«

.n.

,
”‘3“? .
' §.~‘¢3_';;“.V 2.531
xv ‘ 1?“, _ ..
W -

s.‘ ‘

“3'3 ‘ 1 . $473..
ii: 42:.“ ‘ Lu
-.:fi"«r’-‘€

.. v

t

    

‘ .L .-

‘téwmae

.24». . ~‘a¢<

‘ W; 51:54; '
a 3‘3...
- a»

 

gawk. ,.

A"?

l
‘ v-
I

m
‘

1: , 3
39“.:
“H"
*1
' A

 

‘ 1'43.» ‘
. .l.‘-~“"é'.,a,.
‘ “5%.: ‘
' V2:

1:155"

 

 

 

“mm?
4 “‘1 , r '
15353.39" Jun-

kk’fi EX

 

r. A
K w on

57%

if} ‘35-: #11:.“ 1‘-

rHL.

,-;
{a}. -

,

~ ('3.

— i?“
Mfg;
. I' ‘

This is to certify that the

dissertation entitled

Reproduction, Early Life History,
and Recruitment of Rainbow Smelt

in St. Martin Bay, Lake Huron
presented by

Russell W. Brown

has been accepted towards fulfillment
of the requirements for

Ph.D. degree in Fisheries

 

 

 

Major professor

/(¢» 414 7221/
7 fl

Date JUlZ 28, 1994

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

 

 

 

 

liliuliifll iliifi\iillifiiiiili1ifiil

3 1293 01053 50

 

LIBRARY
M‘Chigan State
Unlverslty

 

 

 

PLACE ll RETURN BOXtonmovombclnckoutfvom yournoocd.

TO AVOID FINES mum on or baton dd. duo.

DATE DUE DATE DUE DATE DUE

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

3m

MSU I. An Afflmntlvo Action/Equal Opportunlty Institution
Was-9.1

 

 

 

REPRODUCTION, EARLY LIFE HISTORY, AND RECRUITMENT OF
RAINBOW SMELT IN ST. MARTIN BAY, LAKE HURON

By

Russell W. Brown

A DISSERTATION

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

DOCTOR OF PHILOSOPHY

Department of Fisheries and Wildlife

1994

ABSTRACT

REPRODUCTION, EARLY LIFE HISTORY, AND RECRUITMENT OF
RAINBOW SMELT IN ST. MARTIN BAY, LAKE HURON
By

Russell W. Brown

Rainbow smelt (Osmerus mordax) provide an important source of forage
for introduced salmonines, and are considered to be an important competitor and
predator of native Great Lakes species. Historical records reflect unstable levels
of rainbow smelt abundance in the upper Great Lakes, with much of this
instability in population abundance due to highly variable year class formation and
recruitment. Spawning rainbow smelt were sampled in four tributaries to St.
Martin Bay, Lake Huron to measure population characteristics of spawning
populations and to estimate egg deposition and larval outmigration. Age
composition of tributary spawning rainbow smelt from 1991-1993 was significantly
younger than age compositions reported in the 1970’s. Diel patterns of larval
outmigration were consistent between tributaries with greater than 90% of larval
outmigration occurring between 2100 and 0600 hours. Of the four study
tributaries, the Carp River dominated larval production, producing 72.8%, 74.8%,
and 96.4% of the total larval output from 1991 to 1993, respectively. The four

. tributary streams produced a total of 46 million larvae in 1991, 13 million larvae

in 1992, and 32 million larvae in 1993, with annual differences due to variable
levels of egg deposition and survival between years.

larval and juvenile stages of rainbow smelt were sampled in St. Martin
Bay, Lake Huron from 1991 to 1993 to estimate relative abundance and growth
and survival rates. In 1991, high tributary larval production and high degree of
spatial overlap between larvae and zooplankton prey resources resulted in
production of a strong year class comprised mainly of tributary produced progeny.
In 1992, low tributary production of larvae, low densities of zooplankton prey
resources, and poor spatial overlap between larval populations and prey resources
resulted in relatively poor year class comprised mainly of lake spawned larvae. In
1993, intermediate levels of tributary larval production and a high degree of
spatial overlap with moderate densities of zooplankton prey resources resulted in
production of a moderate year class, with approximately 65% of recruits hatching
during tributary outmigration. Interannual variation in year class strength of
rainbow smelt in St. Martin Bay resulted primarily from variable contributions of

tributary spawned larvae between years.

ACKNOWLEDGMENTS

This publication is a result of work sponsored by the Michigan Sea Grant
College Program, project numbers R/GLF-35 and R/GLF-42, under grant
number NA89AA-D-SG083 from the Office of Sea Grant, National Oceanic and
Atmospheric Administration (NOAA), US. Department of Commerce, and funds
from the state of Michigan. The US. Government is authorized to produce and
distribute reprints for governmental purposes notwithstanding any copyright
notation appearing hereon. Fellowship and scholarship support was provided by
the Canadian Studies Centre Doctoral Fellowship Program and the International
Association for Great Lakes Research, respectively. Research support was also
provided by the Michigan Polar-Equator Club and the Michigan Chapter of Sigma
Xi. Travel support to present results of this research project at professional
meetings was provided by the Ecology and Evolutionary Biology Program at
Michigan State University, the Sea Grant Association, and the American Institute
of Research Biologists.

I thank the members of my dissertation committee, Dr. Richard Hill, Dr.
Gary Mittelbach, and Dr. Niles Kevem for their suggestions and critical review of
this manuscript. I especially thank Dr. William Taylor for his support, guidance,
patience, and friendship during my graduate career at Michigan State.

I am indebted to fellow graduate students John Kocik, Dan Hayes, Bob

‘ Sluka, Paola Ferreri, and Ted Sledge for their insightful suggestions and

iv

recommendations during several phases of this study. I thank graduate students
Bob Sluka, Ted Sledge, and Ed Roseman; and student interns Velda
Hammerbacher, Jim Lucas, Scott Miller, Sam Noffke, Randy Claramunt, and
Mike Winters for their assistance during the intensive field sampling on St. Martin
Bay. I thank student assistants Joe Mion, Heather Sysak, Mandy Dunlap, Sam
Nofflte, Erin Postin, Jeremy Wilder, and Melinda Raths for the assistance in the
preparation and analysis of samples in the laboratory.

I would like to thank my families in Maryland (Mom and Dad), California
(Susan), and Nebraska (Bernie, Sharlene, Sarah, and Nancy) for their love,
encouragement, and patience during my education. Finally, I am grateful for the
love and support of my wife, Amy, who assisted with field sampling, held down
the household while I was up North, and remained steadfastly supportive at times
when I was ready to surrender. She clearly shares in the accomplishments

outlined in this dissertation.

TABLE OF CONTENTS

List of Tables ................................................ xi
List of Figures .............................................. xiv
List of Appendices ........................................... xx
DISSERTATION INTRODUCTION ............................... 1
Study Area ............................................. 5
Recreational Fishery ..................................... .' 7
Dissertation Overview ..................................... 8

CHAPTER 1: VITAL STATISTICS OF RAINBOW SMELT

SPAWNING IN FOUR TRIBUTARIES OF ST.
MARTIN BAY, LAKE

HURON .................................... 11

ABSTRACT ........................................... 1 1
INTRODUCTION ....................................... 12
METHODS ............................................ 14
Sampling of Spawning Adults .......................... 14

Fin Ray Aging ..................................... 15

Data Analysis ..................................... 16
RESULTS ............................................. 18
Length-Frequency Distributions ........................ 18
Length-Age Keys ................................... 29

Age Composition ................................... 29
Length-Weight Regressions ........................... 36

Sex Ratios ........................................ 36

DISCUSSION .......................................... 40

Sex Ratio and Length Frequency Patterns ................ 40
Length-Weight Relationships .......................... 42
Age Composition ................................... 43
Conclusions ....................................... 45

CHAPTER 2: SEASONAL PATTERNS IN EGG DEPOSITION
AND LARVAL OUTMIGRATION OF RAINBOW
SMELT FROM FOUR TRIBUTARIES TO ST.
MARTIN BAY, LAKE

HURON .................................... 47

ABSTRACT ........................................... 47

INTRODUCTION ....................................... 48

METHODS ............................................ 50

Spawner Abundance ................................ 50

Egg Deposition Sampling ............................. 51

Diel Sampling of Larval Outmigration ................... 55
Nightly Sampling of Outmigrating larval Rainbow

Smelt ...................................... 56

Seasonal Estimates of Tributary Mortality ................ 59

Environmental Conditions ............................ 60

RESULTS ............................................. 61

Timing and Magnitude of Spawning ..................... 61

Recreational Fishery Effects on Egg Deposition ............ 69

Seasonal Egg Deposition Estimates ...................... 70

Diel Patterns in Larval Drift .......................... 74

vii

Seasonal and Interannual Patterns in Larval Drift .......... 77

Tributary Mortality ................................. 83
DISCUSSION .......................................... 86
Effect of Temperature on the Timing of Spawning .......... 86
Egg Deposition and Survival .......................... 87
Recreational Fishery Effects on Egg Deposition ............ 89
Diel Outmigration Patterns ........................... 90
Seasonal Larval Outmigration Patterns .................. 91
CONCLUSIONS ........................................ 92

CHAPTER 3: LARVAL GROWTH, SURVIVAL, AND JUVENILE
RECRUITMENT OF RAINBOW SMELT IN ST.

MARTIN’S BAY, LAKE HURON ................ 94

ABSTRACT ........................................... 94
INTRODUCTION ....................................... 96
METHODS ............................................ 98
Spatial and Vertical Distribution Patterns ................ 98
Surface larval Trawling ................... V .......... 100
Handling of larval Samples .......................... 102

Data Analysis of larval Trawling Data ................. 102
Analysis of larval Growth and Mortality Rates ........... 105
Larval Diet ...................................... 107
Zooplankton Sampling .............................. 108
Bottom Trawl Sampling ............................. 109

viii

Otolith Aging and Validation ......................... 110

Scanning Electron Microscopy (SEM) Work ............. 111
Validation of the Formation of Daily Increments .......... 112
Hatch Date Analysis ............................... 115
RESULTS ............................................ 117
Diel Differences in larval Density and Size .............. 117
Density Patterns Between Contours .................... 118
Density Patterns Between Depth Strata ................. 122
Size-Related Patterns Between Depth Contours ........... 122
Size-Related Patterns between Depth Strata ............. 124
larval Diet ...................................... 125
Spatial Distribution of Larvae and Zooplankton Prey ....... 125
Growth and Survival Rates .......................... 139
Otolith Derived Hatch Date Distributions ............... 139
Bottom Trawl Year Class Index ....................... 150
DISCUSSION ......................................... 156
Diel Patterns in larval Distribution .................... 156
Spatial Patterns Between Depth Contours ............... 156
Vertical Distribution Patterns ........................ 157
Size Related Patterns in larval Distribution .............. 158
larval Growth Rates ............................... 159
Relative Contribution to Recruitment .................. 159

Interannual Recruitment Patterns ..................... 162

APPENDICES .............................................. 164

LIST OF REFERENCES ...................................... 168

Table 1.

Table 2.

Table 3.

Table 4.

Table 5.

Table 6.

' Table 7.

Table 8.

Table 9.

Table 10.

List of Tables

Available stream length and area, peak nightly participation,
and average total seasonal effort of recreational anglers
pursuing rainbow smelt on four study tributaries. ............ 7

Subsample sizes and the number of readable fin ray samples used to
estimate age composition of spawning rainbow smelt sampled from
four tributaries to St. Martin Bay, lake Huron from 1991 to

1993. ............................................ 17

Sample size, mean length (mm), standard errors (SE), and range in
lengths of rainbow smelt sampled from four tributaries to St. Martin
Bay, lake Huron in 1991. ............................ 24

Sample size, mean length (mm), standard errors (SE), and
range in lengths of rainbow smelt sampled from four
tributaries to St. Martin Bay, lake Huron in 1992. .......... 24

Sample size, mean length (mm), standard errors (SE), and
range in lengths of rainbow smelt sampled from four
tributaries to St. Martin Bay, lake Huron in 1993. .......... 25

Results of Kolmogorov-Smirnov tests to detect differences in

the length frequency distributions between male and female

rainbow smelt sampled from four St. Martin Bay tributaries

from 1991 to 1993. ................................. 26

Results of Kolmogorov-Smirnov tests to detect differences

between years in the length frequency distributions of

rainbow smelt sampled in four St. Martin Bay tributaries

from 1991 to 1993. ................................. 27

Results of Kolmogorov-Smirnov tests to detect differences in

the length-frequency distributions of spawning rainbow smelt
sampled from four St. Martin Bay tributaries from 1991 to

1993 ............................................ 30

Mean length (mm) of rainbow smelt sampled from lake
Huron in 1975-1976 by Argyle (1982) and 1991-1993 by the
current study ....................................... 31

Percent age composition data for rainbow smelt sampled in
lake Huron during different time periods. ................ 35

Table 11. Length-weight regressions by year and tributary for rainbow
smelt sampled in four St. Martin Bay tributaries from 1991 to
1993. ............................................ 35

Table 12. Sex ratios (percent males) of spawning rainbow smelt
sampled from the Carp River, Nunns Creek, Spring Creek
and St. Martin Creek (Mackinac County, Michigan) from
1991 to 1993 ....................................... 37

Table 13. Length—weight regression coefficients reported for populations
of rainbow smelt in the upper Great lakes. ............... 44

Table 14. Total egg deposition sampling effort on Nunns, Spring, and
St. Martin Creeks, 1991 to 1993. ....................... 53

Table 15. Dates when diel patterns in the outmigration of larval
rainbow smelt were measured on the Carp River, Nunns
Creek, Spring Creek, and St. Martin Creek, 1991 to 1993. . . . . 56

Table 16. Total egg deposition (millions) by rainbow smelt spawning in
three tributaries to St. Martin Bay, lake Huron from 1991 to
1993. ............................................ 74

Table 17. Total outmigration (millions) of larval rainbow smelt from
four tributaries to St. Martin Bay, lake Huron from 1991 to
1993. ............................................ 82

Table 18. Instantaneous mortality (Z) and seasonal survival rates (%)
of rainbow smelt from eg deposition to larval outmigration
on three tributaries to St. Martin Bay, lake Huron from 1991
to 1993 .......................................... 86

Table 19. Split-plot ANOVA designs used to test for differences in
density and mean length of larval rainbow smelt across
contours. (Depth Strata = S; Depth Contours = C;
Sampling Date = D; Year = Y. ...................... 104

Table 20. Results of Wilcoxon Signed Rank tests to compare larval
densities collected in paired day and night samples. N =
number of paired samples (depth contour-depth strata
combinations) compared ............................ 119

xii

Table 21. Results of Wilcoxon Signed Rank tests to compare mean
lengths of larvae collected in paired day and night samples.

N = number of paired samples compared. ............... 120
Table 22. Mean number of prey items in the digestive tract of larval

rainbow smelt and (in parentheses) percentages of larvae

containing prey items. .............................. 128

Table 23. Length-class based estimates of growth for early and late
hatching cohorts of larval rainbow smelt in St. Martin Bay, .
lake Huron from 1991 to 1993. ....................... 144

Table 24. Length-class based estimates of mortality (SE in parentheses)
for early and late hatching cohorts of rainbow smelt in St.
Martin Bay, lake Huron from 1991 to 1993. ............. 144

Table 25. Total catch and combined CPUE of age 0+ rainbow smelt

indexed using a standardized bottom trawl survey in St.
Martin Bay, lake Huron from 1991 to 1993. ............. 154

xiii

List of Figures

Figure 1. Catch per unit effort index of adult (Panel A) and age 0+
rainbow smelt (Panel B) indexed by the National Biological
Survey standardized trawl survey program at five sites on
Lake Huron. ....................................... 4

Figure 2. Map showing the location of the study area in St. Martin Bay,
Lake Huron. ....................................... 6

Figure 3. Length frequency distribution of spawning rainbow smelt
collected from the Carp River (Mackinac County, Michigan)
from 1991 to 1993. ................................. 20

Figure 4. Length frequency distribution of spawning rainbow smelt
collected in Nunns Creek (Mackinac County, Michigan) from
1991 to 1993. ...................................... 21

Figure 5. Length frequency distribution of spawning rainbow smelt
collected in Spring Creek (Mackinac County, Michigan) from
1991 to 1993. ...................................... 22

Figure 6. Length frequency distribution of spawning rainbow smelt
collected in St. Martin Creek (Mackinac County, Michigan)

from 1991 to 1993. ................................. 23
Figure 7. Mean length of spawning rainbow smelt collected from the

Carp River, Nunns Creek, Spring Creek, and St. Martin

Creek (Mackinac County, Michigan) from 1991 to 1993 ....... 28

Figure 8. Length frequency distributions by age of spawning rainbow
smelt sampled from the Carp River, Nunns Creek, and St.
Martin Creek (Mackinac County, Michigan) in 1991. ........ 32

Figure 9. Length frequency distributions by age of spawning rainbow
smelt sampled from the Carp River, Nunns Creek, Spring
Creek, and St. Martin Creek (Mackinac County, Michigan) in
1992. ............................................ 33

Figure 10. Length frequency distributions by age of spawning rainbow
smelt sampled from the Carp River, Nunns Creek, Spring
Creek and St. Martin Creek (Mackinac County, Michigan) in
1993. ........ ' .................................... 34

xiv

Figure 11. Sex ratio (percent males) by 10-mm size class for spawning
rainbow smelt sampled in four St. Martin Bay tributaries
from 1991 to 1993. Sample sizes are shown above each
bar. ............................................. 38

Figure 12. Sex ratio (percent males) by age of spawning rainbow smelt
sampled from four St. Martin Bay tributaries from 1991-1993.
Sample sizes are shown above each bar ................... 39

Figure 13. Sampling design used to measure the impact of dip-net
activity on egg deposition. A plywood cover supported by
four metal rods was used to protect one sampling site in each
strata from dip net activity and wading by recreational
fishers. .......................................... 52

Figure 14. Diagram showing the process for correcting diel outmigration
data to eliminate seasonal trends. When significant positive
or negative slopes in peak nightly outmigration were
detected, data were detrended to eliminate seasonal effects. . . . 58

Figure 15. Spawner abundance of rainbow smelt spawning in the Carp
River, Mackinaw County, Michigan from 1991 to 1993. ...... 63

Figure 16. Water temperatures during spawning and larval outmigration
from the Carp River from 1991 to 1993. Shaded areas
indicate periods of spawning activity and larval outmigration
(arrows indicate peaks). .............................. 64

Figure 17. Spawner abundance (solid line) and mean egg deposition
(bars) of rainbow smelt spawning in Nunns Creek, Mackinac
County, Michigan from 1991 to 1993. .................... 65

Figure 18. Water temperature during spawning and larval outmigration
from Nunns Creek from 1991 to 1993. Shaded areas indicate
periods of spawning activity and larval outmigration (arrows
indicate peaks). ................................... 66

Figure 19. Spawner abundance (solid line) and mean egg deposition
(bars) of rainbow smelt spawning in Spring Creek, Mackinac
County, Michigan from 1992 to 1993. .................... 67

Figure 20. Water temperature during spawning and larval outmigration
. in Spring Creek from 1992 to 1993. Shaded areas indicate
periods of spawning activity and larval outmigration (arrows
indicate peaks). .................................... 68

Figure 21. Water temperature during spawning and larval outmigration
in St. Martin Creek from 1991 to 1993. Shaded areas
indicate periods of spawning activity and larval outmigration
(arrows indicate peaks) .............................. 71

Figure 22. Spawner abundance (solid line) and mean egg deposition
(bars) of rainbow smelt spawning in St. Martin Creek,
Mackinac County, Michigan from 1991 to 1993. ............ 72

Figure 23. Relationship between recreational fishing effort and percent
change in egg density at paired sampling sites in Nunns,
Spring, and St. Martin Creeks sampled in April 1991 to 1993. . 73

Figure 24. Diel larval outmigration patterns occurring in the Carp River,
Nunns Creek, Spring Creek, and St. Martin Creek. Values
are expressed as a percentage of the total larvae captured
during the 24-hour period. ............................ 75

Figure 25. Nightly pattern of larval outmigration (number/hour)
occurring on the Carp River, Nunns Creek and Spring Creek
between 2100 and 0600. .............................. 76

Figure 26. Peak nightly outmigration (number/ hour) of larval rainbow
smelt sampled between 2200 and 0200 on the Carp River,
Mackinac County, Michigan from 1991 to 1993. ............ 79

Figure 27. Peak nightly outmigration of larval rainbow smelt
(number/hour) sampled between 2200 and 0200 on Nunns
Creek, Mackinac County, Michigan from 1991 to 1993. ...... 80

Figure 28. Peak nightly outmigration of larval rainbow smelt
(number/ hour) sampled between 2200 and 0200 on Spring
Creek, Mackinac County, Michigan from 1992 to 1993. ...... 81

Figure 29. Peak nightly outmigration of larval rainbow smelt
(number/hour) sampled between 2200 and 0200 on St.
Martin Creek, Mackinac County, Michigan from 1991 to
1993. ............................................ 84

Figure 30.

Figure 31.

Figure 32.

Figure 33.

Figure 34.

Figure 35.

Figure 36.

Figure 37.

Figure 38.

Figure 39.

Figure 40.

Total peak larval outmigration (number/hour) sampled
between 2200 and 0200 on the Carp River, Nunns Creek,
Spring Creek, and St. Martin Creek from 1991 to 1993. ......

Map showing locations of fixed sampling sites for vertical
larval trawling (Panel A), and stratified sampling design
across depth contours and strata (Panel B). ...............

Map showing locations of fixed sampling sites for surface
larval trawling (Panel A) and bottom trawling (Panel B). . . . .

Results of validation of the formation of daily increments in
larval rainbow smelt otoliths. Panel A shows results for
larvae from age 3 to 20 days, while Panel B gives results for
larvae from age 50 to 70 days. .......................
Densities of larval rainbow smelt sampled along the 2.5-m,
5.0-m, and 10.0-m contours of St. Martin Bay, Lake Huron
from 1991 to 1993. Error bars represent i one standard
error of the mean. .................................

Densities of larval rainbow smelt sampled at depth strata
along the 2.5-m, 5.0-m, and 10.0-m depth contours in St.
Martin Bay, lake Huron in 1992. .....................

Length frequency distributions of larval smelt sampled at
depth strata along the 2.5-m and 5.0-m depth contours in St.
Martin Bay, lake Huron on 27 July, 1993. ...............

Length frequency distributions of larval rainbow smelt
sampled at depth strata along the 10—m depth contour in St.
Martin Bay, lake Huron on 27 July, 1993. ...............

Strauss’ electivity indices for taxa consumed by larval rainbow
smelt in St. Martin Bay, lake Huron in 1991. ............

Site specific densities of larval rainbow smelt and
zooplankton prey in St. Martin Bay, Lake Huron for May 15-
16 and May 30-31, 1991. ............................

Site specific densities of larval rainbow smelt and
zooplankton prey in St. Martin Bay, lake Huron for June 14-
15 and June 24-25, 1991. ............................

85

99

101

121

123

126

127

129

130

131

Figure 41.

Figure 42.

Figure 43.

Figure 44.

Figure 45.

Figure 46.

Figure 47.

Figure 48.

Figure 49.

Figure 50.

Figure 51.

Site specific densities of larval rainbow smelt and
zooplankton prey in St. Martin Bay, lake Huron for July 11-
12 and July 23-24, 1991 ..............................

Length frequency distributions of larval rainbow smelt
sampled in St. Martin Bay, lake Huron from mid-May to
late-July, 1991. ...................................

Site specific densities of larval rainbow smelt and
zooplankton prey in St. Martin Bay, lake Huron for May 20-
21 and May 31-June 1, 1992. .........................

Site specific densities of larval rainbow smelt and
zooplankton prey in St. Martin Bay, lake Huron for June 15-
16 and June 28-29, 1992. ............................

Site specific densities of larval rainbow smelt and
zooplankton prey in St. Martin Bay, lake Huron for July 14-
15 and July 26-27, 1992 ..............................

Length frequency distributions of larval rainbow smelt
sampled in St. Martin Bay, lake Huron from mid-May to late
July, 1992. .......................................

Site specific densities of larval rainbow smelt and
zooplankton prey in St. Martin Bay, lake Huron for May 17-
18 and June 1, 1993. ...............................

Site specific densities of larval rainbow smelt and
zooplankton prey in St. Martin Bay, lake Huron for June 13-
14 and June 27, 1993. ..............................

Site specific densities of larval rainbow smelt and
zooplankton prey in St. Martin Bay, lake Huron for July 12
and July 27, 1993. .................................

Length frequency distributions of larval rainbow smelt
sampled in St. Martin Bay, lake Huron from mid-May to
late-July, 1993. ...................................

Relationship between instantaneous growth and instantaneous
mortality rates for early and late hatching cohorts of rainbow
smelt in St. Martin Bay, Lake Huron. ..................

132

134

135

136

137

138

140

141

142

143

145

Figure 52. Hatch date distributions for recruited juvenile rainbow smelt
sampled in August and September, 1991. Bars represent 7-
day mean numbers (1 3.5 days) representing the 95%
confidence interval for backcalculated hatch dates. ......... 147

Figure 53. Hatch date distributions for recruited juvenile rainbow smelt
sampled in August and September, 1992. Bars represent 7-
day mean numbers (i 3.5 days) representing the 95%
confidence interval for backcalculated hatch dates. ......... 148

Figure 54. Hatch date distributions for recruited juvenile rainbow smelt
sampled in August and September, 1993. Bars represent 7-
day mean numbers (1 3.5 days) representing the 95%
confidence interval for backcalculated hatch dates. ......... 149

Figure 55. Length frequency distribution of juvenile and adult rainbow
smelt sampled during monthly bottom trawl surveys in 1991. . 151

Figure 56. Length frequency distribution of juvenile and adult rainbow
smelt sampled during monthly bottom trawl surveys in 1992. . 152

Figure 57. Length frequency distribution of juvenile and adult rainbow
smelt sampled during monthly bottom trawl surveys in 1993. . 153

Figure 58. Relative contribution of early and late hatching cohorts to
juvenile recruitment in St. Martin Bay, lake Huron from
1991 to 1993. ..................................... 155

List of Appendices

Appendix 1. Age-length key for spawning rainbow smelt sampled from
four tributaries to St. Martin Bay, lake Huron in 1991 ....... 163

Appendix 2. Age-length key for spawning rainbow smelt sampled from
four tributaries to St. Martin Bay, lake Huron in 1992. ..... 164

Appendix 3. Age-length key for spawning rainbow smelt sampled from
four tributaries to St. Martin Bay, lake Huron in 1993. ..... 165

Appendix 4. Length-weight regression equations for male and female
rainbow smelt sampled from four lake Huron tributaries
from 1991 to 1993 .................................. 166

Dissertation Introduction

Rainbow smelt (Osmerus mordax) were first introduced into lake Michigan
through a single stocking in Crystal lake, Michigan in 1912 and spread
throughout the Great lakes by the mid 1930’s (Van Oosten 1937). In the upper
Great Lakes, rainbow smelt rapidly became an important component of the
recreational and commercial fisheries. During the late 1930’s, the recreational
and commercial fisheries harvested 2.5 and 1.8 million kilograms annually in the
Michigan waters of the Great lakes (Van Oosten 1947). Currently, rainbow
smelt support a popular recreational dip net fishery during spring spawning runs
(Raab and Steinnes 1979), although the commercial fishery in the Michigan
waters of the Great lakes has declined significantly from historical levels. In
1993, commercial harvest of rainbow smelt from the Michigan waters of lakes
Superior, Michigan, Huron and Erie represented less than 5% of the total
commercial fishery by weight and approximately 1% by value (Kinnunen 1994).

