f" ' mum u
3.. .J-
u. an. ”5'. ‘
. m. .
M”...
(«In

.w '
mul- .7"-..
.u mgr... .

hr“ .u:

x
‘v

v
“awn",

.4...&"-..4

.mv.

2
’r C
‘ U1?
ﬁring?

3‘! .
5:93.
'u‘t

 

, rrmvuz "'-

wl. up” _.

r 'u‘
.u,

, .
.n a;
(j. ‘

. . . . 523%};

x ’r * Wig-.- 55$ -‘

1
$5?

- X I”

I; "if.
'_ - I

21- ‘

3 x J ;
i u

I C I :
,3}! :IﬁI-I".
' "wig? -
3’. 331'
' 36 “' 1

r'v-<
. , ,,,
a
“w;- '
ruin
Er.

£214 .
..:J
3'3}. ‘
4‘".

I.»
u a. 2

wow
Nw-ml>,1y .

J":

*u.
I

"no-um...-

WA‘M

0.”

. w...-V
in... many.

w}.
.
.1
inn-l"
.. .3221?

T."31‘.‘.-
’1?

7.. -...

2K
"1‘ ‘
~.-.

.00—

 

. ,3...“
3'3:-
“542-“
‘2'.”
t,
. l
m ' *
“—45%
“as.

,w u
p

‘P

«-
..

Ilillllllllllllllilllllllllll’lllllll

31293 0141106 2 1

 

This is to certify that the
thesis entitled
Behavioral, Physiological and Morphological Similarity
Among Populations of Redside Dace, a
Threatened Species in Michigan

presented by

Douglas Craig Novinger

has been accepted towards fulﬁllment
of the requirements for

Master of Science degree inFish. & Wildl.

 

 

    

Major professor

Date Jul) 21, 1995

 

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

 

 

LlBRARY
v‘lichigan State
University

 

l
l
I

 

 

PLACE ll RETURN BOX» romavothb checkoutnom you record.

TO AVOID FINES Mom on or baton dot. duo.

    

DATE DUE DATE DUE DATE DUE

 
 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

MSU I. An mum Adan/Equal Opportunity lnotltwon
W

m1

 

BEHAVIORAL, PHYSIOLOGICAL AND MORPHOLOGICAL SIMILARITY
AMONG POPULATIONS OF REDSIDE DACE, A THREATENED SPECIES IN
MICHIGAN
By

Douglas Craig Novinger

A THESIS

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

MASTER OF SCIENCE

Department of Fisheries and Wildlife

ABSTRACT
BEHAVIORAL, PHYSIOLOGICAL AND MORPHOLOGICAL SIMILARITY
AMONG POPULATIONS 0F REDSIDE DACE, A THREATENED SPECIES IN
MICHIGAN
By

Douglas Craig Novinger

I studied variation in behavior through observations of microhabitat use at sites in
Michigan and New York. Redside dace occupied low velocity, mid-water positions in the
deepest parts of pools. Differences among populations in habitat use corresponded with
differences in habitat availability. I studied variation in physiological traits by measuring
resting routine metabolic rate and critical thermal maximum for four temperature
acclimated New York populations in the lab. Metabolic rate was also measured for
naturally acclimated ﬁsh in the ﬁeld in New York and Michigan. New York redside dace
showed signiﬁcant increases in metabolism and CTM with temperature. Michigan redside
dace tested in the ﬁeld had a signiﬁcantly higher metabolic rate than that predicted for
New York ﬁsh at the same temperature. Morphological variation was studied through a
morphometric and meristic analysis and revealed that Michigan and Wisconsin specimens

were most diﬁ‘erent in body shape and lateral line scale counts.

To my loving wife Sarah, for your unending patience and strength.

iii

ACKNOWLEDGMENTS

I owe the successful completion of this project to the dedicated eﬁ‘ort and
ﬁ-iendship of several individuals. I offer my deepest appreciation to my advisor, Thomas
Coon, for his creativity, insight, and patience, and for his trust in my abilities. Thank you
for your dependable advice and friendship. I also thank the members of my guidance
committee, Thomas Burton and Thomas Getty, for valuable suggestions and discussions
that improved this thesis.

My special thanks to Darryl Hondorp for his ﬁ'iendship and for his reliable
assistance in the ﬁeld and in the lab. I have also beneﬁtted ﬁ'om useﬁil discussions and
technical assistance provided by my Coon lab buddies Edward Baker, Tammy Newcomb,
and John Skubinna and by several other students in the department including Scott Miller,
Shawn and Kristie Sitar, and Ted Sledge. Thank you for memorable moments and lasting
friendships.

I have always been motivated by the steadfast love and support of my wife and
family. This thank you is especially for you Dad, the one who really started me on this
trail.

Funding for this project was provided by the Department of Fisheries and Wildlife

and the Ecology and Evolutionary Biology Program at Michigan State University.

iv

TABLE OF CONTENTS

LIST OF TABLES
LIST OF FIGURES

INTRODUCTION
Background
Objectives .

METHODS . .
Study Area
Behavior . .
Physiology. . . . . . .
Fish Capture and Maintenance
Metabolic Rate
Thermal Tolerance
Morphology
Morphometrics
Meristics .

RESULTS. .

Behavior . .
Habitat Use . .
General Behavior .

Physiology .
Metabolic Rate .
Thermal Tolerance

Morphology
Morphometrics
Meristics .

DISCUSSION
Behavior
Physiology .
Metabolic Rate .
Thermal Tolerance
Morphology

viii

~

11
ll
15
19
19
19
24
26
26
32

34
34
34
46
48
48
51
54
S4
68

73
73
76
76
82
86

CONCLUSIONS .

APPENDIX

BIBLIOGRAPHY .

TABLE OF CONTENTS (cont’d)

91

95

97

LIST OF TABLES

Table Page
1 Description of the 17 landmark points used in the morphometric analysis. . 29
2 Range in stream water quality characteristics of Seeley Drain, Michigan

and Rice and France Brooks, New York, during July, 1994. Sample sizes

aregiveninparentheses. . . . . . . . . . . . 35

3 Fish species co-occurring with redside dace in Michigan and New York.
NamesfollowRobinsetal. (1991). . . . . . . . . . . . 38

4 Sample sizes and standard lengths of redside and rosyside dace morphometric
specimens...... .... ...57

5 Frequency distribution of dorsal ﬁn ray counts for redside dace morphometric
specimens from ﬁve locations. . . . . . . . . . . . . 69

6 Frequency distribution of anal ﬁn ray counts for redside dace morphometric
specimensfromﬁvelocations. . . . . . . . . . . . . 7O

vii

Figure

10

LIST OF FIGURES

Page

A map illustrating the Seeley Drain ﬁeld site located in the Upper River
Rouge watershed of southeastern Michigan. . . . . . . . . . 12

A map illustrating France, Red House, Whig Street, and Rice Brook ﬁeld
sites located in the Allegheny River watershed of southwestern New York. . 14

Diagram of the static respirometer used to collect samples for estimation
of adult redside dace oxygen consumption rates. . . . . . . . . 21

Locations of the 17 landmark points that were digitized around the body

of each redside and rosyside dace museum specimen for morphometric
analysis. Refer to Table 1 for a description of each point. . . . . . 28

Stream bottom substrate percent frequency of availability and use by adult
redside dace in Michigan and New York. Electivity (L) is given above the

bars of each category and sample sizes are given in parentheses. . . . . 39

Instream cover type percent frequency of availability and use by adult redside
dace in Michigan and New York. Symbol explanation follows Figure 5. . 40

Vertical stratum in the water column occupied by adult redside dace in
Michigan and New York. Symbol explanation follows Figure 5. . . . 42

Depth availability and use by adult redside dace in Michigan and New York.
Symbols: vertical bar represents median, box encloses 50% of the distribution,
and horizontal bars mark the range. Sample sizes are given in parentheses. . 43

Mean column velocity availability and use by adult redside dace in Michigan
and New York. Symbol explanation follows Figure 8. . . . . . . 44

Mean oxygen consumption rates (:t 1 SE) by adult redside dace from four
New York streams during trial periods used in the ﬁrst laboratory experiment.
Sample sizes are given in parentheses. . . . . . . . . . 49

viii

11

12

13

14

15

16

17

18

Mean oxygen consumption rates by adult redside dace from New York and
Michigan during 50 min lab and ﬁeld trials in the ﬁrst experiment. The
temperature that each group was tested at and sample sizes (in parentheses)

are given next to the mean consumption rate. Symbols: horizontal bar

represents mean, box encloses 2 SE, and vertical bars mark range. . . . 50

Mean oxygen consumption rates of adult redside dace from New York at

three laboratory acclimation temperatures in the second experiment.

Laboratory data from the ﬁrst experiment is provided for comparison.

