincided with increases in ﬁsh numbers so any explanation of invertebrate
density must include factors other than predation mortality. The general density increases
over the study period may have occurred because invertebrate communities were
recovering from previous low-water conditions, so perhaps relative rates of increase at
different times and places were better indicators of possible mechanisms. Initial wet
meadow densities were quite low, and subsequent increases may have resulted ﬁom a
combination of up-slope immigration, reproduction in the "new" habitat, and reduced ﬁsh

predation. I did not study reproduction, so I cannot evaluate whether recruitment varied

74

over space or time. It is possible that the wet meadow, with its large and relatively-
untouched detritus food source, was a very favorable habitat for reproduction. Also, as
emergent plants became less dense over time, reduced shading probably led to higher
water temperatures, which would favor invertebrate growth.

The largest single density increase occurred in the transition zone from 1997 to

1998, when ﬁsh numbers were relatively high, so recruitment and/or immigration must

 

have been quite high there. Immigration is a likely explanation for this large density F
increase, because the wet meadow was draining in 1998 and many invertebrates could
have been moving down-slope, merging into the "resident" community. As in the

transition zone, the deep marsh density increase from 1997 to 1998 was higher than in the

previous period, and a similar down-slope invertebrate concentration may have been H

responsible. However, the deep marsh increase was dwarfed by that observed in the
transition zone, so perhaps the transition zone was a more favorable habitat. Fish
numbers were relatively high in both zones, but higher plant and coarse detritus density
may have made the transition zone a predation reﬁrge. Again, I did not study
reproduction, so I cannot determine whether reproduction was particularly high in the
1998 transition zone, but the speculations on food source and temperature in the wet

meadow, noted above, probably do not apply to these other zones.

Dispersal mechanisms and immigration rates

The initial goal of this study was to separate the effects of ﬂooding changes from
vegetation zone characteristics by analyzing spatial distribution changes along the coastal
wetland elevation gradient. However, the data suggest that diﬂ‘erences in dispersal
mechanisms among the non-drought—tolerant taxa are very important in determining
community diﬂ‘erences over space and time. Jeﬁies (1994) discussed the importance of
invertebrate dispersal in ﬂood-variable wetlands, but was studying separate pools without

surface water connections. In coastal wetlands such as Mackinac Bay, areas with different

75

ﬂooding conditions are connected by surface water allowing animals without aerial life
stages to invade usually-temporary zones when water is high.

The main changes related to water level variations in Mackinac Bay appeared to
result from dispersal differences. Many taxa shifted their gradient positions up-slope when
water rose, and down-slope when it fell. Most of these taxa were insects with aerial adult
dispersal stages, so they were able to respond quickly to spatial environmental changes.
Meanwhile, non-insects tended to move up-slope during high water, but were less able to
return back down-slope the following year, so many animals were probably left "stranded"
at high elevations, where they would not be able to survive the loss of surface water.
These patterns were equally important in determining the distributions of the dominant
taxa. Chironomidae, which are rapid colonizers, initially dominated all elevations, and
dominated the wet meadow and newly-ﬂooded hummock zone in 1997. Caenidae
mayﬂies, also aerial dispersers, dominated the transition zone in 1997 and became
abundant at higher elevations as well. The more slowly-dispersing isopods and
amphipods, however, became progressively more abundant at successively higher
elevations, with isopods eventually overtaking all other taxa in relative abundance at all
elevations. They were able to do this because water level was almost constantly rising
from late 1996 until mid 1997, and even in 1998 the wet meadow was initially ﬂooded all
the way up to the hummock zone.

While many taxa, mostly non-insects, were unable to respond to the dropping
water levels in 1998, isopods and amphipods remained abundant at lower elevations
because, rather than shifting their gradient positions, they were able to simply expand their
ranges while remaining very abundant at low elevations. Thus, although many individual
isopods and amphipods were undoubtedly "stranded", many more remained at lower
elevations. Other taxa did not fare as well. As mentioned above, dragonﬂy larvae in the
wet meadow appeared to develop slowly in 1998, and I found large numbers of dead and

dying individuals in very shallow water in the wet meadow that year.

76

Implications

Water level ﬂuctuations may be important ecosystem regulators in coastal
wetlands. This study suggests that temporary high-water episodes cause increased
invertebrate production, which may lead to increased detritus processing and may provide
a very large food source for ﬁsh and avian invertivores. While fewer small ﬁsh were
present to exploit the high-water wet meadow prey, larger ﬁsh may have gained from
increases in densities of larger invertebrates. Dragonﬂy larvae densities were high in 1997,
and may have provided an important food source for intermediate-sized ﬁsh (too large for
activity trap sampling, but too small to eat other ﬁsh). Also, related sampling
(unpublished data) indicated that crayﬁsh densities increased over the three year period,
providing food for large ﬁsh.

Falling water levels may have different effects. The high wet meadow invertebrate
densities in 1998 probably led to high mortality as the wet meadow drained and many
animals died. These animals were probably easy prey for predators and scavengers, and
may have led to an important export of energy and material ﬁom the wetland to terrestrial
communities. At the low end of the elevation gradient, dropping water levels force more
animals into smaller wetland areas, where they probably face competition and increased
vulnerability to predators. I can only speculate that this would lead to lower production in
the following year.

Constant water-level variation seems to be the norm in coastal wetlands, so short-
term high or low water conditions are common. However, if changed water-level
conditions last for several years, the spatial positions of vegetation zones shift in the
direction of the water line, so the complete wetland vegetation gradient is re-established at
a new elevation range (Burton 1985). Animals would be expected to adjust to the new
habitat position, but this study suggests that they may do so in a time-lagged fashion, with

a succession of colonists (e. g. midges, mayﬂies, isopods) establishing populations

77

 

consecutively. This constant " catching up " by the invertebrate community may limit total
invertebrate production at a level below the maximum possible under stable conditions,
keeping long-term material and energy ﬂows in dynamic equilibrium determined by water
levels.

Aquatic invertebrate populations in coastal wetlands can only persist if they are

adapted to water level changes, but the changes observed in this study were unusual.

 

Figure 2.2 illustrates the 20-year history of water level changes in northern Lake Huron, rm
showing that water levels can rise or fall relatively quickly, but these changes are usually

sustained for several years. Examination of the 100-year data record indicates that a rapid

water level rise followed immediately by an equivalent drop is historically unusual. In the U

present case, this series of events led to decreased invertebrate diversity because a few
highly-adaptable generalist taxa became dominant. If such rapid water level changes
increase in ﬁ'equency under changing climate conditions, they could lead to sustained
diversity loss in coastal wetlands.

Diversity loss could be exacerbated where human shoreline development interferes
with up-slope migrations of biota, as is common in many Great Lakes wetlands. Road
embankrnents, seawalls, dikes, and residential development interrupt the gently-sloping
wetland substrate, so when water rises, high-elevation temporarily-ﬂooded zones are
eliminated. This study suggests that many taxa rely on this marginal zone as a reﬁrge
during high-water conditions, so loss of this habitat could eliminate these invertebrates
from coastal wetlands. The dynamic hydrology of Great Lakes coastal wetlands leads to
variablity in aquatic animal communities on fairly small spatial and temporal scales.
Further research is needed to determine speciﬁc community-regulating mechanisms, but it
may be more useﬁrl to develop a method to functionally classify invertebrates according to
their dispersal capabilities, which could be used to predict their vulnerabilities to habitat

alteration. In the meantime, human activities that interrupt the upland-to-lake elevation

78

gradients in the few remaining intact coastal wetlands should be prevented to maintain

diversity in these little-studied systems.

79

Table 2.1. Mean invertebrate abundance (#lmz) over all months and transects at each
station during each year.

 

 

 

Year Station 0 Station2 Station 4 Station 6
1996 n.a. 2931.0 5734.9 6697.7
1997 6933.9 8242.0 7388.1 8011.9
1998 55108“ 13,940.5 23,623.8 13,898.1

 

*Only June samples were collected.

Table 2.2. Shannon-Wiener diversity at each station during each year, based on averages
of taxa relative abundances over all months and transects.

 

 

Year Station 0 Station2 Station 4 Station 6
1996 n.a. 3.02 2.71 2.50
1997 2.60 2.30 2.17 2.28
1998 245* ' 2.20 2.14 2.54

 

*Only June samples were collected.

80

 

 

Table 2.3. Top ten most abundant taxa (ranked by relative abundance) at each sampling

station during each year.

 

 

1996 1997 1998
STATION 0
n.a. Chironomidae 28.13% Asellidae 26.41%
Asellidae 14.43% Chironomidae 17.69%
Cyclopoida 8.27% Caenidae 8.82%
Daphniidae 8.09% Cyc10poida 5.96%
Planorbidae 7.46% Isotomidae 5.85%
Macrothricidae 4.83% Planorbidae 4.17%
Sphaeriidae 3.97% Ceratopogonidae 3.92%
Physidae 3.58% Harpacticoida 3.83%
Caenidae 3.25% Sphaeriidae 3.18%
Chydoridae 2.45% Libellulidae 2.40%
STATION 2
Chironomidae 18.56% Chironomidae 32.20% Asellidae 36.89%
Cyclopoida 10.15% Caenidae 17.67% Caenidae 16.49%
Oribatei 8.05% Asellidae 15.93% Chironomidae 15.88%
Planorbidae 6.82% Planorbidae 5.25% Amphipoda 7.83%
Sphaeriidae 6.51% Cyclopoida 4.16% Cyclopoida 6.12%
Asellidae 5.06% Daphniidae 4.08% Isotomidae 1.89%
Ceratopogonidae 4.63% Physidae 3.33% Planorbidae 1.68%
Brachycera 4.37% Naididae 1.77% Ceratopogonidae 1.32%
Harpacticoida 3.84% Ostracoda 1.75% Ostracoda 0.90%
STATION 4
Chironomidae 28.65% Chironomidae 32.59% Caenidae 29.78%
Asellidae 12.48% Asellidae 24.41% Asellid 24.22%
Caenidae 12.34% Caenidae 16.98% Chironomidae 21.57%
Amphipoda 8.25% Amphipoda 4.09% Amphipoda 5.40%
Cyclopoida 4.24% Planorbidae 2.48% Cyclopoida 4.06%
Daphniidae 3.04% Naididae 2.39% Ostracoda 2.04%
Ceratopogonidae 2.96% Daphniidae 2.12% Daphniidae 1.69%
Chydoridae 2.57% Physidae 1.95% Macrothricidae 1.46%
Macrothricidae 2.44% Ostracoda 1.86% Naididae 1.26%
Ostracoda 2.14% Cyclopoida 1.76% Planorbidae 1.03%
STATION 6
Chironomidae 22.19% Asellidae 32.77% Asellidae 26.39%
Asellidae 17.90% Chironomidae 23.77% Caenidae 16.29%
Caenidae 16.60% Caenidae 13.32% Chironomidae 14.79%
Amphipoda 10.45% Amphipoda 10.11% Ostracoda 12.33%
Ostracoda 6.32% , Cyclopoida 3.50% Amphipoda 9.56%
Cyclopoida 2.88% Ostracoda 3.45% Cyclopoida 2.20%
Naididae 2.71% Naididae 1.85% Naididae 2.15%
Macrothricidae 2.63% Sphaeriidae 1.25% Coenagrionidae 1.88%
Daphniidae 1.80% Sididae 0.84% Lymnaidae 1.48%
Ceratopogonidae 1.79% Planorbidae 0.83% Ancylidae 1.37%

 

81

 

 

 

Table 2.4. Relative abundance changes (increase or decrease) of individual taxa between
years at each station, and p-values from paired t-tests.
Station2 Station 4 Station 6
'96-'97 '97-'98 '96-'97 '97-'98 '96-'97 '97-'98
Insecta Ephemeroptera