Currently, rainbow smelt provide an important source of forage for
introduced salmonines (Chinook (Oncorhynchus tshawytscha), coho (0. kisutch),
and pink (0. gorbuscha) salmon, brown trout (Salmo trutta) and steelhead (0.
mykiss) and native predators (lake trout (Salvelinus namaycush) and burbot (Lota
lota)) in the upper Great lakes (Stewart et a1. 1981, Hatch et al. 1981, Johnson et
a1. 1994). Growth and survival rates of introduced salmonines have declined
during the 1980’s (Eck and Wells 1983), resulting in concerns over the possible
depletion of the forage base in the Great Lakes (Stewart et a1. 1981, Johnson et

- a1. 1994).

2

In addition to its role as a prey species, the rainbow smelt is also
considered to be a competitor and predator of native Great lakes species
(Crowder 1980). Rainbow smelt have the potential to compete with native
coregonines because of broad overlaps in distribution and diet with lake herring
Coregonus artediz' (Anderson and Smith 1971), deepwater ciscos (C. sp.), and lake
whitefish (C. clupeafomtis). Although rainbow smelt in the Great Lakes feed
mainly on invertebrates (Gordon 1961, Foltz and Norden 1977, Anderson and
Smith 1971), their diet also includes larval and juvenile fishes (Hale 1960, Price
1963). Predation by adult smelt has been documented for age 0+ alewives (Alosa
pseudoharengus, O’Gorman 1974), lake herring (Selgeby et al. 1978, Loftus and
Hulsman 1986), bloaters (C. hoyi, Stedman and Argyle 1985), slimy sculpin (Cottus
cognatus, Brandt and Madon 1986), and lake whitefish (Loftus and Hulsman
1986). Predation by rainbow smelt and alewife has been hypothesized to be a
major mechanism for the collapse of native coregonines in the upper Great Lakes
(Crowder 1980) and inland waters (Evans and Loftus 1987).

Historical records reflect unstable levels of rainbow smelt abundance in the
upper Great lakes over the past sixty years with periods of marked increases
followed by precipitous declines (Baldwin et al. 1979). Standardized bottom trawl
surveys by the National Biological Survey (Argyle 1994) indicate that lake-wide
indices of adult rainbow smelt abundance have experienced two major peaks
_ (1981 and 1987) since 1973 (Figure 1). Indices of age 0+ rainbow smelt indicate
variable levels of recruitment have occurred, with strong year classes (Catch per

Unit Effort (CPUE) > 800 fish/tow) occurring in 1980, 1981, 1982, and 1986

3

(Figure 1). Recently, the species composition of prey biomass in Lakes Michigan
and Huron has shifted from exotic alewife and rainbow smelt to native bloaters
(Coregonus hoyi; Argyle 1994).

Fluctuations in the abundance of rainbow smelt have been attributed to
cannibalism (Regier et al. 1969), disease (Van Oosten 1947), parasitism (Nepszy
et al. 1978), and climatic factors during spawning including wave action and water
temperatures (Rothschild 1961; Rupp 1965). An annual cycle of alternating year-
class dominance in lake Erie between 1963 and 1974 has been attributed to
cannibalism by yearling rainbow smelt on age 0+ larvae and juveniles (Henderson
and Nepszy 1989). The role of rainbow smelt as a key forage species, predator,
and competitor of other Great Lakes fishes underscores the importance of
identifying and understanding variability in abundance and factors influencing
their recruitment.

Variability in year class strength of many pelagic and demersal fish species
in large aquatic systems is primarily caused by variation in growth and survival
during the early life stages (Houde 1987, Fritz et al. 1990). Relatively small
changes in growth and survival rates of these early stages can translate into large
annual fluctuations in resulting year-class strength (Houde 1987). My hypothesis
is that a combination of abiotic and biotic factors occurring between spawning and
the demersal juvenile stage is important in controlling the year class strength of

rainbow smelt in the upper Great lakes. The goal of this project was to measure

important parameters during the early life history stages to allow for the

 

1,400
1,200 -
1,000 '-
800
600

400

CPUE (Adults/Tow)

200

 

 

 

1973 1977 1981 1985 1989 1993

 

1,400
1,200 -
1,000 '-
800 r
600 -
400 -

200 "

CPUE (Age 0+ lndlvlduals/Tow)

 

 

     

0
1973 1977 1981 1985 1989 1993

Year

Figure 1. Catch per unit effort index of adult (Panel A) and age 0+
rainbow smelt (Panel B) indexed by the National Biological
Survey standardized trawl survey program at five sites on Lake
Huron.

5

identification of causal mechanisms controlling year class formation in rainbow

smelt. My specific objectives are to:

1. Measure vital statistics of tributary spawning rainbow smelt populations
including size and age composition, length-weight relationships, and sex
ratios.

2. Estimate seasonal patterns in egg deposition and larval outmigration from

tributary streams.

3. Measure diel, spatial, and seasonal patterns in abundance, growth and
survival of larval rainbow smelt in lake Huron.

4. Describe patterns of larval distribution in relation to distributional patterns
of larval zooplankton prey resources.

5. Estimate the relative contribution of tributary vs. lake spawned larvae to
overall recruitment.

W

I selected the St. Martin Bay area of northern Lake Huron as a study area
based on estimated strong production of rainbow smelt in this region of lake
Huron (Argyle 1982), historically consistent spawning in area tributaries, and
reported high densities of larval rainbow smelt (O’Gorman 1983). St. Martin Bay
is located in the northwest region of lake Huron along the southern shore of
Michigan’s upper peninsula (Figure 2). This oligotrophic bay has an average
depth of 8-m with gradually increasing depths to a maximum of 30-m. Water
circulates in a clockwise direction which is driven by an offshoot eddy from the
Straits of Mackinac. The bay generally is ice-covered from January to March
‘ (Assel et al. 1983). Surface water temperature ranges from 12 C to 20 C during

the summer, and thermal stratification is rare due to wind and current patterns.

 

 

Lake Huron
84°4 'w

 

 

 

 

 

Figure 2. Map showing the location of the study area in St. Martin Bay,
Lake Huron.

7

Three large (3rd and 4th order) and several smaller (lst and 2nd order)
tributaries enter St. Martin Bay. Based on preliminary sampling, we chose to
sample tributary spawning populations of rainbow smelt in four rivers and creeks,

the Carp River, Nunns Creek, Spring Creek, and St. Martin Creek.
Regreatignal Fishery

The St. Martin Bay area hosts a recreational dip net fishery that targets

rainbow smelt during spring spawning runs. During my study (1991 to 1993), the

majority of recreational fishing effort was focused on the Carp River, where up to

1000 people participated in the fishery on peak weekend nights. The Carp River

supported a seasonal average of 16,542 hours of recreational fishery activity from

1991 to 1993 (Table 1).

Table 1.

Available stream length and area, peak nightly participation, and average total

. seasonal effort of recreational anglers pursuing rainbow smelt on tour study
tributaries.

   
      
    

 

 

 

     

(I A" Tributary ‘ Stream Length . Stream Area Peak Average
‘ , , 1 , Available to Available to Participation Seasonal Total
- . Flshery,(m) , a Fishery (m2) (people) , EtiOrt (hours)

   

 

 

 

 

 

 

 

~~oaron~er > 2000.0 > 10,0000 1043 16.542
(“Nunns Creek 59.5 757.1 221 4,876
isprlng Creek .1 98.0 227.0 34 532
LSt. use... Creek ‘ . 149.3 720.1 47 957

 

 

 

8

Nunns Creek is a second popular dipping area in the bay, although the
fishery is limited to the lake area and a stream length of approximately 60-m due
to private property. Nunns Creek supported a seasonal average of 4,876 hours of
fishery effort from 1991 to 1993 with up to 221 people participating in the fishery
on peak weekend nights. Most of the smaller tributaries in the bay receive at
least token pressure from the recreational fishery. Recreational fishers
participating in the fishery are highly mobile and move between stream locations
until they are able to locate spawning runs of rainbow smelt. I frequently
encountered anglers who had checked streams between Detour and Naubinway,
Michigan, a distance in excess of 150 km, in a single night.

The purpose of this study was to follow three annual cohorts of rainbow
smelt (1991 to 1993 year classes) from spawning and outmigration from tributaries
through the pelagic larval stage until the cohort had become demersal juveniles in
St. Martin Bay. There were three main areas of research effort: 1. measurement
of vital statistics for tributary spawning stocks; 2. estimation of spawner
abundance, egg deposition, and larval outmigration; and 3. estimation of larval
abundance, growth and survival; and juvenile recruitment in St. Martin Bay. I
have organized this dissertation into three chapters to focus on each of these
research areas.

In Chapter 1, I summarize findings regarding vital statistics of tributary
Spawning populations including size composition, age composition, length-weight

relationships, and sex ratios. These data provided information on the status of

9

adult populations in the St. Martin Bay region that could be compared to other
regions in the Great lakes for the same time period and for other time periods in
Lake Huron.

In Chapter 2, I report on findings regarding spawner abundance, egg
deposition, larval outmigration, and egg to outmigrant survival. During three
spawning seasons (1991 to 1993), I indexed the relative abundance of spawners
and measured egg deposition in four representative tributaries in St. Martin Bay.
Because the recreational fishery appeared to have a significant impact on the
density of eggs deposited, I also quantified the effect that the fishery had on egg
densities. I also describe diel and seasonal patterns in larval outmigration from.
each of the four study tributaries, which allowed for estimation of total larval
outmigration from each tributary and egg to hatching estimates of survival.

In Chapter 3, I describe spatial patterns in the distribution of larval
rainbow smelt and their zooplankton prey resources during the pelagic larval
period. Once age 0+ rainbow smelt had become demersal (August-September), I
indexed the relative strengths of annual cohorts through a standardized bottom
trawl survey. One of the focuses of Chapter 3 was to identify unique
characteristics of recruited juvenile rainbow smelt. lake spawning populations
were assumed to spawn later and incubate longer due to colder lake
temperatures, creating temporal segregation in the hatch dates of tributary and

lake spawned larvae. By analyzing sagittal otolith microstructure, I was able to
determine the hatch date distribution of recruited juveniles. I was able to

distinguish between early and late hatching larval cohorts because I had accurate

10

measurements of the timing of spawning and larval outmigration of tributary
spawning rainbow smelt (Chapter 2). Finally, I describe interannual patterns in
larval growth, larval survival, and relative contributions of tributary and lake

spawned larvae, and relate these to juvenile recruitment.

CHAPTER 1: VITAL STATISTICS OF RAINBOW SMELT SPAWNING IN
FOUR TRIBUTARIES OF ST. MARTIN BAY, LAKE

HURON

Abstract
Spawning rainbow smelt were sampled in four tributaries to St. Martin Bay, lake
Huron to estimate population characteristics including size distribution, age
composition, length-weight relationships and sex ratio. Spawners were collected
twice nightly with dip nets from the Carp River, Nunns Creek, Spring Creek and
St. Martin Creek throughout the spawning season from 1991 to 1993. Rainbow
smelt were aged by examining prepared cross-sections of pectoral fin rays. Age
data were used to construct age-length keys to partition the entire sample of fish
into age classes based on length. Length-frequency distributions were not
significantly different (P > 0.05) between the Carp River and Nunns Creek.
Length-frequency distributions of fish sampled from Spring and St. Martin Creeks
were significantly (P < 0.05) larger than those from the Carp River and Nunns
Creek due to the presence of larger, older spawning fish. The slope of length-
weight regression equations decline annually from 1991 to 1993, indicating a
decline in the average condition factor of spawners. Samples of spawning fish
were dominated by males, but the percentage females increased in larger size and
older age classes. Age composition of rainbow smelt from 1991 to 1993 was
_ significantly younger than age compositions reported in the 1970’s. Because only

two age classes (2+ and 3+) of rainbow smelt contain greater than 90% of the

11

12

adult population, production of several consecutive weak year classes could result

in substantial declines in the population abundance of rainbow smelt.

Introduction

The rainbow smelt is an important forage and recreationally-sought species
in lake Huron. Over the past two decades, the absolute biomass and the biomass
of rainbow smelt relative to other forage species have both declined. From 1975
to 1984, rainbow smelt represented approximately 60% of the total biomass of
forage fish in lake Huron (Argyle 1994). Since 1986, the relative biomass of
rainbow smelt has declined steadily from 60% to less than 20% of the total forage
biomass in lake Huron (Argyle 1994). These trends are due to a series of
relatively weak year classes of rainbow smelt since 1986 (Figure 1) causing a
decline in total biomass, and increases in the total biomass of bloaters (Argyle
1994).

Declines in rainbow smelt populations have been attributed to a number of
factors including predation by stocked salmonines, disease (Van Oosten 1947),
parasitism (Schaefer et al. 1982, Henderson and Nepszy 1989), poor recruitment
associated with environmental conditions, and exploitation by recreational
fisheries. Salmonine stocking levels have increased significantly during the past

two decades, resulting in increased forage consumption (Johnson et al. 1994). A
disease outbreak is thought to have resulted in a dramatic die-off of rainbow

smelt in 1943 to 1944 (Van Oosten 1947), while parasitic fungi are thought to

13

cause significant mortality in adult rainbow smelt in lake Superior (Schaefer et al.
1982) and Lake Erie (Henderson and Nepzy 1989). Exploitation by recreational
dip net fisheries has been demonstrated to have significant effects on the
population dynamics of exploited rainbow smelt populations in localized areas.
Frie and Spangler (1985) examined vital statistics of rainbow smelt in South Bay,
lake Huron during a period of intensive fishing (1948 to 1957) and after
exploitation had ceased (1958 to 1970). They found that the modal ages of
spawning rainbow smelt shifted from 2 to 3 during intensive exploitation to 3 to 4
after exploitation had ended. Total annual mortality decreased from 90% during
a period of exploitation to 67% after exploitation ceased, and relative stock sizes
increased sharply after 1957.

Population dynamics of rainbow smelt in the Great lakes have been
assessed from spring spawning runs, when concentrations of sexually mature fish
can be sampled using a variety of sampling gears (Lucy and Adelman 1984, Frie
and Spangler 1985, Gebhardt 1993). Sampling during this time period allows for
representative sampling of the sexually mature portions of tributary spawning
populations. Estimates of vital statistics including size and age compositions, sex
ratios, and length-weight relationships can serve as a measure of the current status
of spawning populations. Comparison of current estimates with estimates from
other areas of the Great lakes during the same time period and estimates
obtained during different time periods provides information about the current

status of spawning populations. Estimates of the age composition allow for

14

assessment of the sensitivity of adult stocks to variation in year class strength and
recruitment.

The goal of this study was to estimate vital statistics of populations of
rainbow smelt spawning in tributaries to St. Martin Bay, lake Huron. My specific
objectives were to:

1. Determine the size and age distributions of spawning rainbow smelt.

2. Identify tributary, size-related, and age related patterns in the sex ratios of
spawning rainbow smelt.

3. Develop length-weight relationships of rainbow smelt from four tributaries
to St. Martin Bay.

Methods

MW

Sampling of spawning adult rainbow smelt was conducted in four
tributaries to St. Martin Bay: the Carp River, Nunns Creek, Spring Creek, and St.
Martin Creek (Figure 2). In each tributary, sampling was conducted nightly to
assess the vital statistics of spawning populations of rainbow smelt. Sampling was
cOnducted from April 12 to May 1, 1991., from April 15 to May 4, 1992, and from
April 14 to May 3, 1993 on all tributaries, except for 1991 when Spring Creek was
not sampled. When rainbow smelt were present, samples of fish were collected
using a long handled (1.3-m) smelt dip net (8-mm mesh).

Previous investigators have found that the age and size composition of

spawners changes on a nightly and seasonal basis. To obtain representative

15

samples of spawning populations on a nightly basis, tributary samples were
collected twice nightly, once between 22:00 and 01:00 hours and again between
01:00 and 04:00 hours. During each sampling period, I sampled for 45 minutes or
until I had collected a timed sample of 100 fish. Length (nearest mm) and weight
(nearest 0.1 g) were recorded from each fish. Because all fish sampled were in
spawning condition, the sex of sampled fish was determined through extrusion of
gametes.

To obtain representative samples for age determination throughout the
spawning season, I collected pectoral fin samples from a maximum subsample of
50 rainbow smelt per sampling period. This allowed for sampling of bony parts to
occur throughout the course of the spawning run (period ranging from 6 to 11
nights). Approximately 500 (range 481 to 504) fin samples were collected on each
tributary in each year.

We

Pectoral fin ray samples were chosen for age analysis for several reasons.
First, scale samples collected from spawning males were difficult to age due to the
presence of nuptial tubercles. Second, some fish lost scales during handling prior
to sample processing. I subsampled collected fin ray samples because of extensive
time involved in the preparation and analysis of fin ray samples. I prepared and
analyzed a subsample of fin rays that was weighted by the overall length and sex

distribution of rainbow smelt collected throughout the spawning season. These

weighted samples were produced for each sampled tributary in each year (N =

16

11). Subsample size was 160 samples and usable samples ranged from 134 to 151
samples per tributary for each year (Table 2).

Dried fin rays were placed in a plastic mold and were embedded in
electron microscopy embedding resin (EmBed-812). Each fin ray was then
sectioned using a low-speed radial arm saw with a wafer-thin diamond blade at
the base on the fin-body margin. Each section was then polished with
progressively finer wet-dry sandpaper (220-, 400-, 600-grit) with final polishing
using aluminum polishing powder and a polishing cloth. Each section was
examined under 10X to 50X magnification to identify annuli. Because fin samples
were collected in early spring prior to significant growth, the margin of the fin ray
was assumed to represent the last annuli.

DataAnahrsis

I used Kolmogorov-Smirnov tests (Siegel 1956, Snedecor and Cochran
1980) to detect differences in the length frequency distribution between males and
females within tributaries and years, between years within tributaries (pooled
sexes), and between streams within years (pooled sexes). When significant
differences were found in the distributions using a two-tailed test, we used one-
tailed tests to determine which distribution had a larger size distribution.

Age data determined from analysis of fin rays was used to construct age-
length keys (Ricker 1975) for each tributary in each year. I tested for differences

between tributaries within years using a Chi-Square contingency tables (Siegel
1956, Snedecor and Cochran 1980). Because I found no significant differences

between tributaries within years, I pooled the age-length keys for all tributaries

Table 2.

17

Subsample sizes and the number of readable fin ray samples
used to estimate age composition of spawning rainbow smelt
sampled from four tributaries to St. Martin Bay, lake Huron

from 1991 to 1993.

 

, Readable 1;
Sammes i

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Carp River 160 139
'   * 1991 Nunns Creek 160 137
w 1991" . St. Martin Creek 160 143

A 19921 Carp River 160 137
1992 . _ Nunns Creek 160 140
_ 1992 ‘ 1' Spring Creek 160 153
j ‘ 1992* St. Martin Creek 160 139
19931 ' ‘ Carp River 160 145
'_ 1993"? ' Nunns Creek 160 140
e "1993 Spring Creek 160 141

L ‘  1' A" St. Martin Creek 160 150

 

 

1564

 

 

18

within years to form a single age-length key for each year. As chi-square tests
indicated significant differences in the age length keys between years, I used age-
length keys specific to each year in subsequent analysis. Age-length keys were
used to partition overall length frequency data into age frequency data for each
year.

Length-weight relationships (Ricker 1975) were determined as follows:

loglo(W) - a + b logloU)

I used analysis of covariance (Steel and Torrie 1980) to test for significant

differences between the slope and intercept of length-weight regression equations.

Results

I l -E D. .1 i
Length-frequency distributions for the Carp River, Nunns Creek, Spring
Creek, and St. Martin Creek are shown in Figures 3 to 6, respectively. Greater
than 90% of fish sampled during spawning runs were between 110 to 170 mm in
total length. Female rainbow smelt had significantly larger mean lengths than
males for all tributary-year combinations (Tables 3 to 5). Length frequency
distributions of female fish were significantly larger than the distributions of males

' in 9 of 11 tributary-year combinations tested (Table 6). I found no significant

19

differences in length-frequency distributions between sexes for 1992 samples
collected at the Carp River and St. Martin Creek.

Length-frequency distributions of rainbow smelt were significantly larger in
1991 than in 1992 or 1993 for all tributaries sampled from 1991 to 1993 (Table 7).
These length-frequencies were significantly larger in 1992 than 1993 in the Carp
River and St. Martin Creek, but not for fish from Nunns and Spring Creeks.
There was a steady decline in the mean length of rainbow smelt on all tributaries
during the three years of this study (Figure 7).

I found significant differences in the length frequencies of spawning
tributaries within years (Table 8). A general pattern that emerged from these
tests was that length frequency differences were as follows: Carp River < Nunns
Creek < (St. Martin Creek = Spring Creek). Mean lengths within years were
always smallest at the Carp River and largest at either St. Martin Creek or Spring
Creek. There was an inverse relationship between recreational fishing intensity
on tributaries and mean lengths of spawning rainbow smelt (i.e. tributaries with
the lowest recreational fishing intensity (Spring and St. Martin Creeks) had the
largest mean lengths, while tributaries with the highest recreational fishing

intensity (Carp River and Nunns Creek) had the smallest mean lengths.

20

 

   

 

 

 

 

 

 

 

 

120

100 . 1991

80 ’ N=1047

60 i )—(=140.6

40 r _

“r l l l

120 r r A MI 'I !| l l! .

100 ~ 1992
3 80 L ' N=1625
g 60 56-1313
3 , .

Z 40 r , j . l
. l W

2° =4 H 'i.

120 .1! .lLl 51M“ ll -n- _!._r -r

100 1993

80 ; .N81763

so ' )_(=‘l36.2

r i] 3
4o , " l l H
2°“ 1
o ? H-li'. M _ I1! it‘ll 4 1

90 100110120130140150160170180190 200
Length (mm)

A Figure 3. Length frequency distribution of spawning rainbow smelt collected
from the Carp River (Mackinac County, Michigan) from 1991 to
1993.

21

 

 

 

    

 

 

60 e

50- i 1991

40 N=1594

30 F L ’ it's143.5

20 - l: i , .

‘° * l l Ill

60 , . m. r I . «I .1.

50 i 1992
g 40 i Nsaeo
E 30 L I _

3 i 1 . i x‘138.0
z 20 It 1.1.
‘° i ll »
60 * 1 ‘l 1.11 1 H ‘ ‘
i

50 ’F 1993

40 than

30 » 76138.3

20 ~ )1 l ‘3; , I

10 - ”1

- , i l l

90 100110120130140150160170180190 200
Length (mm)

9 Figure 4. Length frequency distribution of spawning rainbow smelt collected
in Nunns Creek (Mackinac County, Michigan) from 1991 to 1993.

22

 

 

 

 
 

 

 

 

 

1991
No Samples Collected
50
4o~ 1992
h . ~ N=97O
o 30 - . * -.
g r 3 w x- 143.5
E 20 b F I j l ,
5, * ll
' 1993
4° _" N=481
30 - )‘(= 142.3
20 ~ ,
. i
10 " l1. ‘ W p
» l i ll
0 . .. Illll . Hill ii i , w
90 100110120130140150160170180190 200

Length (mm)

A Figure 5. Length frequency distribution of spawning rainbow smelt collected
in Spring Creek (Mackinac County, Michigan) from 1991 to 1993.

23

 

 

 

 

 

 

 

 

 

50
40 1991
, N=909
30 - i=147.3
20 ~
50 ’1 n n .- git. Hills ii i ll
40; 1992
h . N= 784
g 30 i- 144.6
i ,
§ 20 - , w ,
z _ ’. l[ i! 3
‘°' lll
.. . .. . ill l ,1 1 HI“ ”I
50
40+ 1993
, N=831
3° - i=1462
l
20 r 1
‘°’ II
o L , . .. 1H” ,Ilhll’I l
90 100110120130140150160170180190 200

Length (mm)

Figure 6. Length frequency distribution of spawning rainbow smelt collected
in St. Martin Creek (Mackinac County, Michigan) from 1991 to
1993.

Table 3.

24

Sample size, mean length (mm), standard errors (SE), and range in
lengths of rainbow smelt sampled from four tributaries to St. Martin
Bay, lake Huron in 1991.

 

   

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

§ ,?; 'i'i'ribtnary = V 1‘  -. 3, ‘ZE-Sex V. "Sample . Mean Standard ‘ ' Range
’i g ‘ ‘_ ‘ _»_:_SfiZe>~ Length < «Error ' ’ . '
WL—m—JDLm—M 102 - 198
1991 cam Bug: _Male 715 139-73 (1426 110 - 173
1991 Carp River Female 332 142.58 0.818 107 - 198
.__1————____ __—____——_____———————
fill—1W Both 45194 143.48 0 359 101 -21;
_.1.QQJ___N1mns.Creek___MalL 10110 14? 38 0 404 102 - 197
1991 Nunns Creek Female 594 145.30 0.677 101 - 219
MW Both 909 1&33 n 505 100 - 237
Walt Male 581 145 67 (1572 100 - 237
1991 St. Martin Creek Female 328 150.27 0.943 105 - 216
Table 4. Sample size, mean length (mm), standard errors (SE), and range in

lengths of rainbow smelt sampled from four tributaries to St. Martin
Bay, lake Huron in 1992.