Symbol expanation follows Figure 11. . . . . . . . . . . 52

Mean critical thermal maximum of adult redside dace from New York at three
laboratory acclimation temperatures. Symbol explanation follows Figure 11. 53

Sequence of behaviors observed during critical thermal maximum tests of

adult redside dace from New York at three laboratory acclimation

temperatures (means :1: 28E). OPS= opercular spasms, FLE= ﬁnal loss Of
equilibrium, SUR= surfacing, ESW= eratic swimming, ESW= burst

swimming. See Methods- Physiology -Ihermal tolerance for a description
Ofeachbehavior.................55

Outlines of mean shape conﬁgurations for redside dace ﬁom four regions and
rosyside dace. Outlines have been scaled and superimposed on a baseline
between points 1 and 12. Letters marking interior points represent Michigan
(M), New York (N), 'Wrsconsn- Black River (B), Wisconsin- Wisconsin
River(W),androsysidedace(R).. . . . . . . . . . . 58

Scatter-plot of the uniform factor x- and y-dimensions of the afﬁne
transformation from the mean rosyside dace starting conﬁguration to

individual redside dace shape conﬁgurations. Symbols: (El) rosyside dace at
(0,0); (0) Michigan redside dace; (x) New York redside dace; (0) Wisconsin -
Black River redside dace; (A) Wisconsin - \Ifrsconsin River redside dace. . 60

Cartesian mappings of the thin-plate spline transformation of the rosyside dace
starting conﬁguration to redside dace mean shape conﬁgurations. The

mappings depict the combined effects of the afﬁne and non-unifonn

components of shape change. A) The transformation from the rosyside dace
conﬁguration to the Michigan redside dace conﬁguraton. B) The

transformation ﬁom the rosyside dace conﬁguration to the New York redside
daceconﬁguration................62

Cartesian mappings of the thin-plate spline transformation of the rosyside dace

starting conﬁguration to redside dace mean shape conﬁgurations. The
mappings depict the combined effects of the affine and non-uniform

ix

19

20

21

22

23

24

25

components of shape change. A) The transformation from the rosyside dace
conﬁguration to the \Vrsconsin - Black River redside dace conﬁguraton. B)

The transformation ﬁ'om the rosyside dace conﬁguration to the Wisconsin -
\Vrsconsin River redside dace conﬁguration. . . . . . . . . . 63

Distribution of loadings on the x— and y-dimensions of partial warp 14 for the
thin-plate spline transformation from the mean rosyside dace starting
conﬁguration to individual redside dace shape conﬁgurations. Symbols: ([1)
rosyside dace at (0, 0); (0) Michigan redside dace; (x) New York redside dace;

(0) Wisconsin- Black River redside dace; (A) VVrsconsin- Wisconsin River
redsidedace......... ........64

Distribution of loadings on the x- and y-dimensions of partial warp 11 for the
thin-plate spline transformation from the mean rosyside dace starting
conﬁguration to individual redside dace shape conﬁgurations. Symbols: ([1)
rosyside dace at (0,0); (0) Michigan redside dace; (x) New York redside dace;
(0) Wisconsin - Black River redside dace; (A) Wisconsin - Wisconsin River
redsidedace..................66

Distribution of loadings on the x- and y-dimensions of partial warp 4 for the
thin-plate spline transformation from the mean rosyside dace starting
conﬁguration to individual redside dace shape conﬁgurations. Symbols: (1])
rosyside dace at (0,0); (0) Michigan redside dace; (x) New York redside dace;

(0) Wisconsin - Black River redside dace; (A) Wisconsin - Wisconsin River
redsidedace..................67

Frequency disribution of lateral line scale counts for redside dace
morphometric specimens ﬁom ﬁve regions. Sample sizes are given in
parentheses. . . . . . . . . . . . . . . . . . 72

Oxygen consumption rates (mg g'1 h") associated with resting metabolism
forseveralcyprinidﬁshes.. . . . . . . . . . . . . . 78

Redside dace mean oxygen consumption rates in relation to Krogh’s
“standard curve” (dotted line) adapted ﬁom Winberg (1956). New York data
are indicated by the Open circles and solid line, Michigan by the closed circle. 80

Critical thermal maximum determinations for several cyprinid ﬁshes. . . 84

INTRODUCTION

The redside dace, Clinostomus elongatus, is a threatened minnow species in
Michigan. Successful management of the remnant Michigan populations is impaired by an
inadequate understanding of the ecological requirements of the redside dace and by the
current rarity of the ﬁsh. One solution to this problem is to look to other regions where
redside dace are abundant and more easily observed. However, signiﬁcant heterogeneity
in phenotypic and genetic characters may exist among disjunct populations such as those
of redside dace (Shields and Underhill, 1993; Foster et al., 1992; White, 1988). The
similarity of populations of redside dace in Michigan to geographically separate
populations is not understood. Ifthey are similar, populations of redside dace in other
states may adequately model the habitat needs, reproductive traits, and other attributes of
Michigan ﬁsh. Redside dace from outside of Michigan may also represent future sources
of ﬁsh for stocking into Michigan to reestablish Michigan populations. To assess the
potential of either use, it is necessary to determine the similarity of redside dace in

Michigan to redside dace from other parts of its range.

Background - This project integrates observations of behavioral, physiological, and
morphological similarity and diﬁ‘erences between geographically isolated populations of
redside dace to provide information useful to improving the species’ status in Michigan
and elsewhere. Behavioral, physiological, and morphological traits describe phenotypic

attributes that are critical determinants of an organism’s ecological niche and are at least

2
indirectly related to evolutionary ﬁtness; they are often studied separately ( Foster et al.,

1992; Ehrlich and Roughgarden, 1987; Kerr and Ryder, 1979). Several works suggest
that aspects of the phenotype may evolve together in response to changing environmental
conditions (Foster et al., 1992; Endler, 1986; Fry, 1971). The apparent interrelatedness of
phenotypic traits ﬁrrther suggests that conclusions based on the examination of only one
aspect of phenotypic variation may be misleading, thereby requiring detailed knowledge of
a species’ ecological requirements for the design of successful management policies. In
the same way, approaches that only consider genotypic traits are usually unable to relate
measures of genetic variation to the interactions of an organism with its environment.
Despite their frequent use, molecular techniques have yet to prove their reliability in
phylogenetic analysis over traditional morphological methods (Patterson et al., 1993).
Rather, a genotypic approach to describing ecologically signiﬁcant variation within and
among groups should be complementary to a phenotypic approach.

Fry (1971) provided a framework in which to describe the relationships between
environmental factors and behavioral capacities when he deﬁned the metabolic “scope for
activity” and complementary “zone of tolerance.” The determination of metabolic rates
and thermal tolerances are necessary components of Fry’s models and may be used to
measure the realized or potential ﬁrnctional status of an organism in units that are
observable, common, and important to all organisms in a community (Kerr and Ryder,
1977). Estimation of resting metabolic rates and thermal tolerance have been useful for
the prediction of optimum grth and food consumption rates as well (Jobling, 1983;

Brett and Groves, 1979).

3
Patterns of habitat use have also been described in terms of the bioenergetic

tradeoﬂ‘s between metabolic requirements and the availability of critical resources such as
food and shelter (Hill and Grossman, 1993; Hughes and Dill, 1990). It is generally
understood that habitat use and inter- and intraspeciﬁc behavioral interactions are
important in establishing ﬁsh community structure and species distributions (Grossman
and Freeman, 1987).

Morphology has been linked to the ecology of ﬁshes through intraspeciﬁc variation
in life history and behavior, and has been used to predict patterns of habitat use and
trophic position, and to infer phylogenetic relationships (Shields and Underhill, 1993;
Foster et al., 1992; Reinthal, 1990; Felley, 1984; Gatz, 1979). Morphology has also been
related to physiological performance through measurements of swimming and aerobic
eﬁciency in ﬁshes under a variety of environmental conditions (Brett and Groves, 1979;
Webb, 1971).

The description and measurement of phenotypic attributes provides information
that is readily usable by managers in the design and implementation of resource policies.
Habitat elements including instream structures, channel morphology, and water quality
may be manipulated or regulated to effect a response in aquatic organisms; achieving the
desired result depends on a sound base of ecological knowledge that allows an accurate
prediction of the response of individual species to environmental perturbation. Such
understanding is essential in the case of endangered species which typically have narrow

environmental tolerances and restricted distributions.

4
Redside dace belong to a unique assemblage of glacial relict species that inhabit the

Great Lakes basin and are distributed in disjunct populations characteristic of restricted
headwater stream ﬁshes (White, 1988; Smith et al., 1981). In the western edge of their
range, redside dace are found in southeastern Minnesota and in the upper Mississippi
River and Lake Michigan basins of Wisconsin. East of Lake Michigan, redside dace occur
in southeastern Michigan, eastern Indiana, Ohio, northern Kentucky, West Virginia,
Maryland, and Pennsylvania, and are abundant in south-central New York. Redside dace
are also found in southern Ontario, Canada (Gilbert, 1980). Past geophysical processes
have undoubtedly played a major role in delineating the species’ discontinuous
distribution. Redside dace probably occupied a more continuous northern range that was
disrupted during the Wisconsinan glacial advances as recently as 9 - 14,000 years ago
(Bailey and Smith, 1981). The importance of glacial refugia and routes of redispersal in
structuring the current patterns of distribution has been described for several species of
northern ﬁshes (Underhill, 1986; Bailey and Smith, 1981; Hubbs and Lagler, 1964).
Current distribution of the redside dace is likely a historic product of migration and
changes in watershed boundaries through shifts in drainage outlets among the Great Lakes
and through headwater stream capture (Smith et al., 1981). In recent time, human
activities have further restricted the range of the redside dace and may threaten its
continued survival in Michigan (Coon, 1993; Smith et al., 1981; T rautman, 1981).

Green noted as early as 1935 that the redside dace exhibited a disjunct, extremely
localized, and shrinking distribution that was characteristic of a species threatened with

extinction. Factors identiﬁed as most responsible for the demise of native ﬁshes in

5
southeastern Michigan and throughout North America include widespread degradation

and destruction of instrearn and riparian habitat, deteriorating water quality ascribed to
erosion and chemical alteration, overexploitation, and the introduction of nonindigenous
species (Warren and Burr, 1994; Cain, 1993; Miller et al., 1989; Karr et al., 1985; Smith,
1981). All are symptoms of the rapidly expanding inﬂuence of human activities on stream
ecosystems. Smith et a1. (1981) demonstrated the inﬂuence of agricultural and urban
development and pollution on ﬁsh communities in the Raisin River system of southeastern
Michigan, noting that several species were restricted to the upper part of the drainage and
were threatened by the loss of critical habitat. They ﬁrrther documented the extirpation of
two headwater species from the watershed, the bigeye chub (Hybopsr‘s arnblops) and the
eastern sand darter (Ammocrypta pellucr’da) which, as suggested for redside dace

(T rautrnan, 1981), require clear water and gravel for spawning. The Raisin River drainage
is positioned between the two watersheds that hold the remaining extant populations of
redside dace in Michigan. Kart et al. (1985) presented similar ﬁndings for the Maumee
River drainage of northwestern Ohio and southeastern Michigan where one of the
remaining redside dace populations is located. They found that 50% of the headwater
species in the drainage had declined in abundance since 1850, and four headwater
invertivores that required clear water had been extirpated from the system. Human
impacts, particularly erosional inputs ﬁom intensive agriculture, have been implicated in
the decline of the redside dace in Other parts of its range (Parker et al., 1988; Cooper,
1983; Trautman, 1981; Gilbert, 1980; Eddy and Underhill, 1974; Scott and Crossrnan,

1973).