Caenidae t 0.001 t 0.686 t 0.087 t 0.044 t 0.498 t 0.304
"Baetidae n.a. n.a. na n.a. t 0.039 1 0.047
Diptera
Chironomidae t 0.002 t 0.032 t 0.441 t 0.054 t 0.697 1 0.048
Ceratopogonidae i 0.003 T 0.069 1 0.001 t 0.553 t 0.211 t 0.449
"Tipulidae i 0.044 n.a. i 0.808 1 0.334 n.a. t 0.334
“Dixidae i 0.336 na n.a. n.a. n.a. n.a.
"Culicidae i 0.336 1 0.334 n.a. n.a. n.a. n.a.
“Sciomyzidae l n.a. T 0.334 n.a. n.a. n.a. n.a.
Brachycera t 0.000 t 0.074 t 0.406 t 0.121 t 0.105 t 0.910
Coleoptera
Dytiscidae t 0.006 t 0.013 t 0.171 t 0.083 t 0.334 1 0.334
“Hydrophilidae t 0.336 t 0.334 t 0.334 t 0.334 n.a. n.a.
"Scirtidae i 0.336 T 0.334 n.a. n.a. n.a. n.a.
"Gyrinidae t 0.336 t 0.334 t 0.334 t 0.710 t 0.164 n.a.
Chrysomelidae n.a. n.a. t 0.168 0.613 t 0.182 0.053
"Haliplidae t 0.336 n.a. i 0.334 n.a. n.a. 11.8
“Lampyridae i 0.166 n.a. n.a. n.a. na n.a.
Trichoptera
Leptoceridae t 0.022 1 0.029 1 0.065 t 0.438 t 0.127 t 0.554
Limnephilidae t 0.015 t 0.102 1 0.049 t 0.334 n.a. n.a.
Phryganeidae t 0.101 t 0.087 t 0.079 t 0.514 t 0.214 t 0.402
Polycentropodidae t 0.056 t 0.081 t 0.238 t 0.485 t 0.376 t 0.146
Hydroptilidae t 0.016 1 0.008 t 0.123 t 0.155 t 0.084 t 0.034
Odonata
Aeschnidae — 0.336 t 0.070 1 0.018 t 0.536 J. 0.392 t 0.633
WGomphidae n.a. n.a. n.a. n.a. n.a. T 0.165
Libellulidae t 0.842 t 0.017 1 0.002 t 0.918 t 0.063 t 0.215
Corduliidae n.a. t 0.334 t 0.093 0.684 t 0.229 t 0.064
Coenagrionidae t 0.001 t 0.003 t 0.858 t 0.488 t 0.008 t 0.026
”Lestidae n.a. t 0.334 t 0.334 t 0.334 - 0.334 n.a.
Hemiptera
Corixidae t 0.045 i 0.045 t 0.953 t 0.803 t 0.140 1 0.154
Veliidae 1 0.454 t 0.040 1 0.334 t 0.022 n.a. na
Mesoveliidae n.a. T 0.334 1 0.152 n.a. 1 0.373 T 0.766
"Gerridae n.a. T 0.165 n.a. T 0.334 n.a n.a.
“Hydrometridae n.a. T 0.334 n.a. n.a. n.a. n.a.
Homoptera 1 0.001 t 0.181 1 0.020 t 0.334 t 0.334 11a.
"Hymenoptera Jr 0.180 n.a. n.a. T 0.166 n.a. na
Lepidoptera l 0.082 t 0.995 t 0.169 n.a. t 0.198 t 0.080
Collembola
Isotomidae i 0.001 t 0.012 1 0.086 t 0.184 t 0.092 T 0.324
Sminthuridae i 0.093 t 0.025 1 0.029 t 0.168 na n.a.
I""‘Poduridae 4 0.194 n.a. n.a. n.a. n.a. n.a.

 

82

Table 2.4 (cont'd).

 

 

Station2 Station 4 Station 6
'96-'97 '97-'98 '96.'97 '97-'98 '96-'97 '97-'98
Crustacea
Isopoda t 0.038 t 0.003 t 0.009 1 0.945 t 0.014 t 0.140
Amphipoda t 0.270 1‘ 0.003 1 0.057 t 0.527 t 0.954 1 0.801
Cladocera
Daphniidae t 0.006 1 0.016 t 0.322 l 0.568 1 0.008 t 0.029
Sididae n.a. n.a. 1 0.710 1 0.047 1‘ 0.037 1 0.037
Macrothricidae t 0.191 1 0.443 t 0.103 t 0.232 t 0.171 t 0.337
Chydoridae t 0.164 t 0.080 1 0.068 1 0.952 1 0.030 1 0.725
“Polyphemidae 4 0.376 n.a. n.a. n.a. n.a. n.a.
Copepoda
Cyclopoida 1 0.020 t 0.146 1 0.022 t 0.015 t 0.708 t 0.357
Harpacticoida t 0.280 t 0.946 t 0.236 t 0.076 n.a. t 0.334
“Calanoida n.a. n.a. n.a. 11a. t 0.334 t 0.714
Ostracoda t 0.052 t 0.269 t 0.706 t 0.798 t 0.184 t 0.069
Arachnida
Hydracarina t 0.800 t 0.260 1 0.667 t 0.835 t 0.316 t 0.133
Oribatei t 0.006 t 0.794 t 0.001 t 0.178 1 0.223 t 0.334
Halacaridae i 0.004 t 0.022 t 0.075 t 0.689 1 0.334 n.a
Araneae 4 0.020 T 0.126 t 0.283 t 0.667 t 0.334 t 0.164
Annelida
Oligochaeta
Naididae t 0.249 1. 0.070 t 0.251 t 0.186 1 0.494 t 0.703
"Tubiﬁcidae t 0.336 n.a. t 0.116 n.a. t 0.120 t 0.227
Lumbriculidae t 0.034 t 0.035 t 0.268 t 0.058 t 0.048 t 0.080
Hirudinea t 0.007 1 0.030 t 0.196 t 0.677 t 0.252 0.302
Mollusca
Pelycypoda
Sphaeriidae t 0.105 1 0.030 t 0.906 1 0.148 t 0.342 .1. 0.944
**Dreissenidae n.a. n.a. n.a. n.a. n.a. T 0.344
Gastr0poda
Ancylidae n.a. n.a. t 0.334 t 0.035 t 0.450 t 0.008
Planorbidae i 0.531 t 0.003 t 0.199 t 0.223 1 0.559 t 0.918
Physidae T 0.072 t 0.041 t 0.048 1 0.004 t 0.052 t 0.858
Suocineidae 1 0.001 t 0.055 t 0.017 t 0.104 t 0.165 t 0.191
Bithyniidae t 0.336 t 0.191 n.a. t 0.189 n.a. t 0.165
"Hydrobiidae n.a. n.a. t 0.334 t 0.334 t 0.084 1 0.284
Lymnaidae t 0.111 t 0.088 t 0.812 t 0.370 t 0.200 t 0.156
Nematoda t 0.075 t 0.677 t 0.007 t 0.703 t 0.051 t 0.090
“Platyhelminthes-Turbellaria t 0.336 t 0.334 t 0.334 t 0.334 t 0.334 t 0.334
**Cnidaria-Hydrazoa t 0.336 l 0.334 t 0.334 t 0.334 t 0.334 n.a.
Total abundance T <.001 t 0.001 t 0.175 t <001 t 0.458 t 0.011
Taxa richness i 0.656 t 0.535 t <.001 t <.001 1 0.015 t 0.068

 

83

Table 2.5. Taxa abundance changes (all stations) with paired t-test signiﬁcances, and peak

shifts among years.

 

 

1996-1997 1997-1998 Abundance Peaks*
% change p % change p '96 '97 '98
Insecta Ephemeroptera
Caenidae +197.8 <.001 +212.5 <.001 4-6 2 4
"Baetidae + *** 0.024 -80.2 0.044 - 6 6
Diptera
Chironomidae +828 0.001 +241 0. 198 4-6 2 4
Ceratopogonidae -72.5 <.001 +3013 <.001 2-4 0 2
"Tipulidae -77.1 0.055 0.0 1.000 2 4 6
"Dixidae -100.0 0.323 - - 2 - -
“Culicidae + *** 0.323 -100.0 0.323 - 2 -
“Sciomyzidae - - + *** 0.323 - - 2
Brachycera -48.l 0.078 +69.0 0.125 2 0 2
ColeOptera
Dytiscidae -100.0 0.003 + *** 0.001 2 0 2
"Hydrophilidae -100.0 0.160 + *** 0.160 2-4 - 2-4
"Scirtidae -100.0 0.323 + *** 0.323 2 0 2
"Gyrinidae -73.3 0.268 +1157.1 0.079 6 0 4
Chrysomclidae +57 .4 0.391 -40. 1 0.087 4-6 6 6
"Haliplidae ~100 0. 160 - - 2-4 0 -
"Lampyridae -100.0 0.160 - - 2 0 -
Trichoptera
Leptoceridae +440 0.364 +25.3 0.660 6 2/6 6
Limnephilidae -91.2 0.001 -57.7 0.433 2 2 0
Phryganeidae +8.2 0.659 -9.0 0.844 4-6 2 4
Polycentropodidae -26.4 0.356 +4000 <.001 4 4 4
Hydroptilidae -50.0 0.610 +180.2 0.089 4 2 4
Odonata
Aeschnidae -65.3 0.037 +5642 0.003 4 4 2
"Gomphidae - - + *** 0.160 - - 6
Libellulidae +398 0.360 -11.9 0.689 4 2 4
Corduliidae +2757.1 0.030 +144.5 0.020 4 6 6
Coenagrionidae +27 .2 0.367 +89.6 0.1 15 6 2/6 6
"Lestidae +5286 0.481 -50.0 0.759 6 4 2
Hemiptera
Corixidae +5.0 0.941 -70.3 0.054 6 0-2/6 4-6
Veliidae -46.7 0.520 +4971.4 0.002 4 2 2
Mesoveliidae -87 .0 0. 130 +200 0.462 4 6 2/6
"Gerridae - - + *** 0.092 - - 2
"Hydrometridae - - + *** 0.323 - - 2
Homoptera -100.0 <.001 + *** 0.092 2 - 2
"Hymenoptera -100.0 0.112 + *** 0.160 2 - 4
Lepidoptera +1000 0.393 -77.3 0.069 2 2 2
Collembola
Isotomidae ~95.9 <.001 +6363.6 0.001 2 0-2 0
Sminthuridae -100 0.007 + *** 0.001 4 - 2
"Poduridae -100.0 0.160 - - 2 - -

 

84

 

Table 2.5 (cont'd).

 

 

 

1996-1997 1997-1998 Abundance Peaks"
% change p % change p '96 '97 '98
Crustacea Isopoda +1420 <.001 +167.6 <.001 6 6 4
Amphipoda -16.9 0.818 +2269 <.001 6 6 6
Cladocera
Daphniidae +520 0.347 +43.7 0.289 4-6 0 4
Sididae +64.7 0.225 -46.5 0.090 4 6 6
Macrothricidae -74.2 0.046 +391.5 0.030 6 0 4
Chydoridae -55.5 0.036 +7 8.6 0. 162 4 0 4
"Polyphemidae -100.0 0.323 - - 2 - -
C0pepoda
Cyclopoida +6.3 0.71 1 +1820 <.001 2 0 2
Harpacticoida -38.6 0.527 +92.5 0.627 2 0 0
“Calanoida + *** 0.323 0.0 1.000 - 6 6
Ostracoda +15.l 0.478 +154.7 <.001 6 2/6 6
Arachnida
Hydracarina +61.4 0.474 +121.9 0.015 6 0-6 6
Oribatei -80.9 <.001 +134.8 0.021 2 2 2
Halacaridae -93.5 0.001 +1918.2 0.004 2 - 2
Annelida
Oligochaeta
Naididae +428 0.328 +522 0.159 2 6 6
"Tubiﬁcidae -100.0 0.013 + *** 0.199 4-6 - 6
Lurnbriculidae +48.9 0.250 +229 -0. 183 4 2 0-2
Hirudinea +51.9 0.163 +608 0.261 6 6 0/4
Mollusca
Pelycypoda
Sphaeriidae -21.5 0.228 -4.4 0.883 2 0 0
a""‘Dreissenidae - - + *** 0.323 - - 6
Gastropoda
Ancylidae +1558 0.263 +1103.0 <.001 6 6 6
Planorbidae +124.5 0.036 -.204 0.810 2 0 0-4
Physidae +115.8 0.054 -38.9 0.117 6 0 2
Succineidae -93.5 <.001 +2086.0 0.004 2 0 2
Bithyniidae +2000 0.633 +1918.2 0.024 - 2 2
"Hydrobiidae + *** 0.028 -57.0 0.345 - 6 4
Lymnaidae -24.7 0.281 +3521 0.043 2 4 6
Nematoda -69.4 0.002 +1420 0.089 4 2 0
“Platyhelminthes-Thrbellaria +2000 0. 194 -50.0 0.306 4 2 4
**Cnidaria—Hydrazoa +5286 0.287 -100.0 0.160 6 0-4 -
Total abundance +55.8 0.001 +117.6 <.001
Taxa richness -26.2 0.016 +25.0 0.003

 

* Station(s) where taxon was most abundant: 0=hummock; 2=wet meadow; 4=transition

zone; 6=deep marsh.