 

   

Tributary Sexr 7: Sample 1 ; Mean Standard 3 . Flange '
" :. s; 3 3?- g; i-ebi-y Size Length. 1 - - Error ' ' , - .
AQSLMBLM 1625 13732 OSSL 913-210
4992._Cam_Bbtet—_Mala 1066 13690 0419 96 - 185
1992 Cagg River Female 559 138.14 0.671 103 - 210
1992 Nunns Creek Bpth 375 138 on n 494 93 - 199
4992__Nunns.Craak__Male 593 137 18 n 553 103 -17;
1 Nunns Creek Female 352 139.81 0.905 98 - 199
”M E J
magnum Bpth 970 143 50 n 598 105 - 231
F1992__Spdng.flteeL__Male 557 140.9? 05.96 106 494—.
1992 Sgrlng Cregk Female 413 146.87 0.g_g_1 105 - 231
__1.992_._SL.Manln_Cteek Both JAB 449 m 0 527 107 - 210
ASQLW Male 413 1.4377 0615 114.49- 1
1992 _St. Marti reek _ Female 335 145.69 0.898 107 - 210

 

 

   
  

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

25

Table 5. Sample size, mean length (mm), standard errors (SE), and range in
lengths of rainbow smelt sampled from four tributaries to St. Martin
Bay, Lake Huron in 1993.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

3 Year Tributary , Sex Sample Mean Standard Range

L . - ' _ ~ Size Length Error

E as. W Both 1763 13619 0976 109 - 903

: -.: _QamflmL—Me 1994 135 70 n 399 109 - 903

{1993 Gag River Fem=a=!__e__= 529 137.33 0.528 106 - 193

L we _Nunnsfiraeie Bath 3511 138 95 0 499 101 - 995

: °.: WM 537 137 4a 0 497 101 - 173

L __ Nunns Creek Female ==§87 139.77 0.803 101 - 225

5 2:. W 13pm 481—_142.Zfi___0.591 105 - 189

'9 _Sndnaflaak Male 3.42 14039—4519 105 - 189—,
[__993 Sgring Creek Female 139 145.65 1.191 117 - 183

L 5:: WM 991 14p 19 n 457 111:; - 9:19

. -.: Wm 599 139 95 n 501 105 - 179—
St. Martin Creek Female 309 14.3.43 0.877 107 - 232

 

 

 

26

Table 6. Results of Kolmogorov—Smirnov tests to detect differences in the
length frequency distributions between male and female rainbow

smelt sampled from four St. Martin Bay tributaries from 1991 to

 

  

 

     

   

  

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

1993.
iYea‘r’if? ’Trlbula‘MS)‘   '   Sample-5‘26": ' ~ ~ : Kolmogorov-Smlmov .
4v ‘* “  7_ J}; . ‘. 1 . ‘fig . 1 . Test Results
1 M<>F M>~F~
Males ' Females . P-Value P-Value
1991 Carp River 715 332 0.0370 0.0185
1992 Carp River 1066 559 0.2281 --
1993 Carp River 1233 529 0.0360 0.0180
If 1991-1993 Carp River 3014 1402 0.0006 0.0003 4
1991 Nunns Creek 1000 594 0.0800 0.0117
1992 Nunns Creek 523 352 0.0908 0.0454
1993 Nunns Creek 573 287 0.0805 0.0402
L’ 1991-1993 Nunns Creek 2096 1233 0.0003 0.0001
1992 Spring Creek 557 413 0.0000 0.0000
1993 Spring Creek 342 139 0.0104 0.0052
EL1992-1993 Spring Creek 899 552 0.0000 0.0000
1991 St. Martin 581 328 0.0011 0.0006
1992 St. Martin 413 335 0.1452 --
1993 St. Martin 520 311 0.0038 0.0019
fl‘m St. Martin 1514 974 0.0000 0.0000 .4
1991 All 2296 1254 0.0000 0.0000
1992 All 2559 1659 0.0000 0.0000
1993 All 2668 _ 1266 0.0000 0.0000

 

 

 

 

 

 

 

 

27

 

 

 

 

 

 

 

 

 

 

 

 

 

Table 7. Results of Kolmogorov-Smirnov tests to detect differences between
years in the length frequency distributions of rainbow smelt sampled
in four St. Martin Bay tributaries from 1991 to 1993.
Year ‘ "Tributary : (5) _' L ‘7 7 "Sample Size. ,Kolmogorov - Smirnov
,5; 3 ' _ . ' " Test Results .
' ' L M . . L Y1<>Y2 ‘ Y1<>7Y27
. g _. L L 1 x L‘ 1 Year,- . L Year2 , P-Value P-Value ’ -
_
1991,1992 Carp River 1047 1625 0.0000 0.0000
1991,1992 Nunns Creek 1594 875 0.0000 0.0000
1991,1992 St. Martin Creek 909 748 0.0010 0.0005
E 1991,1992 All 3550 4218 0.0000 0.0000 L
.l
1991,1993 Carp River 1047 1763 0.0000 0.0000
1991,1993 Nunns Creek 1594 860 0.0000 0.0000
1991,1993 St. Martin Creek 909 831 0.0000 0.0000
1991,1993 All 3550 3935 0.0000 0.0000
g.
1992,1992 Carp River 1625 1763 0.0003 0.0001 i.
1992,1993 Nunns Creek 875 860 0.1630 --
1992,1993 Spring Creek 970 481 0.1315 --
1992,1993 St. Martin Creek 748 841 0.0000 0.0000
1992,1993 All 4218 3935 0.0000 0.0000

 

 

 

 

 

 

 

 

28

 

150 All Fish

145

 

140

 

 

135 '

 

 

150

145

 

 

 

 

140

 

135

Mean Length (mm)

150

145

 

140

T

 

U

135

 

 

1

1991 1992 1993

 

Figure 7. Mean length of spawning rainbow smelt collected from the Carp
River, Nunns Creek, Spring Creek, and St. Martin Creek (Mackinac
County, Michigan) from 1991 to 1993.

29

W

Chi-Square tests indicated no significant differences in the length-age
relationships between tributaries within years (P = 0.452), but significant
differences in length-age relationships between years (P = 0.0217). Based on
these results, I pooled age length keys for all tributaries within years, but
developed and used separate length-age keys for 1991 to 1993 (Appendices 1, 2,
and 3, respectively). I found no significant differences in mean lengths at age for
age classes 2 and 3 between years (Table 9). Significant differences between
years were due to differences in the mean length at age for age classes 4 and 5
(Table 9).
Want;

Age composition of rainbow smelt sampled from all tributaries for 1991 to
1993 is shown in Figures 8 to 10, respectively. Spawning populations on all
streams in all years were dominated by age 2 and 3 fish (<95% of fish sampled).
Large overlaps in the length frequency distributions between these two age classes
occurred in the 135 to 150 mm length increment (Figures 8 to 10). Age 2 fish
were dominant in the 1991 and 1993 samples, while age 3 fish dominated in the
1992 samples (Table 10). Age 4 and older fish never accounted for more than

3% of the samples in any year.

30

Table 8. Results of Kolmogorov-Smirnov tests to detect differences in the length-

frequency distributions of spawning rainbow smelt sampled from four St.
Martin Bay tributaries from 1991 to 1993.

 

K°"“°9°'°t Smlmov‘i: 35573722:19;:
- w7'39"F1930!9.5-: .

i Tributary(s) 1 Sample Size ’-
' I t » . . ,, ' A 71’72 “759%?
_ iP-ValuePNalue

   

1991
1992
1993

Carp. Nunns

1 594
875

0.0002 ~—
0.5397 -- ..
0.0006 --

 

Carp, Nunns

 

Carp. Nunns

 

1991-1993 Carp, Nunns 3329 0.0000 -—

 

1992
1993

Carp. Spring 1625 970

1763 481

0.0000 -
0.0000 -

 

Carp, Spring

 

1992-1993 1451

Carp, Spring 0.0000 --

 

1991 Carp, St. Martin
Carp, St. Martin

Carp, St Martin

1047 0.0000 -

 

1992
19%

1625 748
1763 831

0.0000 --
0.0000 --

 

 

 

 

1991-1993 Carp, St. Martin 4435 2488 0.0000 —-— E0.0000 #
F 1992
1993
1992-1993

Nunns, Spring 875 970 0.0000 —

0.0000 --
0.0000 -—-

 

Nunns, Spring

 

Nunns, Spring 1735 1451

 

1991
1992
1993
1991-1993

Nunns, St Martin
Nunns. St Martin

1 594
875

0.0000 -
0.0000 --
0.3007 -— --
0.0000 --

 

748

 

Nunns, St. Martin

 

Nunns. St Martin 3329 2488

 

1992

Spring, St. Martin

970

748

0.0470 --

0.0235

 

1993

Spring, St. Martin

481

831

0.0208

 

 

1992-1993

 

Spring, St Martin

1451

2488

0.1091

 

 

 

 

 

 

 

Fl

Table 9.

31

Mean length (mm) of rainbow smelt sampled from Lake Huron in
1975-76 by Argyle (1982) and from 1991-1993 by the current study.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Spring 1975‘ - 128 149 168 180 191 193

Spring 1975‘I 102 130 152 170 193 208 183

Spring 1991b - 130 149 175 199 - -

Spring 1992'D - 129 152 178 202 - -

Spring 1993b - 132 150 168 181 - -
‘ Argyle (1982). Bottom trawl sampling at five sites in Lake Huron.

Current Study. Dip net sampling in four tributaries to St. Martin Bay, Lake
Huron.

32

 

500
400 r
300 "
200 *
100 *

500 ' ‘ ‘—

400 . Age 3
N = 2180

Age 2
N = 1248

 

300 r
200 *
100 r
500 -*
400 r
300 "
200 *
100 *

Number

 

Age4
N= 100

 

 

. . . . M 9
90 110 130 150 170 190 210

Length (m m)
Figure 8. Length frequency distributions by age of spawning rainbow smelt

sampled from the Carp River, Nunns Creek, and St. Martin Creek
(Mackinac County, Michigan) in 1991.

33

 

600
500 '
400 *
300 '
200 *
100 '
600
500 - Age 3
400 _ N = 1830
300 ~
200 _-
100 -

600 E ' -

500 - A964
400 __ N =19
300 -
200 ~
100 ~

Age 2
N = 2355

 

 

Number

 

 

 

 

0 - ' 4 * r ._ ‘
90 110 130 150 170 190 210
Length (mm)

Figure 9. Length frequency distributions by age of spawning rainbow smelt
sampled from the Carp River, Nunns Creek, Spring Creek, and St.
Martin Creek (Mackinac County, Michigan) in 1992.

Number

Figure 10.

34

 

 

 

 

 

 

 

 

500 — Agez
N=2630
400 .
200 i
0
60° - Age3
- N=1188
400 .
200 .
0 r.
600 . A994
1 N=94
400 ..
200 .
L
0 1 . . . _..|- . 1
90 110 130 150 170 190 210

Length (mm)

Length frequency distributions by age of spawning rainbow smelt

sampled from the Carp River, Nunns Creek, Spring Creek and St.
Martin Creek (Mackinac County, Michigan) in 1993.

 

35
Table 10.

Percent age composition data for rainbow smelt sampled in Lake
Huron during different time periods.

 

 

 

 

 

 

 

 

 

 

 

 

 

Age

IV
1948-1957‘ 0.0 58.3 39.2 2.1 0.3 0.1 0.0
II 1958-1964‘ 0.0 16.7 47.9 25.1 6.8 2.5 0.7
I 1975-1976b 0.4 27.7 45.8 18.8 5.7 1.5 0.2
I 19914993" 0.0 52.7 45.1 1.9 0.3 0.0 0.0

 

Frie and Spangler (1985). Dip net sampling in South Bay, Lake Huron.
Argyle (1982). Bottom trawl sampling at five sites in Lake Huron.

Current Study. Dip net sampling in four tributaries to St. Martin Bay. Lake
Huron.

Table 11.

Length-weight regressions by year and tributary for rainbow smelt
sampled in four St. Martin Bay tributaries from 1991 to 1993.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

1991 All Both 2069 3.10874 -5.49862 0.9304
1992 All Both 2006 3.05284 -5.37280 0.9507

L 1993 All Both 1992 3.02061 5.30575 0.9289
1991-1993 Carp River Both 1631 3.09561 5.46108 0.9432
1991-1993 Nunns Creek Both 1892 3.06223 «5.40057 0.9277
1991-19m Spring Creek Both 982 3.03463 5.33732 0.9460
1991-1993 St. Martin Both 1562 3.05458 -5.37806 0.9428

 

 

36
Qngth-ngght Rggregsions

Length-weight relationships for individual tributaries by year are shown in
Appendix 4, while summary relationships by year (pooled tributaries) and
tributary (pooled years) are shown in Table 11. Females had higher slope values
for 8 of 11 tributary-year combinations, indicating a general trend of greater
weight at size for female fish. Slopes of the length-weight regression equations
declined annually between 1991 and 1993 (Table 11), and were significantly
different between years (ANCOVA, P = 0.0149).
59x Ratigs

Sex ratios of rainbow smelt sampled during spawning were always skewed
toward male fish, but the relative level of skewness was not consistent between
years. Sex ratios were most highly skewed toward males in 1993 (range 67 to
71%) and least highly skewed in 1992 (55 to 66%) on all study tributaries (Table
12). The degree of skewness in sex ratios appears to be related to the size and
age distributions of sampled fish. The relative percentage of male fish showed a
steady decline with increasing size classes of spawners in the three years of this
study (Figure 11). In smaller size classes (< 150 mm), males comprised greater
than 60% of spawners, while in the largest size classes (> 170 mm), females were
more abundant than males (Figure 11). These patterns appear to be related to
the age composition of spawners. The relative abundance of male fish declined
. from greater than 60% for age 2 fish to less than 30% for age 5 + fish (Figure

12).

37

Table 12. Sex ratios (percent males) of spawning rainbow smelt sampled from
the Carp River, Nunns Creek, Spring Creek, and St. Martin Creek
(Mackinac County, Michigan) from 1991 to 1993.

 

 

 

 

iii/“Tributaries; j , 
Carp River 68.0 65.6 70.0
Nunns Creek 62.7 53.6 ‘ 66.6
Spring Creek --- 57.6 71.1

 

 

 

 

 

 

St. Martin Creek 63.9 55.2 68.1
I Total i 64.6 I 59.3 I 69.2 I

38

 

100
1991

972
495 1017 642

343
87

 

1992

1031 1038 956

Percent Males
O)
O

 

100

80- 1993

1 308
940 1067

 

 

 

 

<130130140150160170180>190
Length Class

Figure 11. Sex ratio (percent males) by 10-mm size class for spawning rainbow
smelt sampled in four St. Martin Bay tributaries from 1991 to 1993.
Sample sizes are shown above each bar.

39

 

100
1991

80 ' N=1580

 

N=2587 N=1286

Percent Males
8

 

100

80 ' N=3315

 

 

 

Figure 12. Sex ratio (percent males) by age of spawning rainbow smelt sampled
from four St. Martin Bay tributaries from 1991 to 1993. Sample
sizes are shown above each bar.

40

Discussion

x R ' n h Fr uen Patterns

During the spawning season, sex ratios of samples collected during
spawning runs were male dominated except at peak spawner abundance, when the
percentage of female spawners approached 50%. Over the course of the entire
spawning season, the percentage of males sampled ranged from 59.3% in 1992 to
69.2% in 1993. Male dominance in spawning samples can be attributed to
differential spawning behavior of male and female rainbow smelt. Females tend
to ascend a river and complete spawning in a single night, while males return
repeatedly to spawning streams throughout the spawning season (Bailey 1964,
Scott and Crossman 1985). Most investigators sampling rainbow smelt during
spawning runs report sex ratios that are skewed toward males (Hale 1960, Bailey
1964, and Jilek at al. 1979, Gebhardt 1993).

Several investigators have reported that adult rainbow smelt sampled
during non-spawning periods were dominated by female fish (Beckrnan 1942;
Gordon 1957; Burbidge 1969; Robinson 1973; Argyle 1982; Gebhardt 1993). Frie
and Spangler (1985) found that the proportion of males in spawning run samples
were consistently higher than in gill net surveys made later in the summer. These
patterns are probably attributable to two factors: 1. males are more vulnerable
. to capture during spawning because they ascend tributaries on multiple nights to
spawn, while females normally spawn on a single night, and 2. males may

experience higher mortality associated with spawning activities (McKenzie 1964,

41
Burbidge 1969, Murawski and Cole 1978, Nsembukya-Katuramu et al. 1981,

Schaefer et al. 1982). Frie and Spangler (1985) found that the percentage of
males in spawning run samples increased from 60% to 78% after cessation of
recreational exploitation in South Bay, Lake Huron. Schaefer et al. (1982) found
that post spawning die-offs of rainbow smelt in Lake Superior contained a
significantly higher proportion of male fish than would be expected from
measured spawning ratios. Schaefer (1979) found that total annual mortality of
males was higher than females for rainbow smelt in Lake Superior.

Sex ratio and length frequency differences in this study indicate that female
rainbow smelt may experience a lower total annual mortality rate when compared
with males. If spawning related mortality due to natural predators or the
recreational fishery is significant, males will experience higher mortality rates
during this period. Differences in vulnerability between sexes to dip-net fishing
were evident in my data set where the ratio of males to females was 1.79:1 over
the entire study period. Declines in the proportion of male fish in larger size and
age classes provides further evidence for differential mortality between sexes.
Observed larger mean lengths of female rainbow smelt were not caused by
differences in growth rates between sexes as mean lengths between male and
female rainbow smelt within age classes were not significantly different from one
another. Significant differences in mean lengths between sexes were due to the
presence of a greater percentage of older female fish when compared to the age

composition of male fish.

42

Tributary differences in length frequency distributions and mean lengths
may be related to recreational fishery intensity. The most heavily exploited
tributaries (Carp River and Nunns Creek) had the smallest mean lengths and
length frequency distributions observed. These tributaries also have higher levels
of recreational fishery effort. Differences in mean size were due to differences in
age composition rather than differential growth rates between sexes or tributaries.
If tributary-related differences are due to changes in age composition resulting
from differing mortality rates, these differences may indicate that spawning
populations at the Carp River, Nunns Creek, and Spring/ St. Martin Creeks may
represent distinct spawning stocks. Genetic analysis would be needed to provide
support for this hypothesis.

n - ' h l ' hi

The slope of length-weight relationships reported in this study were within
the range of coefficients reported for rainbow smelt during the same time period,
and in the Great Lakes during other time periods. Length-weight relationships
were more variable for female rainbow smelt because of the presence of gravid,
partially spawned, and completely spawned females collected in samples.

The slope of length-weight relationship declined annually from 1991 to
1993; however, the decline was not consistent with changes in sampled age
composition or sex ratio. The negative trend in length-weight slopes may indicate

a decline in the average condition of spawning rainbow smelt.

43
W

Age 2 fish were dominant in the 1991 and 1993 samples, while age 3 fish
dominated in the 1992 samples. It appears that a relatively strong year class
(1989) dominated the 1991 samples as age 2 fish and the 1992 samples at age 3.
A relatively strong 1991 year class appears to dominate the age composition as
age 2 fish in 1993. Age composition data indicated broad overlaps in the length
distribution at age of rainbow smelt sampled in this study. Bailey (1964) reported
large ranges in length at age with broad overlaps in the length distributions of
ages 2 to 6. Such differences may be caused by variation in individual growth
rates or the production of multiple cohorts within a year resulting in significantly
different first year growth.

Mean lengths at age did not vary between years for younger fish (ages 2
and 3), but there were significant differences in the mean lengths of older fish
(ages 4 and 5) between years. These differences may reflect growth differences
because older fish utilize different food resources (Mysis, Pontaporea, and fish)
than younger fish (primarily zooplankton) (Gordon 1961, Foltz and Norden 1979).

Mean lengths at age appear to have remained fairly constant during the
last two decades when compared to mean lengths at age from Argyle (1982).
However, there have been significant in shifts the age composition from the late
1940’s to the 1991 to 1993 time period. Frie and Spangler (1985) reported age
. composition data for dip net sampled rainbow smelt from South Bay, Lake Huron
during a period of intense recreational fishing (1948 to 1957) and period when

recreational fishing was absent (1958 to 1964). During the period of intense

44

 

Table 13. Length-weight regression coefficients reported for populations of
rainbow smelt in the upper Great Lakes.
1 ILYears Months Sex N Slope Interce

i I ;f~_ Location

   

pt

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Carp River, Nunns Both 6067 3.06 -5.39
Creek
St. Martin Bay 1991-1993 April Males 3825 3'04 -5.35
Lake Huron Females 2242 3.06 -5.38
r
Days River Males 85 2.97 -5.16
Green Bay 1990-1992 April
Lake Michigan Females 80 2.93 -5.09
Squaw Creek Males 73 3.08 -5.38
Green Bay 1990—1992 April
Lake Michlo ; n Females 72 3.08 534
Point Beach Males 72 3.05 -5.33
Lake Michi n 1990-1992 ril
ga Ap Females 80 3.00 -5.22
Red Arrow Park Males 88 3.00 527
Green Bay 1990-1992 April
Lake Michio : n Females 75 2.94 -5.13
Turtle Creek Males 74 3.26 -5.85
Green Bay 1990-1992 April
Lake Michigan Females 89 3.30 -5.91
Whitefish Bay Creek Males 68 2.82 -4.85
Lake Michigan 1990-1992 April
Females 83 3.31 -5.91
f Two Rivers, Wisconsin 1971-1972 March, May Both Not 3.28 -5.88
' Lake Michigan Given
Zion, lllinois 1971-1972 March, May Both Not 3.27 -5.81
Given

 

Lake Michigan

 

 

45

recreational fishing (1948 to 1957, total annual mortality = 86 to 90%), age
compositions of spawning fish were dominated by age 2 and 3 fish (97.5%).
During a period without recreational fishing (1958 to 1964, total annual mortality
= 66 to 67%), age compositions of spawning fish contained a greater proportion
of older fish. Argyle (1982) reported that younger fish (< age 4) represented
73.9% of the adult population in 1975 to 1976 at five sites in Lake Huron. In the
current study, these younger age classes represented 97.0%, 99.3%, 97.1% of the
adult population in 1991 to 1993, closely resembling the age composition reported
by Frie and Spangler (1985) for heavily exploited populations in South Bay.
These shifts in the age composition toward a younger age composition reflect
higher total annual mortality rates during the 1991 to 1993 time period. The
current (1991 to 1993) population is vulnerable to significant declines or collapse
if the population were to experience 2 to 3 years of recruitment failure.
990911151905

Age composition of tributary spawning rainbow smelt in St. Martin Bay is
largely comprised of younger fish (ages 2 and 3). Although variable year class
strength was responsible for alteration of the age composition of spawning
rainbow smelt, age 2 and 3 fish comprised greater than 95% of spawners during
this study. Age composition of spawning fish is much younger than was reported
for Lake Huron populations in the 1970’s and closely resembles the age
. composition reported for heavily exploited South Bay populations in the late

1940’s and 1950’s (Frie and Spangler 1985).

46

Young age distributions, low sex ratios of male fish, and declining length-
weight regression slopes are indications of stressed populations. Poor recruitment
of rainbow smelt since 1987 has resulted in a decline in the relative abundance of
rainbow smelt in Lake Huron. In addition, high adult mortality rates associated
with salmonine predation, recreational fishery extractions, and other unidentified
sources have resulted in a young age composition typical of a less stable
population. The recreational fishery in St. Martin Bay appears to have a localized
impact on heavily exploited tributary spawning populations, but its effect on the
overall population in northern Lake Huron is probably negligible. If current
levels of low recruitment continue to occur, populations of rainbow smelt can be
expected to decline further. It is important to recognize that rainbow smelt
populations are resilient, and that populations in South Bay, Lake Huron survived
annual mortality rates of 90% during the 1940’s and 1950’s without collapse (Frie
and Spangler 1985). Current estimates of annual mortality rates for rainbow
smelt populations in Lake Huron are approximately 80% (Ralph Stedman,
National Biological Survey - Great Lakes Science Center, Personal
Communication). Species and population specific life history characteristics
including variable spawning locations, early maturation, and high fecundity all

contribute to the resiliency of rainbow smelt populations.

CHAPTER 2: SEASONAL PATTERNS IN EGG DEPOSITION AND
LARVAL OUTMIGRATION OF RAINBOW SMELT FROM
FOUR TRIBUTARIES TO ST. MARTIN BAY, LAKE
HURON

ABSTRACT

To investigate seasonal and interannual patterns in the timing and
magnitude of egg deposition, egg survival, and larval outmigration, I sampled
spawner abundance, egg deposition, and outmigration of rainbow smelt from four
St. Martin Bay tributaries. Indices of spawner abundance indicated that spawning
generally occurred two to five days earlier in the Carp River and Nunns Creek
than on Spring and St. Martin Creeks. Water temperature at peak spawning was
consistent within tributaries, but varied between 3.5 and 7.0 C between tributaries.
Spawning in all tributaries was delayed by prolonged cold water temperatures (<
2 C) in 1992. Nightly egg deposition was strongly correlated with spawner
abundance, except during the early portions of spawning runs when males
comprised greater than 80% of the spawning fish. Egg densities on streams were
negatively affected by recreational fishery activity with high levels of fishery effort
resulting in significant nightly declines in egg density.

Larval outmigration occurred five to nine days earlier on the Carp River
and Nunns Creek than on Spring and St. Martin Creeks due to earlier spawning
times and warmer water temperatures during egg incubation. Diel patterns were
consistent between tributaries with greater than 90% of larval outmigration

occurring between 2100 and 0600 hours. Nightly peak larval outmigration

47

48

occurred between 0000 and 0100 hours. Of the four study tributaries, the Carp
River dominated larval production, producing 72.8%, 74.8%, and 96.4% of the
total larval output from 1991 to 1993, respectively. The four tributary streams
produced a total of 46 million larvae in 1991, 13 million larvae in 1992, and 32
million larvae in 1993, with annual differences due to variable levels of egg
deposition and survival between years. The magnitude of differences in tributary
larval production measured during this study was sufficient to cause significant

interannual variation in recruitment from tributary spawning areas.

INTRODUCTION

Rainbow smelt populations in Lake Huron have demonstrated variable
levels of year class formation that appear to fluctuate independently of stock size
(Argyle 1994). An important first step in the identification of causes for variable
year class formation is to quantify production of eggs and larvae in principal
spawning areas. In the upper Great Lakes, rainbow smelt spawning occurs during
mid to late April and early May in tributary streams and nearshore lake areas
(Scott and Crossman 1985). Anadromous spawning occurs primarily at night in
the lower portions of rivers and streams, with spawning fish usually returning to
lake areas prior to dawn (Bailey 1964). Eggs are released in clusters and become
adhesive, attaching to substrate or vegetation after fertilization (McKenzie 1964).
~ Downstream displacement or drift of rainbow smelt eggs may occur as a

consequence of (1) spawning activity, (2) loss of adhesiveness of the egg

49

membrane, (3) increased stream velocity, (4) attachment of eggs to floating
vegetation or (5) dislodgement of eggs by animal or human activities (Johnston
and Cheverie 1988).

Cold water temperatures or stormy weather during the normal spawning
period may increase the incidence of shoal or shoreline spawning by rainbow
smelt (Scott and Crossman 1985). Shore spawning by rainbow smelt has been
documented in Crystal Lake, Michigan (Lievense 1954) and Branch Lake, Maine
(Rupp 1965), and the occurrence of shore spawning in Lakes Michigan and Huron
are believed to be significant (Personal Communication, R. Stedman, NBS-Great
Lakes Science Center, Ann Arbor, MI).

Egg incubation times are inversely related to water temperature, ranging
from 5 (at 16 C) to 30 (at 3 C) days (Hoover 1936; McKenzie 1964; Cooper
1978). Immediately after hatching, larvae have large yolk sacs and limited
swimming ability (Cooper 1978). Larvae are transported downstream by tributary
currents to receiving waters where they grow and develop into juveniles (Johnston
and Cheverie 1988). Outmigration of rainbow smelt follows a diel pattern with
peak drift rates occurring at night (Ouellet and Dodson 1985; Johnston and
Cheverie 1988; Winnell and Jude 1991).

To further understand factors affecting year class formation of rainbow
smelt, information is needed concerning seasonal and interannual patterns of
spawning, egg deposition, and larval outmigration from tributaries. Quantification
, of egg deposition and larval production and estimation of survival rates from eg

deposition to larval outmigration from tributary streams represent important step

50

needed to identify factors controlling the early survival of rainbow smelt. The

goal of this study was to identify and quantify factors important in determining

larval production of rainbow smelt in tributaries to northern Lake Huron. My

specific objectives were to:

1. Estimate the timing and magnitude of spawner abundance and egg
deposition in St. Martin Bay tributaries.

2. Evaluate the impact of the recreational dip net fishery on egg deposition.

3. Measure the timing and relative contribution of larval rainbow smelt
outmigration from St. Martin Bay tributaries.

4. Estimate survival rates between egg deposition and larval outmigration.

METHODS

52313113900030“:

The abundance of spawning rainbow smelt in tributary streams was indexed
by measuring catch per unit effort (CPUE) of fish captured using a wire mesh dip
net (45 cm diameter, 40 cm depth, 1 cm mesh). Sampling was conducted to
measure CPUE at fixed sites on each tributary twice nightly, once between 2200
and 0100 and again between 0100 and 0400. Preliminary sampling in 1990 and
1991 indicated that spawning fish were rarely present in streams prior to 2130 or
after 0400. During each sampling period, I recorded the number of spawning fish
captured during a timed sampling period ranging from 5 and 60 minutes.