6
Though locally abundant in certain regions, isolated populations of redside dace

are especially vulnerable to chance catastrophic events because mortality in a local
population is unlikely to be offset by immigration ﬁ'om other populations. Individual
rosyside dace, Clinostomusﬁmduloides, the closest relative of redside dace, have a small
home range (Hill and Grossman, 1987). In addition, a disjunct distribution may have
important implications for inter- and intrapopulation genetic diversity that have also been
described for rosyside dace inhabiting headwater streams in southern Ohio (White, 1988).
White (1988) found signiﬁcant heterogeneity among populations at two polymorphic
allozyme loci, which he interpreted as a minor degree of interpopulation diﬁ‘erentiation.
The most signiﬁcant discovery, though, was of unexpectedly low levels of intrapopulation
variability that supported an hypothesis of routine bottleneck episodes and coinciding
founder effects (Hartl and Clark, 1989). Similar conditions may prevail among
populations of redside dace that exhibit a similar, restricted pattern of distribution.

Several jurisdictions have provided protection of the redside dace. They are listed
as species of special concern in Kentucky and West Virginia, and are under legal
protection in Indiana (possibly extirpated; Gilbert, 1980). Redside dace have been
extirpated ﬁ'om Iowa (Harlan and Speaker, 1987) and have decreased dramatically in Ohio
(T rautman, 1981) and Ontario (Parker et al., 1988). Furthermore, Becker (1983) reported
the “disappearance” and “severe decimation” of several populations in southeastern
“Wisconsin.

In other parts of their range, redside dace appear to be abundant and persistent.

Populations in southeastern Minnesota, northern Wisconsin, and western New York have

7
not experienced the declines reported for redside dace elsewhere (I. Underhill, I. Lyons,

and R Daniels, personal communication).

The redside dace is known from the headwaters of the Upper and Middle River
Rouge, a single tributary of the Huron River, and a tributary of the Maumee River near the
Ohio border (Carl Latta, personal communication) in southeastern Michigan and is listed
as a threatened species in the state. Redside dace have also been introduced into Fleming
Creek, a tributary of the Huron River near Ann Arbor. Collection records of the
University of Michigan Museum of Zoology suggest that the redside dace may have been
more widely distributed in the River Rouge watershed in the recent past and survey efforts
have demonstrated a decline in populations in the Upper River Rouge (G. Smith, personal
communication); however, the full extent of their historical Michigan range is unknown.
There is currently no evidence to suggest that redside dace were ever more widespread
among southeastern Michigan watersheds nor has the complete loss of populations from a
watershed been documented.

Adultredsidedaceaverage65 ~70 nuninstandardlengthandareeasilyidentiﬁed
from other cyprinids by their large eyes and large, upturned mouth with a projecting lower jaw.
They are dark olive along the top of the back with a bright, golden stripe that stretches ﬁ'om
theupperoperculumtothebaseofthetail. Adark, rnidlateral stripe extendsalongthe
posterior portion of the body, and becomes mixed with silvery pink anterior of the dorsal ﬁn.
'l'hisregionbecomesastrildngerimsoninbreedingmales.

The biological literature offers little in the way of comprehensive life history research

concerningtheredsidedace. Habitatusebyredsidedacehasbeendescribedindetailatone

8
sitebyKoster(l939)andingeneraltermsbyseveral authorsreferencedbelow. Redsidedace

arefoundineool, dearheadwatersu'eamswithnrodaategradiernthatoﬁ‘erambctureofpool
and riﬂle habitats (Meade et al., 1986; Becker, 1983; McKee and Parker, 1982; Trautman,
1981;Koster, 1939). Thoughadultsaremostoﬁencollectedinpools, theyreportedlyuse
riﬂes for spawning (Koster, 1939).

Ripmianvegaaﬁonalsoappearswbeanhnponamamihrteofredsidedaoehabnat
pahapsmremdwspedaﬂndfeedinghabhsofredsidedacemmﬁdsdemﬁewsld, 1994;
Parker et al., 1988; Schwartz and Norvell, 1958). In summer, redside daoe fwd primarily on
terrestrialinsectsthatlandorfalltothewater’ssurface, andregulariyleap severalinchainto
the air to catch hovering insects (personal observation; Daniels and Wrsniewski, 1994;
Schwartz and Norvell, 1958; Koster, 1939). Redside dace stomachs analyzed by Daniels and
Wrsniewski (1994) and Schwartz and Norvell (1958) contained 77 - 88% aerial terrestrial
insects by number, primarily adult Diptera

Information on the physiology and phylogeny of redside dace is more limited. Evans
and Deubler (1955) gave a detailed description of the process of pharyngeal tooth replacemmt
in redside dace. Bailey (1951) discussed the relationship ofClimstonms to Richardsonius and
Gila, suggesting that the groups be combined into the single genus Gila. In contrast, Koster
(1939) remarked on similarities ofredside dace to Chrosomus (Phwcinus) spp. as well as to
ngwrlsc-rrsmmgan'ta. The rosysidedaee isthe only congener ofredside dace and has a

more southern distribution (Gilbert, 1980); the ranges of the two species are adjacent but do

notoverlap.

9
DiﬁarhquesﬁonscomgﬂwconﬁmedwrvivalofredsidedacemMiclﬁganare

irmninem InatlasttwoofthefompopulationsinMichigan,immediatethreatsareposedby
reddarﬁalandcmmnadaldevdoprnanmarburmearohmdbytheMOduaionofa
potenﬁalpredator,browntrout(Sabnotmtta),anon-nativesportﬁsh. Researcheﬁ‘ortsthat
aredirectedtohnpmveourunderstandingofﬂreecologicalrequiranartsofredsidedacein
Mchiganarewarrantedforseveralreasons. First, redsidedaceappeartohavenarrowhabitat
mlaameswhichnukeﬂranavahrablemdicatmofstreamquaﬁty(Snﬁﬂraal, 1981). More
mmtanﬂyﬁedsidedaceﬁﬂaumquemleasadgmﬁcammsdvorwspredatmmaaial
terrestrialinsects. Thespecializcdforagingstrategyoftheredsidedaceisdistinctamong

manbasofﬂwnﬁnmwfamﬂyaswdlasamongothaNonhAmaicanstreamﬁshes.

Objectives. - The purpose of this research project was to evaluate the potential for using
redsidedaceornsideofMichiganasmodelsbycompanngthephenotypic similarityof
populations ﬁ'om Mchigan, New York, and Wisconsin, populations representing the breadth
of the species’ distribution. This study describes ecological requirements of the redside dace
and provides infonnation that is immediately useful in formulating management plans for
redside dace in Michigan and for declining populations threatened in other states.

My speciﬁc objectives were to describe (1) behavioral similarity between regions by
comparing instream observations of microhabitat use, (2) physiological similarity among
populations and regions by comparing metabolic rates; I also provided information on the

thermal tolerance of New York redside dace as it related to variation in acclimation

10
temperatures, and (3) morphological similarity among populations and regions based on a

comparison of morphometric and meristic traits.

To complete the three objectives, the project coordinated activities in the ﬁeld, lab, and
museum to make comparisons within and among geographic regions. ﬁeld work
encompassed observations of behavior and microhabitat use of redside dace (ﬁrst objective) in
one stream in Michigan and two streams in New York. Physiological comparisons (second
objective)weremadeintheﬁeldthroughtheinstreammeasurementofmetabolicratesatone
siteinMichiganand one siteinNewYork. Labworkalso focused on comparisons of
metabolic rates and thermal tolerances of redside dace (second objective) collected from four
populations in New York that were returned to Michigan for study. I was restricted by state
regulations ﬁ'omtransportingMichigan redsidedacetothelab. ResearchattheUniversity of
MelﬁganMusannonoologywasperfonnedwithpreservedspecimensto satisfythethird
objective, a comparison of morphology between redside dace originally collected ﬁom one

watershedinMidriganmdtwowatershedseachinNewYorkandWrsconsin.

METHODS

Study Area

I performed ﬁeld work at one site in Michigan and two sites in New York. Seeley
Drain is a 7 km long headwater of the Upper River Rouge that ﬂows southeasterly
through the suburbs of Oakland County in southeastern Michigan (elevation 260 m,
latitude 42°30’; Figure 1). The Upper Rouge joins with the Middle and Lower branches
to ﬂow east through the center of urban Detroit and into the Detroit River.

The study area included a 2 km reach between 12 and 13 Mile Roads
approximately 2 km ﬁ'om the strearn’s origin. Seeley Drain ﬂowed through an extensive
wetland area just upstream of the study site and received inputs in the form of runoff from
residential neighborhoods, parking lots, city streets, and construction sites as well as
treated wastewater eﬂluent from a facility upstream of the wetland (Coon, 1993; personal
observation). In the study section, the stream was bordered by a golf course on the west
and homes on the east. At a few points where riparian vegetation had been altered or
eliminated, lawns and fairways met the stream bank.