** Rare taxa (10 or fewer individuals taken in all samples).

*** Increased from zero.

85

v .

 

 

 

 

 

 

 

-, . "" 3 { Water
I? " / { Lads
i" l l 1
" 2"}. ~10
ﬁlm‘rﬁ' , |'/ *"l/ . ‘ ‘l - -- , July 1997
a“ .A.‘ \u, 3;. -- -- July 1996
E] 1 l -- -- i - Julyl998
1
El . " ' ' H.425
Sample 3 “ﬁgﬁﬁ ' '
Stations: E] .41
5 E] E

 

Wet Meadow Cattail/ Deep Marsh
Zone Transition Zone Zone

 

Figure 2.1. Schematic cross-section through representative sample transect in study site.

86

 

 

 

 

 

 

 

 

l77-51TIIIIIIIIIIIIIIIII
Project
F O --
staL»;
ft} 177.0 - __
7’ a
> : -
2 a
” a
"a s
’4 176.5 — ‘ 5 l ""
1760 L]- I l4 l l 1 J. l ! l l l l j 99’\! I I
8°9°° 9" W99 \9" “H93" 9‘56 ”9‘9 $0"? 9"“ \‘99 (”09 (309 99 89589 MW
Year

Figure 2.2. Lake Huron monthly average water levels in meters above International Great
Lakes Datum, measured at DeTour Village, MI, 1980 through 1998 (Data source: US.
Department of Commerce, NOAA/NOS, Silver Spring, MD).

87

 

177.20

 

 

 

 

 

 

 

177.10J
g 177.oo~
a d Hummocks
:1 176.90 A
i
Q 176.60-
5 Wet Meadow
.... 17670-1
99’
.8 176.60‘
46 Cattail Zone
2 176.50—
2
g 176.401/J \
‘ Deep Marsh
176.30-1
176.20 IITTIIﬁIII FIIIInI I I I I
8i 33 8i 3’2 '03 3. '03 5. 8i 3 3i 3i
c r- — 13 c - — *5 c - —
a a 5 o a a 5 o g .9 e g

'lntemational Great Lakes Datum

Figure 2.3. Water level changes at Mackinac Bay over the study period.

88

 

 

 

 

 

80,000 _
"e .
g 60,000
8
'8
5 40,000 —
>
.E
“a
0
4!:
20,000 L
0 I I I I I I I I I

6/96 7/96 8/96 6/97 7/97 8/97 6/98 7/98 8/98

Time (m/yy)

Figure 2.4. Average density of invertebrates among all sample stations, by sampling date
(means with standard errors).

89

THEM-nun... viii. «Ii!

 

80,000 '-

60,000 —
— — 1998
40,000 — / l/ \l _ .. .. 1997

20,000—Ik—/"+--" "'—

# of invertebrates/m2

 

 

 

 

0 I r I I

0 2 4 6
Sampling station

Wet Deep

Meadow ‘ ’ Marsh

Figure 2.5. Invertebrate density differences among sampling stations, by year (means with
standard errors). -

9O

 

 

DIMENSION 2

 

 

 

 

l
-1 0

A—0

DIMENSION 1

Figure 2.6. Correspondance analysis ordination of communities at each sampling station
during each year, with year-to—year shiﬁs indicated by arrows.

91

 

 

 

 

 

 

 

DIMENSION 2

 

 

 

 

DIMENSION 1

Figure 2.7. Correspondance analysis ordination of invertebrate communities with
connecting lines representing community gradients during each year.

92

AJ .

 

 

 

 

 

 

 

1.2 -

1.1 '-
3; Sggling
:3 1.0 — 0“
>
:5 — - '- 6
I:
g ...... 4 I;
S
.c: 0.9 P 2
U)

0.8 L- L

0.7 I I I J I

1996 1997 1998
YEAR

Figure 2.8. Shannon diversity of invertebrate samples at each sampling station, by year
(means with standard errors).

93

100%
90%
80%
70%
60%
50%
40%
30%
20%
1 0%

Relative abundance

25

20

15

# taxa/sample

10

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

- all others
I:I Chironomidae
- Caenidae
IIIII hmmﬂb
—i

m J i

1996 I997 1998

YEAR

Figure 2.9. Dominant taxa relative abundances, and changes in taxa richness by year

(means over all stations, with taxa richness standard errors).

94

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

I 1996
2 __
.3 I 1997
e
a I—
:5 F E] 1998
2.:
§
‘ " r
Wet . Deep
“mm Meadow cm“ Marsh

 

*small ﬁsh up to ~50 mm

Figure 2.10. Mean ﬁsh catch per unit eﬂ‘ort among stations and years (means with
standard errors).

95

 

- 1996
C] 1997

I 1998

   

i'liI'Ij'lm =i‘ mm 1.;‘: .
'IIjliilIiﬂIIIIIIII’iIIIII' I; n .I

I
I

   

IIIIIIIiIIIIi

  

# Mature dragonﬂy larvae/sample

 

 

 

 

 

 

 

Hummock Meadow

Figure 2.11. Mean densities of mature dragonﬂy larvae among stations and years (means
with standard errors) ..

96

 

I mature larvae

l —' .
D early-Instar larvae

# individuals/sample

 

 

 

 

 

 

 

 

1996 1997 1998

YEAR

Figure 2.12. Densities of mature and early-instar dragonﬂy nymphs by year (means with

standard errors).

97

Chapter 3

BIOTIC CONTROLS IN AN ABIOTICALLY-VARIABLE HABITAT:
EXPERIMENTS ON THE EFFECTS OF PREDATORS ON A GREAT LAKES
COASTAL WETLAND AQUATIC INVERTEBRATE COWUNITY.

INTRODUCTION

The relative importance of biotic factors in controlling community composition is
believed to vary depending on habitat stability, which often depends on ﬂooding regimes
in shallow lentic systems (Batzer and Wissinger 1996). When placed on a continuum from
permanent to ephemeral ﬂooding, communities in systems that experience permanent or
near-permanent ﬂooding are expected to be structured primarily by predation, while
systems with more temporary ﬂooding are mostly inﬂuenced by the constraints imposed by
ﬂuctuating abiotic conditions (Figure 3.1; Wellbom et a1. 1996). According to this
model, ﬁsh are the dominant predators when present, and large invertebrate predators
dominate in the absence of ﬁsh. In very temporary systems, however, populations of
predators cannot become established well enough to become dominant inﬂuences on the
invertebrate community.

The importance of top-down regulation of invertebrate communities in vegetated
shallow waters can vary among and within sites, depending on ﬂooding regimes and
vegetation structural complexity, among other factors (reviewed in Batzer and Wissinger
1996). Wetlands with ﬁsh usually have lower invertebrate diversity and biomass, but in
vegetated lake littoral zones, ﬁsh can strongly reduce abundances of some invertebrate
taxa and weakly reduce others (Hall et al. 1970, Johnson et al. 1996). When the effects
are weak, it may be because of reduced feeding eﬂiciency in structurally-complex
vegetation (Gilinsky 1984), or because prey communities are "experienced", hence
adapted to their local predator assemblage (Wellbom and Robinson 1991). Invertebrate

predator control can also vary, but is apparently not as effective or as general as ﬁsh

98

predation. Wissinger and McGrady (1993) were able to demonstrate predation effects of
dragonﬂy larvae in lab experiments, but it has been more diﬁicult to demonstrate strong
dragonﬂy larva eﬁ‘ects on invertebrate communities in the ﬁeld (Johnson et al. 1996).
Several studies have shown few or no signiﬁcant reductions of prey densities by these
predators (e. g. Johnson et al.1987), but this may be because high prey turnover balances
predation with recruitment (Benke 1976). Prey may also beneﬁt indirectly ﬁom dragonﬂy
larva predation on other invertebrate predators (Larson 1990), and ﬁsh predation may
have similar eﬂ‘ects.

The Wellbom et al. (1996) model described above is a useful generalization, but
may not apply to all systems. Coastal wetlands of the Great Lakes are characterized by
continuous gradients in ﬂooding regimes along elevation gradients from lake to shore
(Gathman et al. 1999). High-elevation fringing wet meadows are ﬂooded seasonally as
lake levels rise and fall somewhat predictably, so ﬁsh and invertebrates from adjacent low-
elevation permanent zones have direct aquatic access to these temporary zones. This
creates a situation that differs from the model in that the irnperrnanence of ﬂooding does
not fully exclude permanent-water predators ﬁom wet meadows, and distributions of ﬁsh
and large invertebrate predators overlap along the elevation gradient (Chapter 1). In such
cases, it is diﬁicult to predict whether temporary-water invertebrate communities will be
structured by predation pressures from immigrants, or whether the ﬂuctuating abiotic
conditions prevent invading predators fr0m regulating communities.

A further complication in coastal wetlands is that distributions of individual ﬁsh
species vary over the elevation gradient (Cardinale et al. 1997, Gathman and Keas 1999;
Chapter 1). Fish cannot be considered a monolithic functional group because they vary in
size and feeding preferences ontogenetically (Werner and Hall 1988) and taxonomically.
This means that spatial variation in coastal wetland ﬁsh assemblages should create spatial
variation in predation pressures as well. While many ﬁsh predation studies in lake littoral

zones have focused on relatively large invertivores such as Lepomis spp. (e. g. Bohanan

99

 

and Johnson 1983, Gilinsky 1984), fewer (e. g. Johnson et al. 1996) have tried to
determine the effects of the many small-bodied species, and juveniles of larger species, that
are oﬁen abundant in the vegetated shallows of lakes and wetlands.

The conceptual model described above includes a version of the "environmental
sieve" idea (sensu van der Valk 1981), in which abiotic environmental factors prevent
some organisms from occupying more harsh or variable habitats. Conversely, it states that
inhabitants of harsh habitats are excluded ﬁ'om more "benign" habitats by biotic
interactions (predation and competition). However, in the middle of the harsh-to-benign,
or temporary-to-permanent, gradient, both biotic and abiotic factors can be important.
This study focused on predation in the middle of the gradient, with a particular focus on
how predation, rather than abiotic conditions, may prevent permanent-water taxa from
occupying temporary-water wetland zones. I studied predation eﬁ‘ects in a coastal
wetland in which small ﬁsh, especially mudminnows (Umbra limi), and dragonﬂy larvae
(families Libellulidae and Aeschnidae) were common in the lower-elevation portions of the
temporarily-ﬂooded wet meadow zone. I hoped to determine whether these two diﬁ‘erent
types of predators could signiﬁcantly inﬂuence wet meadow invertebrate community

composition by excluding taxa that were characteristic of deeper, more permanent zones

METHODS

Two experiments were run in the Lake Huron coastal wetland described in earlier
chapters. Because the lake level was slowly dropping in 1998, I placed mesocosms in the
transition zone of the wetland, at a lower elevation than the wet meadow, to ensure that
suﬁcient water depth would be present for the duration of the experiments. The majority
of the transition zone was occupied by a dense cattail-dominated plant community, but
openings could be found in the band of emergent plants, so I chose one of these as an
experiment site to minimize vegetation damage. Both experiments were run for 28 days in

late summer, 1998.

100

 

Predator assemblage and ﬁsh density estimates

The ﬁsh and invertebrate predator assemblage varied along the elevation gradient,
though only invertebrate predators (mostly Dytiscidae beetle larvae and adults) were
common at the highest elevations in the shallowest water (Chapter 2). Below this shallow
extreme, the lower wet meadow was dominated by dragonﬂy larvae and small ﬁsh.
Mudminnows (Umbra limi) and Iowa darters (Etheostoma exile) were the most common
"resident" species, while juvenile largemouth bass (Micropterus salmoides), yellow perch
(Percaﬂavescens), and pumpkinseed (Lepomis gibbosus) ﬁequented the wet meadow as
well as lower elevations. Just below the wet meadow was the transition zone, dominated
by cattail plants. Fish density tended to remain high in this zone, but dragonﬂy larva
density and ﬁsh density both tended to decrease with lower elevation (greater depth)
where the transition zone plants gave way to an emergent marsh dominated by bulrushes.
In this zone, larger piscivores were common, small ﬁsh were more sparse, and the most
common dragonﬂy larvae were very sessile, sprawling types (mostly Corduliidae:
Epitheca).