, Sampling was normally conducted until 100 individuals had been collected, or

until the sampling period had elapsed. Sampling was conducted for at least 5

51

minutes at times when spawner densities were high (CPUE > 5000 fish/hour) to
provide an accurate estimate of CPUE.
E i i I'm

Egg deposition was sampled on three of the study tributaries, Nunns Creek,
Spring Creek, and St. Martin Creek (Figure 2). The Carp River was not sampled
due to unwadable depths and drastically fluctuating discharge during spawning.
During this study, spawning generally occurred between April 15 and May 2. Egg
deposition was monitored on Nunns Creek in 1991, and sampling was expanded to
include Spring and St. Martin Creeks in 1992 and 1993. Commencement and
termination of egg deposition sampling was determined by examination of
artificially placed substrate for egg deposition and detection of spawning female
rainbow smelt through dip net sampling. Total annual sampling effort ranged
from 9 to 14 days depending on the duration of spawning (Table 14).
Upstream fish movement on Spring and St. Martin Creeks was restricted to within
150 m of the mouth by road culverts. In 1991, fish spawning in Nunns Creek had
unrestricted access upstream, although most fish remained within 150 m of the
stream mouth. Between 1991 and 1992, a lowhead barrier dam was installed on
Nunns Creek restricting fish movement to within 150 m of the stream month. In

all years, recreational fishing activity on Nunns Creek was limited to 59 m

Figure 13.

  
   

52

 

 

a Protected Site
Unprotected Site

 

Stream Margin

 

 

 

 

 

' Lake Huron

Sampling design used to measure the impact of dip-net activity on
egg deposition. A plywood cover supported by four metal rods was
used to protect one sampling site in each strata from dip net activity

and wading by recreational fishers.

53

Table 14. Total egg deposition sampling effort on Nunns, Spring, and St.
Martin Creeks, 1991 to 1993.

 

. Nunns Creek Spring Creek St. Martin Creek

 

 

 

 

 

 

 

ilijig1991 i . f -. 9 nights
I 4 1992 * ~ ‘ 11 nights 10 nights 10 nights
1993 14 nights 14 nights 14 nights

 

 

upstream from the month by private property, providing a spawning refuge area
(82 m long, 495 m2) between the property line and the barrier dam. On Spring
and St. Martins Creeks, I divided the spawning area into six equidistant strata, five
upstream from the tributary mouth and a sixth located immediately downstream
from the tributary mouth (Figure 13). On Nunns Creek, three sites were located
upstream of the mouth and three were located in the stream plume in St. Martin
Bay. Observations of spawning in 1990 indicated that rainbow smelt tended not
to spawn in stream margins given the water velocities (0.1 to 1.1 m/sec) available
in our study tributaries. Accordingly, fixed sampling sites were randomly located
within the middle 60% of the thalweg of the tributary within each strata.
Sampling sites remained fixed during the study, except that two sites were
relocated in 1992 because of changes in tributary morphology (shifting sandbars)
at the mouth of Nunns Creek. At each site, two paired sample locations were
identified that had similar depth, velocity, distance from tributary mouth, and

substrate characteristics. One sample location at each site was randomly chosen

54
and protected with a 30 X 30 cm plywood cover (Figure 13), which protected the

area beneath from wading and dip-net activity.

I used 19.2 X 9.2 cm concrete bricks covered with burlap to estimate egg
deposition at each site. At each sampling site, this artificial substrate was placed
flush with the tributary bottom (Figure 13). Artificial substrates were placed at
each sampling site before each sampling night. On retrieval, the burlap covers
were removed and eggs were counted in the laboratory under 3x magnification.
Eggs were identified using criteria developed by Cooper (1978).

On each sampling night, egg density estimates at the six protected and six
unprotected sites were pooled to produce density estimates at protected and
unprotected sites. The Wilcoxon Signed-Rank Test was used to compare paired
samples on each sampling night. To evaluate the effect of protective shelters on
egg deposition, I made 13 measurements of egg deposition on nights when there
was no recreational fishing effort. I was also able to obtain 4 additional nights of
observation by placing additional sampling sites above the private property line on
Nunns Creek where they were not vulnerable to the fishery.

To quantify the effect of recreational fishing activity on egg density, I
evaluated the relationship between fishing effort on nights when the mean egg
deposition was greater than 2,831 eggs/m2 (equal to a mean of 50 eggs/brick) and
fishing effort. I determined that egg densities less than 2,831 eggs/m2 were too
low to allow accurate measurement of fishery effects because of highly variable
- spatial distribution of eggs at low spawner abundance. Generally, levels of

recreational fishing effort were low (less than 5 people/ 100 m2) on nights when

55

eg densities were less than 2,831 eggs/m2 (n = 24 nights), because of low
densities of spawning rainbow smelt.
Qigl Sampling 9f Larval Qgtmigggtign

I conducted diel sampling surveys of larval rainbow smelt to determine
daily patterns in outmigration from tributary streams. Drift samples were
collected every three hours (0000, 0300, , 1800, 2100) over a 24 hour period to
estimate diel patterns in larval outmigration. On the Carp River, a total of six 24
hour samples were collected from 1991 to 1993 (Table 15). On the Nunns,
Spring, and St. Martin Creeks, four 24 hour samples were collected.

Diel patterns were interpreted by estimating the proportion of the total 24
hour drift (collected in all eight samples) that was collected during each of the
eight time periods. Because seasonal trends of increasing or decreasing larval
drift were incorporated into the raw data, I tested for significant relationships in
the two 0000 samples preceding and following the 24 hour sample date. I found a
significant trend in the seasonal patterns for 14 of the 18 diel sampling dates.
When significant patterns were found, the diel drift data were detrended by
correcting for the slope of the regression equation between the four sample dates
(see Figure 14). Because greater than 90% of the larval outmigration occurred
between 2100 and 0600, I also sampled hourly between 2100 and 0600 for one
night each on the Carp River and Spring Creek in 1992, and one night on Nunns
Creek in 1993.

56

Table 15. Dates when diel patterns in the outmigration of larval
rainbow smelt were measured on the Carp River, Nunns
Creek, Spring Creek, and St. Martin Creek, 1991 to 1993.

 

= 1991.Sample 1992 Sample 1993 - .
Dates 1. 1.1 Dates _gsamplc._j

  
  
  

  
 

  

 

 

 

 

 

 

 

 

)NunhsiCreek ‘

{LL31 ~'   = ‘ ,   May 22 -- May 16
gspringICreek-g 4 -- May 29 May 26
I ” 7 -- May 30 May 27
StMan int}; ’   4 -- May 29 May 26
Creek f2}, it. -- May 30 May 27

 

Variance for diel patterns were estimated as the standard error of the diel
patterns between sampling dates. Results of diel sampling were used to estimate
the daily outmigration and obtain seasonal estimates of larval production for each
tributary.

N' 1i ' i rv l R in w m l

Larval rainbow smelt production for four St. Martin Bay tributaries was
estimated through drift net surveys. I sampled one large river (Carp River), a
medium size creek (Nunns Creek), and two smaller creeks (Spring and St. Martin
Creeks) nightly between May 6 to June 12, 1991 to 1993 (Table 1). On the Carp
‘ River, a 0.5-m diameter, 2.5-m long, 363-11 mesh conical net was used for

sampling in 1991, while a 1.0-m diameter, 3.0—m long, 363-11 mesh conical net was

57
used in 1992 to 1993. All nets on the Carp River were fished at mid-depth in the

water column. On Nunns Creek, sampling was conducted using a 0.5-m diameter,
2.5-m long, 363-11 mesh conical net that generally fished from the surface to
bottom. On the smaller creeks, 610 mm by 305 mm square drift nets with 363-11
mesh were deployed across the entire water column to sample outmigrating
larvae.

Conical nets were fitted with a General Oceanics flow meter with a low
speed rotor to estimate the volume of water filtered during sampling. Shallow
water depths at the sampling sites on smaller tributaries prevented the use of
flowmeters to estimate volume filtered. On these tributaries, I used a Marsh-
McBirney digital flowmeter to develop regression relationships between sampling
site velocity and stream staff gauge (discharge) measurements. Volume filtered
was then estimated based on the sampling time and staff gauge reading taken just
before and after each sampling period.

On Nunns, Spring and St. Martin Creeks, drift net sampling was conducted
at the mouth of each tributary. On the Carp River, drift net sampling was
conducted at the first point that had consistent unidirectional downstream flow
that was unimpeded by lake seiche activity (approximately 200 m upstream from
the mouth). The majority of spawning activity on the Carp River occurs in two

riffle areas, approximately 800 and 1200 m upstream from the mouth.

58

Seasonal Distribution

 

[00%

d
V

 

Larval Drift (10 6larvae/hour)
0

Figure 14.

sted Larval Drift
6 larvae/hour)

Adju
(1 0

Larval Drift (10 Glarvae/hour)

(0

N

A

O

(O

curl

 

 

 

 

LJIA n A L 9‘9

10 11 12 13 14 15716 17 18 19 2O 21 22 23 24 25 26 ‘27 28

L/Uncorrected Diel Drift Pattern

\

 

 

 

 

 

 

 

 

 

L8"
..... Mayzi
....... 0000
--"I- ..... May22
....... 0000
r “‘I .........
D I . m
I I I I
0 I 0369121518210 0
Transformation
V Corrected Diel Drift Pattern
ll.MIyzo W21 Mayzz M23
0000 0000 0000 0000
-r_I- --------------- I- --------------- I- --------------- Ii
1- I -
I
' I I I
0 0369121518210 0
Time

Diagram showing the process for correcting diel outmigration data
to eliminate seasonal trends. When significant positive or negative
slopes in peak nightly outmigration were detected, data were
detrended to eliminate seasonal effects.

59
Diel sampling in 1991 and previously published studies (Clifford 1972, Gale

and Mohr 1978, Winnell and Jude 1991) indicated that larval outmigration peaked
at night, so all sampling was conducted between 22:00 and 02:00 hours. At each
site, three replicate samples were collected by deploying nets at three fixed sites
to estimate the mean outmigration rate and associated variance. Because larval
drift rates varied by several orders of magnitude during the period of larval
outmigration, I adjusted sampling time on each night between 2 to 10 minutes
depending upon the prevailing larval drift rates. Sampling time was adjusted to
maintain larval catch rates that would allow for accurate estimate of outmigration
rates without sampling excessive numbers of larvae (capture rates ranged as high
as 1500 larvae/ minute). I adjusted sampling times to maintain larval capture
rates at a minimum of 50 larvae per net set. I used a maximum sampling time of
10 minutes at each site to allow for sampling on four tributaries nightly during the
peak period of diel outmigration. When larval outmigration rates were low,
capture rates fell below 50 larvae per 10 minute net sets. All captured material
was preserved in a 5% formalin solution. In the laboratory, larval rainbow smelt
were separated from other collected material, counted, and preserved in 95 %
ethyl alcohol.
5 l E . {I 'l l i ii

Estimates of instantaneous and seasonal mortality from eg deposition to
larval outmigration were calculated based on estimates of total seasonal egg
' deposition and seasonal larval outmigration. I estimated total egg deposition by

multiplying total seasonal egg deposition (estimated at protected sites only) by the

60

area of available spawning habitat in each tributary. Estimates of total larval
outmigration were calculated from nightly peak outmigration estimates and diel
patterns in larval outmigration. Instantaneous mortality estimates were estimated

by methods outlined in Ricker (1975) as follows:

ln Nc-ln N0
Z-
t

 

where
z = instantaneous mortality rate;
Nt = total seasonal larval outmigration;
N0 = total seasonal egg deposition; and
t = time in days between peak spawning and peak outmigration.
n ' l n i '

Tributary temperatures were measured by deploying Ryan recording
thermographs on tributary streams from April through early June of each year. I
did not deploy a thermograph on Spring Creek in 1991 and the thermograph on
Spring Creek failed in 1993. Temperature patterns for Spring Creek in 1993 were
estimated by developing a multiple regression relationship for 1992 when
temperatures were measured on all tributaries. Temperature patterns between
these tributaries follow similar patterns and the multiple regression relationships

used to estimate temperatures were robust (Rz’s > 0.97).

61

Discharge estimates were obtained by establishing staff gauges and
developing discharge-staff gauge reading relationships for each tributary. Fixed
staff gauges were established on each tributary at a point upstream from the
influence of lake seiches. On each sampling night, staff gauge readings were
taken prior to and just after sampling has been completed. Stream discharge was
measured across a range of staff gauge readings to a discharge curve relationship
related to fixed staff gauges on each stream. Staff gauge readings were used to

estimate discharge during each sampling period.

RESULTS

I. . 1 l l . 1 E S wni

Spawning activity of rainbow smelt was concentrated between April 12th
and May 4th on tributaries to St. Martin Bay, Lake Huron. Egg deposition was
generally closely correlated with spawner CPUE on all tributaries. On all
tributaries spawning occurred later and was temporally compressed in 1992.
Although there were consistent trends in the water temperature at peak spawning
within tributaries, peak spawning across tributaries occurred over a range of
temperatures from 3 to 7 C.

From 1991 to 1993, spawning rainbow smelt were generally present in the
Carp River between April 14th and May 2nd (Figure 15). In 1991, concentrations
' of spawning fish were present from April 14th to April 26th. In 1992, cold water

temperatures resulted in delayed and temporally compressed spawning runs

62

compared to 1991 (Figure 16). In 1993, concentrations of spawning fish were
present from April 18th to April 28th, with the highest CPUE of adults occurring
from April 18th to April 20th. In 1991 and 1992, peak spawner abundance
occurred when water temperatures were 5 to 7 C, while in 1993 peak spawner
abundance when water temperatures were only 3.5 C.

In Nunns Creek, the timing and duration of spawning activity closely
resembled patterns observed for the Carp River. In 1991, extensive spawning
occurred over a two week period from April 16th to April 30th, although the
majority of egg deposition occurred from April 215i to April 26th (Figure 17). In
1992, cold water temperatures (< 3 C) resulted in delayed and temporally
compressed spawning in Nunns Creek (Figure 18). Egg deposition was closely
correlated with spawner CPUE with peak spawning activity occurring on April
26th (Figure 17). In 1993, spawner abundance and egg deposition were
biomodally distributed with peak activity occurring on April 20th and April 27th.
(Figure 17). In all three years, peak spawner abundance and egg deposition were
associated with water temperatures of 5 C (Figure 18).

On Spring Creek, water temperature, spawner abundance, and egg
deposition data were not collected in 1991. Egg deposition was closely correlated
with spawner CPUE in 1992 and 1993 (Figure 19). Peak spawning activity in 1992
and 1993 were associated with water temperatures of 4 C (Figure 20). In 1992,
spawning activity ranged from April 23rd to May 4th with peak spawner

abundance and egg deposition occurring on April 26th (Figure 19). In 1993,

63

 

 

 

N-bmm
0000
0000

 

 

O

1 993

E (Number/Hour) of Spawn'
00
O
0

CU
Ad)
00
CO

200 r

O l n l l A 1 n n n
1012141618202224262830 2 4
April May

 

 

 

 

 

 

, Figure 15. Spawner abundance of rainbow smelt spawning in the Carp River,

Mackinaw County, Michigan from 1991 to 1993.

 

25
20 '
15 '
1O '

 

25
20 '
15 '
10 '

Temperature (C)

 

25

 

20'

10'

    

1 5 913172125293 71115192327314 81216
April May June
Figure 16. Water temperatures during spawning and larval outmigration from

the Carp River from 1991 to 1993. Shaded areas indicate periods of
spawning activity and larval outmigration (arrows indicate peaks).

 

 

 

65

 

600 50

500 ’ 1991 - 40
400 r
" 30
300 '
20

200 ’

 

 

 

 

 

400 ’

00
O

100 . m

600 50 8

50° ' 1992 - 40 5
400 -

30 8

300 - g:

200 r 20 3

100 . 10 a

600 1 50 to”

$3

3

J)’

CPUE (Number/Hour) of Spawning Adults

 

 

 

 

 

300 "

200 . 2°

100 . ‘ 10
0

1012141618202224262830 2 4
April May
. Figure 17. Spawner abundance (solid line) and mean egg deposition (bars) of

rainbow smelt spawning in Nunns Creek, Mackinac County,
Michigan from 1991 to 1993.

66

 

 

Temperature (C)

       

 

 

 

l ”19

0 n . .13:- :3:;1: 9 .
5 91317212529 3 71115192327314 81216

April May June

—L

Figure 18. Water temperature during spawning and larval outmigration from
Nunns Creek from 1991 to 1993. Shaded areas indicate periods of
spawning activity and larval outmigration (arrows indicate peaks).

67

 

800 25

1991 ,
600 ’ 20
400 . No Samples Collected 15
10

200 ’

 

800

600 '

400 '

 

200 '

 

 

800

600’

CPUE (Number/Hour) of Spawning Adults
(aw/8663801) uoiiisodeq 663

400 "

200 ’

 

 

 

 

April May

. Figure 19. Spawner abundance (solid line) and mean egg deposition (bars) of
rainbow smelt spawning in Spring Creek, Mackinac County,
Michigan from 1992 to 1993.

68

 

25
1991
20 -
Water Temperature and Spawning

15 Data not Collected in 1991
10 -

 

25

1 992
20 -

15'

Temperature (C)

 

 

25' ' T
20'
15’

     

 

o M A ‘71:: -'
1 5 913172125293 71115192327314 81216
April May June

 

 

Figure 20. Water temperature during spawning and larval outmigration in
Spring Creek from 1992 to 1993. Shaded areas indicate periods of
spawning activity and larval outmigration (arrows indicate peaks).

69

water temperatures in Spring Creek rose above 3 C approximately 7 days earlier
than in 1992, resulting in an earlier and prolonged spawning period (Figure 20).
In 1993, spawning activity ranged from April 17th to May 2nd with peak spawning
and egg deposition occurring on April 28th and 3151 (Figure 19). Spawning
generally occurred 3 to 7 days later on Spring Creek than on the Carp River and
Nunns Creek due to slower warmer of tributary water temperatures (Figures 16,
18, & 20).

The timing and duration of spawner abundance and egg deposition on St.
Martin Creek closely resembled the pattern on Spring Creek due to similar water
temperatures (Figure 21). In all three years, peak levels of spawner activity were
associated with water temperatures of 5 to 6 C. Spawning activity in 1991 ranged
from April 16th to April 27 with peak spawner abundance occurring on April 21st
(Figure 22). In 1992, spawning was delayed by cold water temperatures and
ranged from April 22 to May 3rd (Figures 21 & 22). In 1993, spawning activity
was biomodally distributed with peak spawner abundance occurring on April 19th
and April 28th (Figure 22). This pattern was similar to the bimodal distribution
of spawning observed in Nunns Creek in 1993 (Figures 17 & 22).
WW

I found that protected sites had significantly higher egg densities (Wilcoxon
Signed Rank Test, Normal T Approximation 4.564, P < 0.0001) than unprotected
sites, suggesting that recreational fishery activity caused a decline in the density of
eggs nightly. To evaluate the effect that the protective shelters may have had in

either increasing or reducing egg deposition, I compared egg densities between

70

protected and unprotected sites on nights when there was no recreational fishing
effort. Comparison of these paired observations demonstrated no significant
differences between protected and unprotected sites (Wilcoxon Signed Rank Test,
Normal T Approximation T = 0.349, P = 0.7268) on nights when recreational
fishing activity was absent.

I found a negative relationship between recreational fishing effort and
percent change in egg deposition between protected and unprotected sites
(adjusted r2 = 0.633, P < 0.0001, Figure 23). Levels of recreational fishing effort
below 5 people/100-m2 had little effect on egg densities, while levels of effort
ranging from 12 to 17 people/ 100-m2 resulted in 30 to 70% reductions in egg
density.

i i i

Because recreational fishery activity resulted in significant declines in egg
density, I used eg deposition estimates from protected sites to estimate total
seasonal egg deposition. Nightly egg deposition estimates were summed over the
course of the spawning season to produce a seasonal egg deposition estimate. On
Nunns Creek, total egg deposition was highest in 1991 and declined annually
between 1991 and 1993 (Table 16). Total egg deposition on Spring Creek was
slightly less than the 3-year average on Nunns Creek in 1992, but declined
substantially in 1993 (Table 16). Total egg deposition on St. Martin Creek was
approximately equal to egg deposition on Spring Creek in 1992, but also declined

between 1992 and 1993 (Table 16).

71

 

 

Temperature (C)

 

     

 

 

1 5 91317212529 3 71115192327314 81216
April May June

Figure 21. Water temperature during spawning and larval outmigration in St.
Martin Creek from 1991 to 1993. Shaded areas indicate periods of
spawning activity and larval outmigration (arrows indicate peaks),

CPUE (Number/Hour) of Spawning Adults

. Figure 22.

400

300 '

200'

100’

400

300 ’

200

100

400

300 '

200

100’

72

 

  

40

1991
' 30

Egg Deposition Data
Not Collected in 1991 . 20

 

 

 

 

 

 

 

1012141618202224262830 2 4
April May

 

(aw/$663 8o 1) uoirisodea 663

Spawner abundance (solid line) and mean egg deposition (bars) of

rainbow smelt spawning in St. Martin Creek, Mackinac County,

Michigan from 1991 to 1993.

73

 

 

Percent Change in Egg Density

 

 

'80 l l I
O 5 1 O 15 20

Recreational Fishing Effort (People/100 m2)

Figure 23. Relationship between recreational fishing effort and percent change
in egg density at paired sampling sites in Nunns, Spring, and St.
Martin Creeks sampled in April 1991 to 1993.

74

Table 16. Total egg deposition (millions) by rainbow smelt spawning in three
tributaries to St. Martin Bay, Lake Huron from 1991 to 1993.

 

Total Egg Deposition (x 106)

 

 

 

 

 

 

 

 

 

Nunns Creek 113.4 84.4 70.9
Spring Creek . ----- 69.4 19.0
st. Martin creek 1 ----- 60.3 8.4
L_— * fil
Di 1 in rv lDri

All rainbow smelt larvae sampled during drift net surveys were at the
prolarval stage (yolk sacs present). Diel sampling of larval outmigration
demonstrated that the peak and vast majority of larval outmigration occurred at
night (Figure 24). Diel patterns were remarkably similar between sampling dates
as evidenced by small standard errors (Figure 24), indicating that diel patterns
were stable between tributaries on a seasonal basis. Diel patterns between
tributaries were consistent and not significantly different from one another (Chi
Squarezl = 2.153, P > 0.30). The Carp River, Nunns Creek, and Spring Creek
were each sampled hourly on one night between 2100 and 0600 to produce a finer
resolution of the outmigration pattern during nightly peak events. Results of this
sampling indicate that the nightly distribution of larval outmigration was
approximately normally distributed with peak outmigration occurring between

2300 and 0100 (Figure 25).

Percent of Total Sample

80

GOP

4O
20

O
60
4O

20-

0
60

40-
20’

0
60

40»
20.

0

75

 

 

Carp River

 

Nunns Creek N = 4

 

Spring Creek

 
    

 

 

St. Martin Creek

 
     

 

 

 

1500 1800 2100 0000 300
Time

1 200 600 900

Diel larval outmigration patterns occurring in the Carp River, Nunns
Creek, Spring Creek, and St. Martin Creek. Values are expressed as
a percentage of the total larvae captured during the 24-hour period.

Figure 24.

76

 

500,000 . Carp River May 14,1992

 

400,000 .

200,000 -

 

1,600 n n n n n n n n n n
. Nunns Creek May 15, 1993

1,200 .

 

800 -

400 P

 

o L
50,000 . Spring Creek May 29, 1992

Larval Outmigration (number/ hour)

40,000 r

20,000 -

 

L l

o A n n n 1 n n n
2100 2300 0100 0300 0500

Time

 

Figure 25. Nightly pattern of larval outmigration (number/ hour) occurring on
the Carp River, Nunns Creek and Spring Creek between 2100 and
0600.

 

77
59959991 and Intgrgnnggl Pattgrns in Larval Drift

The Carp River dominated larval production on the four study tributaries,
producing 72.8%, 74.8%, and 96.4% of the total larval output from these
tributaries in 1991 to 1993, respectively (Table 17). High larval output from the
Carp River is associated with the large area of available spawning habitat and
large spawning populations in the river system. In 1991, larval outmigration
occurred from May 2nd to May 25th, with peak hourly outmigration of
approximately 1.2 million larvae/hour on May mm (Figure 26). Larval
outmigration in 1992 occurred between May 11th and May 28th, with the peak
hourly larval outmigration of approximately 860,000 larvae/ hour on May 13th. In
1993, larval outmigration occurred between May 2nd and May 25th, with the peak
hourly larval outmigration of approximately 900,000 larvae/ hour on May 11th. In
1992 and 1993, peak larvae outmigration coincided with a peak in tributary water
temperature. Total larval production in the Carp River was highest in 1991 (33.85
million larvae) and lowest in 1992 (9.88 million larvae, Table 17). Discharge on
all tributaries declined throughout the larval outmigration periods, and there was
no consistent relationship between discharge levels and outmigration on any
tributary.

Nunns Creek was a minor contributor to the total larval production of
these tributaries, producing 0.7%, 1.6%, and 1.7% of the total larval production in
1991 to 1993, respectively (Table 17). The start, peak, and end of larval
outmigration on Nunns Creek occurred several days after the corresponding

events on the Carp River (Figures 26 & 27). In 1991, larval outmigration

78

occurred between May 4th and May 28th, with two peaks in hourly outmigration
on May 13th (13,700 larvae/hour) and May 21st (11,780 larvae/hour, Figure 27).
Larval outmigration in 1992 occurred between May 13th and May 26th, with the
peak hourly migration of 22,000 larvae/hour on May 18th. In 1993, larval
outmigration occurred between May 6th and May 26th with the peak hourly
migration of 53,000 larvae/hour on May 13th (Figure 27). Total larval production
in Nunns Creek was highest in 1993 (550,000 larvae) and lowest in 1992 (210,000
larvae, Table 17).

Spring Creek produced the latest larval outmigration of the four study
tributaries in all three years of the study due to colder water temperatures that
delayed spawning and increased egg incubation time (Figure 28). Spring Creek
had relatively high levels of larval production in 1991, producing almost 10 million
larvae and representing 21.4% of the total larval production of the four study
tributaries (Table 16). In 1991, larval outmigration occurred between May 14th
and June 1st with the peak hourly outmigration of 773,000 larvae/ hour occurring
on May 24th (Figure 28). Larval outmigration in 1992 occurred between May
21st and June 6th with the peak hourly migration of 83,000 larvae/ hour on May
28th (Figure 28). In 1993, larval outmigration occurred between May 20th and
June 15th with the peak larval outmigration of 23,560 larvae /hour on May 26th
(Figure 28). Total larval output from Spring Creek was highest in 1991 (9.97

million larvae) and lowest in 1993 (390,000 larvae, Table 16).