Through most of the reach, a dense zone of grasses, willows, and woody shrubs
overhung the stream. Seeley Drain was channelized at least twenty years ago and has
been maintained as a public drain by the county, though it displayed a pool-rime-nrn
sequence dominated by pools and shallow runs. Sediments were relatively homogeneous:

sandy with a thin layer of silt. Gravels were restricted to riﬂles and fast runs and there

11

 

 

 

 

OAKLAND CO
Seeley Drain E
5 o G :
______' _____ F®€reo _ __ -__-___'. ________________
i 086 ‘8},
“is
. ' c o
Mlddle Roug ea
1 s
1 ' b
i
i Louver Rouge :5?
1 N
.......... i__‘l";‘.*.”.E_C_°._-_- __ _ , r——.
. ’ ‘ X 5 km

 

 

Figure l. A map illustrating the Seeley Drain ﬁeld site located in the Upper River Rouge
watershed of southeastern Michigan.

13
was little instrearn woody structure or aquatic vegetation. Width was less than 2 m and

depth was less than 0.75 m.

In New York, I conducted ﬁeld work in two streams, Rice and France Brooks
(Figure 2), both headwater streams in the Allegheny River drainage of Cattaraugus
County at the southwestern edge of the state. The streams were within the boundaries of
Allegany State Park and drained steep valleys of mature maple and oak forest. Human
impacts to the streams were much more subtle than in Seeley Drain. Areas adjacent to the
streams have undergone little modiﬁcation other than the construction of seldom traveled
gravel roads and associated bridge crossings. Oil and natural gas ﬁelds and timber
resources have been exploited in the streams’ watersheds, and brook trout (Salvelinus

fontinalis) and brown trout have been introduced into both streams.

France Brook is a tributary of Red House Brook, a small stream that ﬂows
northwesterly into the Allegheny River. The 0.7 km reach that I studied (elevation 510 m,
latitude 42°03’) was 1 km upstream of the mouth and ﬂowed northeast through a wetland
dominated by grasses and shrubs that overhung stream margins. There were few trees
near the stream, though a large beaver dam and pond were Situated near the lower end of
the study section. The reach was primarily pool and run habitat with large, rocky
substrates covered with a ﬁne layer of silt. Stream width was less than 1 m and depths
under 0.5 m. .

Rice Brook was the largest stream studied and ﬂowed eastward through a more
heavily wooded region. Just outside of park boundaries the stream joined Tonquin Creek

to ﬂow north to the Allegheny River. The study reach was a 1.2 km section, 4 km

14

‘

 

 

 

 

 

 

 

   

NewYork
,i ____________________
| CATTARAUGUS CO. I
E WhigStreetBrook
RedHouseBrook - 1
: RiceBrook N
FranceBrook ‘ V : 9
I ‘ |
__________ ___-"___.’
|———'l
15km
was
we?

 

 

 

Figure 2. A map illustrating France, Red House, Whig Street, and Rice Brook ﬁeld sites
located in the Allegheny River watershed of southwestern New York.

l 5
upstream of the mouth (elevation 430 m, latitude 42°03’), composed of numerous deep

pools (widths > 7 m, depths > 1 m) separated by riﬂles and runs. Substrates were
dominated by pebble and cobble that were blanketed with a very thin layer of silt. Sand
ﬁlled the interstitial spaces in most areas. Trees and shrubs overhanging the stream were
less common and there was little instream woody structure.

I collected redside dace from two additional New York streams. Red House
Brook was much larger than France Brook but contained similar habitat attributes. It
ﬂowed through heavily wooded areas and has been described by Daniels and Wisniewski
(1994). Whig Street Brook was a small stream ﬂowing ﬁ'om the north in the Allegheny
River drainage and was bordered by roads and farmlands. The stream had been
straightened and received direct runoff from cultivated row crops, though water clarity
was high and depths up to 0.5 m were common. Substrates were rocky, lightly covered

with silt, and there was an abundance of small, overhanging willows and shrubs.

Behavior
I studied microhabitat use by redside dace in Seeley Drain, Michigan, during June
through August, 1994 and in Rice and France Brooks, New York, in mid-July, 1994. All
observations were made during daylight hours between 7:00 am and 7 :00 pm. Focal
animal observations of individual ﬁsh (Altmann, 1974) lasted approximately 10 minutes
and were conducted ﬁom the stream bank in Seeley Drain and by snorkeling in New York.
Stream bank focal animal observations followed the covert tactics of Reighard (1910) and

Hankinson (1920) and were facilitated by the use of polarized sunglasses and binoculars.

16
Good water clarity and the presence of high stream banks made this an effective method

of observation in Seeley Drain where the small size of pools, shallow depths, and risks
posed by urban inputs inhibited snorkeling. Direct observation of redside dace by
snorkeling in an upstream direction proved most useful in the New York streams that I
studied (Bovee, 1986; Grossman and Freeman, 1987).

The ﬁrst minute of observation was sufﬁcient to allow estimation of ﬁsh standard
length to the nearest 5 mm, sex, location over the stream bed, and vertical position in the
top, middle, or bottom third of the water column. Sex was visibly determined by
coloration and by the relative length of the pectoral ﬁns, males having a more intense red
stripe and longer ﬁns than females (Schwartz and Norvell, 1958). For the remainder of
the observation period, I made a primarily qualitative description of behavior that focused
on the use of instream cover, feeding activities, and intra- and interspeciﬁc interactions.
Cover types were distinguished by their assumed function as providing either visual shelter
(overhead structure), velocity shelter (structure projecting above the stream bottom), a
combination (most woody debris), or no cover (Monahan, 1991). Information that I
obtained while snorkeling was communicated through the snorkel to an assistant on the
bank. Focal animal observations were terminated either by the end of the 10 minute
period or by a behavioral response of the ﬁsh to the presence of the observer. Only data
collected ﬁom undisturbed ﬁsh were included in ﬁrrther analysis.

At the end of a focal animal observation, a weighted, numbered marker was placed
on the stream bottom beneath the location of the ﬁsh. I recorded microhabitat

characteristics at each observation location later the same day that included the dominant

17
substrate particle type (diameter < 4 mm = sand/silt; 4-16 mm = gravel; 16 - 64 mm =

pebble; 64 - 256 mm = cobble; > 256 mm = boulder), distance to the nearest physical
structure, water column depth, stream width, mean water column velocity, and velocity at
the midpoint of the vertical stratum occupied ( hereafter termed focal velocity). I also
recorded the distance to and the mean column velocity, width, and depth of the nearest
fast water area. Riparian attributes measured included estimation of percent canopy
coverage and the presence or absence of overhead structure within 1 m of the strearn’s
surface at each marker location. Depths were measured to the nearest 0.01 m by use of a
calibrated wading rod. Current velocities were measured by counting the rotations of a
pygmy Gurley Meter and translating counts into m s’1 using a standard calibration. Mean
column velocity was measured at 0.6 depth for depths < 0.75 m; at greater depths, mean
column velocity was taken as the average of measurements at 0.2 and 0.8 depth (Bovee
and Milhous, 1978). Focal velocity was measured in the middle of the vertical stratum
occupied corresponding to 17% depth in the upper third, 50% depth in the middle third,
and 83% depth in the lower third stratum.

I assessed the availability of habitat components by performing the measurements
described above at 39 points in Seeley Drain and 59 points in the New York streams
distributed randomly within the pools where I observed redside dace (Grossman and
Freeman, 1987). Therefore, the data that I collected reﬂected only the relative distribution
of habitat components found exclusively within pools used by redside dace and did not

describe the availability of habitat attributes throughout the stream reach.

1 8
I also measured water temperature, pH, alkalinity, conductivity, and turbidity at

each sampling site in addition to making a qualitative description of stream channel and
bank condition. Water temperatures were measured with hand-held thermometers and pH
with a Hanna Instruments waterproof pH tester that was calibrated before each use.
Alkalinity was also measured in the ﬁeld with a Hach portable water quality test kit and
conductivity was measured with a YSI Model 33 conductivity meter. Samples for
determination of turbidity were collected ﬁ'om mid-water. Sample bottles were
immediately placed in an ice chest and analyzed at the lab within 24 hours by use of a
Hach Model 2100A turbidometer calibrated with Hach gel standards prior to each use.

Frequency data for categorical habitat variables were analyzed with a 12 analysis of
multidimensional contingency tables (Zar, 1984). I computed the electivities (L) for each
category using Strauss’ index (Strauss, 1979) to more clearly demonstrate selection or
avoidance of speciﬁc habitat attributes in relation to their availability. Electivity scores of
-1.00 to -0.50 indicated strong avoidance of the habitat attribute, -0.49 to -0.26 moderate
avoidance, -0.25 to 0.25 neutral selection, 0.26 to 0.49 moderate selection, and 0.50 to
1.00 indicated strong selection of the habitat attribute (Baltz, 1990). Continuous habitat
variables were tested using two-factor AN OVA (PROC GLM, SAS Institute Inc., 1988)
on rank-transformed data (Potvin and Roff, 1993 ), and Tukey-type multiple comparisons
were performed among mean ranks (Zar, 1984). For both procedures, or = 0.05 was

selected to judge the signiﬁcance of test statistics.

19
Physiology

Fish capture and maintenance. - I collected a total of 100 adult redside dace in
late October, 1993, from France, Red House, Rice, and Whig Street Brooks in
southwestern New York by use of a battery-operated Wisconsin ABP-4 backpack
electroﬁshing unit. The ﬁsh were immediately sealed in plastic bags that contained stream
water and pure oxygen gas, placed in coolers, and returned to lab facilities at Michigan
State University within 24 hours. We exercised extreme care in handling and transporting
the ﬁsh and were rewarded by the 100% survival of the redside dace we captured. In the
lab, ﬁsh were held in 90 L glass aquaria at densities of 25 ﬁsh per aquarium. The aquaria
received a continuous ﬂow of untreated, aerated well water that maintained a temperature
of 12 i 0.2 °C throughout the study. Dissolved oxygen was maintained at levels above
80% saturation and pH varied between 7.8 and 8.2. A combination of natural and
artiﬁcial lighting conditions were controlled at 14:10 hours of daylightzdarkness, and ﬁsh
were fed daily to satiation on a diet of freeze-dried krill and prepared pellets,
supplemented twice weekly with ﬁ'ozen adult brine shrimp. Fish were held for 5 months
before experiments were initiated. Because of their protected status, Michigan redside
dace could not be included in the lab study.