Previous years' sampling indicated that the wet meadow ﬁsh assemblage was
dominated numerically by mudminnows (U. limi), occurring in three size classes: young-
of-year (YOY): ~30 mm mean standard length by late summer; presumed yearlings: ~55
mm mean standard length; larger ﬁsh: ~90 mm mean standard length. Gut analyses
(unpublished data) indicated that YOY mudrrrinnows fed exclusively on meiofauna, while
larger size classes ate macroinvertebrates also. The dragonﬂy assemblage in the site
consisted of several genera of Aeschnidae, Libellulidae, and Corduliidae, distributed
diﬁ‘erently along the elevation gradient. The lower wet meadow was occupied by Aeschna
(Aeschnidae), and three genera (Libellulidae: Libellula, Corduliidae: Dorocordulia,
Somatochlora) that were morphologically indistinguishable in the ﬁeld.

To estimate ambient mudrrrinnow densities, I carefully enclosed three areas (~9.3

m2) of wet meadow using lead-line netting from ﬁsh trap nets (Fyke nets). Nets had ﬂoats

101

along the top edge and I slowly and quietly worked the weighted lower edge into the
bottom detritus so it was ﬂush against the underlying substrate, being careful not to step in
the enclosed area. I then distributed twelve activity traps within the enclosure, laying them
horizontally on the substrate.

I emptied the traps twice daily until two successive checks resulted in no
mudminnows caught in any traps, which occurred within three days in each enclosure.
This method probably resulted in a somewhat conservative estimate of mudrninnow
density because some ﬁsh may have ﬂed the sample area before I ﬁnished the enclosure.
However, previous experience indicated that, when startled, mudminnows tended to hide
in detritus rather than ﬂee, so losses due to disturbance during initial net placement were
probably relatively low. The procedure was effective in catching individuals of several
species other than mudminnows, predominantly young-of-year pumpkinseed, but also
including young brown bullhead (Ameiurus nebulosus) and bowﬁn (Amia calva).
Estimated mean densities based on total numbers of ﬁsh removed from the enclosed areas

were 1.39 ﬁsh/m2 or 1.02 U. limi/mz.

Experiment 1 design and setup

The ﬁrst experiment was a 2x2 factorial randomized complete block design with
10 blocks. The two factors were: 1. presence or absence of 10 dragonﬂy nymphs (5
Aeschnidae and 5 Libellulidae, representing the ambient assemblage), and 2. presence or
absence of 2 young-of-year (YOY) mudminnows (representing a very high density relative
to ambient conditions). The experiment was primarily designed to test: 1) whether
ambient densities of dragonﬂy larvae could signiﬁcantly alter invertebrate abundance
and/or relative abundances of individual taxa; and 2) whether the presence of ﬁsh would
aﬁ'ect predation eﬁ‘ects of dragonﬂy larvae. I did not expect that these small ﬁsh could
signiﬁcantly aﬁ‘ect macroinvertebrate community composition, because preliminary gut

content analysis revealed that their diets mostly consisted of microcrustaceans. I "loaded"

102

the ﬁsh-addition treatments with high ﬁsh densities to test the hypothesis that even a high
density of these ﬁsh could not strongly aﬂ‘ect macroinvertebrates.

Mesocosms consisted of circular enclosures (0.48 m diameter, 0.185 m2 basal
area) made of ﬁberglass window screen (approximately 1mm x 1.5mm mesh size) attached
to plastic hoops at the top and bottom. Screen bottoms were attached to the lower hoops
so experimental units were enclosed by screen on all sides except the top. I installed the
enclosures in ~30 cm deep water, but they were >1m in height to ensure that they could
not be over-topped by waves if a storm surge occurred during the experiment. I installed
10 blocks of 4 closely-spaced enclosures, with each of the 4 treatment combinations
randomly assigned to enclosures in each block (Figure 3.2). I worked one block at a time,
installing enclosures as follows. I ﬁrst decided on the position of the ﬁrst enclosure, and
dropped a separate weighted plastic hoop (same size as enclosure bottoms but without
attached mesh) to the substrate. I used a D-ﬁ'ame dipnet to collect detritus and
invertebrates until I had cleared the area within the hoop down to the underlying dense
root—mat/soil matrix. I quickly transferred collected material to a large plastic water-ﬁlled
tub for temporary keeping. I then removed the hoop and installed the enclosure in its
place. The resulting enclosure was a vertical mesh cylinder with its mesh bottom lying
ﬂush on the substrate and its top extending well above the water surface and held in place
by three wooden stakes around its perimeter. The t0p hoop of the enclosure was stapled
to the stakes, which were ﬁmily planted in the substrate.

I repeated the installation procedure for the remaining three units in the block,
collecting all detritus and animals in the same plastic tub. I carefully examined all
collected detritus, removing dragonﬂy nymphs and retaining them for use in dragonﬂy-
addition treatments. As an attempt to homogenize initial conditions in enclosures within

the block, I carefully mixed and stirred the material in the tub by hand, poured it into a 250
I um mesh sieve-bottomed bucket, and distributed it among the four units in the block,

using a standard brass sieve as a measuring device to ensure that approximately equal

103

portions of material were added to each unit in the block. In all I installed ten blocks of
four units each, in three rows of 4, 4, and 2 blocks. The four treatment combinations were
randomly assigned to experimental units in each block, and predators were added to the
appropriate units. For dragonﬂy-addition treatments, I determined density, and
Aeschnidae/Libellulidae ratio, by counting all the collected larvae (removed from detritus,
as described above) and dividing by the number of enclosures. This resulted in 5
Aeschnidae and 5 Libellulidae larvae collected per enclosure, so I used these densities for
treatments. For ﬁsh additions, I collected mudminnows with activity traps during the

previous two nights.

Eigoeriment 2 design and setup

The second experiment was designed to test whether ambient densities of post-
YOY mudminnows could alter invertebrate abundances and/or relative abundances of
individual taxa. The experiment was a 2-treatment completely randomized design with 5
replicate enclosures per treatment (Figure 3 .3). The ﬁsh-addition treatment consisted of
the addition of two intermediate size-class mudminnows (~55 mm standard length)
collected from the lower wet meadow. This resulted in a 1.57 ﬁsh/m2 mudminnow
density, which was a bit higher than estimates of ambient density, but not unrealistically
so. Previous sampling over three years indicated that mudminnow densities could be more
than double, and total ﬁsh densities more than triple, the estimates at the beginning of this
study (based on comparisons of catches-per-unit-eﬂ‘ort from activitytraps). The circular
mesh enclosures were similar to those used in the ﬁrst experiment, but they were larger
(1.27 m diameter, 1.27 m:2 basal area). Enclosure placement followed the same method
used for the ﬁrst experiment, except that detritus was not mixed before it was added to

enclosures because the experiment was not set up in blocks.

104

Data collection and analysis

At the close of each experiment, I collected all detritus ﬁ'om each enclosure and
transported it ﬁ'om the ﬁeld in plastic bags. The same day, I removed all very coarse
material by hand under running water and preserved the remaining detritus/invertebrates in
jars in 95% ethanol for transport to the lab and subsequent sampling. For sampling I used
the same radial sample divider described in Chapter 2. To sample Experiment 1 enclosure
contents, I thoroughly mixed the detritus/Invertebrate mixture in water, divided the
material in halfusing the sample splitter, and randomly selected 3 one-twelﬁh samples
(one-sixth samples from each half, randomly selected), for a total of one-fourth of all
material sampled. I washed each sample in a 500 um sieve and picked all invertebrates
ﬁom the sample under 10X magniﬁcation. I examined distributions of individual taxon
abundances, relative abundances (as proportions of sample total), total invertebrate
abundance, and taxa richness among treatments and decided to use a log(x+1)
transformation. I analyzed data from the more common taxa using ﬁrlly-factorial 2-way
AN OVAs with blocks. The rarer taxa required Kruskal-Wallis non-parametric tests for
diﬁ‘erences among treatments.

For Experiment 2, I collected 2 one-twelfth samples, for a total of one-sixth of all
material sampled. I pooled the two samples and washed the combined sample through a
graded series of sieves. I picked invertebrates under 10X magniﬁcation from two size
fractions: material retained in a 2 mm mesh sieve (the largest size used), and material
retained in a 1 mm mesh sieve. Data analyses were performed on individual taxon
abundances and relative abundances (calculated as proportion of all invertebrates
represented by each taxon) ﬁom each size ﬁaction and from the two combined. I
examined data distributions for each dependent variable and decided to use log(x+l)
transformations. I analyzed the data using t-tests for the more common taxa and non-

parametric tests for the more rare taxa.

105

RESULTS

The ambient community at the experiment site was dominated by Caenidae
(Caenis) mayﬂy larvae, isopods 'of two genera (Caecidotea and Lirceus), Chironomidae
larvae, and amphipods which were almost exclusively Hyallela azteca (Table 3.1).
Leptoceridae caddisﬂy larvae (0ecetis, T rianodes, and many small unidentiﬁable
specimens) and several gastropod genera were also common. Various other taxa occurred
at lower abundances, including dragonﬂy larvae of diﬁ‘erent sizes, Sphaeriidae clams,
Dytiscidae beetle larvae, and Ceratopogonidae ﬂy larvae. Most of the abundant taxa were
typical of deeper-marsh communities observed during previous years, while most of the
less abundant groups were previously more associated with higher elevations. This

community mixture was a result of dramatic water level changes over the previous years.

Experiment 1

Table 3.2 lists individual ANOVA results by indicating main effects (% abundance
change with predator addition) and signiﬁcance levels for all model terms. I considered
test results weakly signiﬁcant if p-values were between 0.1 and 0.05 and strongly
signiﬁcant if they were less than 0.05. I did not hold strictly to a 0.05 or level because
previous experience indicated fairly high within-group variability in these invertebrate taxa.
Block terms were signiﬁcant in most tests, indicating that either my attempts to
homogenize initial conditions within blocks was successﬁrl or that spatial variation in
environmental factors caused differences among blocks. Because blocks were spatially
segregated, blocks diﬁered in water depth and wave exposure, although these diﬁ‘erences
were very subtle within the ~10m by ~10m experiment area. Nonetheless, adding
dragonﬂies resulted in only a few signiﬁcant differences in taxon abundances (summarized
in Table 3.3), largely because of highly variable dragonﬂy effects among blocks. For
example, dragonﬂy larvae eﬁ‘ects on isopods were strongly negative in some blocks, and

strongly positive or very small in others. However, I found a few negative effects of

106

dragonﬂy addition over all blocks, including weakly signiﬁcant reductions in relative
abundances (proportions of all invertebrates) of Lirceus isopods (—3 0.8%; p=0.079),
damselﬂy larvae (-20%; p=0.079), and Lymnaidae snails (-50%; p=0.093). Also, caddisﬂy
larvae (Trichoptera) in dragonﬂy-addition treatments were 27.3% (p=0.004) less abundant
than in dragonﬂy exclusions. This was mainly because the most abundant family,
Leptoceridae ( T rianodes and 0ecetis), was strongly reduced by dragonﬂy larvae
(p=0.004). The other notable dragonﬂy effects occurred among two less common taxa
(requiring non-parametric tests): Dytiscidae beetle larvae, and Turbellaria (p=0.049 and
p=0.024, respectively; Wilcoxon rank sum test) were signiﬁcantly reduced by dragonﬂy
addition.

Small mudminnows had no negative effects on macroinvertebrates, but did have
positive eﬁ‘ects on two macroinvertebrate groups. Leeches (Hirudinea) and limpets
(Ancylidae) were 100% and 47.7% (p=0.052 and p=0.047, respectively; Wilcoxon rank
sum test) higher in ﬁsh addition treatments than in those without ﬁsh. Also, densities of
small, unidentiable Leptoceridae caddisﬂy larvae were 50.0% higher in ﬁsh treatments
than in no-ﬁsh treatments (p=0.018), which'was the opposite of the dragonﬂy effect on
this group. Negative effects of ﬁsh presence were conﬁned to very small animals
(meiofauna): a 30.4% reduction of cyclopoid copepods (p=0.074, though may not be
meaningful because of interaction eﬁ‘ects, described below), a 41% reduction of Naididae
worms (p=0.012) and a 43.8% reduction of Oribatei soil mites (p=0.032). Meiofauna
were not the focus of this study so sampling methods did not reﬂect their absolute
densities, though relative abundances among samples probably represent abundances of
large-bodied individuals reasonably well.