79

 

1991

00

1000 -

300 L
100 -

1992

HQ
00

0)

-I
00°
oo
00
-I
CD
0
00

Larval Outmigration (10 aiarvae/ hour)
3
O

at»
0°

00

 

I I I I I I I I I I

1 4 710131619222528313 6 9121518
May June

Figure 26. Peak nightly outmigration (number/hour) of larval rainbow smelt
sampled between 2200 and 0200 on the Carp River, Mackinac
County, Michigan from 1991 to 1993.

 

100

0|
0

m9? 9

O _LN d”
III-IN 010° O—IN (ll-IN UIOO

Larval Outmigration (10 aiarvae/ hour)
or .0 P .
O —| N

.410
AM 0100

P
or

.0.
AN

Figure 27.

80

 

 

 

 

 

* 1991
;

’ 1992
Z 1993

 

 

1 4 710131619222528313 6 9121518
May June

Peak nightly outmigration of larval rainbow smelt (number/ hour)
sampled between 2200 and 0200 on Nunns Creek, Mackinac County,
Michigan from 1991 to 1993.

81

 

1000

300 - 1991
100 -

d0
00

P
(AI-‘0

D
d
in
L
i.

3

Larval Outmigration (10 larvae/ hour)
3

30° ' 1992
100 -

(d
0

PP
dud”

3“ ' 1993
100 r

.0 at»
(1)-500°

I j I I I I

1 4 710131619222528313 6 9121518
May June
Figure 28. Peak nightly outmigration of larval rainbow smelt (number/hour)

sampled between 2200 and 0200 on Spring Creek, Mackinac County,
Michigan from 1992 to 1993.

 

 

P
.3

82

 

 

 

 

 

 

Table 17. Total outmigration (millions) of larval rainbow smelt from

four tributaries to St. Martin Bay, Lake Huron from 1991 to
1993.

” w 1992 1993 ‘ ~

L Carp River . ' , » » 33.85 9.88 30.58

Nunns Creek ' ¥ , 0.32 0.21 0.55

Spring Creek _- A ” 9.97 0.99 0.39

St. Martins Creek *6 ’ 2.35 2.13 0.20

"Total ', . ' 46.49 13.21 31.72

 

 

 

 

St. Martin Creek was a minor contributor to total larval production in 1991
and 1993, but produced 16.1% of the total larval production of the four study
tributaries in 1992 because of relatively low larval production from other
tributaries. In 1991, larval outmigration occurred from May 13th to May 29th
with the peak hourly outmigration of 66,000 larvae/ hour on May 15th (Figure 29).
Larval outmigration in 1992 occurred from May 17th to May 31st with a peak
hourly outmigration of 225,000 larvae/hour on May 22nd (Figure 29). In 1993,
larval outmigration occurred from May 12th to June mm with no discemable
peak in hourly outmigration (Figure 29). Total larval output from St. Martin
Creek was highest in 1991 (2.35 million larvae) and lowest in 1993 (200,000
larvae, Table 16).

Total larval production from the four study tributaries was highest in 1991
and lowest in 1992. Earliest larval outmigration began on approximately May 3rd

in 1991 and 1993, but cold water temperatures delayed the initiation of larval

83
outmigration until May 11th in 1992 (Figure 30). In all three years, larval

outmigration from tributaries had subsided by approximately June 5th (Figure 30).
T ' M li

Tributary mortality rates were estimated for the period from eg deposition
to larval outmigration for tributaries and years when egg deposition estimates
were available. Instantaneous mortality was highest in Nunns Creek (Table 18),
while mortality rates were similar between Spring and St. Martin Creeks (Table
18). Patterns between 1992 and 1993 were consistent among tributaries, with
higher survival occurring in 1993. Because instantaneous mortality rates were also
higher in 1992, low survival in 1992 was not strictly a function of increased

incubation time (Table 18).

1000

300
100

do:
duct:

99
d“

Larval Outmigration (1 0 3larvae/ hour)
-I 00
O O
O O

‘00
00

PP
dad“

300
100

A”
HQOO

99
d0

Figure 29.

84

 

 

 

 

 

 

 

_ 1991
- 1992
i 1993

 

1 4 710131619222528313 6 9121518
- May . June
Peak nightly outmigration of larval rainbow smelt (number/hour)

sampled between 2200 and 0200 on St. Martin Creek, Mackinac
County, Michigan from 1991 to 1993.

 

1000

300
100

‘0
DO

00

1000

larvae/ hour)

3

gration(10
d (A)
8 8 8 S

00

_s
O
O
O

Larval Outmi
-I 00
O O
O O

‘00
DO

00

Figure 30.

85

 

1991

 

. 1993

 

I I I I I I

1 4 710131619222528313 6 9121518

May June

Total peak larval outmigration (number/hour) sampled between
2200 and 0200 on the Carp River, Nunns Creek, Spring Creek, and
St. Martin Creek from 1991 to 1993.

 

86

Table 18. Instantaneous mortality (Z) and seasonal survival rates (%)
of rainbow smelt from eg deposition to larval outmigration
on three tributaries to St. Martin Bay, Lake Huron from 1991
to 1993.

Instantaneous Mortality Rates (Z)

     
  
 

.;_,,2:—rnbntnyi ’ 7   . w ' ‘ . .. , V ' ’ ; 1991 1992

 
 
 
   

*NunnsCreek , ‘ ' ‘

 

Springcreek I L, L i i. -----
SLMartinCreek ' e * _____

 

 

 

 

Seasonal Survival Rates (%)

 

 

 

 

 

 

 

 

 

LL L _ ___- fl 7h L . ., . ., . . . L . . L i” 7"“ - _ “Vi—LT-J—LL
l; Tributary ~.   _   e   1991 ~ 1992 1993. —  i
l’NttmrsCreek . x "f * , f , 0.300 0.249 0.776
[spring-creek _ ” ' ' ‘ ----- 1.426 2.053
~St.Martin Creek _ . - . . » . 1.875 2.371
DISCUSSION
f r r n h Timi f 'n

Normally, spawning activity commenced in the four tributary streams once
snowmelt associated runoff subsided and tributary water temperatures reached 3
to 5 C. In some years, initial spawning activity was curtailed by a late snowfall
(e.g. April 19 to 20, 1993), which quickly melted resulting in depressed tributary

water temperatures. In this study, peak spawning occurred at temperatures of 3

87

to 7 C, and varied between tributaries. In mid-April, adults (primarily males)
were often present in tributaries when water temperature were less than 1 C.
Peak spawning in the all four tributaries occurred at colder water temperatures (3
to 7 C) than has been previously reported (8.9 C) by Scott and Crossman (1985).
i i n rviv

Egg deposition was highly correlated with spawner abundance, except
during the earliest portion of spawning runs when spawning populations were
dominated (>80%) by males. Egg to outmigrating larvae survival rates were
approximately an order of magnitude lower on Nunns Creek than on Spring or St.
Martin Creeks. There are three factors that might adversely affect survival rates
of eggs and larvae on Nunns Creek. First, water temperatures on this tributary
were generally warmer than on the other two tributaries. Warmer water
temperatures during incubation may lead to increased growth of saphrolitic fungi
which has been previously documented as a source of mortality for rainbow smelt
eggs (Rothschild 1961). I did observe fungal grth in areas with dense egg
densities on Nunns Creek. Second, Nunns Creek hosts a dense spawning run of
white (Catostomus commemoni) and longnose (C. catastomus) suckers, which
generally occurred during egg incubation of rainbow smelt. Spawning activity by
suckers may increase mortality by dislodging eggs and by declines in dissolved
oxygen due to dense concentrations of eggs. During periods of intense spawning
by suckers, I observed increased numbers of drifting rainbow smelt eggs during
drift net surveys. Adult suckers may also feed directly on rainbow smelt eggs,

although this would be difficult to document due to the rapid digestion rates of

88

fish eggs. Third, recreational fishing effort is significantly higher on Nunns Creek
than on Spring or St. Martin Creeks. Because recreational fishing activity was
Shown to significantly reduce egg densities in these study streams, higher levels of
activity on Nunns Creek probably contribute to higher mortality rates between egg
deposition and larval outmigration.

Egg survival rates reported in this study were within the range of survival
rates reported by other investigators. Rupp (1965) reported egg survival rates
ranging from 0 to 2% (average 1.06%) for shoreline spawning rainbow smelt in
Branch Lake, Maine. Rothschild (1961) reported egg survival of 24% to the
prehatching stage, and 0.55% to the prolarval stage for rainbow smelt spawning in
Dean Brook, Maine.

At high egg densities, egg survival is negatively related to egg density
because the close proximity of eggs results in limited dissolved oxygen and
facilitated spread of fungal growth. McKenzie (1947) found a negative
relationship between egg density and hatching success in an enclosed stream area.
He reported hatching success ranging from 0.05% at an egg density of 581,000
eggs/tn2 to 3.7% at egg densities of 5,200 eggs/m2. Rothschild (1961) reports that
maximum prolarval production occurred at egg densities of approximately 126,000
eggs/m2. Average seasonal egg deposition densities in the current study were
generally less than 126,000 eggs/m2, indicating that study tributaries may not be
receiving optimal levels of egg deposition reported to produce maximum larval

outmigration (Rothschild 1961).

89
Rggrgatiggal Fishgry Effggs 9n Egg Dgpgsitign

Differences in egg densities between protected and unprotected sites may
have resulted from localized fish removal prior to spawning, egg dislodgement, or
egg destruction at unprotected sites. Dislodged eggs were likely transported
downstream to lake areas where they were vulnerable to wind and wave action
(Rupp 1965). Although I cannot identify the ultimate fate of dislodged eggs, I
noted that wave action often deposited wind rows of eggs on shore following
major spawning events. In addition, Johnston and Cheverie (1988) found that
41% of drifting rainbow smelt eggs in West River, Prince Edward Island were
dead. Wading activity associated with the recreational fishery also may have been
an important source of direct and indirect mortality of deposited eggs. Previous
research found that angler wading had a significant effect on survival of egg and
fry survival of rainbow trout (911mm M155) and cutthroat trout (Q, glam;
Roberts and White 1992). In addition, Rupp (1965) reported destruction of
rainbow smelt eggs along the Branch Lake (Maine) shoreline associated with
wading and boat launching.

While results of this study measured localized effects of fish removal and
egg dislodgement/destruction, the experimental design does not address whether
the fishery deters schools of spawning fish from entering tributaries. Further, I
did not measure cumulative effects of multiple nights of recreational fishing
activity at sites over the course of the spawning season. In 1991, I attempted to
measure cumulative reduction in egg density by placing egg sampling substrate

during the entire spawning season on Nunns Creek. The majority of these bricks

90

were either transported downstream by repeated "capture and discarding" by dip
nets, or buried under 50 cm of substrate due primarily to dislodged gravel and
sand from recreational fishing activity. If any of these factors is important, our
results will underestimate the total impact that the fishery has on egg deposition
and survival in tributary habitats.

Our results indicate that disturbance of egg deposition by the recreational
fishery can be significant in areas where effort is high (> 10 people/ 100 m2).
Rainbow smelt have been introduced into a number of inland lakes to provide
forage for a variety of gamefish species. In systems where Spawning habitat is
limited, recreational fishery disturbance may impair recruitment of rainbow smelt.
In such cases, management actions such as regulations to limit fishing effort or
provide spawning refuges may be necessary to enhance recruitment of rainbow
smelt.

i l ' i

Diel patterns observed in this study were similar to those observed by
Johnston and Cheverie (1988) for rainbow smelt in the West River, Prince
Edward Island, Canada. In both studies, the majority of larval outmigration
occurred at night, with peak larval outmigration occurring between 0000 and 0200.
Sampling of yolk sac rainbow smelt larvae in the St. Lawrence estuary indicated
that larval drift also peaked at night (Ouellet and Dodson 1985). They found that
although light conditions influenced the vertical distribution of yolk sac larvae, the
influence of current speed appeared to be more important in determining riverine

drift patterns.

91

There are two possible explanations for the appearance of high densities of
larvae at night. First, hatching may occur throughout the 24 hour period and
upward movement by larvae at night results in increased drift of larvae. This
explanation is unlikely because of the relatively poor swimming abilities of
prolarval rainbow smelt in relation to prevailing stream velocities. A more
plausible suggestion is that peak hatching of eggs occurs at night resulting in
higher drift rates. Nocturnal hatch may represent an adaptive mechanism to
allow for displacement into river estuaries at a time when visually feeding
predators are limited by light conditions (Johnston and Cheverie 1988).

During the current study, I was unable to detect a nightly "bigeminous"
(two peaks occurring after sunset and before sunrise) drift pattern described by
other investigators studying larval fish outmigration (Clifford 1972; Reisen 1972;
Armstrong and Brown 1983; Iguchi and Mizuno 1990). In this Study, larval
outmigration was typically normally distributed with peak outmigration occurring
between 2300 and 0100. Johnston and Cheverie (1988) were also unable to detect
bigeminous drift patterns for larval rainbow smelt outmigrating from tributaries to
the St. Lawrence River.

5 l I i Q . . E

The relative seasonal timing in larval outmigration between streams
followed a constant pattern in all three years with outmigration occurring on the
Carp River, then Nunns Creek, followed by St. Martin Creek and Spring Creek.
Seasonal timing of outmigration in these tributaries is controlled by water

temperature, which affects the timing of spawning and the incubation time of

92

eggs. In 1992, spawning activity and larval outmigration were both temporally
compressed into relatively short time periods on all tributaries. The compressed
spawning period was caused by delayed warming of tributary waters before
Spawning. The compressed outmigration periods resulted from both the
compressed spawning period and rapid warming during the late stages of
incubation and hatching. Rapid warming associated with low discharge levels
during this period resulted in rapid development and hatching of rainbow smelt
eggs. In all three years, larval outmigration had essentially ceased by the end of

the first week in June.

CONCLUSIONS

Spawner abundance and egg deposition levels varied between tributaries
and between years. Egg deposition was higher when suitable water temperatures
were present in tributaries during spawning. Recreational fishery activity
significantly reduced nightly egg densities in tributaries when effort exceeded 5
people/m2. Timing of egg deposition and larval outmigration was related to
tributary water temperatures. Cold water temperatures during spawning and early
egg deposition, coupled with warm water temperatures during late incubation and
hatching, resulted in a compressed period of larval outmigration on all tributaries
in 1992. Of the four study tributaries, the Carp River dominated larval
production, producing 72.8%, 74.8%, and 96.4% of the total larval output. The

four tributary streams produced a total of 46 million larvae in 1991, 13.21 million

93

larvae in 1992, and 31.72 million larvae in 1993. Variation in tributary larval
production was caused by interannual differences in spawner abundance, egg
deposition, and egg survival during incubation. Interannual variation in egg
deposition, egg survival and larval outmigration were sufficient to independently
cause a three-fold variation in recruitment levels of rainbow smelt in St. Martin

Bay.

CHAPTER 3: LARVAL GROWTH, SURVIVAL, AND JUVENILE
RECRUITMENT OF RAINBOW SMELT IN ST. MARTIN
BAY, LAKE HURON
ABSTRACT

Larval and juvenile stages of rainbow smelt were sampled in St. Martin
Bay, Lake Huron from 1991 to 1993 to estimate relative abundance and grth
and survival rates. To determine distributional patterns of larval rainbow smelt, I
sampled larval populations using surface and vertically stratified larval trawl
surveys. I used a vertically stratified larval sampling design to evaluate temporal
and spatial patterns in the distribution of rainbow smelt larvae across time of day,
season, depth contour, and depth strata in St. Martin Bay, Lake Huron from 1991
to 1993. Day samples contained significantly lower densities and significantly
smaller larval size distributions once larvae reached a length of 10-mm (early to
mid-July). I found significant trends in larval density and larval size distribution
with increasing depth strata once mean larval size exceeded 10-mm in July.
During late July and August, I found larger larval size distributions with increasing
depth along the 2.5-m, 5.0-m, and 10.0-m depth contours.

Estimates of larval abundance, grth and survival were used to determine
the relative contribution of early hatching cohorts (larvae hatching during
tributary outmigration) and late hatching cohorts (lake spawned larvae hatching
after tributary outmigration had ceased). In 1991, high tributary larval production
(46.0 million larvae) and high degree of spatial overlap between larvae and

zooplankton prey resources resulted in relatively high grth rates (G = 0.0314

94

95

d'l) and relatively low mortality (2 = 0.098 d'l) of early hatching larvae. Otolith
derived hatch date distributions indicate that early hatching, tributary-produced
larvae comprised 88% of juvenile recruits in 1991. Juvenile recruitment in 1991
(CPUE = 25.0), as indexed by a standardized bottom trawl survey, was the
highest recorded during the three year period of this study. In 1992, low tributary .
production of larvae (13.2 million larvae), low densities of zooplankton prey
resources, and poor spatial overlap between larval populations and prey resources
resulted in relatively low growth rates (G = 0.0182 d'l) and relatively high
mortality rates (2 = 0.184 d’l) of the early larval cohort (hatching prior to June
5). Juvenile recruitment was poor (CPUE = 2.9), indicating that recruitment of
late spawning larvae made a relatively small contribution to recruitment. In 1993,
intermediate levels of tributary larval production (31.7 million larvae) and a high
degree of spatial overlap with moderate densities of zooplankton prey resources
resulted in intermediate growth (G = 0.0209 d'l) and mortality (2 = 0.137 d‘l) of
early hatching larvae. Late hatching larvae also encountered intermediate
densities of prey resources, resulting in moderate growth (G = 0.0284 d'l) and
intermediate mortality (2 = 0.126 d'l) of this cohort. Juvenile recruitment in 1993
was moderate (CPUE = 10.9), indicating a moderate year class of rainbow smelt
was produced. Although the relative contribution of late hatching larvae to
juvenile recruitment ranged between 12 and 65 percent, interannual variation in
year class strength of rainbow smelt in St. Martin Bay resulted primarily from

variable contributions of tributary spawned larvae between years.

96

INTRODUCTION

Historical records reflect unstable levels of rainbow smelt abundance in the
upper Great Lakes over the past Sixty years with periods of marked increases
followed by precipitous declines (Baldwin et al. 1979). Variability in the stock
abundance of many fish species is often related to interannual variability in
recruitment (Fritz et al. 1990). Standardized bottom trawl surveys by the National
Biological Survey (NBS)-Great Lakes Science Center indicate variable levels of
year class strength in Lake Huron since 1973 (Figure 1). Fluctuations in the
abundance of rainbow smelt have been attributed to variable recruitment caused
by cannibalism (Regier et al. 1969), disease (Van Oosten 1947), parasitism
(Nepszy et al. 1978), and climatic factors during spawning including wave action
and water temperatures (Rothschfld 1961; Rupp 1965). An annual cycle of
alternating year class dominance of rainbow smelt in Lake Erie between 1963 and
1974 has been attributed to cannibalism by yearling on age 0+ rainbow smelt
(Henderson and Nepszy 1989).

In the upper Great Lakes, rainbow smelt spawn in tributary streams and
lake environments (Scott and Crossman 1985). Although the relative contribution
of tributary vs. lake spawning to overall recruitment has not been estimated,
spawning by rainbow smelt in lake habitats has been shown to be significant in
other inland lakes (Rupp 1965, Legault and Delisle 1968, Plosila 1984). Because

the timing of spawning and hatching of larval rainbow smelt differs between these

97

two habitats (Tin and Jude 1983), the relative contribution of each larval source
may contribute to variability in year class formation of rainbow smelt. Although
distribution patterns of larval rainbow smelt have been previously described for
Lake Michigan (Jude et al. 1980, Tin and Jude 1983), Lake Huron (Emery 1973,
O’Gorman 1983, 1984), Lake Erie (Ferguson 1965), and Lake Ontario (Dunstall
1984), there are no published estimates of larval grth or survival rates of
rainbow smelt in the upper Great Lakes. Overall patterns of larval distribution,
growth and survival of rainbow smelt in the Iaurentian Great Lakes are difficult
to interpret due to differences in time (day vs. night), date, depth contours and
depth strata sampled by previous investigators. Year class strength of rainbow
smelt appears to be determined by early fall, once juvenile fish become demersal
at a size of 30 to 40 mm (Henderson and Nepszy 1989). Variability in year class
strength of rainbow smelt translates directly into variability in the adult standing
stock because greater than 90% of adult spawning stocks in Lake Huron are
currently comprised of only two age classes (Chapter 1).

The importance of rainbow smelt as a key forage species, predator, and
competitor of other Great Lakes fishes underscores the importance of identifying
and understanding the variability in the abundance of adult rainbow smelt in the
Great Lakes. To understand the underlying causes for the variability in
recruitment of rainbow smelt, the factors influencing their year class strength need

to be identified. The overall goal of this study was to investigate factors

98

influencing growth and survival of larval rainbow smelt in St. Martin Bay, Lake

Huron. My specific objectives were to:

1. Describe spatial and temporal patterns in the distribution of larval rainbow
smelt.
2. Measure the abundance, growth, and survival of larval rainbow smelt

during the pelagic larval stage.

3. Describe the relative distribution of important zooplankton prey resources
in relation to the spatial distribution of larval rainbow smelt.

4. Index the year class strength of age 0+ juvenile rainbow smelt through
bottom trawl surveys.

5. To estimate the relative contribution of early (hatching during tributary

outmigration) vs. late (hatching after tributary outmigration had subsided)
larval cohorts to overall recruitment.

METHODS

5.! ill'lD'l'E

I conducted depth-stratified oblique trawling surveys on a monthly basis
from May to August, 1991 to 1993 to determine Spatial and vertical distribution
patterns of larval rainbow smelt in the nearshore regions of St. Martin Bay, Lake
Huron (Figure 31). I used a weighted net frame modified from a design by
Nester (1987) to sample larval fish at depth. This frame eliminated obstructions
at the front of the net (bridle and towing) cable that create a pressure wave
resulting in avoidance of the net by larval fish. Net frames were fitted with 0.5-m,

2.5-m long, $00-11 mesh nets, which have previously been demonstrated to be

99

 

 

 

 

 

 

 

Depth Strata

 

Figure 31. Map showing locations of fixed sampling sites for vertical larval
trawling (Panel A), and stratified sampling design across depth
contours and strata (Panel B).

100

effective in collecting larval rainbow smelt (O’Gorman 1984). General Oceanics
flowmeters mounted in the mouth of each net were used to measure the volume
of water filtered during each sample.

On each sampling date, samples were collected during daylight (1200 to
2000 hours) and at night (2200 to 0300 hours). Sampling sites were fixed
throughout the study (Figure 31) and sites were located using Loran-C (1991 and
1992) and Global Positioning System (1993). During each sampling period, 10
minute oblique tows at an average speed of 1.17 (i 0.05) m/sec were made at
constant depths over the 2.5-m, 5.0111, and 10.0-m contours. Replicated tows were
collected at each of ten contour-depth combinations, with depths ranging from the
surface (0 to 0.5-m depth) to within l-m of the bottom at 2.0-m depth intervals
(Figure 31). Two nets were deployed during each tow and depths were staggered
so that replicate tows at a given depth strata were never collected simultaneously.
We

I conducted surface trawl surveys to determine spatial distributions of
larvae on alternate weeks from May to August, 1991 to 1993. Because previous
investigations and preliminary sampling indicated that larvae were most
vulnerable to surface trawling at night, all sampling was conducted between 2200
and 0300. Twelve fixed transects were established along the 2.5-m, 5.0-m, and
10.0-m contours at each of four sites in St. Martin’s Bay (Figure 32). Each site
was located using Loran-C coordinates (1991 and 1992) and a Global Positioning

System (1993). At each site, paired 0.5-m diameter, 2.5-m long, 363 p. mesh

101

 

 

 

 

Figure 32. Map showing locations of fixed sampling sites for surface
larval trawling (Panel A) and bottom trawling (Panel B).

102

conical push nets were deployed at the bow of the boat at an average speed of
1.04 (:1; 0.0056) m/sec for a period of 10 minutes (O’Gorman 1984, Dunstall
1984).
H 1i f m 1

All captured material was preserved in 95% ethanol (Butler 1992). Larval
fish were separated, preserved in fresh 95% ethanol within 24 hours of capture,
and identified using a larval fish key by Auer (1982). Rainbow smelt were
measured to the nearest 0.1-mm using a computerized digitizing system (Brown
and Taylor 1992). Shrinkage due to preservation in alcohol was determined by
obtaining fresh and preserved measurements of total length of 263 larvae ranging
in Size from 5 to 32 mm. Shrinkage in length due to preservation in alcohol was

corrected for using the following equation:

CL-0.08726+1.06373 PL (r2 - 0.936, n - 263)

where

CL corrected length (mm) and

PL = preserved length (mm).

W
Numbers of larvae caught during each trawl were adjusted to number of
larvae per 1,000 m of water filtered. I used a logc(x+ 1) transformation for all

ANOVA tests (O’Gorman 1984), because variance of both larval density and

103

mean length increased with higher values and larval densities had a non-
parametric distribution.

I tested for diel differences in larval density and mean larval length for
paired day/night samples using the Wilcoxon Signed Rank test (Siegle 1956,
Snedecor and Cochran 1980). I compared samples collected for each of ten
contour-depth strata combinations when there was adequate data (minimum 10
larval rainbow smelt in each paired sample) to test at least 3 of the 10 sampled
contour-depth strata combinations for a given sampling date. There were
adequate sample sizes of larval length data to test diel differences in mean larval
length for 9 of 11 sampling dates.

I used only the night samples to test for differences in larval density and
size across depth contours and depth strata between dates and years because of
inadequate sample sizes from day samples. I tested for differences in larval
density and mean length between depth strata, date and year using a Split-Plot
ANOVA design (Table 19, Maceina et al. 1993). Separate ANOVA analyses
were performed for each depth contour because the number of depth strata
sampled over each depth contour differed (2 depth strata over the 2.5-m contour,
3 depth strata over the 5.0-m contour, and 5 depth strata over the 10.0-m
contour). I also used a Split-Plot ANOVA design (Table 19) to test for
differences in larval density and mean length between depth contours, dates, and

years. When significant differences were detected for depth strata or sampling

104

Table 19. Split-plot ANOVA designs used to test for differences in density and
mean length of larval rainbow smelt across contours. (Depth Strata
= S; Depth Contours = C; Sampling Date = D; Year = Y)

 

Model A: Used to test for differences in density and mean length between
depth strata.

129591191103 Dggzggs 9f Frgeggm
Main Plot A

Depth Strata (S - 1)

Date (D - 1)

Depth Strata * Date (S - 1)(D - 1)
Subplot B

Date (D ' 1)

Date ‘ Year (Y - 1)(D - 1)

Error S(D - 1)(Y - 1)
Corrected Total (SDY - 1)

Model B. Used to test for difi‘erences in density and mean length across
depth contours.

Description 12923951134999.0111
Main Plot A

Depth Contour (C - 1)

Date (D ' 1)

Depth Contour “ Date (C - 1)(D - 1)
Subplot B

Date (D - 1)

Date ‘ Year (D - 1)(Y - 1)

Error C(D - 1)(Y - 1)
Corrected Total (CDY - 1)

105

date, I used Tukey’s multiple comparison of means methods to test for differences
in individual values (Snedecor and Cochran 1980, Steel and Torrie 1980).
r M 1'

Growth rates of larval rainbow smelt from late May through mid-July were
estimated using a length-based method (Hackney and Webb 1978, Zigler and
Jennings 1993). I calculated an abundance-weighted mean date of capture for
each l-mm length group. Growth rates were calculated separately for an early
hatching cohort (larvae hatching during tributary outmigration) and a late
hatching cohort (larvae hatching after tributary outmigration had ceased). Growth

rates for early and late hatching cohorts were then estimated by:

L-ath

where
L = lower limit of each length interval (M);
a = the length axis intercept (m);
G = coefficient of instantaneous growth (day'l); and

t = time in days.