Metabolic Rate. - I used static-type respirometers and the sealed vessel method
(Ccch, 1990; Winberg, 1956) to measure the resting routine metabolic rates of mature
redside dace. Oxygen consumption may be related to metabolic rate or the rate of energy
exPenditure through application of an appropriate oxycaloriﬁc coefﬁcient that is

dependent on ﬁsh condition and diet (F eldmeth and Jenkins, 1973; Elliot and Davison,

20
1975). Cech et al. (1985) deﬁned resting routine metabolism as characteristic of quiescent

ﬁsh, though not necessarily the lowest rate during the diurnal cycle (Cech, 1990). Resting
routine metabolism is a slightly higher level of metabolism than the standard rate that was
described by Fry (1971) as the minimum rate required to maintain life (zero activity).

I measured metabolic rate in two sets of experiments. In the ﬁrst, I compared the
metabolic rates of the laboratory-acclimated populations ﬁom New York with metabolic
rates measured in the ﬁeld for naturally-acclimated redside dace in New York and in
Michigan. Lab experiments were carried out in May - June and ﬁeld tests performed in
late September and early October, 1994. Individuals from the four lab populations (n =
29) were tested for oxygen consumption rate in consecutive 10, 20, 30, 40, and 50 min
trials. Each ﬁsh was tested once. In the ﬁeld, 6 individuals from Rice Brook, NY and 12
from Seeley Drain, MI, were tested in single, 50 min trials.

In a second set of experiments, the lab populations were acclimated to
temperatures of 6, 12, and 20 °C (11 = 16, 29, and 31 respectively) and individual ﬁsh were
tested for oxygen consumption rate in single, 50 min trials. A small number of the ﬁsh
tested in experiment 1 were retested; these individuals were evenly represented among the
temperature treatments. From a starting temperature of 12 ”C, water temperature was
adjusted 1 °C per day until the target acclimation temperatures were reached. Fish were
then held at the new acclimation temperature for 14 days before testing.

The respirometers that I used were constructed from glass pint jars with a
gasketed lid that was ﬁtted with polystyrene inlet and outlet tubes (Figure 3). Water
samples were taken from a short length of airline tubing connected to the outlet tube in the

lid of the jar. Prior to testing, redside dace were starved for 40 - 48 h. Individual test ﬁsh

21

water
source

 

 

 

  

 

 

 

 

— thermometer
water bath
respirometer
f =
inlet tube
outlet tube air stone sample

 

 

Figure 3. Diagram of the static respirometer used to collect samples for estimation of
adult redside dace oxygen consummion rates.

22
were gently netted and sealed in the respirometers and submerged in glass aquaria that

received a continuous ﬂow of aerated, constant temperature well water. The respirometers
were carefully debubbled and a siphon was started to maintain the circulation of fresh
water through the respirometers. Test ﬁsh were then allowed an overnight acclimation
period to become accustomed to the chambers. In addition, one blank respirometer was
always set up and tested as a control to estimate oxygen consumption due to microbial
metabolism. All oxygen consumption measurements were corrected for consumption
recorded with the corresponding control. The aquaria were shielded from the movements
of experimenters, and small mirrors allowed me to observe the behavior of ﬁsh in the
respirometers.

To measure oxygen consumption, I took an initial water sample ﬁom the outlet
tubing in a 60 ml glass bottle. I then plugged the end of the tube, thereby interrupting
ﬂow for the duration of the trial period. The amount of dissolved oxygen (mg L") in the
sample was immediately measured using a YSI Model 50B Dissolved Oxygen Meter and a
Model 5730 polarographic probe with an automatic stirrer. At the end of the trial period,
the outlet tube was unplugged and a ﬁnal water sample taken and measured as above.
Oxygen consumption rates (mg 02 h") were calculated by dividing the difference in
dissolved oxygen concentration (initial - ﬁnal) by the duration of the trial (h) and adjusting
for the respirometer volume. The concentration of dissolved oxygen in the respirometers
was not allowed to drop below 70% saturation to avoid the potential effects of hypoxic
stress (Brett, 1964) and all measurements were taken between 8:00 am and 5:00 pm. At

the conclusion of each trial, I measured standard length of the test ﬁsh to the nearest 0.1

23
mm and wet weight to the nearest 0.01 3. Test ﬁsh were also given a small ﬁn clip to

insure that they were tested only once. Sex was determined by coloration and the ratio of
head length to pectoral ﬁn length as outlined by Schwartz and Norvell (1958).

Field procedures were identical to those followed in the lab (described below) with
a few exceptions. First, redside dace were captured by seine and held instream in plastic
cages perforated with 3 mm openings prior to testing. Second, the respirometers were
submerged in coolers placed just below natural beaver and debris dams. Stream water
was directed over the dams and into the coolers with a piece of rain gutter to allow
maintenance of the siphons that ﬂushed the respirometers. Measurements were made on
the stream bank with the same equipment used in the lab.

The oxygen meter was calibrated to Winkler titration (azide modiﬁcation: APHA
et al., 1989) on a regular basis, daily during periods of ﬁ'equent use. Precision of the
meter was reported by YSI to be within i 0.02 mg L", a claim supported by consecutive
measurement of samples from the same source (range of differences = 0.004 - 0.04, mean
= 0.018, n = 25). However, the degree of accuracy of the meter was more difﬁcult to
evaluate. Calibration records suggest that the absolute diﬂ'erences between the meter and
titrations varied from 0.02 - 0.57 mg L", with a mean deviation of 0.23 (n = 24). Because
the data submitted to analysis was the difference in initial and ﬁnal readings, precision was
considered to be of primary importance. Deviations in accuracy were assumed to be
consistent within a trial period, an assumption validated by the minuscule rates of
consumption measured during trials with the blank respirometer. To account for error in

the precision of the meter, all oxygen consumption rate standard errors were assigned a

24
minimum value of 0.01, the error associated with measurements using the pint-sized

respirometers (0.473 L; Cech et al., 1985). No attempt was made to correct for
measurement accuracy other than required maintenance and calibration of the meter.

Oxygen consumption rates were adjusted by an allometric relationship to account
for mass-speciﬁc diﬁ'erences among the ﬁsh tested:

Y=ax‘

where Y = 0; consumption rate (mg min"), a = mass coeﬁcient, X = mass (g), and b =
mass exponent (Cech, 1990). The parameters a and b were estimated from the linear
regression of the log-transformed oxygen consumption rates on log mass:

log Y = -2.7364 + 0.7451 log X
(12 = 0.732, SE = 0.0448). The raw consumption rate data were then divided by (mass")
to create “mass-independent” metabolic rates that were analyzed in log-transformed form
(Winberg, 1956). The signiﬁcance of treatment and group effects were assessed through
factorial AN OVA and Tukey’s HSD (PROC GLM, SAS Institute Inc., 1988).

Brenna! Tolerance. - I also studied a second physiological attribute of redside
dace, upper thermal tolerance, using the critical thermal maximum method (CTM)
originated by Cowles and Bogert (1944) and presented as a standardized methodology by
Hutchison (1961). Several authors have modiﬁed the procedure for application to ﬁshes
(Cox, 1974; Hutchison and Maness, 1979; Becker and Genoway, 1979; Paladino et al.,
1980). Becker and Genoway (1979) deﬁned CTM as a method for measuring the thermal
tolerance of poikilotherms by raising the temperature from ambient acclimation at a linear

rate so that no signiﬁcant time lag exists between the temperature of the media and the

25
internal tissue of the test organism. Raising the temperature too fast may induce thermal

shock, and a heating rate that is too slow may allow partial acclimation during the CTM
test. Becker and Genoway (1979) further deﬁned CTM as a parameter: the temperature
where the test organism loses the ability to escape ﬁ'om conditions that promptly lead to
its death.

I measured the CTM for a total of 23 redside dace from France and Rice Brooks,
two of the New York lab populations used in the metabolic rate experiments. The ﬁsh
were already acclimated to temperatures of 6, 12 and 20 °C and had been tested
previously for metabolic rate. Individual redside dace were heated from ambient
acclimation temperatures in a large beaker containing 3 L of water. The heating rate
varied between 0.3 - 0.5 °C rnin'l (Becker and Genoway, 1979), and was measured at
regular intervals with an AS'I'M certiﬁed mercury thermometer (0.1 °C) to insure linearity
of heating rate. The beaker was aerated to provide uniform mixing and adequate
dissolved oxygen levels. Heating was initiated immediately after careﬁrl introduction of
the ﬁsh to the beaker, and observations were made from behind a blind with a viewing slit.

I identiﬁed the CTM end point as the ﬁnal loss of equilibrium (FLE), marked by a
total loss of righting ability. At this point, the ﬁsh had lost the ability to escape from the
thermal conditions that would promptly lead to its death (Becker and Genoway, 197 9). I
also identiﬁed four other behavioral markers that appeared to be related to increasing

thermal stress:

26
1. BSW - sporadic burst swimming occurred when a continuous, consistent level

of swimming became interspersed with sudden bursts of activity usually
directed at the wall ofthe beaker.

2. SUR - surfacing, breaking the water’s surface with the snout.

3. ESW - erratic swimming, characterized by sudden, directionless bursts, muscle
spasms, and initial loss of equilibrium.

4. OPS - opercular spasms indicated by ﬂaring and spasmodic movements of the
gill covers and gulping motions (F eminella and Matthews, 1984). OPS soon
followed FLE; if not immediately removed from the test conditions, the ﬁsh
quickly died.

I recorded the time and temperature upon recognition of each behavioral landmark. At
the conclusion of a test, ﬁsh were weighed and measured for standard length. It was
determined ﬁ'om a small number of ﬁsh that immediate return to acclimation temperature
following OPS allowed at least temporary recovery. No attempt was made to assess long
term eﬁ‘ects of the thermal tolerance tests.