The more interesting results of ﬁsh addition were signiﬁcant interactions between
predator types that aﬁ‘ected several taxa. Interactions aﬁ‘ected amphipods (p=0.042) and
water mites (Hydracarina; p=0.078) in similar ways. For example, either dragonﬂy larvae

or ﬁsh alone reduced amphipod abundances by 21.5% and 29.1%, respectively, but

107

together these predators had the opposite eﬁ‘ect (Figure 3 .4). Both predators in the same
enclosures resulted in an average 5.7% increase in amphipod abundance, compared to no-
predator controls. When analyzed in a post-hoc contrast, combining dragonﬂy-alone with
ﬁsh-alone treatments vs. the other two, amphipod reduction by solo predator types
averaged 27.4% (p=0.033). The Fish X Dragonﬂy interaction was also signiﬁcant in tests
on cyclopoid copepod abundances @0014), but the effect was the opposite of the eﬁ‘ect
on amphipods. Either predator type alone resulted in higher cyclopoid abundances than in F“
the no-predator controls (ﬁsh alone: +23.1%; dragonﬂies alone: +94.9%), while both
predator types together caused an 17.9% reduction. A weak Fish X Dragonﬂy interaction

 

aﬁ‘ected Sphaeriidae clams similarly (p=0.081). b

Experiment 2

The eﬁ‘ects of larger mudminnows were more substantial than those of dragonﬂies
or small mudminnows, but results varied among size fractions of sample material (Table
3.4). When data ﬁ'om both fractions were combined, mean total invertebrate density was
29.2% lower (p=0.054) among ﬁsh-added enclosures (677 animals/sample, or 3198/m2)
than among no-ﬁsh enclosures (479 animals/sample, or 2263/m2). This difference was
mostly a result of a 42.1% reduction in large-ﬁaction abundance (p=0.017), while the
small-ﬁaction difference was non-signiﬁcant. Mean taxa richness was also signiﬁcantly
lower (p=0.06) in the ﬁsh treatment (19.2 taxa per sample) than in the no-ﬁsh treatment
(15.4 taxa per sample). Tests on the combined size-ﬁaction data revealed that ﬁsh
signiﬁcantly (p<0.05) reduced abundances of Diptera, Mollusca, and Trichoptera, though
diﬁ‘erences within most lower taxonomic groups within these taxa were not signiﬁcant.
Weakly signiﬁcant diﬁ'erences (0.1>p20.05) between treatments occurred in total
abundances of Caenidae mayﬂy larvae and damselﬂy larvae. Abundances of the
numerically-dominant isopods and amphipods were non-signiﬁcantly lower in ﬁsh

treatments, but, conversely, relative abundance of all isopods was signiﬁcantly higher in

108

 

the ﬁsh treatment (p=0.006). Higher relative abundances of both isopod genera occurred
in the ﬁsh treatment, though only the Caecidotea diﬁ‘erence was signiﬁcant (p=0.085).
Individual-taxon data collected from the larger size fraction of samples produced

more signiﬁcant differences among treatments than those from small ﬁactions (Table 3.5).
Among the large-fraction variables, ﬁsh caused strongly signiﬁcant (p<0.05) reductions of
amphipods and Lirceus isopods (Table 3 .6), though these were not signiﬁcantly reduced in
the small fraction. In fact, mean Lirceus density in the small fraction was non-signiﬁcantly TI
higher in the ﬁsh—added treatment, leading to a 50% larger (p=0.013) relative abundance

of Lirceus in small fractions from this treatment. Large-fraction mayﬂy and caddisﬂy

 

larvae were also strongly reduced by ﬁsh, but the small-fraction diﬁ‘erences were not [
signiﬁcant. Isopods, molluscs, damselﬂy larvae, and limpets (all large fraction) were all
less abundant in the ﬁsh treatment, though differences were weakly signiﬁcant. Relative
abundances of isopods in general and speciﬁcally Caecidotea were higher in ﬁsh
treatments (both size fractions), while mayﬂy relative abundance was lower (large fraction

only).

DISCUSSION

These results suggest that larger mudminnows at ambient densities can aﬂ‘ect the
structure of wetland invertebrate communities in Mackinac Bay, but that the effects of
dragonﬂy larvae are weaker. Fish effectively reduced overall densities of invertebrates,
and signiﬁcantly lowered densities of the numerically-dominant taxa, except Caecidotea
isopods. They failed to signiﬁcantly reduce most of the less abundant taxa. However,
almost all taxa occurred at lower abundances (whether statistically signiﬁcant or not) in
ﬁsh enclosures than in no-ﬁsh enclosures, so relative magnitudes of change and signiﬁcant
changes in relative abundances may be more informative measures of potential
mudminnow effects on community composition. Percent reductions suggested that mayﬂy

larvae and amphipods were particularly vulnerable to ﬁsh predation, although relative

109

 

abundance (proportion of all invertebrates) of amphipods was only non-signiﬁcantly
reduced. The substantal effects of these ﬁsh raise the question of whether there was an
"enclosure eﬁ‘ect", or a tendency of ﬁsh to eat more invertebrates ﬁom a 1.27 m2 than
they normally would if not penned. While this possibility cannot be completely reﬁrted
without a separate study, it seems fairly unlikely because mudminnows are known to be
relatively sedentary "ambush" predators (Scott and Crossman 1973 ), so they probably do
not patrol a large area in search of food. 1:

Fish exhibit size-selectivity in prey selection (Mittelbach 1981), favoring larger

 

macroinvertebrates (W ellbom 1994), so it is not surprising that the majority of signiﬁcant
reductions occurred among prey collected from larger size fractions of samples. Among I,
the smaller ﬁactions of samples, no invertebrate taxa were signiﬁcantly reduced by ﬁsh,

though taxa richness was. However, small-fraction relative abundances of isopods (total
and each genus individually) were higher in ﬁsh enclosures, so all other taxa in this size
fraction must have been reduced collectively. This reﬂected the general increase in
relative abundance of isopods in both size fractions and collectively, despite the fact that,
like most other taxa, their absolute abundance was reduced by ﬁsh.

The inability of dragonﬂy larvae to signiﬁcantly alter abundances of the most
common invertebrate taxa was consistent with the fact that previous studies have shown
weak predation effects of dragonﬂies at best (F auth 1999). Dragonﬂies have low
metabolic demands compared to ﬁsh, and because of their sit-and-wait feeding mode they
probably have lower metabolic requirements than more mobile invertebrate predators as
well. Nonetheless, their high densities in the study site initially suggested that their eﬁ'ects
could be substantial. However, dragonﬂy larvae often‘exhibit negative density dependence
in feeding and development rates (Anholt and Werner 1998, Johnson et al. 1995), which
probably results ﬁom interference competition and cannibalism (Crowley and Martin
1989, Hopper et al. 1996, Van Buskirk 1989). Also, it has been suggested that dragonﬂy

larvae reduce densities of other invertebrate predators, leading to indirect survival

110

enhancement in other invertebrate "prey" (Blois-Heulin et al. 1990). In this study,
dragonﬂy larvae signiﬁcantly reduced densities of all other invertebrate predators
collectively by 25% =0.002), so indirect beneﬁts to prey may have counteracted direct
predation losses.

Interactions between small ﬁsh and dragonﬂies aﬁ‘ected abundances of some taxa,
as seen in some other studies (e. g. Swisher et al. 1998). Such interactions could occur
through direct mechanisms, such as dragonﬂy detection of chemical signals from ﬁsh, as
Koperski (1997) observed in damselﬂies, or aggressive behavior by dragonﬂies toward

ﬁsh, but indirect mechanisms are also possible. Some animals, such as nematodes and
copepods in this study, may be able to ﬁnd refuge from one type of predator, but the
presence of two predator types occupying diﬁ‘erent microhabitats may increase the risk of
being eaten. Amphipods, on the other hand, were apparently safer from predation when
both predator types co-occurred. These animals' high mobility and relatively large size
may make them especially vulnerable to both types of predators, but the small size of the
ﬁsh used in the factorial experiment suggests that predation by ﬁsh was not a likely risk for
amphipods. Perhaps direct interactions between predator types enhanced amphipod
survival when both predators were present.

According to the model proposed by Wellbom et al. (1996), taxa from more
permanently-ﬂooded habitats are excluded from temporary habitats by abiotic forces, but
in coastal wetlands it may be biotic forces that prevent many invertebrate taxa ﬁ'om
reaching high densities in wet meadow zones. The experiments in this study consisted of
the addition of wet meadow predators to invertebrate communities that were characteristic
of lower elevations during a previous "average" water-level year (1996; see Chapter 1),
and these predators reduced the abundances of invertebrate taxa that had not co-occurred
with them during that year. Figure 3.5 is the correspondance analysis plot of 1996 data
ﬁom Chapter 1, including the plotted positions of the most common taxa. Those that

were aﬁ‘ected by predators in these experiments are underlined. The combined eﬁ‘ects of

111

larger mudminnows and dragonﬂy larvae reduced the abundances and relative abundances
of most of the deep-marsh-associated taxa (Lirceus isopods, amphipods, and damselﬂy,
mayﬂy, and caddisﬂy larvae), while those taxa most associated with the temporarily-
ﬂooded lower wet meadow (e. g. Chironomidae, snails) were not signiﬁcantly affected.
Taxa that were previously generally conﬁned to the shallowest high-elevation areas (6. g.
larvae of Dytiscidae beetles, Sphaeriidae clams, and Ceratopogonidae ﬂies) were also
negatively aﬁ‘ected by predators. Many smaller taxa that were formerly common in the
wet meadow (copepods, nematodes, Oribatei soil mites, Naididae worms) were apparently
reduced by small ﬁsh, but the high ﬁsh densities used in the factorial experiment were
unrealistic, so these effects would presumably be much lower in natural conditions. A
particularly interesting result was that Caecidotea isopods were not signiﬁcantly reduced
by either predator type, and their proportional representation (relative abundance) in the
community was enhanced by ﬁsh presence. These generalist animals migrated readily into
the wet meadow in previous years, despite mudminnow presence, and their abundances
grew notably over the entire elevation gradient over the three-year period ending with this
study.

Further examination of the changes seen in the previous three-year period
(described in previous chapters) lends more evidence to the roles of predators suggested
here. During the "average" year, invertebrate densities, especially in the wet meadow,
were fairly low. Caecidotea isopods were one of the most successﬁrl immigrants from
low elevations to the mudminnow-occupied wet meadow. In the next year the water level
in the site was much higher, causing permanent wet meadow ﬂooding, and almost all
invertebrate taxa that had previously dominated only the deeper marsh invaded the wet
meadow. Some of these taxa grew in abundance prodigiously, concurrent with a lowered
ﬁsh density (Figure 3.6) which probably relieved predation pressure and "opened" the wet

meadow to previously-excluded taxa. The lower ﬁsh densities at higher elevations may

112

have been a result of invasion by piscivorous ﬁshes (as noted in Chapter 2), suggesting
temporal shiﬁs in food chain dynamics.

The year after high water (which was when these experiments were performed)
saw reduced wet meadow abundances of many previously deep-marsh associated taxa,
which shifted back down the gradient with the lower water levels. However, isopods
(both genera), amphipods, and mayﬂies continued to increase in the wet meadow, which
may have partially resulted from the fact that wet meadow ﬁsh densities were very low.
By late that year most ﬁsh, including mudminnows, were concentrated in the slightly
deeper water just below the wet meadow. Interestingly, however, invertebrate densities
also continued to increase in these lower-elevation zones, despite relatively high ﬁsh
numbers. Some of this increase may have resulted from down-slope migrations as many
invertebrates vacated the wet meadow by migrating or ovipositing in the transition and
deep marsh zones.