Length groups with mean larval densities less than 10 larvae per 1000 m3 were
not used to estimate growth or mortality. There was generally adequate data to

estimate larval growth and mortality rates for larval size classes from 5 to 20 mm.

106

I used a cohort-specific catch curve method to obtain within-season
estimates of larval rainbow smelt mortality (Campana and Jones 1992). Size
related changes in catchability (vulnerability to sampling gear) made it difficult to
determine instantaneous mortality rates using methods by Crecco et al. (1983) and
Essig and Cole (1986). To utilize cohort-specific catch curve methods, I needed
to estimate the difference in catchability coefficients (q) between early and late
spawned larvae. 5"

I used an analysis technique by Hoenig et al. (1990) to compare the i

 

relative survival of two groups of larvae spawned in the same season. This
analysis technique remains robust when catchability (q) is unequal between groups
and when larval patchiness is problematic (Hoenig et al. 1990). I used the ratio
of catches of early to late spawned larvae sampled at times t1 .. tn (where n is the
total number of samples) to calculate ratios of late to early spawned larvae (R).
I regressed the log of Rt against t. The slope of this regression line represents an
estimate of the difference in instantaneous mortality rates between the cohorts of
fish (C).

Once I had estimated a catchability correction factor, I was able to
estimate time period-specific instantaneous mortality rates with CPUE data based
on previous methods by Crecco et al. (1983) and Essig and Cole (1986) using the

following equation:

 

—ln ( fl)
2 - C N°
t:
where
Z = instantaneous mortality rate (day'l);
= estimate of catchability correction factor;

Nt = larval density (number/1000 m3) at the end of the time

period;
N0 = larval density (number/1000 m3)at the beginning of the time

period; 1
t = days during the time period.

I used only the descending limb of the catch curve to estimate larval mortality
rates because late May larval trawl samples were collected prior to peak larval
abundance.
M91

I used larvae collected during surface larval trawling surveys in 1991 to
determine zooplankton prey species utilized by larval rainbow smelt in St. Martin
Bay. I analyzed stomachs containing ingested prey items from 168 larval rainbow
smelt ranging in size from 6 to 30 mm collected from May to early August.
Larvae used for gut analysis were collected in supplemental samples during both
surface and depth-stratified larval trawling surveys. To prevent egestion of gut

contents, all larvae were anesthetized with tricaine methanesulfonate, and

108

preserved in 10% buffered formalin. I measured total length (nearest 0.1-mm) of
each larvae, and contents of the entire digestive tract were removed (Mchllough
and Stanley 1981). Identifiable gut contents were identified to the lowest possible
taxa using keys by Balcer et al. (1984) and Torke (1974), counted, and measured
for total length.

ZOOpIankton prey availability was obtained from zooplankton samples
collected in conjunction with corresponding surface and depth-stratified larval
trawl samples. For each larval rainbow smelt, I calculated selectivity using the

Strauss electivity index (Stauss 1979, 1982) as follows:
L - I1 — p 1

where L represents the difference between the relative abundance of prey item i
in fish gut (ri) and the relative abundance of that prey item in the field (pi).
Values of each zooplankton taxon range between -1 and + 1; -1 indicating
complete rejection and +1 indicating complete positive selection of a prey taxon.
Zonalanktnnmizling

Zooplankton samples were collected in conjunction with surface larval
trawling surveys to assess the relative abundance of principal prey taxa required
by rainbow smelt. Paired zooplankton samples were collected with each surface
larval tow. Samples were collected using a 0.5-m diameter, 2.5-m long, 64-lr mesh
net. At each site replicate samples were collected from the top 1.0-m of water to
correspond with the depth strata sampled during surface larval trawling.

Collected material was preserved in 5% buffered formalin.

109

Zooplankton were examined using a projecting microscope and identified
using taxonomic keys by Balcer et al. (1984) and Torke (1974). All cladocerans
were identified to species, while copepods were identified as QIclopoida,
Calanoida, copepodites, or nauplii. Rotifers were identified to genus. Cladocera
and copepods were measured for total length using a computerized digitizing
system (Brown and Taylor 1992). Zooplankton samples were subsampled using
either a sample splitter or a Henson-Stemple pipette to obtain subsamples .
containing a minimum of 100 measurable cladocerans and copepods. Rotifers
were counted, but not measured. I used lake water strained (0.126 m3) and the
subsample volume to estimate overall densities of zooplankton (#/l) at specific
sites in the bay. I used the results from larval diet studies to produce estimates of
prey abundance of taxa and sizes of zooplankton selected by rainbow smelt.

I used site specific estimates of larval density and zooplankton prey density
to describe the spatial overlap between the distribution of larval rainbow smelt
and prey resources. To describe the relative abundance and spatial overlap, I
produced maps of relative densities for sampling dates from late May through
July, 1991 to 1993
W

The abundance of juvenile (age 0+) and yearling rainbow smelt was
indexed through a standardized bottom trawl survey. Bottom trawl surveys were
conducted on a monthly basis from June to October 1991, and from May to

September 1992 to 1993 at each of five index stations in St. Martin Bay (Figure

110

32). At each index station, two replicate trawl samples were collected during each
sampling period. Trawl index stations were located using Loran C coordinates in
1991 to 1992 and a Global Positioning System in 1993. Trawl routes were made
along standardized compass headings across depth contours. Because preliminary
sampling indicated that yearling and adult rainbow smelt vertically migrated at
night and were not susceptible to bottom trawling after dark, all bottom trawl
sampling was conducted during daylight hours.

Bottom trawls were made using a 4.9-meter, 2-seam slingshot balloon trawl
(33 mm body mesh, 5 mm codend mesh) that was fished for 10-minutes (bottom
time) at an average speed of 2.7-knots. For each tow, I recorded the starting and
ending location coordinates, depth, engine RPM range, and speed for each trawl
sample. All captured rainbow smelt were measured, weighed, and preserved in
95% ethanol to preserve sagittal otoliths for subsequent age analysis. Age 0+ and
age 1+ rainbow smelt were easily distinguished on the basis of length. Age 0+
CPUE data in August and September were used to index year class strength of
rainbow smelt.
thlith Aging and Vandalism

Otoliths were removed from juvenile rainbow smelt captured in August and
September bottom trawl samples for determination of daily age and assessment of
hatch date distributions (Secor et al. 1991, Epperly et al. 1991). Sagittae were
removed from juveniles, cleaned with 5% sodium hypochlorite and dried for

storage. I embedded sagittae in electron microscopy embedding resin (Embed-

111

812) to prevent cracking of the otolith during sectioning and polishing (Secor et
al. 1992). Embedded otoliths were sectioned using a frontal sectioning plane with
a low speed radial-arm saw (300-400 rpm) using parallel diamond wafer blades
(0.015 mm) to produce thin sections (<05 m). These sections were then
mounted on microscope slides with thermoplastic resin (CrystalBond). Each
section was progressively polished with 220, 400, and 600-grit wet/ dry sandpaper
to produce a thin section (< 50 u) at the otolith core (Secor et al 1992). Final
polishing was completed using aluminum polishing powder (Buehler Micropolish
H 0.3 it alumina) and a metallurgical polishing cloth.

Polished otolith sections were examined under a light microscope and the
number of growth increments from the point of initial increment formation to the
edge of the otolith were counted. Each otolith sample was read three times by a
single reader and mean increment counts were used for subsequent analysis.

I i M W rk

To identify the point at hatching and search for subdaily increments in
rainbow smelt otoliths, I prepared sagittae for inspection using SEM (Secor et al.
1992). All SEM work was completed at the Belle W. Baruch Institute for Marine
Biology and Coastal Research at the University of South Carolina. To prepare
otoliths for SEM examination, I mounted and polished each otolith to expose the
core region. I etched each otolith using a 5% solution of tri-sodium

ethylenediaminetretraacetate (EDTA) for a period of 30 seconds to 2 minutes to

112

enhance daily increments. Samples were then mounted on an SEM stub and
sputter coated with gold (100 Angstroms) and examined under SEM at 15 kV.
Validatigg gf 1h; FQrmgtiQn gf Daily Ingrgmgnts

Validation was defined by Wilson et al. (1983) as "the confirmation of the
temporal meaning of an increment". There are two areas of otolith related
assumptions that needed to be validated in this study:
1. Time of initial increment formation (i.e. the age of larvae when regular

increment formation begins and how well ring structure relates to

incubation and hatching).

2. Deposition rate (i.e. how increment structure can be used as a
chronological record of daily scale).

To validate the formation of initial increment formation, I took two
independent approaches. First, I visually examined the core of several otoliths
using SEM to determine the nature of initial increment formation. At the core,
there was several subtle increments located inside of an obvious and consistent
increment. From my analysis, it appeared that rainbow smelt form otoliths in the
egg stage (prior to hatching), and that there was a sharp, consistent increment
corresponded to hatching.

To confirm this hypothesis, I designed an experiment to determine the
point in the otolith structure that corresponds to hatching. In 1993, I incubated
and hatched rainbow smelt larvae in the laboratory. Incubation water was
changed daily to remove previously hatched larvae and to ensure that larvae had

hatched on the date sampled. On May 18th, otoliths of newly hatched larvae (age

113

0 days) were marked with Alizarin Complexone by immersing larvae into a 200
mg/l solution for a period of 6 hours (Tsukamoto et al. 1989). Marking with
Alizarin Complexone results in a clear area of red fluorescence appearing on the
otolith corresponding with the time of marking (Geffen 1992). Larvae were
maintained under laboratory conditions; harvested after 3 to 10 days posthatch;
sacrificed; and dissected to remove sagittae for examination. I examined larval
otoliths from this study and was able to confirm that formation of the first clear
increment corresponded with hatch date, and that increments formed inside of
this increment were formed during the egg stage.

To validate the formation of daily increments, I examined otoliths from 112
known age larvae held under laboratory conditions and in field enclosures. Newly
hatched larval rainbow smelt were collected adjacent to the lowhead dam on
Nunns Creek in 1993 using drift nets. Because this was the upstream barrier for
spawning migrations in this system, all larvae were assumed to be newly hatched.
Bony structures in all captured larvae were marked with Alizarin Complexone
(200 mg/l exposure for 6 hours) to provide a reference point for subsequent
aging. Approximately 200 larvae captured on a single night (May 13, 1993) were
held under laboratory conditions for a maximum period of 10 days. Larvae
collected on May 13 and 17, 1993 were marked with Alizarin Complexone (200
mg/l exposure for 12 hours) and added to enclosures in St. Martin Bay to grow

under field conditions.

 

114

Larvae raised under laboratory conditions were harvested, sacrificed, and
preserved in ethanol on a daily basis between 3 and 10 day posthatch.
Approximately 50 larvae (true ages 14 to 18 days) and 36 larvae (true ages 55 to
68 days) were harvested from enclosures, sacrificed, and preserved in ethanol for
subsequent analysis. I removed and prepared otoliths using methodology that was
previously described. I was able to prepare and read 60 otoliths from laboratory
raised samples (known ages 3 to 10 days), 27 enclosure raised samples (known
ages 14 to 18 days), and 25 enclosure raised samples (known ages 55 to 68 days).
Neither length nor age of otoliths were known to the reader during analysis. All
samples were counted three times by a single reader and estimated age was
considered to be the mean increment counts for each sample.

I found a significant linear relationship (r2 = 0.951, P < 0.001) between
the mean number of sagittal increments and the known chronological age (Figure
33), indicating the age estimates from increment counts represented a valid
estimate of true age. The slope of this relationship was greater than the expected
1.0, indicating that increment counts increasingly underestimated true age at
older ages. This underestimation may occur because readers fail to identify and
correctly interpret increments or because larvae do not produce otolith increments
on a daily basis. Because increment counts appear to consistently underestimate
true age, I used the regression equation between estimated age and true age to

correct age estimates for older larvae. Maximum error in the analysis of sagittal

 

 

115

increments from 112 known age larvae was i 4 days, and 95% confidence
intervals around increment counts were generally i 3.5 days.
W156

I used estimated daily ages of juvenile rainbow smelt to backcalculate
hatch date distributions of successfully recruited juveniles. Individual larvae
hatching earlier experience a greater cumulative mortality through the date of
collection than larvae hatching later in the season (Campana and Jones 1992).
Therefore, early season larvae are under represented in the backcalculated hatch
date distribution relative to late hatching larvae. This effect is observed
irrespective of the mortality rate, length of hatching period, and interval collection
(Campana and Jones 1992).

Two factors generally control the magnitude of bias observed due to
differential hatch date: (1) the relative mortality rate at the life stage when fish
are collected, and (2) the duration of the spawning period from which larvae are
recruited (Campana and Jones 1992). For larval populations of rainbow smelt in
this study, juveniles used for hatch date estimation were relatively old and had a
relatively low mortality rate compared to larval stages. This would effectively
minimize the observed bias in hatch date distributions. However, the assumed
hatching period of both tributary and lake spawned larvae was extensive (greater
than 40 days) in my Study. This effect would tend to enhance the relative bias of
hatch date distributions calculated from otolith increments. Because of the

extended hatching period of larvae, I used estimated instantaneous mortality rates

 

Figure 33.

116

 

 

 

 

 

 

 

 

20
11
1;; 112 1
$15L 1 ii
9 2L111
8; 1
‘10-
g 1222.
11
g 1 111
3215’ 11
a, 1
Lu
0 l l I
0 5 10 15 20
70 1
7,? 1
>.
to .
965 1 1
8) 1
1
<60- .2 .
E 1
1 1
f 1. l.
“5'55 1 1
m 11
50 11 l l
50 55 60 65 70
True Age (Days)

Results of validation of the formation of daily increments in larval
rainbow smelt otoliths. Panel A shows results for larvae from age 3
to 20 days, while Panel B gives results for larvae from age 50 to 70

days.

117

estimates to correct hatch date distributions for differential mortality between
early and late hatching larvae.

Hatch date backcalculations are subject to error because variation in
increment counts increase with increasing age. Using the validation data collected
in this study, a 95% confidence interval for estimated ages of older larvae (55 to
68 days) was approximately 15 3.5 days, after correcting for the slope between
increment count and known age. Therefore, I used 7 day moving averages to

achieve a better representation of backcalculated hatch date distributions.

RESULTS

i ' r ' rv l D i iz

I found significant differences in the densities of larval rainbow smelt
collected during day and night (Table 20). Larval densities estimated from night
samples were significantly higher (P < 0.05) in all years and for eight of the
eleven sample dates (Table 20). There were no diel differences in density
estimates for late May and early June (4 June, 1992 and 30 May, 1993) samples
collected just after peak hatching. Day/night differences in density on 28 July,
1992 were marginally significant (P = 0.06).

I found significant diel differences in the mean lengths of larval rainbow
smelt collected in paired day/night samples (Table 21). Mean lengths of larvae

collected from night samples were significantly longer (P <0.05) in all years and

118

for seven of nine sample dates (Table 21). I collected insufficient numbers of
larvae during day samples to compare mean lengths for diel sampling on 21
August, 1991 and 24 August, 1992. The two sampling dates when I found no
significant differences in larval length both occurred in late May to early June,
immediately after peak hatch when mean larval lengths were 6 to 7-mm and no
larvae exceeded 10 mm.
D i P B e n n r

Split-plot ANOVAS revealed significant differences in larval density
between depth contours (P = 0.0519), date (P = 0.0917) and year (P < 0.0001).
Significant differences in larval densities along depth contours were due to high
larval densities along the 5.0-m contour on July and August sampling dates
(Figure 34). Tukey’s comparisons indicated that larval densities along the 5.0-m
contour were significantly (P = 0.023) higher than densities along the 2.5 and
10.0-m contours. Significant differences in density between dates were generally
due to declining larval densities through the season (Figure 34). Highly significant
year effects were associated with later sampling dates in 1991 resulting in lower
densities of larval fish and differences between years due to variable year class
strength (Figure 34). Tukey’s comparisons indicated that larval densities were
significantly (at = 0.05) higher in 1993, but that there was no significant difference
between larval densities in 1991 and 1992. The date'year interaction term was
also significant (P = 0.0045) in our analysis indicating that density-date patterns

were not consistent between years. This significant interaction term is indicative

Table 20.

   
 

119

Results of Wilcoxon Signed Rank tests to compare larval densities

collected in paired day and night samples. N = number of paired

samples (depth contour-depth strata combinations) compared.

 

   

Normal T

i Approximation

 

 

 

 

 

  
   

July 24 10 Night 2.242 0.0249
August 8 10 Night 2.344 0.0191
August 24 10 Night 2.223 0.0213
All 30 Night 3.661 0.0003

June 4

 

   

June 30

 

  

July 28

 

 

   

August 24

 

 

   
 

All

 

  

 

    

 

 

 

 

 

 

 

 

May 30 10 Day 1.937 0.0528
June 30 10 Night 2.752 0.0059
July 27 Night 2.039 0.0415
Night 2.650 0.0080
Night 2372 0.0177

 

 

of shifts from nearly uniform distributions across contours in May and June

samples to concentrations of larval density along the 5.0—m contour by late July in

1992 and 1993 (Figure 34).

120

Table 21. Results of Wilcoxon Signed Rank tests to compare mean lengths of
larvae collected in paired day and night samples. N = number of

paired samples compared.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

.L l ‘ jDate  . LN HighestflfiE‘l= ‘ Normal T -1
i5], ' = (Density ‘ Approximation L
L 1991 July 24 Night 2.336 0.0275 L
g 1991 August 8 4 Night 2.183 0.0304 l
l 1991 August 21 2 Insufficient ----------- L
Sample
L 1991 All 9 Night 1.659 0.0472 1
1992 June 4 6 Day 1.048 0.2945 L
l 1992 June 30 5 Night 1.888 0.0591
1992 July 28 8 Night 2.450 0.0143 L
1992 August 24 3 Insufficient ----- mm L
Sample L
1992 All 22 Night 3.117 0.0018 L
i 1993 May 30 9 Night 1.422 0.1551 1
1993 June 30 7 Night 2.282 0.0225 L
L 1993 July 27 Night 2.606 0.0092
I 1993 August 11 7 Night 2.775 0.0069
fl 1993 All 32 Night 4.348 0.0001 _l

 

 

 

 

 

200
1991 7/24 I 2.5-m
150 - 5.0-m
10.0-m
100 — 8/8
A 50 _ No Sample No Sample
a: Collected Collected
8 0
O
b _ 1992 8/24
8 400 l
E
g 0 6/04 6/30 7/28
V 200 -
g No Sample
a, Collected
D 0 ...............
To
a - 1993 7/27
3 2000
1,000
No Sample
Collected
0 ........................

Figure 34.

121

 

 

 

 

 

    
 
   

 

  
 
    
 
  

 

 

    

Late Late Late Early Late
May June July August August

Densities of larval rainbow smelt sampled along the 2.5—m, 5.0-m,
and 10.0—m contours of St. Martin Bay, Lake Huron from 1991 to
1993. Error bars represent _+_ one standard error of the mean.

122

D i P B en h r

I found significant density differences between depth strata for the 2.5-m (P
= 0.0154), but not the 5.0-m (P = 0.5686) and 10.0-m contours (P = 0.6563).
Results of Tukeys comparisons of means for individual depth strata on the 2.5-m
contour indicated that larval densities were significantly higher at the 2.0-m depth
strata than at the surface. Larval densities were lowest at the surface depth strata
for 15 of the 16 depth strata-date combinations in 1992 (Figure 35). Larval
densities were generally lowest at the surface strata across all contours; however,
these differences were not significant due to high levels of variance and
inconsistent patterns between deeper depth strata.

- B D

I found significant differences in the mean length of larvae sampled
between dates (P = 0.0497) and years (P < 0.0001), but not between contours (P
= 0.4592). These results indicate that contour-specific distributional patterns
were not related to larval size. Significant differences between dates were due to
larval growth occurn'ng through the sampling season. Significant differences
between years was primarily due to high growth rates and late sampling times
during 1991, and low growth rates during 1992. The date‘year interaction term
was not significant (P = 0.7852) indicating that time-related patterns between

yeam were consistent.

'- _l

123

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

250
. 2.5-m Contour
200 -
150 - S‘i'iice
200.4“
00
E I
8 """""""""""" I ...................
o __ _
E 250 5.0-m Contour ‘ 3mg
9 200 ' 23:19
3 100 "
g 50
g 250
— 10.0-m Contour
g 200 -;..<_._...L...L ....................... A Su c9
'J mm
100 @521"
8.0-
50 _..-......l_.T'
0 6/04 6/30 7/28 8/24

Sampling Date

Figure 35. Densities of larval rainbow smelt sampled at depth strata along the
2.5-m, 5.0-m, and 10.0-m depth contours in St. Martin Bay, Lake
Huron in 1992.

124
Size-Related Pattems between Depth Strata.

I found significant differences in the mean length of larval rainbow smelt
between depth strata for the 2.5-m (P = 0.0488), 5.0-m (P = 0.0392), and 10.0-m
contours (P = 0.0036). Tukey’s comparisons indicated a general pattern of
increasing larval size with depth across all contours. For the 2.5-m contour, mean
larval length was significantly higher for the 2.0—m depth strata than for the
surface strata (P < 0.05). At the 5.0-m depth contour, mean larval length was
significantly higher (P < 0.05) for the 4.0-m depth strata than the surface strata,
although I detected no significant differences between the surface 0 and 2.0-m,
and 2.0-m and 4.0-m depth strata combinations (P > 0.05). For the 10.0-m
contour, mean larval lengths on the surface strata were significantly smaller (P <
0.05) than the 6.0-m and 8.0-m depth strata. I found no significant differences in
mean lengths among the 2, 4, or 6-m depth strata over the 10-m contour.

Examination of length-frequency distributions for July and August samples
indicates that larger larvae are distributed at increasing depths throughout the
summer. To demonstrate size-related patterns observed later in the growing
I season (July and August), I present length frequency distributions across depth
strata for the 2.5 and 5.0-m depth contours (Figure 36) and the 10.0-m contour
(Figure 37). On all three contours, the mean length and overall size distribution
shifted toward larger average lengths at deeper depth strata across all three depth

contours. This pattern was not present on sampling dates earlier in the season

125

(May and June), and was not observed on sampling dates when small sample sizes
of larvae were obtained.
rv l i

Larval rainbow smelt demonstrated ontogenetic shifts in their preference
for zooplankton prey taxa as larval size increased. Immediately after hatching and
yolk sac absorption, larval rainbow smelt less than 10mm selected copepod
nauplii, smaller cyclopoid copepods, and three species of rotifers, Kertatella,
Polyarthra, and Tn'chocerca (Table 22, Figure 38). Rotifer species and copepod
nauplii declined in importance with increasing larval size between 10 and 16-mrn.
Cyclopoid copepods (primarily Cyclops bicuspidatw thomasi) increased in
importance as larvae grew from 6 to 14-mm and gradually declined in importance
as larvae grew from 14-mm to 22-mm (Table 22, Figure 38). Once larvae had
reached ~16-mm, calanoid copepods (primarily Diaptomus sicilzls) gradually
increased in diet importance. Bosmina longirostms became an important diet item
once populations became abundant in July and August.

i n rv nk n Pr

High output of larvae from tributaries in 1991 (Chapter 2) resulted in high
initial densities of larval rainbow smelt in most areas of St. Martin Bay (Figure
39). High densities of larval rainbow smelt overlapped significantly with high
densities of preferred prey species resulting in high survival and maintenance of

high densities through the first 12 weeks post hatch (Figures 40 and 41). Length

 

Number

10

80

40'

12

Number

20’
15'

10

Figure 36.

126

 

 

20’

Surface
N = 509

2.0 m
N = 230

 

 

 

O 5 1 O 1 5 20 25 3O

 

* 3 Surface

N =503

2N
IIO
33

 

 

4.0 m
N =149

 

 

O 5 1O 15 20 25 30
Length (mm)

Length frequency distributions of larval smelt sampled at depth
strata along the 2.5-m and 5.0-m depth contours in St. Martin Bay,
Lake Huron on 27 July, 1993.

 

127

 

Surface
N = 399

   

 

2.0 m
N = 195

 

 

 

Number
8

 

 

 

 

 

 

0 . 1O 15 20
Length (mm)

Figure 37. Length frequency distributions of larval rainbow smelt sampled at
depth strata along the lO-m depth contour in St. Martin Bay, Lake
Huron on 27 July, 1993.

 

 

128

 

.mm.

 

.nm. Qm

.00. 0.5

 

.00.
v. p

.00. mé

.00. 9m

 

.vv.
0.0

.50. m.N

.05. 0.”

 

 

“268

 

.m5. Tn

.Nm. TV

8.8. a... .

 

.mm. m.N

.mm. m.N

.N: «.0

.0 . N.0

5p

..l¢l

W,.____._......0...:

 

.0. F0

.50. 0...

.nm. v6

.08 5.0

N—

_.H__ we . .68.

 

.0. ".0

.3. 0.0

.m5. Pd

.0»... m. p

N.

 

mm .M t 0..., F, .

 

.0: N0

:3.
.30

.0... m4

.m5. Md

.0... F0

 

.0. P0

.0:
—.0

.5N. 0.0

:0. 5.»

.5n. #0

:

 

.3”. 0.0

as m...

.00.
v.0

.N0. ...0

.m: N0

.03 .4. P

.No. Nd

mp

 

 

.nn. 5.0

.vc. v.0

.50.
0.0

.mm. 0.—

.mm. _.._.

2:;

 

.ON. N0

.03 n0

.5!
0.0

.5N. V0

.53 0.0

we

2 2

 

 

.mm. v.0

.mN.
ad

.0. «.0

 

.nn. 0.0

N—

am om

 

...... _._.u....... . .. using“. .. w1.....3,...1._.._3«
2.3265 6838 ‘ .3962 2.. 1,...23 .26»

a A .3: .826 8.8

8.98.: M £88 1,. «2.6581.

 

_ .8659 .. .1._,66u666...6§ . 15.6%“... .

 

 

 

 

 

 

 

588. he... 323:8 outs. .o

mowsfiouoa 09.0553“. E. new :25 38:5 Rim. .0 8a.: 338»? 05 a. mEB. 59:. .8 Bass: :82 .mm 2%...

129

 

 

 

 

 

 

Strauss’ Electivity Index

 

 

 

 

 

 

 

 

 

 

 

Strauss’ Electivity Index

I“.
/ \1 \-"° "\
\

 

 

4.0 1 1 LI 1 1 1 1 “y 1 1 1 1
6 81012141618202224262830

Fish Length (mm)

 

Figure 38. Strauss’ electivity indices for taxa consumed by larval rainbow smelt
in St. Martin Bay, Lake Huron in 1991.