Untransfonned end point temperatures were analyzed by factorial AN OVA and

Tukey’s HSD (PROC GLM, SAS Institute Inc., 1988).

Morphology
Morphometrics - I performed a landmark-based morphometric analysis
(Bookstein, 1992) to examine variation in external body shape among populations of

redside dace collected in “Wisconsin, Michigan, and New York and with a congener, the

27
rosyside dace, that served as an outgroup. The preserved specimens were provided by the

University of Michigan Museum of Zoology (App.) and included lots that were collected
between 1925 and 1953. The redside dace specimens were selected to represent
populations ﬁom two watersheds in each state with the exception of Michigan, where
available material was limited to the River Rouge drainage (primarily individuals from the
Upper River Rouge). Included in the analysis were redside dace collected in the
Allegheny River and Susquehanna River watersheds in New York and the Black and
“Wisconsin River watersheds in Wisconsin. Rosyside dace specimens from North Carolina
and Tennessee were pooled to obtain an adequate sample size of adult specimens.

Rosyside dace were selected as the outgroup as the only other member of the
genus Clinostomus. Patterns of habitat use, social, and foraging behaviors of the rosyside
dace (Freeman, 1990) are in many ways analogous to those of the redside dace (Koster,
193 9), yet rosyside dace have a non-overlapping, more southern distribution and display
morphological diﬂ‘erences from redside dace that have been summarized by other authors
(Page and Burr, 1991). The rosyside dace form provided a convenient reference ﬁ'om
which to judge the scale of shape differences among the groups of redside dace.

For the shape analysis, I selected 17 corresponding landmark points around the
bodies of the ﬁshes (Figure 4, Table 1) chosen because they were representative of overall
body shape and could be identiﬁed with conﬁdence on each specimen. The images of each
specimen and the x- and y-coordinates of each landmark were captured and digitized to
the nearest 0.1 m using a Cohu RS-170 video camera (resolution 565x350) and Sony

Trinitron HR color monitor interfaced with a micro-computer and image-processing

28

 

10 11

15 14
“' 13

Figure 4. Locations of the 17 landmark points that were digitized around the body of each
redside and rosyside dace museum specimen for morphometric analysis. Refer to Table l
for a description of each point.

29

Table 1. Description of the 17 landmark points used in the morphometric analysis.

Point

PWSP‘P‘FPN!"

Hr—Ir—Ir—Iu—I—Ir—Ir—I
SP‘MFP’NFFD

Landmarlrllescn'm
Furthest anterior projection of the lower jaw.
Furthest posterior projection of the mouth.
Center position of the anterior naris.
Most anterior edge of the eye.
Most posterior edge of the eye.
Occiput.
Dorsal insertion of the operculum.
Furthest posterior projection of the operculum.
Anterior insertion of the dorsal ﬁn.
Posterior insertion of the dorsal ﬁn.
Dorsal insertion of the caudal ﬁn.
Posterior base of the vertebral column.
Ventral insertion of the caudal ﬁn.
Posterior insertion of the anal ﬁn.
Anterior insertion of the anal ﬁn.
Anterior insertion of the pelvic ﬁn.
Dorsal insertion of the pectoral ﬁn.

30

software (OPTIMAS, Optimas Corp.). In order to minimize measurement error within
groups of specimens, I digitized the landmark coordinates of each ﬁsh on two
nonconsecutive occasions (Y ezerinac et al., 1992). The mean of these two estimates
provided the working data set. For analysis, the landmark points were transformed to
shape coordinates through the two point registration technique (Bookstein, 1986): the
points were rotated and scaled to a common baseline drawn between landmarks 1 and 12
that were given arbitrary coordinates of (0,0) and (1,0). In this way, the method may be
regarded as a size-independent analysis of shape differences. Only redside dace > 45 mm
SL and rosyside dace > 40 mm SL were included in the analysis.

The untransformed shape coordinates were submitted to a multivariate test (W ilks
A; PROC GLM, SAS Institute Inc., 1988) as dependent variables with group (a
categorical variable) as the independent variable in the model. Experimentwise error was
controlled by adjusting the signiﬁcance level for the univariate tests of each coordinate
(Hayes, 1993; Sokal and Rohlf, 1981). To maintain an experimentwise error rate of or“, =
0.05 for the 30 individual tests, a signiﬁcance level of or = 0.0017 was used.

I assessed the magnitudes and directions of shape change between groups by
application of the thin-plate spline decomposed by its partial warps, a morphometric
method recently presented by Bookstein (1989; 1992) and summarized in Rohlf and
Bookstein (1990). The thin-plate spline is a multivariate ﬁrnction that may be used to
interpolate surfaces over scattered data and is an effective tool for modeling biological
shape change that is described by change in the conﬁgurations of landmark coordinates.

The thin-plate spline can be visualized as the bending of an inﬁnitely thin metal plate or

3 1
grid over one set of landmarks, the reference conﬁguration or “starting form”, so that its

height over each landmark is equal to ﬁrst the x-, then the y- coordinate of the
corresponding landmark in another set, the “ﬁnal form” (Zelditch et al., 1995). In this
way, the thin-plate spline may be thought of as a distance function that maps the
conﬁguration of one object onto another, displayed as a deformation of the starting form
(Bookstein, 1990). The thin-plate spline mapping quantiﬁes the metaphorical “bending
energies” associated with changes in the shape of the metal plate as the starting form is
ﬁtted to the ﬁnal form.

The shape deformation described by the thin-plate spline mapping is decomposed
into a uniform or linear part, the afﬁne transformation, and a non-uniform, non-linear part.
The aﬁne transformation is a uni-directional change in the position of all landmarks that is
proportional to the landmark’s perpendicular distance from the selected baseline. By
deﬁnition, it is a homogenious transformation over the entire form that keeps parallel lines
parallel and is recognized as an apparent tilting or rolling of the thin-plate. The uniform
transformation has zero bending energy. In contrast, non-uniform shape variation is
localized at progressively smaller spatial scales as changes diﬂ‘ering ﬁom one region to
another due to the displacement of individual groups of landmarks.

The pattern of non-unifonn bending is summarized in a bending energy matrix with
the eigenvectors of the matrix termed “principal warps.” Principal warps, Bookstein’s
(1992) “descriptors of deformation” are computed by the same subroutines that compute
principal components in Principal Components Analysis. The bending energy depends on

the location and spacing of the landmarks of the starting form and measures the

32
displacement of a landmark relative to its distance ﬁ'om other landmarks. Principal warps

may be localized by eigenvalue (bending energy per unit summed squared landmark
displacement) which corresponds to the inverse geometrical scale of the deformation. The
largest eigenvalue corresponds to the smallest scale or local change, and the smallest
eigenvalue corresponds to the largest scale change. The two-dimensional contribution of
each principal warp to the shape deformation is described by a corresponding number of
“partial warps,” vectors that are computed by multiplying the principal warps by the
landmark coordinates of the ﬁnal form (Zelditch et al., 1995). Partial warps provide a
graphical representation of how each principal warp alone bends the thin-plate. The
principal warps represent the geometric terms in which shape differences can be described,
and partial warps are the values assigned to those terms (Swiderski, 1993). Thus, the
primary advantage of the method of thin-plate splines is that shape change, modeled as a
deformation, may be attributed to uniform and non-uniform components of variation, and
the non-uniform component may be ﬁrrther localized as changes at smaller spatial scales.
For this analysis, the mean shape coordinates of rosyside dace, the outgroup, were
compared to individual redside dace shape coordinates using the TPSPLINE software
package (version 9/14/94) developed by F .J . Rohlf. The x- and y-components of the
uniform transformation were estimated using an algorithm provided in Bookstein (1990).
Meristics. - I also made three types of meristic counts on redside dace using
specimens ﬁ'om the same lots examined in the morphometric study. The number of dorsal
ﬁn rays, anal ﬁn rays, and scales in the lateral line were determined. Counts were made

according to Hubbs and Lagler (1964). Scale counts were made at least twice on each

33
specimen to minimize counting error. Accuracy of ﬁn ray counts were randomly veriﬁed

by a second investigator (D. Hondorp).

RESULTS

Behavior

Habitat Use. - Stream water quality characteristics of Seeley Drain were different
in several respects ﬁom characteristics of the New York streams (Table 2). The
temperature data presented in Table 2 is a summary of readings collected in July, 1994 in
Seeley Drain and in mid-July in New York. A comprehensive temperature record for
either of the New York streams was not available. Water temperatures were generally
higher in Seeley Drain. Alkalinity, pH, and conductivity were all higher in Seeley Drain,
perhaps reﬂecting regional geologic differences between Michigan and New York and
differences in land and water uses. Turbidity tended to be somewhat higher in the New
York streams, perhaps the result of deeper, richer soils in southwestern New York that do
not have the high sand content of southeastern Michigan soils. However, all of the
streams were very clear.

I searched for redside dace in pool, run, and riﬁle habitats, but only found them
associated with pools. The Seeley Drain study reach was composed of approximately
50% pool, 30% run, and 20% riﬂle compared to the Rice Brook site which was 45 %
pool, 30% run, and 25% riﬁle, and the France Brook site which was 40% pool, 25% run,
35% riﬁle. Observation was most difﬁcult in rifﬂes where turbulence and glare at the
stream surface were greatest and runs were generally too shallow to navigate while

snorkeling. Pools consequently received the greatest amount of Observation effort.

34

35

Table 2. Range in stream water quality characteristics of Seeley Drain, Michigan and Rice
and France Brooks, New York, during July, 1994. Sample sizes are given in parentheses.

l D 'n Rice Brook France Brook

Highest recorded 24.4 19.4 21.1
Temperature (°C)

Conductivity (umhos) 820 (1) 27 (1) 32 ( 1)
pH 7.8 - 8.3 (5) 7.2 - 7.3 (6) 7.6 (5)

Alkalinity(mg/LCaCO3) 210-268(4) 17-22(3) 15-17(4)
Turbidity(ntu) 1.5-2.6(9) 1.0-1.2(9) 3.9-4.7(9)

36
Redside dace were observed in 5 pools each in Seeley Drain and France Brook and 6

pools in Rice Brook.