These observations and the results of the experiments in this study suggest a model
of community regulation that diﬁ‘ers somewhat from the Wellbom et al. (1996) model.
This new model, illustrated in Figure 3.7, is an attempt to explain invertebrate community
variation along the coastal wetland elevation gradient as seen in an "average" water-level
year ("average" must be used cautiously because year-to-year changes in water level are
very common in the Great Lakes). This model depends on a hypothesized series of effects
beginning with regulation of dissolved oxygen levels along the elevation gradient. As
described in Chapter 2, I measured a dissolved oxygen gradient consisting of saturation
levels in the deep marsh, where oxygen production was presumably high, and a fairly steep
drop through the transition zone to hypoxia in the wet meadow, where oxygen
consumption was probably higher than production (Figure 3.8). Incoming waves probably
mixed some of the oxygenated water into the wetland, setting up the gradient. This would
be similar to the mechanisms driving water chemistry gradients measured by Cardinale et

al. (1998) in a Saginaw Bay coastal wetland, resulting ﬁom a gradual energy reduction in

113

 

waves that moved through the dense vegetation of the wetland. Also, dissolved oxygen
levels rose and fell with seiche-induced water level oscillations, so it appeared that
dissolved oxygen levels were directly related to water levels. Thus dissolved oxygen at
any given point on the gradient is largely determined by water level and the cumulative
plant density through which waves pass to reach the point in question. Following the
chain of mechanisms (Figure 3.9), dissolved oxygen levels determine which species of ﬁsh
occupy a region of the gradient and predation effects of these ﬁsh cascade through the
food web.

Interpreting the experimental and previous observational data in light of this new
conceptual model leads to the following. Wet meadow communities usually consist of
invertebrate taxa that are rapid colonizers of newly-ﬂooded areas, or those that can endure
de—watering and can overwinter in un-ﬂooded substrates (Chapter 1). These animals must
also be able to tolerate very low dissolved oxygen levels that are common in the wet
meadow. Mudminnows dominate the wet meadow ﬁsh community because of their ability
to "breathe" air at the water surface. The upper edge of the ﬂooded wet meadow is
characterized by taxa that can tolerate alternating ﬂooding and draining (caused by
constant seiche activity), but are vulnerable to the predators at lower wet meadow
elevations (mudminnows and dragonﬂy larvae). By contrast, the lower wet meadow
community consists of animals whose populations can withstand culling by these
predators.

Below the wet meadow, dissolved oxygen levels are higher so more ﬁsh species
can occupy these wetland zones. Also, ﬂood duration is longer, so more invertebrate taxa
can occupy the habitat without risk of losing surface water before completing their life
cycles. In this middle portion of the gradient, there is a gradual change in the food web
with greater depth, as piscivorous ﬁsh become more common at lower elevations.
Invertebrates in deep marsh areas may be released from heavy predation pressures because

smaller ﬁsh are reduced in number by piscivores where habitat structure is less complex

114

(Werner and Hall 1988, and reviewed therein). Dragonﬂy larva abundance can also be
reduced by larger ﬁsh (Morin 1984). Deep marsh "prey" invertebrates may be excluded
ﬁom higher elevations because they are more vulnerable to the predators encountered
there, hence their reductions by predators in this study.

This model suggests a spatial shift in trophic cascade eﬁ‘ects, from a very shallow
primary-consumer community with little top-down control, to a mid-range secondary-
consumer-dominated group, and ﬁnally to a deeper tertiary-consumer-dominated food
web. In this way, it is similar to the Wellbom et al. (1996) model, except that the
boundaries between compartments are permeable: there are no distinct boundaries
between habitats, so no boundaries between community members. Because the system is
open, animals can respond to habitat change by relocating, so community membership at
diﬁ‘erent points on the gradient is ﬂuid and mediated by abiotic factors. It is also mediated
by biotic factors, but these are themselves subject to abiotic control.

This temporally-dynamic nature may, on one hand, allow the persistence of some
"ﬁrgitive" taxa that would be eliminated from the community in more stable conditions;
on the other hand, however, it may exclude many taxa that cannot endure or respond to
the abiotic changes characteristic of the system. Thus, the spatial stability of animal
populations implied in the Wellbom et al. (1996) model is not present in this new model.
Not only are communities not spatially stable in this model, but there are no true
communities because their compositions vary constantly. Thus, this model suggests a
Gleasonian, or individualistic, view of communities in which individual invertebrate
populations constantly shift their spatial distributions in response to environmental changes
and largely independent of each other.

This model depends on temporal instability, so it may be less applicable to more
slowly-changing, but otherwise similar, habitats. Wetlands with richly-developed and
expansive wet meadow habitat may be very rare on shores other than the Great Lakes

because: 1) water level ﬂuctuation is usually required for such systems to develop, and 2)

115

human desire for lakeshore properties leads to destruction of wetland habitat above the
mean water level. However, the model may be at least partly helpﬁrl in ﬁrrthering
understanding of littoral wetlands that don't experience regular water-level changes
because it depends on depth and dissolved oxygen as much as ﬂooding duration. The
added eﬁ‘ect of ﬂood duration is to restrict high-elevation invertebrate community
membership to only those taxa that can withstand de-watered periods. Thus, Great Lakes
coastal wetlands may diﬁ‘er ﬁ'om littoral communities of inland lakes because of the unique

suite of hydrologic factors that are characteristic of the Great Lakes.

116

Table 3.1. Ambient invertebrate community composition at experiment locations
(proportions of community comprised by individual taxa).

 

 

Experiment 1 Experiment 2
Caenidae 0.521 Caenidae 0.281
Caecidotea 0.130 Caecidotea 0.204
Macrothricidae 0.061 Chironomidae 0.114
Chironomidae 0.060 Macrothricidae 0.07 8
Cyclopoda 0.055 Amphipoda 0.072
Mphipoda 0.046 Ostracoda 0.040
Lirceus 0.022 Lirceus 0.036
Ostracoda 0.014 @clopoda 0.03 l
Ceratopogonidae 0.009 Naididae 0.019
Planorbidae 0.009 Coenagrionidae 0.014
Naididae 0.008 Leptoceridae 0.010
Hydracarina 0.007 Planorbidae 0.009
Chydoridae 0.006 Sphaeriidae 0.008
Corduliidae 0.005 Chydoridae 0.008
Coenagrionidae 0.005 Lurnbriculidae 0.007
Daphniidae 0.005 Hydracarina 0.007
Polycentropodidae 0.004 Daphniidae 0.007
Lumbriculidae 0.003 Ancylidae 0.006
Isotomidae 0.002 Corduliidae 0.005
Hirudinea 0.002 Ceratopogonidae 0.005
Sphaeriidae 0.002 Hirudinea 0.003
Aeschnidae 0.001 Polycentropodidae 0.003
Sminthuridae 0.001 Sididae 0.001
Limnephilidae 0.001 Nematoda 0.001
Leptoceridae 0.001 Limnephilidae 0.001
Phryganeidae 0.001 Phryganeidae 0.001
Hydroptilidae 0.001 Hydroptilidae 0.001
Libellulidae 0.001 Libellulidae 0.001
Harpacticoida 0.001 Harpacticoida 0.001
Oribatei 0.001 Oribatei 0.001

 

117

Table 3.2. Experimentl main eﬂ‘ects (% difference between factor levels) and
signiﬁcance levels of model terms from ANOVAs based on log-transformed abundances.

 

 

Draggnﬂy larvae [Ls]; DragXFish Block
% change p % change p
Ephemeroptera: Caenidae +18.7 0.276 -7.4 0.731 0.478 <0.001
Diptera: Chironomidae +0.2 0.713 -10.1 0.139 0.802 <0.001
Ceratopogonidae +7.5 0.579 -5. 1 0.982 0.895 0.081
Coleoptera: Dytiscidae“ -64.0 0.049 -33 .0 0.252
Trichoptera: Leptoceridae -27.3 0.009 +125 0.302 0.776 <0.001
Phryganeidae“ -42.0 0.738 +40 0.558
Polycentropodidae“ -46.0 0. 137 -9.0 0.47 l
Odonata: -l7.6 0.103 -0.9 0.931 0.287 0.006
Coenagrionidae
Hemiptera: Corixidae* -50.0 0.553 -50.0 0.553
Collembola: Isotomidae“ +20.0 0.780 +44.0 0.780
Sminthuridae” +50 0.637 -33.0 0.637
C mstacea
Isopoda
Caecidotea -7.7 0.613 -0. 3 0.708 0.965 0.046
Lirceus -23.1 0.272 +14.6 0.181 0.423 0.003
Amphipoda +7 .8 0.860 - l. 1 0.383 0.042 0.002
Cladocera“ +80 0.661 -9.0 0.431
Copepoda: Cyclopoida +24. 1 0.400 -30.4 0.074 0.019 <0.001
Ostracoda +1 1.0 0.385 +8.0 0.516 0.876 0.022
Arachnida
Hydracarina -8.4 0.455 +1.3 0.999 0.208 0.129
Oribatei +289 0.3 16 43.8 0.032 0.737 0.226
Annelida '
Oligochaeta: Naididae +32.7 0.205 -41.0 0.012 0.848 0.446
Lurnbriculidae +12.8 0.367 +4.1 0.666 0.714 0.002
Hirudinea + 10.0 0.480 +100.0 0.034 0.042 0.226
Mollusca
Pelycypoda: Sphaeriidae* +30.0 0.350 +17 .0 0.289
Gastropoda: Ancylidae“ -65.0 0. 136 +48.0 0.047
Planorbidae 0.0 0.996 +8.0 0.516 0.327 0.072
Physidae“ -45.0 0.248 +22.0 0.773
Lymnaidae“ -43 .0 0.180 +20 0.771
Hydrobiidae“ -69.0 0.427 -40.0 0.460
Nematoda -20.6 0.555 -24.6 0.476 0.078 0.411
Platyhelminthes-Turbellaria* -82.0 0.024 +17 0.929
Total invertebrate predators -25.0 0.002 +12. 1 0. 135 1.000 0.000
Total abundance +1 1.3 0.241 -6.0 0.493 0.234 0.003
Taxa richness -3.5 0.178 +0.7 0.784 0.414 0.034

 

*signiﬁcanoe levels represent Kruskal-Wallis test results

118

Table 3.3. Signiﬁcant effects of dragonﬂy larvae: percent changes in dependent variables
when dragonﬂy larvae were added to enclosures.

 

 

Dependent variable % difference p
Trichoptera abundance ‘ -27. 6 0004*
Leptoceridae abundance -27 .3 0004*
Lirceus relative abundance -3 0.8 0079*
Coenagrionidae relative abundance -20.0 0079*
Lymnaidae relative abundance -50.0 0093*
Dytiscidae abundance -64.0 0.049“
Turbellaria abundance -82.0 0.024"

 

* ANOVA of log-transformed data
** Wilcoxon rank-sum test

119

Table 3.4. Eﬁ‘ects of ﬁsh addition on taxa abundances and signiﬁcance levels from
Experiment 2 t-tests on log-transformed abundances (effects are % diﬁ‘erences relative to
no-ﬁsh controls) .

 

 

Size ﬁaction: Large M M
% change p % change p % changre p
Insecta -
Ephemeroptera: Caenidae -70.4 0.017 -34.9 0.246 -47.1 0.078
Diptera: -67 .9 0.159 -3 1.2 0.171 -63.6 0.043
Chironomidae -51.3 0.271 +0.5 0.898 -35.4 0.320
Ceratopogonidae -100.0 <0.001 -89.7 0. 161 -94.1 0.015
Coleoptera: Dytiscidae* -37 .5 0.365 +343 0.825 -21.1 0.480
Trichoptera: 45.5 0.025 -35.5 0.448 41.9 0.044
Leptoceridae -70.0 0.048 -49.6 0.242 -62.2 0.014
Phryganeidae* -20.0 0.410 -32.9 0.861 -18.7 0.405
Polycentropodidae“ -54.4 0. 174 +3 .8 0.718 -26.0 0.211
Odonata: Coenagrionidae -36.8 0.078 +0.4 0.970 -30.9 0.087
Libellulidae -55.9 0.254 - - -55.9 0.254
Crustacea
Isopoda: -31.2 0.066 +11 0.522 -16.4 0.267
Caecidotea -29.1 0.163 -3.3 0.887 ~19.9 0.315
Lirceus -35.5 0.029 +32.5 0.134 -12.1 0.386
Amphipoda -57.8 0.037 -26.8 0.208 -34.6 0.110
Mollusca -66.3 0.1 14 -38.7 0.396 -54.3 0.047
Pelycypoda: Sphaeriidae“ -82.3 0.074 +9.3 0.823 -50.5 0.331
Gastropoda: -62.8 0.178 -40.8 0.382 -52.6 0.750
Ancylidae -87.6 0.081 +3 1.22 ' 0.855 -77.8 0.108
Planorbidae -60.9 0.153 -43 .0 0.435 -51.2 0.104
Physidae -38.1 0.380 - - ~38.1 0.380
Lymnaidae* -1.6 0.147 -46.7 0.451 -61.9 0.343
Hydrobiidae* + 14.1 0.906 +1597 0.234 +79.6 0.203
Total abundance -42.1 0.017 -17 .0 0.273 -29.2 0.054
Taxa richness -30.7 0.060 -19.0 0.086 -19.8 0.060

 

*signiﬁcance levels represent Kruskal-Wallis test results

120

Table 3.5. Eﬁ‘ects of mudminnows: differences in number of signiﬁcant results between
prey of diﬁ‘erent sizes

 

 

 

Sample # of taxa showing signiﬁcant
size fraction differences between treatments
Large (>2mm) 4*, 6**
Small (1mm-2mm) None
Both combined 2*, 5**
* 0.1 > p 2 0.05
** p < 0.05

121

 

Table 3.6. Eﬂ‘ects of mudminnows: % changes in some large-fraction dependent variables
when intermediate size-class ﬁsh were added to enclosures.