 

130

 

.82 .38 32 2a 2-2 .32

Lo. :95... 8.3 Sam 552 ..m E huh. cos—5308 van :25. 30958 its. .0 82.33 058% 03m

18A I
:omv E
:mmv I
:omv
:va I

:o—v I
:m v D

>9n. :oECMEOON

 

 

 

 

 

.3 63mm

mEBdA I

 

 

 

 

52 .58 >62

 

A

 

 

 

 

m”5.2V '

 

meadv
meadv I
«Etov I
«.556 D

 

 

52 9-2 >62

:95 385% _623

131

.32 .mm-v~ 25.. can 2.: 0:3

no. :83. 8:3 Sam 552 gm 5 >20. cot—5308 v5 :25 32.58 .93. .o 8:65.. 2.28% can

:omA I

 

:mmv I

 

:omv
:va I
.3: I
:m v D

>3“. co§:m_QO0N

 

 

 

 

 

 

 

mew .mmém 0:31

 

 

 

 

 

 

 

meF .m 7.; 0:31

.9. £sz

winch I
minov I
mitov
oE\m.ov
mEadv I
m”95.3 I
head D

 

zoEw 2625mm EBB

132

5.3 8. :05... 8.3 .58 552 an 5 5o... =05§Eoo~ 9:. :25. 32.58 RE. .0 852% 2.58% SE

:omA I

 

:8v

:va

 

:omv

:2v I
:2v I
:m v D

 

59m co.xcm_a00N

.32 .vmtmm 5.2. Ea NT:

 

filu

 

 

 

 

 

 

 

 

 

 

 

 

52 .2-.. 55..

.3 28E

age? I

MUEnvov I

msixov

 

 

minov
meadv -
m"sidv I
mead D

=95 3856”. _623

133

frequency distributions indicate high rates of larval growth and a small pulse of
smaller, late hatching larvae occurring in mid to late June (Figure 42).

In 1992, low output of larval rainbow smelt from tributaries (approximately
1/3 of 1991 levels, Chapter 2) resulted in low initial densities of larval rainbow
smelt (Figure 43). The distribution of larval rainbow smelt did not overlap
significantly with high densities of preferred prey species concentrated in the I
southeast portion of the bay resulting in low survival, and gradual deterioration in

the larval densities throughout the bay (Figures 44 and 45). Length frequency l

 

distributions indicated low. rates of larval growth and a pulse of small, late
hatching larvae occurring in late June and early July (Figure 46).

In 1993, moderate output of larval rainbow smelt from tributaries
(approximately 2/3 of 1991 levels, Chapter 2) resulted in intermediate initial
densities of larval rainbow smelt (Figure 47). The distribution of larval rainbow
smelt did overlap significantly with moderate densities of preferred prey species
resulting in moderate survival, and gradual decline in the larval densities
throughout the bay (Figures 48 and 49). Length frequency distributions indicate
moderate rates of larval growth and a significant pulse of late hatching larvae
occurring in mid to late June (Figure 50). These larvae comprised a significant

proportion of the larval population in the bay during July (Figure 48).

134

 

1 20 . May 16
80
4O :

40 May 30
20 r

30 : June 15

 

12 I June 25

Number
5;

 

July 11

VfT‘VV

July 24

 

5 10 15 20 25 30
Length (mm)

Figure 42. Length frequency distributions of larval rainbow smelt sampled in
St. Martin Bay, Lake Huron from mid-May to late-July, 1991.

 

» 135

.82 .. 252m :2 as 8-2 a:

8. :23. 8.3 .53 5882 ..m 5 58.. 5.5.5.508 can :25 32.58 .33. .o 8520.. 2.58% 05m

:8A .

 

:mmv I

 

:oNv MM
:2v I
:2v .
:m v D

5an. COECMEOON

 

 

 

 

.8 2:3".

555...? I

m55...? I

 

 

 

 

m8. .. 6:2. - 5 56:

 

 

 

 

 

 

=95 38

N8. .58 >92

  

meadv ___
0Eadv I
H.55.? I
mead D

 

sum .98..

136

dam. .3th 25.. was 3-2 25..

.5. :05... 8.3 .53 588.). gm 5 58.. 58.5.38 v5 :85 32.58 at“. .o 8520.. 2.58% 05m .3. 8:3".

 

:mmv I
Zomv
:va .
:Ev I
:m v D

 

 

   

3431.3. . . Wm mEBdA-

.Hfiuuum.yxmuum..uu mF.\m.OV.

 

5.5.4.3

 

 

 

 

 

 

Nmmw .mm-wm mF5...

 

meadv

 

mimov I
mgidv I
mead D

 

 

 

 

 

 

 

58m coicmaofi Nmmw .0 Tmr 0:3... :35 32.5mm $20..

137

.82 .2le :3 ea 2-:

5.3. .8 :82: 8.3 .53 :88). am 5 58.. 88.5.2.8 cam :85 32.58 .33. .o 852% 2.58% 85

 

 

:omv .fi

:va m

 

 

:omv
:8v .
:2v l
:m v D

 

  

 

 

 

 

.2. 88E

553.? I
oE:m.ov I

95.5.3

 

 

N2: .52 22.

 

m=5de

 

 

 

 

 

 

mENdv a
muEFov l
mead D

 

 

58a. :oESwEOON N02. 6.1.1. >5... =me 32.5mm Eta...

138

 

 

 

 

 

 

 

 

 

 

400
. May 20
200 -
300 L 1 1 1 1
.. June 1
100 r
12 - June 15
g 23 '. . . . .
E
a ' June 28
Z 10 ..
I. .II IIIIIIIIIIII.._ . .
July 14
I fil—I—L—a 1 I
July 28
1O -
I l I

 

 

 

5 10 15 20 25 30
Length (mm)

Figure 46. Length frequency distributions of larval rainbow smelt sampled in
St. Martin Bay, Lake Huron from mid-May to late-July, 1992.

 

139

Winn/11832195

Instantaneous growth rates of the late hatching cohort were higher than the
early hatching cohort in 1992 and 1993, but not in 1991 (Table 23). Growth rates
for the early cohort were highest in 1991 and lowest in 1992 (Table 23). Growth
rates appear to be less variable between years for the late hatching cohort, when
compared to the early cohort of larval rainbow smelt.

Instantaneous mortality rates differed between early and late cohorts for
1991 to 1993 (Table 24), and were inversely correlated with the relative growth
rates exhibited by these cohorts (Figure 51). In 1991, high growth rates (G =
0.314 d'l) of the early hatching cohort corresponded with low instantaneous
mortality rates (2 = 0.098 d'l). In 1992, the early hatching cohort exhibited low
growth rates (G = 0.182 d'l) and high mortality rates (z = 0.184 d'l), while the
late hatching cohort exhibited intermediate grth rates (G = 0.251 d'l) and
lower mortality rates (2 = 0.169 d‘l). In 1993, the early cohort exhibited marginal
growth rates (G = 0.209 d'l) and moderate mortality (2 = 0.137 d'l), while the
late hatching cohort demonstrated higher growth (G = 0.284 d'l) and reduced
mortality (2 = 0.126 6'1).

1' 'v i ri i

Otolith-derived hatch date distributions indicated differential contribution
of tributary vs. lake spawned larval contribution to juvenile recruitment. In 1991,
backcalculated hatch dates indicate that the majority of the 1991 year class was

composed of fish outmigrating from tributary sources (Figure 52). Later spawned

 

140

_ .82 .. 82 ea 2-: 32
.8 :83. 8.3 .53 588). ..m 5 58.. 58.8.38 can :85. 32.58 is. .o 8528 2.58% SE

 

ZomA -
:omv .m.

:mmv D

 

 

 

 

 

 

 

 

Zomv -

mmmw ._. 0:2.

 

:mwv .
:2v I
:m v D

 

 

 

C

 

 

 

 

 

58.". :o§:0_aooN

82 8-: >55.

.2. 2:26.

558.? I
aEmov I

 

meadv I
mitov I
aead D

=0Ew BOD—har— BENI—

141

.8... .5m 0:3. .28 Eta. 0:3.

5.. 58:... 8.3 .53 588). gm 5 502. 58.5.58... 55 52% 32.58 .98. .o 8.520.. 05.8% 85

Zomv

:mmv

:omv

:2v I
:2v I
:m v D

58.". :otEmEOON

 

 

 

 

 

 

mmmw .5m 0:2.

 

 

 

 

 

 

 

 

 

mam. .5 70.. 0:35.

.2. 2.5..

high I
meadv I

mitov

 

meiov
mE:~.ov I
mcidv I
«866 D

 

:85 .5858 _uafi

142

.82 .8 2.; 9a a
33 he 5:5 33 $3 Eta—2 4m 5 xoa Sig—meow v5 :05 3858 REE .«o 8222. 058% BE

 

:omA.

 

:omv

:mmv.

 

 

 

 

 

 

 

:va.
:Sv.
:m v_H_

 

 

 

 

 

 

«a 8:9".

min? I

meadv

 

m

 

msadv

mafidv I
M”EFov I
«866 D

. >9n_ co§:m_Q00N mam? .N_. >51 =oEm 26352”. 323

143

 

 

 

 

 

 

 

 

 

 

300 I May17
200 z
100 C
200 ‘ ‘ ‘ ‘ ‘
. June1
100 ’
June13
June 27
: July12
20 r II ll
40 : July27
20 ,
o L

O 5 10 15 20 25 30
Length (mm)

Figure 50. Length frequency distributions of larval rainbow smelt sampled in
St. Martin Bay, Lake Huron from mid-May to late-July, 1993.

144

 

 

 

 

 

 

Table 23. Length-class based estimates of growth for early and late hatching
cohorts of larval rainbow smelt in St. Martin Bay, Lake Huron from
1991 to 1993.
f’ Year Cohort * Instantaneous ' Standard
“ ' I ‘ - Growth Rates ~ . Error
1991 Early (May-early June) 0.0314 0.0031
. Late (mid June-July 0.0247 0.0078
Early (May-early June) 0.0182 0.0063
Late (mid June-July) 0.0251 0.0097
Early (May-early June) 0.0209 0.0049
. Late (mid June-July) 0.0284 0.0062

 

 

 

 

 

 

145

0.200

:2 0.180 -

tY (day'
.0
3
O
I

0.140 '-

, 0.120 -

 

 

Instantaneous Mortali

I Early Cohort

0100 - 9 Late Cohort I

 

 

 

 

 

 

0.080 I l I l I l A 1
0.016 0.020 0.024 0.028 0.032

Instantaneous Growth (day'l)

Figure 51. Relationship between instantaneous growth and instantaneous
mortality rates for early and late hatching cohorts of rainbow smelt
in St. Martin Bay, Lake Huron.

146

tributary larvae (hatch dates from May 20 to 28, 1991) had higher survival rates
than early hatch larvae (May 6 to 15, 1991). Approximately 88% of the
backcalculated hatch date distribution occurs during the period of tributary larval
outmigration prior to June 1, 1991.

By contrast, the hatch date distribution in 1992 was dominated by larvae
hatching from mid-June to mid-July in 1992 (Figure 53). Tributary spawned
larvae hatching between May 11 and June 2 experienced poor survival (Table 24)
and made a relatively small contribution to the 1992 year class. Larvae with hatch
dates after June 5th (the end of tributary outmigration) comprised approximately
65% of the 1992 year class.

In 1993, the hatch date distribution was dominated by larvae hatching
during May (Figure 54). The hatch date distribution of recruited juveniles was
highly correlated with larval outmigration measured in four St. Martin Bay
tributaries, indicating that larval survival of tributary-produced larvae with
differing hatch dates was relatively constant. A smaller, but significant group of
juveniles was recruited from larvae hatching between June 10th and July 9th, after
tributary larval outmigration had subsided. Recruited juveniles with batch dates
before June 5th (end of tributary outmigration) comprised approximately 74% of

the 1993 year class.

147

.823 :32
3330—88—08 8.. 3385 3:03:00 *3 05 newsman». A93“. On «3 23:5: :88 >51. 382%..
mam .82 525.com 98 33:4. 5 BEES :25. 33:3 2202:. 3:202 28 8023:66 23 :8“: .mm Eswi

 

 

 

 

 

 

 

 

>5... 0:3... >55.
«.meonouumfiiopm«mumuputfimm.o 1..
1d d d d‘ w
A N A
. t W
o m
U
. Q N
arm u 2 . 2 n
20:33:55 3mm :03: .939:sz m M.
N. e
«Poconomuumpzopowamuputnpmmw 11
q d d d u 1 d d d d d F ) e
L .A.
. o O... W
. 2 l
m m
4 on m 1
n a w-
oo. / 6
. 8» u. m
o
cos—«5.5.30 _mim... 333...... < n n
. . . . 820 0
U

148

.003: :83
33:28:03 :8 3225 00:09.28 9» mm 05 w:::0m2:2 Gan: On 3 23:5: 5.2: 53.: 822:2
2mm .82 £35033 :5 “2&3: :m :0388 :08... 3352 0:52;. 3:22 :8 80:32.2: 08: :03: .8 2:2,:

2:: 0:3... >22

2 a 0 882232 a N 33822 a m .o
. u
. e
. o

mmuz .m

5:35.05 Ban. :30: :BmEzmm m . o—

 

 

 

NF m #onmNNNmp¢Fo_. o NQNmNFNthp a mp

‘ d d d d d l d fl d I d d P

 

 

o
8

e
uanerfigwmo |elue1

O
0

. 00—.
. com

5:299:30 $20.. 333:... < . 82

(Jnoq/eelueI 0|.) .qutunN “99W ABP-l.

 

 

 

149

.005: :05:
3550—00:03 :8 5205 00:03:00 :0 no 05 5:522:02 aha: ma 3 235:: 5005 .95.: 522:2

 

 

 

 

20m .82 £35053 :5 0554. 5 3558 ..050 33:5: 2502:. :2552 .5: 0:25:32: 05: :05: .vn 05wEM
>5: 0:3... >05. m .w.
20000382320«mumuputnpomF 2mm
, O - r.
p m...
. N e .m m
0A d O
. e W mm
a g
0 e om
4 U Wm
. Q N wh
mi n z . 2 n m... n.
8:352: 2a: :30... 85250.". 0 w m .m m
«F m. .m yw
200800002320«$352200F JMWW
d d d J C d W 4 1 J P U NE “0...
e B
.. n ObA n
10mm.
. e p. m M.
. on A m 9.
W R .m
cop VIM.
. 30 52
5322550 .020.— >55n_:._. < m mu"
. . 820mb

 

 

 

F
U

150

B mT wlY r 1a In ex

Bottom trawl surveys from May to September indicated that age 0+ and
age 1+ rainbow smelt were most abundant in St. Martin Bay. In 1991, age 1+
rainbow smelt were abundant inshore during May and June, but migrated outside
of the bay by July (Figure 55). Age 0+ rainbow smelt from the 1991 year class
were recruited to bottom trawling gear by mid-July and were most abundant in
August 1991 (Figure 55). In 1992 and 1993, age 1+ rainbow smelt remained in
St. Martin Bay throughout the summer (Figures 56 and 57). In 1992 and 1993,
age 0+ rainbow smelt were not fully recruited to bottom trawling gear until
September (Figures 56 and 57) because of relatively slow growth rates (Table 23).

The relative sizes of the 1991 to 1993 year classes of rainbow smelt were
indexed by bottom trawl surveys in August and September. In 1991, large
numbers of age 0+ juvenile rainbow smelt were captured in bottom trawls and
the combined August/ September CPUE was 25.0 fish/tow (Table 25). In 1992,
relatively few age 0+ rainbow smelt were captured and the combined
August/ September CPUE was only 2.9 fish / tow (Table 25). In 1993, intermediate
numbers of age 0+ juveniles were captured with a combined August/ September
CPUE of 10.9 juveniles / tow (Table 25). Differences in juvenile recruitment
between years were due primarily to variable levels of recruitment from early
hatching larval cohorts (Figure 58). While the contribution of early hatching
cohorts was highly variable, late hatching cohorts made a relatively constant

contribution to juvenile recruitment between years (Figure 58).

151

 

May
N = 46

 

 

June
N=260

 

 

 

July
N = 264

 

 

August
N = 384

 

 

September
N = 125

 

 

 

 

0 - 5 L . _._ h . l .
o 20 4o 60 80 100 120 140 160 180
Length (mm)

Figure 55. Length frequency distribution of juvenile and adult rainbow smelt
sampled during monthly bottom trawl surveys in 1991.

12

120

80’
40*

120

Number

Figure 56.

152

 

 

 

80’
40*

Age

 

20'

10*

 

30 r
20

 

e1

 

 

Ae 2+ May?

N=136§

§A9e 2+ June
. gr-N =697

 

 

 

September

N=128

 

o 20 4o 60 80 100120140160160
Length (mm)

Length frequency distribution of juvenile and adult rainbow smelt
sampled during monthly bottom trawl surveys in 1992.

 

 

 

 

 

 

 

 

 

 

 

 

 

2 .
1
0 4
20 -
10 ’
L. 60 5 5
(D ’ E E
_Q 40 _ Age1 AgeZ+ é ENJul!4
E r '
2 2° ;
80 bn __ E E A 1: E . __
60 L ' ge AUQUSt
. ' N=242
September
=257

 

 

 

 

 

O ‘ : 5 . ‘ A _
o 20 4o 60 80 100 120 140 160 180
Length (mm)

Figure 57. Length frequency distribution of juvenile and adult rainbow smelt
sampled during monthly bottom trawl surveys in 1993.

 

154

Table 25. Total catch and combined CPUE of age 0+ rainbow smelt indexed

using a standardized bottom trawl survey in St. Martin Bay, Lake
Huron from 1991 to 1993.

 

 

 

 

 

-.-:Yearf August Age 1 August , September September Combined
j 7 0+ Catch Effort” v Age 0+ ~ ‘ Effort ' CPUE
.' " , a g (tows) ‘f Catch ' ‘ (TOWS) ' w ' ,
51991 7 376 10 123 10 25.0
.1992 f i; 21 10 34 9 2.9

J. ,i 2'

”“1993 ~ 11 10 196 9 10.9

 

 

 

 

 

 

 

155

 

 

El Early Cohort (Tributaries)
I Late Cohort (Lake)

 

 

 

 

Bottom Trawl CPUE
a

 

 

1991 1992 1993
Year

Figure 58. Relative contribution of early and late hatching cohorts to juvenile
recruitment in St. Martin Bay, Lake Huron from 1991 to 1993.

156

Discussion

Diel differences in the density and size distribution of larval rainbow smelt
are most likely a function of vertical migration and net avoidance. The influence
of both of these factors on diel differences escalates with increasing larval size
because of enhanced swimming ability of larger larvae. In May and early June,
when the average size of larvae is less than 10-mm, I found no significant
differences in the density or size distributions between day and night samples.
However, once larvae reached a size exceeding 10-mm, vertical migration patterns
became apparent. Tin and Jude (1983) found similar patterns, reporting that
larval rainbow smelt exhibited no particular pattern of vertical distribution in the
water column from May until early August, but that age 0+ were closely
associated with the bottom in late August and September. As larvae grow and
become more easily detected by predators, vertical migration may reduce the risk
of predation by visually feeding predators (Miller et al. 1988).
WWW

During late May and June, larvae appeared to be uniformly distributed
across depth contours in the study area. Dunstall (1984) found that densities of
rainbow smelt larvae in lake Ontario were generally greatest in surface waters
A over the 3-m depth contour immediately after hatching. He found decreasing

densities of larval rainbow smelt with increasing depth from the 3.0-m to the 13.0-

157

m contour indicating dispersal of larvae from shallow water regions. Larval
densities in nearshore regions of our study area increased late in the season in
1992 and 1993, indicating an onshore migration of age 0+ rainbow smelt. Tin
and Jude (1983) noted a similar pattern in Lake Michigan, reporting higher
densities of larval rainbow smelt in nearshore waters in late summer samples.

I found several consistent trends in the vertical distribution of larvae.
Larval densities at the surface depth strata were usually the lowest densities
recorded on each sampling date. Highest larval densities were associated with
mid-depth strata in May and June, and with near-bottom strata in July and
August. Emery (1973) found that rainbow smelt larvae in Georgian Bay, Lake
Huron were closely associated with the bottom, and reported that larvae vertically
migrated into upper depth strata at night from July to September.

O’Gorman (1983) sampled larval rainbow smelt in nearshore areas (5 to
10-m depths) of western Lake Huron during daylight hours to determine
geographic distribution and seasonal occurrence and abundance. Abundances of
rainbow smelt were highest from 2 to 6-m beneath the surface with abundances
ranging as high as 2.5 larvae/m3. O’Gorman (1983) found that peak abundance
of larval smelt in St. Martin Bay (467 larvae/1000 m3) occurred in late May to
early June. lobserved approximately the same densities in late May to early June

. in 1992 and 1993, but larval densities in 1991 were approximately three times as

 

158

high as were observed by O’Gorman (1983). These differences reflect interannual
variation in the year class strength of rainbow smelt in this area.
R ' rv l is ri i

Length-frequency data from larval rainbow smelt collected on 27 July, 1993
demonstrate two important depth-related patterns. First, the mean size and
length frequency distribution is shifted toward larger size in the deeper depth
strata along all three contours. Second, at the 6.0-m and 8.0-m depth strata on
the 10-m contour, the size distribution appears to be bimodal indicating a second
cohort of larger larvae. I believe that the smaller size classes (6 to 20 mm)
represent larvae hatching from lake spawning, which are spawned later and
incubate longer due to colder lake water temperatures. The larger size classes
(22 to 30 mm) represent tributary spawned larvae, which hatched in May and
early June in St. Martin Bay tributaries.

There are several explanations for size-related patterns in the depth
distribution of larval rainbow smelt. Larger larvae may inhabit deeper water as a
predator avoidance mechanism. As larvae grow, they become more detectible by
visual predators including adult rainbow smelt and alewives (Crowder 1980).
larger size distributions collected in night samples indicate that larger larvae may
vertically migrate toward the surface at dusk. Nightly vertical migrations into the
upper water column may allow larger larvae to feed on larger zooplankton, which
‘ also migrate up in the water column after dark. However, it is difficult to

determine whether length frequency differences between day and night samples

 

159

are due to vertical migration or gear avoidance by larger larvae during day
sampling.

It is unlikely that size-related patterns are due simply to dispersion away
from spawning tributaries. The sample sites utilized in this study are located at
least 2 km away from primary egg deposition areas and larvae dispersing from
these areas will encounter deeper depth contours prior to reaching the shallower

areas. In addition, I was unable to detect differences in the mean size of larvae

 

between contours. .
MW

Larval growth rates were related to the spatial overlap of larval rainbow
smelt and appropriate prey taxa and size classes. High densities of prey resources
in late May and early June resulted in rapid growth of the early cohort of rainbow
smelt in 1991. In 1992, low densities of prey availability during the same period
and lack of spatial overlap between larval rainbow smelt and prey resources
resulted in lower larval growth rates. By late June and J uly, higher densities of
prey resources resulted in higher growth rates for the late hatching cohort. In
1993, moderate densities of prey resources and a high degree of spatial overlap
between larvae and prey resources is reflected by intermediate growth rates.

ri i i nt

Otolith-derived hatch date distributions indicate that 12% to 65% of

. juveniles contributing to recruitment were hatched after larval outmigration from

tributaries had ceased. Other investigators (Tin and Jude 1982) have suggested

160

that later hatching larvae originate from eggs spawned in lake environments. To
reach this conclusion, it is necessary to eliminate other possible explanations for
the patterns observed in this study.

The tributaries selected for this study covered a wide range in size and
average water temperatures. While other St. Martin Bay tributaries contributed
to the larval populations in the bay, it is unlikely that the timing of outmigration
occurred later than the four tributaries. Of the study tributaries, Spring Creek.
consistently had the coldest water temperatures. This tributary hosted the latest
spawning runs and latest dates for larval outmigration in all three years. Spot
sampling of water temperatures on 7 other St. Martin Bay tributaries (Pine River,
Paquin Creek, McCloud Creek, and 4 unnamed tributaries) and 4 tributaries just
west of the Straits of Mackinac (Martineau, Rabbit Back, Hoban, and Moron
Creeks) demonstrated that Spring Creek had the coldest water temperatures
during late April and May. Because spawning, incubation, and hatching of
rainbow smelt is controlled by water temperature, it is reasonable to assume that
Spring Creek had the latest larval hatch dates of St. Martin Bay tributaries.
Because the nearest upcurrent tributaries outside of the St. Martin Bay-Mackinac
Straits area are located in Lake Michigan, it is highly improbable that significant
numbers of tributary spawned larvae would immigrate from sources outside St.
Martin Bay.

The timing of spawning and hatching of lake spawned larvae was not

directly measured in this study. However, differences in the timing of spring

 

161

warming of tributary and lake areas control the timing of spawning and larval
hatching in lake areas. Because the bay warms more slowly than its tributaries, it
takes longer for lake areas to reach preferred spawning temperatures causing lake
spawning to occur later than tributary spawning. Shoreline seining conducted by
personnel from the Chippewa-Ottawa Treaty Fishery Management Authority and
bottom trawling conducted for this study during May and early June routinely
captured spawning condition fish, sometimes weeks after tributary spawning had
ceased. Because the rate of warming during incubation is slower in lake areas,
the incubation time of lake spawned eggs is longer than for tributary spawned
eggs. later spawning and longer incubation of lake spawned eggs result in a later
and temporally distinct hatching period for these larvae.

Because this study clearly identified the period of larval outmigration from
tributaries, it is reasonable to conclude that recruited juveniles with hatch dates
after the end of tributary outmigration originated from lake spawned larvae. It is
less clear whether juveniles with hatch dates that correspond to periods of larval
outmigration originated from tributary or lake spawned fish. However, given
temperatures in St. Martin Bay and temperature-incubation duration information
from the literature (Hoover 1936; McKenzie 1964; Cooper 1978), it was possible
to estimate an approximate spawning period needed by lake spawning fish to have
larval hatching temporally overlap with hatching of tributary spawned larvae: For
_ lake spawned larvae to hatch by late May (the end of tributary outmigration),

spawning would have to occur in early to mid April when water temperatures

 

162

were less than 2 C. Not only has spawning never been documented at
temperatures this low, temperatures of 2 C have been shown to cause high
mortality of rainbow smelt eggs under laboratory conditions (McKenzie 1964).
Therefore, I conclude that juveniles with hatch dates corresponding with periods
of tributary outmigration originate from tributary spawned larvae, while juveniles
with batch dates after the end of tributary outmigration represent lake spawned
larvae.

r i n P In

The strongest year class produced during this study occurred in 1991, when

the August/ September bottom trawl CPUE of age 0+ juveniles was 25.0
juveniles/tow. Tributary larval production in 1991 was the highest recorded in
this study (46 million larvae, Chapter 2), and hatch date distributions indicated
that the majority (approximately 88%) of recruited juveniles were recruited from
tributary spawned larvae. A weak year class was recruited in 1992 when the
bottom trawl CPUE of age 0+ juveniles was only 2.9 juveniles/ tow. Tributary
larval production in 1992 was the lowest of the three years of the study (13.2
million larvae, Chapter 2), and the majority (approximately 65 %) of recruited
juveniles were recruited from lake spawned larvae. A moderate year class was
recruited in 1993 when the bottom trawl CPUE of age 0+ juveniles was 10.9
juveniles/tow. Tributary larval production in 1993 was at intermediate levels
’ (31.7 million larvae, Chapter 2), and significant contributions of both tributary and

stream spawned larvae comprised the 1993 year class.