Methods for observing habitat use were generally successful, and viewing
conditions allowed conﬁdent identiﬁcation of individual redside dace and other co-
occurring ﬁsh species (Table 3). In Seeley Drain, 10 species were collected or observed
to occupy the same pools used by redside daoe Creek chub (Semotilus atromaculatus),
blacknose dace (Rhinichthys atratulus), and central mudminnow (Umbra limi) were most
common. In the New York streams, 19 species were observed with redside dace of which
blacknose dace, central stoneroller (Campostoma anomalum), creek chub, and common
Shiner (Luxilus comutus) were most prominent. Maintaining an observation period of
suﬁcient length prior to discovery by focal animals was difﬁcult while snorkeling. The
quiet conditions of most pools required extreme care on the part of the observer, and
individual redside dace did not always orient upstream in the larger, slower pools.
Therefore, discovery was not easy to avoid but usually allowed at a minimum the
identiﬁcation and determination of the location of the undisturbed ﬁsh over the stream
bed. Disturbance of individual ﬁsh was apparent by characteristic escape responses or
more often by exploratory movements towards the snorkeler.

Data from the two New York streams were pooled to obtain adequate sample
sizes for comparison to habitat use data from Seeley Drain, Michigan. The statistical
analysis of three categorical habitat variables, substrate, instream cover, and canopy, was
hindered by small cell sizes that violated the assumptions of x2 . In the case of substrate

and canopy coverage, categories were meaningfuﬂy combined when necessary to eliminate

37

Table 3. Fish species co-occurring with redside dace in Michigan and New
York. Names follow Robins et al. (1991).

SREQLES MEGAN NM
Campostorna anomalum x
Catostomus commersoni x x
C ottus bairdi x
Culea inconstans x
Etheostoma caeruleunr x
Etheostomaﬂabellare x
Etheostoma nigrum x x
Hypentelium nigrum x
Lepomis cyanellus x x
Luxilus cornutus x x
Nocomis biguttatus x
Percina caprodes x
Percina maculata x
Phoxinus eos x
Pimephales notatus x
Pimephales promelas x x
Rhinichthys atratulus x x
Rhinichthys cataractae x
Salmo trutta x
Salvelinusfontinalis x
Semotilus atromaculatus x x

Umbra limi x

38
empty cells (Zar, 1984). However, this was not possible in the analysis of instream cover,

therefore 12 computations were considered biased.

Several of the habitat characteristics that I measured differed signiﬁcantly in
frequency of use by Michigan and New York redside dace; however, few of the same
characteristics differed when frequency of use was compared to the inherent availability of
the attribute within the pools that I sampled. The relationship between use and availability
of substrate types illustrated this pattern (Figure 5). Substrate types were combined into
two categories, ﬁne (sand and clay) and coarse (gravel - boulder) for the x2 tests, and
these conﬁrmed that there were clear diﬂ‘erences between Michigan and New York in both
the use and availability of substrate types (P < 0.001). Pools in Seeley Drain had an
overwhelmingly sandy bed in contrast to the New York streams where substrates were
rocky and more diverse. However, the use of substrates in Seeley Drain and New York
showed insigniﬁcant deviation from what was expected based on availability (P > 0.25).
Electivity scores indicated that redside dace at both study sites exhibited neutral selection
of substrate types.

There was a slight tendency for New York redside dace to be associated with the
boulder substrate type (11.6 % use, 5.1% available), a ﬁnding that was related to
observations of cover use (Figure 6). New York ﬁsh were associated with velocity
shelters (41.9% use), usually boulders, more often than predicted by random choice (22%
available; P < 0.10) and more frequently than redside dace in Seeley Drain; the electivity
score suggested neutral to moderate selection by New York redside dace for this
component as well. Seeley Drain contained more structures assumed to provide visual

cover such as vegetation hanging in the water and small tree roots. In Michigan and in

39

 

 

     
   
 

 

 

 

 

 

 

 

 

MICHIGAN NEWYORK
0-02 I] USE (38) [I USE (43)
A 80" so - AVL (59)
s: 0.11
40 - 4o - -0. 14
20- .005 20- / 0.07
-005 0.010.01 i If -005
L 1;, ,1, i": all h a I // I?”
>n
3 E it it a g 2 E
m e a eg “ e

 

Figure 5. Stream bottom substrate percent ﬁ'equency of availability and use by adult
redside dace in Michigan and New York. Electivity (L) is given above the bars of each
category and sample sizes are given in parentheses.

 

 

MICHIGAN NEWYORK
E] USE (38) '0’29 I] USE (43)
I AVL (39) AVL (59)

   
  

 

0.20

 

 

 

 

 

 

 

 

Figure 6. Instream cover type percent ﬁequency of availability and use by adult redside
dace in Michigan and New York. Symbol explanation follows Figure 5.

4 1
New York, use of locations with no cover was also frequently observed (52.6% MI and

41.9% NY). This lack of association with structure appeared to be directly related to the
pelagic positions occupied by most of the redside dace: 73.7% of Michigan and 81.4% of
New York ﬁsh held positions in the middle third of the water column (no difference; P >
0.05; Figure 7), suggesting moderate to strong selection of this position. Focal animals
were usually oriented upstream. Michigan and New York redside dace were also similar
in their proximity to structure (P > 0. 50), irrespective of cover use (median distance 0.20
m for both Michigan and New York), and use did not differ signiﬁcantly from availability
(P > 0.20).

The New York stream pools offered signiﬁcantly greater (P < 0.001) median
depths (0.27 m) and widths (3.10 m) than Seeley Drain (0.18 m and 1.30 m), and redside
dace in both locations were consistently found in the deepest (P < 0.005; Figure 8) and
widest (P < 0.05) locations (Michigan use: depth = 0.42 m, width = 2.00 m; New York
use: depth = 0.40 m, width = 5.00 m). Yet, differences in stream dimensions did not
equate to signiﬁcant diﬂ‘erences in the median mean column velocities (P > 0.35; Figure 9)
available in Seeley Drain (0.072 m S“) or in New York (0.079 m s'l); the available
velocities also were not signiﬁcantly different ﬁom the velocities associated with use
(Michigan: 0.054 m s", New York: 0.063 m 8"). Corresponding median focal velocities
encountered by redside dace also did not differ signiﬁcantly (P > 0.05) between Michigan
(0.050 m s“) and New York (0.083 m s“) and were not signiﬁcantly different from the
median mean column velocities associated with use described above (Mann-Whitney U, P

> 0.5 for Michigan and New York).

42

 

 

 

 

 

 

 

 

E MI (38)
- NY(43)
80 I
35‘:
a) an
E3
E 40 .1
20 a -0.20 40.20 '0'”
TOP 1/3 MIDDLE 1/3 BOTTOM 1/3

Figure 7. Vertical stratum in the water column occupied by adult redside dace in
Michigan and New York. Symbol explanation follows Figure 5.

USE

AVL

USE

AVL

43

 

 

 

 

ﬂ: MI (33)
W M (39)
—_E1: NY (43)
W NY (59)

 

 

 

 

l I I I l
0.2 0.4 0.6 0.8 1

DEPTH (m)

Figure 8. Depth availability and use by adult redside dace in Michigan and New York.
Symbols: vertical bar represents median, box encloses 50% of the distribution, and
horizontal bars mark the range. Sample sizes are given in parentheses.

 

 

 

 

 

 

 

 

 

 

 

 

 

USE -l i F MIC“)
MI 39
AVL W ‘ ’
r NY(43)
USE -l 1 ‘ﬁ
, , NY(59)
AVL /////////,%/////////2
I I l I
0.1 0.2 0.3 0.4
VELOCITY (m see")

Figure 9. Mean column velocity availability and use by adult redside dace in Michigan and
New York. Symbol explanation follows Figure 8.

45
I also evaluated four characteristics of the fast water areas that were nearest to the

focal animal observation points. Designation of a fast water area was somewhat
subjective; the requirements included a sudden and signiﬁcant increase in velocity along a
cross-channel transect, displayed by the formation of shear lines on the stream surface.
Redside dace in New York were typically found farther ﬁom fast water areas than ﬁsh in
Seeley Drain (P < 0.025; medians of 5.70 m and 3.60 m, respectively), a result expected
ﬁ'om the greater width of the pools in New York than in Michigan. Like other pool
dimension measurements, fast water areas were also wider (P < 0.001) in the New York
streams (3.80 m) compared to Seeley Drain (1.00 m). However, median fast water depths
(0.12 m for Michigan and New York) and velocities (0.256 m 8'1 MI, 0.308 m s'lNY)
were not found to be signiﬁcantly diﬁ‘erent (P > 0.20 and P > 0.40) between the sites.
Similar to instream habitat attributes, stream bank and riparian characteristics also
differed greedy between Seeley Drain and the New York streams. Riparian vegetation
along the pools in Seeley Drain was dense and close to the stream surface, with more than
75% of the points sampled having some form of structure within 1 m above the water.
There was no diﬂ‘erence between frequencies of use and the availability of locations
associated with overhanging structure within 1 m of the stream surface in Seeley Drain.
Riparian structure along Seeley Drain was composed mostly of tall fescue and Short
willow trees that hung in and over the stream. The New York streams possessed
signiﬁcantly fewer points associated with overhanging structures that were sampled for
availability compared to Seeley Drain (42%; P < 0.005). Overhanging structure along the

New York streams was represented by woody shrubs and a variety of small trees.

46
However, at these sites New York redside dace showed moderate selection of points

beneath overhanging structure, using them more often than expected based on random
choice (72%; P < 0.025).