 

 

Dependent variable % difference p

Isopoda abundance -3 1.2 0.066
Amphipoda abundance -57 .8 0.037
Caenidae abundance -70.4 0.017
Chironomidae abundance -51.3 0.271
Trichoptera abundance -45.5 0.025
Coenagrionidae abundance -36.8 0.078
Sphaeriidae abundance -82.3 0.074
Gastropoda abundance -62.8 0. 17 8
Total invertebrate abund. -42.l 0.017
Taxa richness -30.7 0.060
Isopod relative abundance +25.0 0.067

 

* ﬁom ANOVA of log-transformed data

122

 

Temporary I Permanent,

 

 

 

 

 

I
Habitats I ﬁshless habitats I “it“! with “I
Periodic drying I Winter oxygen stress I
I "L ~N
I Predatory

/ l 54.21

~L
I ‘ Invertebrate A ’r
I top Predators I / :
I
’I’
I

 

 

 

 

 

 

 

 

 

 

A’
Very active, rapidly Large, moderately I Small, inactive
developing prey -' active prey i prey
I I
Permanence Predator
transition transition

Gradient ﬁom small ephemeral ponds to large permanent lakes
’

 

Figure 3.1. Wellbom et al. (1996) conceptual model of community structuring forces in
ﬁeshwater systems.

123

 

 

Treatment combinations
randomly assigned within blocks:

 

 

 

 

 

 

 

 

 

oo oo
82‘... oo oo

oo oo oo 00
D... 00 oo oo oo
””8“ oo oo oo 00
l 00 0000 00

 

 

Figure 3.2. Plan view of Experiment 1 in-site block layout (d = no dragonﬂy larvae; D =
dragonﬂy larvae added; f = no ﬁsh; F = ﬁsh added).

124

 

 

   

~lOm

 

 

 

Figure 3.3. Plan view of Experiment 2 in-site enclosure placement, indicating ﬁsh addition
and exclusion.

125

without ﬁsh with ﬁsh

 

 

 

 

 

 

 

 

47 47
o
o
5
13 37 — 37 r
r:
:r
.o
a
'o
o
a.
3‘;
E 27 " 27 “

17 I I 17 I I
without with without with
dragonﬂy dragonﬂy dragonﬂy dragonﬂy
larvae larvae larvae larvae

Figure 3.4. Effects of Fish X Dragonﬂy-larvae interactions on abundances of amphipods
in experimental enclosures in Experiment 1 (means with standard errors).

126

 

 

 

 

 

Sphaeriidae
1 _.
1
N M Qitiscidae
g 6 Caecidotea
r; . Chironomidae Plano rbi dae
E 0 _ Leptoceridae M919 2
'-‘ Lymnaidae
: Hyalellidae 3 Umbra limi
_ Libellulidae
Caenidae
Coenagrionidg
-1 l l l
-l DINIENSION l 1

Figure 3.5. Correspondance analysis plot of 1996 data including plotted positions of
predators used in experiments and their affected prey. Underlined taxa are those that were
signiﬁcantly aﬁ‘ected by predators in enclosures.

127

 

— 1996

 

 

—— All Fish
A, ------ Umbra Iimi

 

 

# caught per 48 hrs.

 

 

 

 

 

’

 

Meadow Marsh

Figure 3.6. Spatial distributions of mudminnows (Umbra limb and all ﬁsh species
combined during each of 3 years ending with the experiment year (means with
standard errors).

128

 

 

Seasonlly-ﬂooded Intermediate “MM
high elevations elevations low elevations

Low D.O. Shallow water

I I
Periodic drying : a low 0.0. ’
I

 

 

 

 

 

 

I
High-elevation .
. . Lowelevatron
speculrstprey : 1 ‘L spe - l' '
Midelevation
? ——————— b generalist prey ‘- - - - - ?

 

 

 

Figure 3.7. Conceptual model of regulation of spatial community variation in Mackinac
Bay wetland. Solid arrows represent strong negative effects and dashed arrows represent
weaker eﬁ‘ects (following diagram format of Wellbom et al. 1996).

129

Waves (height depends on water depth)

 

 

ir‘ﬂtgv::"r A I E ‘I ‘ . ‘

 

 

 

 

 

 

. I U "‘ I .- —
High D.0. ' f , l . High 0.0.
consrnnption «of. m :53? I. production
~ ..
Wave Energy
1 l l l I l I

 

 

 

 

 

Figure 3 .8. Observed spatial patterns in certain abiotic factors in Mackinac Bay wetland.

 

130

 

 

Dissolved

[ Water level I (H
I Oxygen

 

  
   
 

 

 

 

 

 

 

 

 

(-)
Cumulative plant -
density Small ﬁsh (and Piscivores
* dragonﬂy larvae)
(-)

 

Prey abundance

 

 

Figure 3.9. Hypothesized chain of mechanisms describing how primary wetland abiotic
factors regulate biotic controls on prey invertebrates.

 

131

 

CONCLUSIONS

The results presented in the previous chapters suggest that Great Lakes coastal
wetland communities are very dynamic, as are the hydrologic conditions in which they are
found. Spatial and temporal complexity in these systems appears to be very high
compared to adjacent systems, requiring much further work to fully understand how
communities are aﬁ‘ected by environmental factors. Chapter 1 provides a substantial data
set upon which future studies in the Les Cheneaux Islands region may be based, while
Chapters 2 and 3 indicate ﬁJture research directions for such studies.

Many environmental factors vary along the lake-to-upland coastal wetland
ecotone, but ﬂooding characteristics appear to be the most important ultimate driving
force behind animal distributions in these complex habitats. Animal communities in wet
meadows are affected by periodic draining and, conversely, by the inﬂux of immigrating
taxa during high-water periods. Those animals surviving in temporarily-ﬂooded wet
meadows have adapted to the ﬂooding regime through drought-tolerance or drought-
avoidance mechanisms. A shift from temporary to year-round (or longer) ﬂooding allows
many taxa from lower elevations to take advantage of wet meadow habitat, which may
negatively aﬁ‘ect the temporary-adapted taxa through predation and competition.
However, as seen in this study, predation pressures from small ﬁsh may actually decrease
during high-water periods (probably because piscivorous ﬁsh increase in number),
allowing many invertebrate taxa to ﬁrlly exploit the newly-available habitat. Lower-
elevation communities appear less affected by water level changes, but extreme lows, not
seen in this study, should have effects that are somewhat predictable. Iflower-elevation
zones are temporarily drained, the invertebrate community there would probably tend to
resemble that seen in the 1996 wet meadow, as described in Chapter 1, because the

animals usually found at low elevations are not adapted to temporary-ﬂooding conditions.

132

 

 

Ifhigh water is sustained for several years, vegetation zones would be expected to
shiﬁ up-slope, converting wet meadows to transition and deep marsh zones (Burton
1985). Sustained lows should have opposite effects. While vegetation zonation may not
have strong direct effects on many animals, it's indirect effects may be very important. As
cumulative plant density increases along an imaginary transect from low to high elevations,
wave energy and water mixing are reduced, creating water chemistry gradients (Cardinale
et al. 1997). These gradients may affect animals in several ways, but the dissolved oxygen
gradient in particular is probably largely responsible for ﬁsh distributions within coastal
wetlands. Chapter 3 suggests that small ﬁsh can affect invertebrate communities, so
factors affecting ﬁsh should also affect their prey. The suggestion of a spatial shift in the
wetland trophic web should be studied if we are to develop a complete understanding of
community dynamics in coastal wetlands.

This study suggests the importance of different migratory abilities among
invertebrate taxa in structuring communities found at different wetland elevations.
Because some animals (particularly non-insects) migrate more slowly than others, some
invertebrate community changes lag behind water-level changes. Slow migrators follow
multi-annual rising water levels, but when levels fall, the receding water line appears to
leave many slow—responders stranded in rapidly-draining habitats. Only those taxa that
can withstand such draining will survive these conditions, while others must depend on
more broad habitat selection for their populations to persist. If global climate changes
cause more extreme water-level variation in the Great Lakes, as suggested by Mortsch
(1998), wetland invertebrate diversity may decrease as only generalist species will be able
to survive.

Other threats to coastal wetland communities, particularly in the Les Cheneaux
region, include the effects of shoreline development at the upland end of the wetland
continuum, and increased boating activity and channel dredging at the lower end

(Gathman and Keas 1999). These two factors can create barriers to up-slope and down-

133

 

s10pe migrations of plant and animal communities, reducing habitat and community
heterogeneity across the gradient by truncating its ends. This study suggests that the ﬁrll
range of habitat conditions may be required to maintain wetland diversity, so regional
planning must be based on long-term probabilities of lake-level variations. This study,
combined with other recent work, should be a useful starting point for developing such a
long-term outlook, and also may be a foundation for the development of a complete model
of the mechanisms driving community variation in some types of Great Lakes coastal

wetlands.

134

 

LITERATURE CITED

 

Literature Cited

Albert, D. A, S. R. Crispin, G. A. Reese, L. A. Wilsmann, and S. J. Ouwinga. 1987. A
Survey of Great Lakes Marshes in Michigan's Upper Peninsula. Report to the
IVﬁchigan Department of Natural Resources, Land and Water Management Division.

Anholt, BR, and BE. Werner. 1998. Predictable changes in predation mortality as a
consequence of changes in food availability and predation risk. Evolutionary Ecology
12:729-738.

Barlocher, F., R.J. Mackay, and GB. Wiggins. 1978. Detritus processing in a temporary
vernal pool in southern Ontario. Archiv ﬁrr Hydrobiologie 81:269-295.

Bataille, K.J., and GA. Baldassarre. 1993. Distribution and abundance of aquatic
macroinvertebrates following drought in three prairie pothole wetlands. Wetlands
13(4):260-269.

Batzer, DP, and VB. Resh. 1992. Macroinvertebrates of a California seasonal wetland
and responses to experimental habitat manipulation. Wetlands 12: 1-7.

Batzer, D. P., and S. A. Wissinger 1996 Ecology of insect communities in nontidal
wetlands. Annual Review of Entomology 41: 75- 100.

Benke, AC. 1976. Dragonﬂy production and prey turnover. Ecology 57:915-927.

Blois-Heulin, C., P.H. Crowley, M. Arrington, and D.M. Johnson. 1990. Direct and
indirect effects of predators on the dominant invertebrates of two freshwater littoral
communities. Oecologia 84:295-306.

Bohanan, RE, and D.M. Johnson. 1983. Response of littoral invertebrate populations to
a spring ﬁsh exclusion experiment. Freshwater Invertebrate Biology 2:28-39.

Brady, V.J. 1992. The invertebrates of a Great Lakes coastal marsh. MS. Thesis.
Michigan State University, East Lansing, Michigan.

Brady, V.J., and TM. Burton. 1995. The rnicrocrustacean community of a Saginaw Bay
coastal emergent marsh. Pages 325-341 in M. Munawar, T. Edsall, and J. Leach,
editors. The Lake Huron ecosystem: ecology, ﬁsheries and management. SPB
Academic Publishing, Amsterdam, The Netherlands.