 

163

Based on these observations it appears that, although lake spawned larvae
contribute significantly to recruitment, variation in recruitment in the St. Martin
Bay system is primarily controlled by the contribution of tributary spawned larvae.
The relatively strong 1991 year class was produced as a result of high tributary
production of larvae, high growth rates and low mortality rates of this early
cohort. In 1992, low tributary production of larvae, low growth rates and high
mortality of early hatching larvae resulted in poor recruitment of this cohort. .
Larvae hatching after the end of tributary outmigration (assumed to be lake
spawned fish) had relatively high growth and low mortality rates. Although late
hatching larvae represented a significant portion of recruited juveniles relative to
early hatching larvae, the overall magnitude of recruitment was low compared to
1991. During the three years of this study, late hatching cohorts made a relatively
minor contribution to overall levels of recruitment.

It is important to recognize that the St. Martin Bay area has a relatively
large number of tributary areas suitable for spawning by rainbow smelt. In areas
where suitable tributaries are less prevalent (central and southwest Michigan
coasts) and in onshore areas, the relative contribution of lake spawned larvae to

recruitment may be significantly greater than was estimated for St. Martin Bay.

 

APPENDICES

164

Appendix 1. Age-length key for spawning rainbow smelt sampled from four tributaries to St.
Martin Bay. Lake Huron in 1991.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

   
    
 

 

 

 

 

 

Length Interval

MlnlmumMaXimum
9O - 94 0 - - - - - -
ll 95 - 99 0 - - - - - -
II 100 -104 g o 100 o o o 0
u 105 -109 1 0 100 0 0 0 0
ll 110 - 114 6 0 100 O 0 0 0
II 115 -119 8 O 100 0 0 O 0
I 20 -124 17 O 100 O O O 0
125 -129 25 0 100 0 0 0 O
I 130 -134 44 0 98 2 O 0 0
135 -139 55 0 84 16 0 0 0
140 -144 57 O 37 63 O O 0
145 -149 52 0 16 84 0 0 0
150 -154 40 0 3 97 0 0 0
155 -159 27 O 0 100 0 0 O
160 -164 go 0 0 100 0 0 O
165 - 169 16 0 0 100 0 O 0
170 - 174 13 0 0 6g 38 0 0
175 - 179 8 O 0 63 37 0 0
180 - 184 7 0 0 71 gs 0 O
185 - 189 8 0 0 13 87 O 0
l 190 - 194 3 O O 0 100 O O
3 O 0 O 67 33 0
g 0 0 0 50 50 0
1 O 0 O 0 100 O
210 - 214 i 0 0 O O 100 0
215 - 219 2 - - - - - -
220 - 224 0 - - - - - -
225 - 229 0 - - - - - -

 

 

 

 

 

 

 

165

Appendix 2. Age-length key for spawning rainbow smelt sampled from iour tributaries to St.
Martin Bay, Lake Huron in 1992.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Length Interval ,
’ Minimum-Maximum
90 - 94 O - - - - - -
I 95 - 99 2 0 100 0 0 0 0
100 -104 g o 100 o o o o I
105 -109 12 0 100 0 O O 0
110 - 114 15 0 100 O 0 O O
115 -119 21 0 100 0 0 0 0
120 -124 30 0 100 0 0 0 0
125 -129 55 0 100 0 0 0 0
130 -134 63 0 100 0 0 0 0
135 -139 62 O 100 0 0 O 0
140 -144 66 0 85 15 O 0 0
145 -149 57 0 32 68 0 0 O
150 -154 43 0 0 100 0 0 0
155 -159 27 0 0 100 0 0 O I
160 -164 2_g 0 0 100 O 0 0 I
165 - 169 0 0 100 0 0 0 l
170 - 174 15 0 O 100 0 0 0 II
175 -179 15 0 0 100 0 O O I
180 -184 10 0 0 100 0 0 0
185 -189 8 O 0 100 0 0 O
190 -194 11 O 0 36 64 O 0
195 -199 0 - - - - - - I
200 - 204 4 0 0 0 50 50 0
205 - 209 1 0 O 0 0 100 0 I
210 - 214 3 0 0 0 0 100 0 I
215 - 219 0 - - - - - -
220 - 224 o - - - - - - l

 

166

Appendix 3. Age-length key for spawning rainbow smelt sampled from four tributaries to St.
Martin Bay, Lake Huron in 1993.

 

 
   

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

3 Length Interval
Minimum-Maximum
95- 99 -
100-104 0 100 0 0. 0 0
105-109 12 0 100 0 0 0 0
110-114 12 0 100 o 0 0 0
115-119 22 0 100 0 0 0 0
120-124 33 0 100 0 0 0 0 I
125-129 51 0 100 0 0 0 0 I
130-134 71 0 100 0 0 0 0 I
135-139 95 0 99 1 o 0 o '
140-144 75 0 as 15 0 0 o
145-149 64 0 75 25 o 0 0 E
150-154 41 0 10 90 0 0 o E
155-159 31 0 0 100 0 o 0 E
160-164 g7 0 o 96 4 0 0
165-169 0 0 80 20 0 0 i
1 170-174 16 0 0 2_8____zg 0 0 i
I 175-179 5 0 0 0 100 0 0
I 180-184 a 0 0 0 33 67 0 l
I 185-189 1 0 0 0 0 100 0 E
I 190-194 1 0 0 0 0 100 0 E
195-199 0 - - - - - - E
200-204 _2 o 0 0 0 100 0 E
205-209 0 - - - - - - 3
210-214 0 - - - - - - .
215-219 0 - - - - - - E
220-224 0 - - - - - - E
225-229 1 0 0 0 0 0 100 f
__ .9 - «___- .. ____ .c--. 0 _v__ _ .3- . 04 LE

 

 

 

 

 

I
i
I

 

167

Appendix 4. Length-weight regression equations for male and female rainbow smelt sampled
from four Lake Huron tributaries from 1991 to 1993.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

E Year - Tributary V. _ Sex Sample Size-7 81009 Intercept _E
E 9.: Migrates 203 3 10268 M7225 0 w E
L-199_ ,Ca_.___9'______-__ EM 0- ,r
5: WM 560 306988 -5-.1... : m.
E 1.: WM 326 318790_ - .l- “E
E_1.=...=_____=__=.==.= 886 __12____-_-554 _@l
E as St Martin Jam—343 301161 - :2 ° 1 °_9 . E
E a: 31 Martin _Eemales_ 214 302525‘ - ; “E
L _991________ St. Martin Both __ _3. -5. 35111
E w :.- _Males_ .. ”mm
w m - WWW
l@_ _ =..B=ot.11___ ~9____-__, __ _I-l
' 4:. Mm: 305 3122721“
E ‘13? mm 137 306781 -. -.1 .3. m
r- _9921 Arms Creek Both 502 3.092;.3_l _ 5.45966 m
E w WM 203 301650 53 - ME
3 w _SpdnoflaeLiamalas 233 304530_ - - 1-.--~ j
MEMJLAAWA 996—211 _- .4
E°¢r__St_Manin__Maias 256 292060_--111--
. 161' Mam—WA 305910 31:.- “E
E Martin __H_________ 500_ _ _______ __3.0_2§_75___ __-_5.3189_7_ -
E ~61 0:1. :1 .. _Malm 315 311659__;5.5m_42_E 1 n... E
E '13.‘ 4.. :. 1 _Eemales 137 I a15295__.5_51234_ 1-.1... E
1&1 Caeu ==Bgt_g=n _-___________. .____2956_____-_____ ___-3:“ l
1 ~94 _NunnsfireeLJalm 9:13 3017 .. - 26:. 1 .1
E -.: mm 1711-.- -1. 1;»- E
1999 M40111 504__ 27 _ -s_977__ _ 01.91 1
‘ 0.3-1 W Males 342 ° - 1.3: 1°11. i

'41 WWJL 4 - 1 1 '11
1 WM___,____.____ ________ _ 300615 '5-27709 _1
' .1. 4 ,. MW“

.1. 4 .4- mum-m.
E1993 St.‘Ma__n_ __oth 505 3_.0_0§1_2 E

 

 

 

LIST OF REFERENCES

LIST OF REFERENCES

Anderson, ED. and LL. Smith, Jr. 1971. Factors affecting abundance of lake
herring (Coregonus artedii Iesueur) in western Lake Superior. Transactions
of the American Fisheries Society 100:691-707.

Argyle, R.L. 1982. Alewives and rainbow smelt in Lake Huron: midwater and
bottom aggregations and estimates of standing stocks. Transactions of the
American Fisheries Society 111:267-285.

Argyle, R.L. 1994. Status of prey fishes in 1992, p. 59-67. In M.P. Ebener, [ed.]
The state of the Lake Huron fish community in 1992. Great Lakes Fishery
Commission Special Publication, 93-?, Ann Arbor, MI. In Press.

Armstrong, ML. and AV. Brown. 1983. Diel drift and feeding of channel catfish
alevins in the Illinois River, Arkansas. Transactions of the American
Fisheries Society 112:302-307.

Assel, R.A., F.H. Quinn, G.A. Leshkevich, and SJ. Bolsenga. 1983. Great Lakes
Ice Atlas. NOAA - Great Lakes Environmental Research Laboratory, Ann
Arbor, MI.

Auer, N.A. (ed). 1982. Identification of larval fishes of the Great Lakes basin
with emphasis on the Lake Michigan drainage. Great Lakes Fishery
Commission Special Publication 82-3:744p.

Bailey, M.M. 1964. Age, growth, maturity, and sex composition of American
smelt, Osmerus mordax (Mitchill), of western Lake Superior. Transactions
of the American Fisheries Society 93:382-395.

Balcer, M.D., N.L. Korda, and 8.1.. Dodson. 1984. Zooplankton of the Great
Lakes: a guide to the identification and ecology of the common crustacean
species. The University of Wisconsin Press, Madison, WI. 174 p.

Baldwin, N.S., R.W. Saalfield, MA. Ross and HJ. Buettner. 1979. Commercial
fish production in the Great Lakes 1867-1977. Great Lakes Fishery
Commission Technical Report No. 3.

168

169

Beckman, W.C. 1942. Length-weight relationship, age, sex ratio and food habits
of the smelt (Osmerus mordax) from Crystal Lake, Benzie County,
Michigan. Copeia 2:120-124.

Brandt, SB. and SP. Madon. 1986. Rainbow smelt (Osmerus mordax) predation
on slimy sculpin (Cottus cognatus) in Lake Ontario. Journal of Great
Lakes Research 12(4):322-325.

Burbidge, RC. 1969. Age, growth, length-weight relationship, sex ratio, and food
habits of American smelt, Osmerus mordax (Mitchill), from Gull Lake,
Michigan. Transactions of the American Fisheries Society 98:631-640.

Butler, J.l. 1992. Collection and preservation of material for otolith analysis,
pages 13-17. In D.K. Stevenson and S.E. Campana [ed.] Otolith
microstructure examination and analysis. Canadian Special Publication of
Fisheries and Aquatic Sciences 117.

Brown, R.W. and W.W. Taylor. 1992. Effects of egg composition and prey
density on the larval growth and survival of lake whitefish, Coregonus
clupeaformis (Mitchill). Journal of Fish Biology 40:381-394.

Campana, SE. and CM. Jones. 1992. Analysis of otolith microstructure data,
pages 73-100. In D.K. Stevenson and SE. Campana [ed.] Otolith
microstructure examination and analysis. Canadian Special Publication of
Fisheries and Aquatic Sciences 117.

Clifford, H.F. 1972. Downstream movements of white sucker, Catostomus

commemoni, fry in a brown-water stream of Alberta. Journal of the
Fisheries Research Board of Canada 29:1091-1093.

Cooper, J.E. 1978. Identification of eggs, larvae, and juveniles of the rainbow
smelt, Osmerus mordax, with comparisons to larval alewife, Alosa
psuedoharengus, and gizzard shad, Dorosoma cepedianum. Transactions of
the American Fisheries Society 107:56-62.

Crecco, VA. and T.F. Savoy. 1985. Effects of biotic and abiotic factors on
growth and relative survival of young American shad, Alosa sapidim’ma) in
the Connecticut River with changes in year-class strength. Canadian
Journal of Fisheries and Aquatic Sciences 40:1719-1728.

Crowder, LB. 1980. Alewife, rainbow smelt and native fishes in Lake Michigan:
competition or predation? Environmental Biology of Fishes 5:225-233.

170

DeAngelis, D.L., B.J. Shuter, S.M. Ridgway, and M. Scheffer. 1993. Modeling
growth and survival in an age-0 fish cohort. Transactions of the American
Fisheries Society 122:927-941.

Dunstall, T. M. 1984. Distribution of rainbow smelt and alewife larvae along the

north shore of Lake Ontario. Journal of Great Lakes Research 10:273-279.

Eck, G.W. and L. Wells. 1983. Biology, population structure, and estimated
forage requirements of lake trout in Lake Michigan. U. S. Fish and
Wildlife Service Technical Paper 111.

Emery, AR. 1973. Preliminary comparison of day and night habitats of
freshwater fish in Ontario Lakes. Journal of the Fisheries Research Board
of Canada 30:761-774.

Epperly, S.P., D.W. Ahrenholz and RA. Tester. 1991. A univeral method for
preparing, sectioning and polishing fish otoliths for daily ageing. U.S.
Dept. Commerce. N OAA Technical Memorandum N MFS-SEFSC-283,
15p.

Essig, RJ. and CF. Cole. 1986. Methods of estimating larval fish mortality from

daily increments in otoliths. Transactions of the American Fisheries
Society 115:34-40.

Evans, D.O. and DH. Loftus. 1987. Colonization of inland lakes in the Great
Lakes region by rainbow smelt, Osmerus mordax: Their freshwater nich and
effects on indigenous fishes. Canadian Journal of Fisheries and Aquatic
Science 44(Suppl. 2):249-266.

Ferguson, KG. 1965. Bathymetric distribution of American smelt Osmerus
mordax in Lake Erie. Proceedings of the Conference on Great Lakes
Research 13:47-60.

Frie, RV. and GR. Spangler. 1985. Dynamics of rainbow smelt during and after
exploitation in South Bay, Lake Huron. Transactions of the American
Fisheries Society 114:713-724.

Fritz, E.S., LB. Crowder, and RC. Francis. 1990. The National Oceanic and
Atmospheric Administration Plan for Recruitment Fisheries Oceanography
Research. Fisheries 15:25-31.

Foltz, J.W. and CR. Norden. 1977. Food habits and feeding chronology of
rainbow smelt, Osmems mordax, in Lake Michigan. Fisheries Bulletin
75:637-640.

 

 

 

171

Gale, W.F. and H.W. Mohr, Jr. 1978. Larval fish drift in a large river with a
comparison of sampling methods. Transactions of the American Fisheries
Society 107:46-55.

Gebhardt, KJ. 1993. Population dynamics and stock identification of rainbow
smelt in Green Bay and the adjacent waters of Lake Michigan. Masters
Thesis, University of Wisconsin-Stevens Point.

Geffen, AJ. 1992. Validation of otolith increment deposition rate, pages
101-113. In D.K. Stevenson and SE. Campana [ed.] Otolith
microstructure examination and analysis. Canadian Special Publication of
Fisheries and Aquatic Sciences 117.

Gordon, W.G. 1957. Biology of the American smelt, Osmerus mordax (Mitchill)
of Saginaw Bay, Lake Huron. Masters Thesis. University of Michigan.

Gordon, W.G. 1961. Food of the American smelt in Saginaw Bay, Lake Huron.
Transactions of the American Fisheries Society 90:439-443.

Hackney, PA. and J .C. Webb. 1978. A method for determining growth and
mortality rates of ichthyoplankton, p. 115-124. In L.D. Jensen, [ed.]
Fourth National Workshop on Entrainment and Impingement. Ecological
Analysts Communications, Melville, New York.

Hale, J. 1960. Some aspects of the life history of the smelt (Osmerus mordax) in

western Lake Superior. Minnesota Fish and Game Investigations, Fish
Series 2:25-41.

Hatch, R.W., P.M. Haack, and EH. Brown, Jr. 1981. Estimation of alewife
biomass in Lake Michigan, 1967-178. Transactions of the American
Fisheries Society 110:575-584.

Henderson, BA. and SJ. Nepszy. 1989. Factors affecting recruitment and
mortality rates of rainbow smelt (Osmerus mordax) in Lake Erie, 1963-
1985. Journal of Great Lakes Research 15:357-366.

Hoenig, J .M., P. Pepin, and W.D. Lawing. 1990. Estimating relative survival rate
for two groups of larval fishes from field data: do older larvae survive
better than young? Fishery Bulletin (U.S.) 88:485-490.

Hoover, E. 1936. The spawning activity of freshwater smelt, with special
reference to sex ratio. Copeia 1936:85-91.

172

Houde, ED. 1987. Fish early life dynamics and recruitment variability.
American Fisheries Society Symposium 2:17-29.

Iguchi, K. and N. Mizuno. 1990. Diel changes of larval drift among
amphidromous gobies in Japan, especially Rhinogobius brunneus. Journal
of Fish Biology 37:255-264.

Jilek, R., B. Cassell, D. Peasce, Y. Garza, L.Riley and T. Siewart. 1979.
Spawning population dynamics of smelt, Osmerus mordax. Journal of Fish
Biology 15:31-35.

Johnson, J.E., G.M. Wright, D.M. Reid, and CA. Bowen. 1994. Status of the
salmonine fish community in 1992, p. 9-35. In M.P. Ebener, [ed.] The
state of the Lake Huron fish community in 1992. Great Lakes Fishery
Commission Special Publication 93-?, Ann Arbor, MI.

Johnston, CE. and J .C. Cheverie. 1988. Observations of the diel and seasonal
drift of eggs and larvae of anadromous rainbow smelt, Osmerus mordax,

and blueback herring, Alosa aestivalis, in a coastal stream on Prince
Edward Island. Canadian Field-Naturalist 102(3):508-514.

Jude, D.J., G.R. Heufelder, H.T. Tin, N.A. Auer, S.A. Klinger, P.J. Schneeberger,
T.L. Rutecki, C.P. Madenjian, and PJ. Rago. 1980. Adult, juvenile and
larval fish populations in the vicinity of the J .H. Campbell Power Plant,
eastern Lake Michigan, 1979. University of Michigan, Great lakes
Research Division, Special Report 79, Ann Arbor, Michigan, USA.

Kinnunen, RE. 1994. Michigan 1993 Commercial Fish Harvest. Commercial
Fisheries N ewsline. Michigan State University Cooperative Extension
Service 13(2):3-5.

Legault, R. and C. Delisle. 1968. The spawning of a population of giant smelt,
Osmems eperlanus mordax, in Lake Heney, Gatineau County, Quebec.
Journal of the Fisheries Research Board of Canada 25:1813-1830.

Lievense, SJ. 1954. Spawning of American smelt, Omerus mordax, in Crystal
Lake, Benzie County, Michigan. Copeia 1954(3):232-233.

Loftus, DH. and RF. Hulsman. 1986. Predation on larval lake whitefish
(Coregonus clupeaformis) and lake herring (Coregonus an‘edii) by adult
rainbow smelt (Osmems mordax). Can. J. Fish. Aquatic Sci. 43:812-818.

173

Luey, J .E. and LR. Adelman. 1984. Stock structure of rainbow smelt in western
Lake Superior: population characteristics. Transactions of the American
Fisheries Society 113:709-715.

Maceina, M.J., P.W. Bettoli, and DR. DeVries. 1994. Use of a split-plot analysis
of variance design for repeated-measures fishery data. Fisheries 19(3):14—
20.

Mchllough, RD. and J.G. Stanley. 1981. Feeding niche dimensions in larval
rainbow smelt (Osmems mordax). Rapports et Proces-Verbaux des
Reunions, Conseil International pou l’Exploration de la Mer 178:352-354.

Murawski, S.A. and CF. Cole. 1978. Population dynamics of anadromous
rainbow smelt, Osmerus mordax, in a Massachusetts river system.
Transactions of the American Fisheries Society 107:535-542.

McKenzie, RA. 1964. Smelt life history and fishery of the Miramichi River, New
Brunswick. Fisheries Research Board of Canada Bulletin 144:77p.

Miller, T.J., LB. Crowder, J .H. Rice, and EA. Marshall. 1988. larval size and
recruitment mechanisms in fishes: toward a conceptual framework.
Canadian Journal of Fisheries and Aquatic Sciences 45:1657-1670.

Nepszy, SJ ., J. Budd, and AD. Dechtiar. 1978. Mortality of young-of-the-year
rainbow smelt (Osmerus mordax) in Lake Erie associated with the
occurrence of Glugea hertwigii. Journal of Wildlife Diseases 14:233-239.

Nester, RT. 1987. Horizontal ichthyoplankton tow-net system with unobstructed
net opening. North American Journal of Fisheries Management 7:148-150.

Nsembukya-Katuramu, S., E.K. Balon and R. Mahon. 1981. A comparison of
spawning, harvested, and die-off rainbow smelt, Osmerus mordax, in eastern
Lake Erie. Journal of Great Lakes Research 7: 144-154.

O’Gorman, R. 1974. Predation by rainbow smelt (Osmerus mordwr) on young-of
-the-year alewives (Alosa psuedoharengus) in the Great Lakes. Progressive
Fish-erturist 36:223-224.

O’Gorman, R. 1984. Catches of larval rainbow smelt (Osmems mordax) and
alewife (Alosa pseudoharengus) in plankton nets of different mesh sizes.
Journal of Great Lakes Research 10(1):?3-77.

174

Ouellet, P. and J .J . Dodson. 1985. Tidal exchange of anadromous rainbow smelt
(Osmems mordax) larvae between a shallow spawning tributary and the St.
Lawrence estuary. Canadian Journal of Fisheries and Aquatic Sciences
42:1352-1358.

Plosila, BS. 1984. Spatial distribution of rainbow smelt spawning in the New
York waters of Lake Champlain. New York Fish and Game Journal
31:109-110.

Price, J.W. 1963. A study of the food habits of some Lake Erie fish. Bulletin of
the Ohio Biological Survey, New Series 2(1):1-89.

Raab, R.L and D.N. Steinnes. 1979. The economics of smelting as a recreational
activity. Minnesota Sea Grant Publication 04-8-M01-26, Final Report,
Duluth, Minnesota.

Regier, H.A., V.C. Applegate, and RA. Ryder. 1969. The ecology and
management of walleye in western Lake Erie. Great Lakes Fishery
Commission Technical Report 15.

Reisen, W.K. 1972. The influence of organic drift on the food habits and life
history of the yellowfin shiner, Notropis lutipinnis (Jordan and Brayton).
American Midland Naturalist 88:376-383.

Ricker, W.B. 1975. Computation and interpretation of biological statistics of fish
populations. Bulletin of the Fisheries Research Board of Canada 191:382p.

Roberts, BC. and R.G. White. 1992. Effects of angler wading on survival of
trout eggs and pre-emergent fry. North American Journal of Fisheries
Management 12:450-459.

Robinson, RD. 1973. Age, growth, and sex composition of the American smelt,
Osmerus mordax (Mitchill), from along the western shore of Lake
Michigan. Master’s thesis. University of Wisconsin, Stevens Point.

Rothschild, BJ. 1961. Production and survival of egg of the American smelt,
Osmerus mordax (Mitchill), in Maine. Transactions of the American
Fisheries Society 90:42-48.

Rupp, RS. 1965. Shore-spawning and survival of eggs of the American smelt.
Transactions of the American Fisheries Society 94:160-168.

Scott, W.B. and E. J. Crossman. 1985. Freshwater Fishes of Canada. Fisheries
Research Board of Canada, Bulletin 184:966p.

175

Secor, D.H., J.M. Dean, and EH. Laban. 1991. Manual for otolith removal
and preparation for microstructural examination. Electrical Power
Research Institute, 85p.

Secor, D.H., J .M. Dean, and EH. Laban. 1992. Otolith removal and preparation
for microstuctural examination, p. 19-57. In D.K Stevenson and SE.
Campana [ed.] Otolith microstructure examination and analysis. Canadian
Special Publication of Fisheries and Aquatic Sciences 117.

Schaefer, W.F. 1979. Population dynamics of rainbow smelt in western Lake
Superior. Doctoral Dissertation. Brigham Young University. Provo, Utah.

Schaefer, W.F., R.A. Heckrnann, and WA. Swenson. 1982. Post spawning
mortality of rainbow smelt (Osmerus mordax) in western Lake Superior,
USA. Journal of Great Lakes Research 7:37-41.

Siege], S. 1956. Non-parametric statistics for the behavioral sciences. McGraw-
Hill, New York.

Selgeby, J.H., W.R. MacCallum and D.V. Swedberg. 1978. Predation by rainbow
smelt (Osmerus mordax) on lake herring (Coregonus artedii) in western
Lake Superior. Journal of the Fisheries Research Board of Canada
35:1457-1463.

Snedecor, G.W. and W.G. Cochran. 1980. Statistical methods, 7th edition. Iowa
University Press, Ames, IA.

Stedman, RM. and R.L Argyle. 1985. Rainbow smelt (Osmerus mordax) as

predators of young bloaters (Coregonus hoyi) in Lake Michigan. Journal of
Great lakes Research 11(1):40-42.

Steel, R.G.D. and J .H. Torrie. 1980. Principals and procedures in statistics, 2nd
edition. McGraw-Hill, New York.

Stewart, DJ ., J .F. Kitchell and LB. Crowder. 1981. Forage fishes and their
salmonid predators in Lake Michigan. Transactions of the American
Fisheries Society 110(6):751-763.

Strauss, RE. 1979. Reliability estimates for Ivlev’s electivity index, the forage

ratio, and a proposed linear index of food selection. Transactions of the
American Fisheries Society 108:344-352.

176

Strauss, RE. 1982. Influence of replicated subsamples and subsample
heterogeneity on the linear index of food selection. Transactions of the
American Fisheries Society 111:517-522.

Tin, H.T. and DJ. Jude. 1983. Distribution and growth of larval rainbow smelt
in Eastern Lake Michigan, 1978-1981. Transactions of the American
Fisheries Society 112:517—524.

Torke, B.G. 1974. An illustrated guide to the identification of the planktonic
crustacea of Lake Michigan with notes on their ecology. Center for Great
Lakes Studies, University of Wisconsin-Milwaukee, Special Report 17.

Tsukomoto, KH., H. Kuwadad, J. Hirokawa, M.Oya, S. Sekiya, H. Fujimoto, and
K. Imaizumi. 1989. Size dependent mortality of red sea bream, Pagrus
major, juveniles released with fluorescent otolith-tags in News Bay, Japan.
Journal of Fish Biology 35(Suppl. A):59-69.

 

Van Oosten, J. 1937. Dispersal of smelt, Osmerus mordax (Mitchell), in Great
Lakes region. Transactions of the American Fisheries Society 66:160-171.

Van Oosten, J. 1947. Mortality of smelt, Osmerus mordax (Mitchell), in Lakes

Huron and Michigan during the fall and winter of 1942-43. Transactions of
the American Fisheries Society 74:310-337.

Wilson, C.A., E.B. Brothers, J.M. Casselman, C.L Smith, and A. Wild. 1983.
Glossary, p. 207-208. In E.D. Prince and L.M. Pulos [ed.]. Proceedings of
the international workshop on age determination of oceanic pelagic fishes:
tunas, billfishes, and sharks. NOAA, NMFS Technical Report 8.

Winnell, M.H. and DJ. Jude. 1991. Northern large-river benthic and larval fish
drift: St. Marys River, USA/ Canada. Journal of Great Lakes Research
17:168-182.

Zigler, SJ. and CA. Jennings. 1993. Growth and mortality of larval sunfish in
backwaters of the upper Mississippi River. Transactions of the American
Fisheries Society 122:1080-1087.