Brazner, J .C. 1997. Regional, habitat, and human development inﬂuences on coastal

wetland and beach ﬁsh assemblages in Green Bay, Lake Michigan. Journal of Great
Lakes Research 23: 36-51.

136

 

Burton, TM. 1985. The effects of water level ﬂuctuations on Great Lakes coastal
marshes. Pages 3-13 in H. H. Prince and F. M. D'Itri, editors. Coastal wetlands.
proceedings of the ﬁrst Great Lakes coastal wetlands colloquium. Lewis Publishers,
Inc., Chelsea, Michigan.

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

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
changes in the abiotic environment? Can. J. Fish. Aquat. Sci. 54: 2314-2322.

Cole, R. A., and D. L. Weigman. 1983. Relationships among zoobenthos, sediments, and
organic matter in littoral zones of western Lake Erie and Saginaw Bay. Journal of
Great Lakes Research 9: 568-581.

Collinson, N.H., J. Biggs, A Corﬁeld, M.J. Hodson, D. Walker, M. Whitﬁeld, and P.J.
Williams. 1995. Temporary and permanent ponds: an assessment of the eﬁ‘ect of
drying out on the conservation value of aquatic macroinvertebrate communities.
Biological Conservation. 74: 125-133.

Comer, P.J., D.A. Albert, H.A. Wells, B.L. Hart, J .B. Raab, D.L. Price, D.M. Kashian,
R.A. Corner, and D.W. Schuen. 1995. Michigan's native landscape, as interpreted
from General Land Oﬁice Surveys 1816-1856. Michigan Natural Features Inventory
Report to Water Division, US Environmental Protection Agency, and Wildlife
Division, Michigan Department of Natural Resources. Lansing, Michigan. 74:125-
133.

Cowardin, L.M., V. Carter, F.C. Golet, and ET. LaRoe. 1979. Classiﬁcation of wetlands
and deepwater habitats of the United States. US. Fish and Wildlife Service
Publication FWS/OBS-79/31. Washington DC.

Crowley, PH, and BK. Martin. 1989. Functional responses and interference within and
between year classes of a dragonﬂy population. Journal of the North American
Benthological Society 8:21 1-221.

Duffy, W.G., and TR. Batterson. 1987. The St. Mary's River, Michigan: an ecological
proﬁle. US. Fish and Wildlife Service Biological Report 85 (7.10). Washington,
DC.

Fauth, J.B. 1999. Identifying potential keystone species ﬁom ﬁeld data: an example from
temporary ponds. Ecology Letters 2:36-43.

137

 

Gathman, J .P., TM. Burton, and 3.]. Amritage. 1999. Coastal wetlands of the upper
Great Lakes: Distribution of invertebrate communities in response to environmental
variation. In: Invertebrates in ﬁeshwater wetlands of North America: Ecology and
management. John Wiley and Sons, Inc. New York.

Gathman, JP. and BE. Keas. 1999. Les Cheneaux Coastal Wetland Project: A
Synthesis. Report submitted for The Nature Conservancy, Michigan Chapter to the
Michigan Coastal Management Program. Project number 99-3 099-24.

Gilinsky, E. 1984. The role of ﬁsh predation and spatial heterogeneity in determining
benthic community structure. Ecology 65:45 5-468.

Greenacre, HG. 1984. Theory and applications of correspondance analysis. Academic
Press. London.

Hall, D.J., W.E. Cooper, and BE. Werner. 1970. An experimental approach to the
production dynamics and structure of ﬁeshwater animal communities. Lirnnology and
Oceanography 15:839-928.

Hanson, BA, and GA. Swanson. 1989. Coleoptera species inhabiting prairie wetlands
of the Cottonwood Lake area, Stutsman County, North Dakota. Prairie Naturalist
~21(1):49-57.

Hilsenhoﬂ‘, W.L. 1988. Rapid ﬁeld assessment of organic pollution with a family-level
biotic index. Journal of the North American Benthological Society 7(1):65-68.

Hopper, K.R., PH. Crowley, and D. Kielman. 1996. Density dependence, hatching
synchrony, and within-cohort cannibalism in young dragonﬂy larvae. Ecology 77: 191-
200.

Hubbs, CL, and KP. Lagler. 1967. Fishes of the Great Lakes region. The University of
Michigan Press. Ann Arbor, Michigan.

Jaworski, E., and ON. Raphael. 1978. Coastal wetlands value study in Michigan. Phase
I and II. Fish, wildlife, and recreational values of Michigan's coastal wetlands. Report
to the Michigan Coastal Management Program, Michigan Department of Natural
Resources, Lansing, Michigan.

Jeﬁiies, M. 1994. Invertebrate communities and turnover in wetland ponds affected by
drought. Freshwater Biology. 32:603-612.

Johnson, D.M., T.H. Martin, P.H. Crowley, and LB. Crowley. 1996. Link strength in

lake littoral food webs: net eﬂ‘ects of small sunﬁsh and larval dragonﬂies. Journal of
the North American Benthological Society 15:271-288.

138

Johnson, D.M., T.H. Martin, M. Mahato, L.B. Crowder, and PH. Crowley. 1995.
Predation density dependence, and life histories of dragonﬂies: a ﬁeld experimnet in a
freshwater community. Journal of the North American Benthological Society 14:547-
562.

Johnson, D.M., C.L. Pierce, T.H. Martin, C.N. Watson, R.E. Bohanan, and PH. Crowley.
1987. Prey depletion by odonate larvae: combining evidence ﬁ'om multiple ﬁeld
experiments. Ecology 68: 1459-1465.

Jude, DJ, and J. Pappas. 1992. Fish utilization of Great Lakes coastal wetlands. Journal
of Great Lakes Research 18:651-672. ‘

Koperski, P. 1997. Changes in feeding behavior of the larvae of the damselﬂy Enallagma
cyathigerum in response to stimuli from predators. Ecological Entomology 22: 167-
175

Krecker, F. H. 1939. A comparative study of the animal population of certain submerged
aquatic plants. Ecology 20:553-562.

Krieger, K.A. 1992. The ecology of invertebrates in Great Lakes coastal wetlands:
current knowledge and research needs. Journal of Great Lakes Research 18(4): 634-
650.

Larson, DJ. 1990. Odonate predation as a factor inﬂuencing dytiscid beetle distribution
and community stmcture. Quaestiones Entomologicae 26: 151-162.

Liston, CR, and S. Chubb. 1985. Relationships of water level ﬂuctuations and ﬁsh.
Pages 121-134 in H. H. Prince and F. M. D'Itri, editors. Coastal wetlands. .
proceedings of the ﬁrst Great Lakes coastal wetlands colloquium. Lewis Publishers,
Inc., Chelsea, Michigan.

Maynard, L., and D. Wilcox. 1997. Background Report on Great Lakes Wetlands
prepared for the State of the Great Lakes 1997 report issued jointly by the US
Environmental Protection Agency Great Lakes National Program Ofﬁce, Chicago,
Illinois, and Environment Canada, Toronto, Ontario.

McLaughlin DB, and HI. Harris. 1990. Aquatic insect emergence in two Great Lakes
marshes. Wetlands Ecology and Management 12111-121.

Merritt, R.W., and K.W. Cummins. 1996. An introduction to the aquatic insects of North
America. Kendall/Hunt Pub. Co. Dubuque, IA. Third edition.

Mine, L. 1996. Michigan's Great Lakes coastal wetlands: deﬁnition, variability, and

classiﬁcation. Report to Michigan Natural Features Inventory. East Lansing,
Michigan.

139

 

Mitsch, W.J. and J .G. Gosselink. 1993. Wetlands. New York: Van Norstrand Reinhold.
Second edition

Mittelbach, G.G. 1981. Foraging eﬂiciency and body size: a study of optimal diet and
habitat use by bluegills. Ecology 62: 1370-1386.

Morin, R]. 1984. The impact of ﬁsh exclusion on the abundance and species
composition of larval odonates: results of short-terrn experiments in a north carolina
farm pond. Ecology 65:53-60.

Mortsch, L.D. 1998. Assessing the impact of climate change on the Great Lakes
shoreline wetlands. Climatic Change 40: 391-416.

Neckles, H.A., H.R. Murkin, and J .A. Cooper. 1990. Inﬂuences of seasonal ﬂooding on
macroinvertebrate abundance in wetland habitats. Freshwater Biology. 23:311-322.

Nelson, J .W., J .A Kadlec, and HR. Murkin. 1990. Responses by macroinvertebrates to
cattail litter quality and timing of litter submergence in a northern prairie marsh.
Wetlands. 10:47-60.

Pennak, R.W. 1989. Freshwater invertebrates of the United states: Protozoa to
Mollusca. J. Wiley & Sons. New York.. Third edition.

Prince, H.H., and TM. Burton. 1996. Final Report, 1995: Wetland restoration in the
coastal zone of Saginaw Bay. Report to the Michigan Department of Natural
Resources. Michigan State University, Department of Fisheries and Wildlife.

Prince, H.H. and FM. D'Itri. 1985. Coastal wetlands. Proceedings of the ﬁrst Great
Lakes coastal wetlands colloquium. Lewis Publishers, Inc., Chelsea, Michigan.

Prince, H.H., P.I. Padding, and R.W. Knapton. 1992. Waterfowl use of the Laurentian
Great Lakes. Journal of Great Lakes Research 18:673-699.

Schneider, D.W., and TM. Frost. 1996. Habitat duration and community structure in
temporary ponds. J. N. Am. Benthol. Soc. 15:64-86.

Scott, W.B. and El Crossman. 1973. Freshwater ﬁshes of Canada. Bulletin 184.
Fisheries Research Board of Canada, Ottawa.

Suzuki, N., S. Endoh, M. Kawashima, Y. Itakura, C.D. McNabb, FM. D'Itri, and TR.

Batterson. 1995. Discontinuity bar in a wetland of Lake Huron's Saginaw Bay.
Journal of Freshwater Ecology 10:111-123.

140

Swisher, B.J., D.A. Soluk, and DH. Wahl. 1998. Non-additive predation in littoral
habitats: inﬂuences of habitat complexity. Oikos 8:30-37.

van Buskirk, J. 1989. Density dependent cannibalism in larval dragonﬂies. Ecology
70: 1442-1449.

van der Valle AG. 1981. Succession in wetlands: a Gleasonian approach. Ecology
62:688-696.

Voights, D.K. 1976. Aquatic invertebrate abundance in relation to changing marsh
vegetation. American Midland Naturalist 95:313-322.

Wellbom, GA 1994. Size-biased predation and prey life histories: a comparative study
of ﬁ'eshwater amphipod populations. Ecology 75:2104-2117.

Wellbom, GA, and J .V. Robinson. 1991. The inﬂuence of ﬁsh predation in an
experienced prey community. Canadian Journal of Zoology 69:2515-2522.

Wellbom, G.A., D.K. Skelly, and EE. Werner. 1996. Mechanisms creating community
structure across a ﬁ’eshwater habitat gradient. Annual Review of Ecology and
Systematics 27:337-363.

Werner, BE, and DJ. Hall. 1988. Ontogenetic habitat shifts in bluegill: the foraging
rate-predation risk trade-off. Ecology 69: 1352-1366.

Wiggins, G.B., R.J. Mackay, and 1M. Smith. 1980. Evolutionary and ecological
strategies of animals in annual temporary pools. Archiv fur Hydrobiologie-
Supplement 58:97-206.

Williams, DD. 1987. The ecology of temporary waters. Timber Press. Portland, OR.
Originally Croom Helm Ltd. Beckenham, England.

Wissinger, S., and J. McGrady. 1993. Intraguild predation and competition between
larval dragonﬂies: direct and indirect effects on shared prey. Ecology 74:207-218.

Zedler, PH. 1981. Microdistribution of vernal pool plants of Keamy Mesa, San Diego

County. In: Vernal pools and intermittent streams. Jain, S., and P. Moyle (eds).
Institute of Ecology, U. California, Davis, Pub. No. 28. pp 185-197.

141

a .l
.15.... trier:
sill}:-
3; .1: o I.

 

r 2.
i3... {In-I
1., I555:

 

 

 

 

 

 

£4.15 3
Jun»... 181
i... .

ls

 

.61
bit. .
r. a):
Sieve-B".
. .i-i d

s

i i.e..

. FE... 8.35.5.
. at! . .
53-...

.5
.

 

 

 

 

 

 

.

r

or

 

.u
8.2... Te.