Percent vertical canopy coverage high above the streams was also different
between Seeley Drain and the New York sites (P < 0.001), but coverage associated with
points of use were similar to proportions available (P > 0.10) indicating neutral selection
of canopy density. Overall, the New York streams were located in regions of thicker
forest; however, the larger New York streams created a wider opening in the tree
coverage than Seeley Drain where smaller trees provided a dense canopy at most
locations. For the New York streams, percent canopy coverage of less than 50% made up
more than 80% of the observations associated with habitat use and availability. In
contrast, coverage greater than 50% characterized 70 - 80% of the use and availability
observations in Seeley Drain.

General Behavior. - I observed remarkably few interactions among redside dace or
between redside dace and other species. Foraging dominated redside dace activities in
Michigan and New York. During 35 out of 40 focal animal observations (88%) that
lasted for the entire 10 min observation period, redside dace were observed holding
stationary positions from which they intermittently struck at items in the water column or
at the surface. They did not exhibit the aggressive, territorial behaviors that I observed
simultaneously in blacknose dace, common shiners, and creek chubs. Individuals of these
other species frequently came into physical contact with each other while attempting to

maintain or regain positions in the pools. Lunges or even a brief chase often followed

47
these contacts. The other minnow species often held positions in the lower halfof the

water column, below the positions selected by most redside dace. When a mixed
assemblage was present, the deepest pockets of the pool were usually held by large creek
chubs and white suckers (Catosromus commersonr) that were seldom challenged by the
other ﬁshes. Redside dace did not respond to the locations or actions of other ﬁshes and
the only indication of interaction that I recognized was in competition for particles that
entered the pool, especially items ﬂoating at the surface. Redside dace usually arrived at
ﬂoating particles ﬁrst, though it was not always possible to determine if the item was
successfully eaten. Redside dace were not as successful at securing objects ﬁ'om the water
column. Typically, small creek chubs or blacknose dace reached suspended particles ﬁrst.
Redside dace displayed a characteristic feeding strategy that set them apart from
the other minnows. From a mid-water position, redside dace launched sudden strikes at
ﬂoating food particles, often leaping out of the water with the force of their momentum.
They performed this behavior with greater frequency than other minnows sharing the pool
and more often than they attempted to capture suspended particles. For 17 foraging
redside dace, I quantiﬁed the number of strikes made at the surface compared to the
number of strikes made at suspended items during the 10 min observation period. On
average, the ﬁsh made 1 strike at the surface every 2.5 min, almost four times the
ﬁ'equency of strikes at particles in the water column (1 strike in 10 min). Two individuals
made as many as 8 strikes at the surface during the 10 min period. Overall, redside dace

struck at items once every 2 min and as often as once every 50 sec.

48
Physiology

Metabolic Rate. - Metabolic rates measured in the lab during the ﬁrst experiment
decreased in difference among the four populations as the length of trial periods increased
(Figure 10). In several cases, variability in the data was less than that expected due to
measurement error of the dissolved oxygen meter, so most SE were assigned a value of
0.01 to reﬂect the meter’s sensitivity. Repeated measures AN OVA of the log-transformed
oxygen consumption rates revealed a signiﬁcant population effect (P = 0.034) but no mean
difference between sexes (P > 0.30). Further inspection showed that the Whig Street
Brook population had a signiﬁcantly lower consumption rate than the other groups during
the 20 min trial (P < 0.05); however, there was no detectable difference among
populations in any of the other trial periods. The 50 min trial length appeared to offer the
greatest potential for future repeatability of results among the trial lengths tested, and I
selected it for use in later lab and ﬁeld experiments. Redside dace consumed a mean of
0.114 mg 02 g‘” h'1 during the 50 min trial.

Mean oxygen consumption rates of redside dace tested in the ﬁeld in Michigan and
New York (Figure 11) were not different (Tukey HSD: P > 0.34; 0.130 and 0.143 mg 02
3* h"), but New York ﬁeld rates were found to be signiﬁcantly higher than the rates
measured in the lab (0.114 mg 02 g'” h"; P < 0.005). Michigan ﬁeld rates also tended to
be higher than lab rates but were not shown to be signiﬁcantly different (P > 0.05). The
standard errors shown in Figure 11 were assigned the value assumed for instrument error
(0.01); however, the ANOVA procedure did not account for this adjustment. Assignment

of these SE values negates the difference found between New York ﬁeld and laboratory

49

 

 

 

 

 

 

"-= 0.16 -
ice
on -
E 0.14
g 0.12 -
E 0.10 '
m 0.08 -'
é
o 0.06 a
E ----- FRANCE (7)
0.04 - — — RED HOUSE (8)
:2 RICE (7)
o 0.02- - - -WHIGST(7)
l I l l I
10 20 30 4O 50

TRIAL LENGTH (min)

Figure 10. Mean oxygen consumption rates (:1: 1 SE) by adult redside dace from four
New York streams during trial periods used in the ﬁrst laboratory experiment. Sample
sizes are given in parentheses.

50

 

 

 

 

 

 

 

 

 

 

 

7.1: 0.18 -
ice
5’ 0.16- -
g 014 "'" 14 0C
E ' .. (6) _ _ 10°C
.. 12
cg 0.12" 12 °C ( )
o (28) i
U
"4 1
ti
o 0.08 "
NYLAB NYFIELD MIFIELD

Figure 11. Mean oxygen consumption rates by adult redside dace ﬁom New York and
Michigan during 50 min lab and ﬁeld trials in the ﬁrst experiment. The temperature that
each group was tested at and sample sizes (in parentheses) are given next to the mean
consumption rate. Symbols: horizontal bar represents mean, box encloses 2 SE, and
vertical bars mark range.

5 1
measurements by AN OVA (Figure 11). There was, however, a deﬁnite tendency for

estimates from ﬁeld measurements to range higher than lab rates. Redside dace tested in
the ﬁeld in New York were acclimated to a slightly warmer temperature (14°C) compared
to ﬁsh in the lab (12°C).

I designed a second experiment to explore the importance of temperature on
metabolic rate in which redside dace were acclimated to two additional temperature
treatments, one warmer and one cooler than the ambient temperature used in the ﬁrst
experiment. As in the ﬁrst experiment, I was unable to ﬁnd a difference between sexes (P
> 0.37) or between the four New York populations (P > 0.48) included in the study, but
treatment temperature had a pronounced effect (P < 0.001) on resting routine metabolic
rate (Figure 12). Redside dace acclimated to 6°C had a signiﬁcantly lower mean
consumption rate (0.068 mg 02 g" h"; P < 0.05) than the control group maintained at
12°C (0.123 mg 02 g” h") and the warm treatment group acclimated to 20°C (0.227 mg
02 g'” h"). A comparison of the 12 °C consumption rates between the two experiments
failed to show a signiﬁcant difference (two-tailed t-test, P > 0.20) and provided conﬁdence
in the repeatability of the results.

Ihermal Tolerance. - Acclimation temperature also had a marked effect on the
critical thermal maximum of New York redside dace (Figure 13). The CTM endpoint,
FLE, differed signiﬁcantly among all three temperature treatments (P < 0.05). Variation
ﬁ'om the mean decreased as acclimation temperature increased. Bartlett’s test and
multiple comparisons (Zar, 1984) showed that the variance in CTM of redside dace

acclimated to 6 °C (0.7079) was signiﬁcantly higher (P < 0.005) than redside dace

52

 

 

 

 

 

'7': 0.35 "
e °° 0 3
. 0 a
' 3
E 0.25 -
g 0 20 {i} (31)
. 1
5
E 0.15 -
(16) (28)
,2 0.10 -
8 0.05 e (29)
6°C 12°C 12 °C 20°C
(lst Exp.)
ACCLIMATION TEMPERATURE

Figure 12. Mean oxygen consumption rates of adult redside dace from New York at three
laboratory acclimation temperatures in the second experiment. Laboratory data from the
ﬁrst experiment is provided for comparison. Symbol explanation follows Figure 11.

53

 

33 4 $" (8)

32"
31‘
30“
29"

28 " ﬂ (7)
27 "
26 " (8)

25'
24"

CRITICAL THERMAL MAXIMUM ( C)

 

 

 

6 °C 12 °C 20 °C

ACCLIMATION TEMPERATURE

Figure 13. Mean critical thermal maximum of adult redside dace ﬁ'om New York at three
laboratory acclimation temperatures. Symbol explanation follows Figure 11.

54
acclimated to 20 °C (0.0564). Variance was especially low for the latter group where the

SE was equal to the degree of measurement error in reading the thermometer (0.1 °C). I
detected no differences between the two populations (P > 0.49) used in the experiment. A
single individual in the control treatment died of unknown causes before completion of the
acclimation period.

The order of occurrence of the four other behaviors that I identiﬁed, BSW, ESW,
SUR and OPS, was consistent among temperature treatments (Figure 14). Expression of
the behaviors was more constricted for the 20 °C treatment group, for which the
behavioral events were expressed at temperatures closer to the CTM than in the other
treatments. Only one individual failed to show all of the behaviors by not surfacing. The
occurrence of each behavior was also accompanied by a general increase in the level of
excitement and activity. During CTM tests, ﬁsh swam continuously against the slight
current produced by the aerator. Burst swimming activity (BSW) generally occurred well
before other apparent stress indicators and was the earliest sign of discomfort. Three

individuals died immediately after showing signs of OPS.

Morphology
Morphometrics. - Sample sizes and mean standard lengths of the specimens used in
the morphometric analysis are shown in Table 4. Preliminary viewing of the specimens did
not suggest any obvious differences among the ﬁve groups Of redside dace that I studied
(River Rouge watershed, Michigan; Allegheny River and Susquehanna River watersheds,

New York; Black River‘and Wisconsin River watersheds, Wisconsin). There were,

55

 

31-
29-
27-
25.1
23-1
211
19‘

 

CRHICALIHERMALMAXIMUMiC)

17-1

 

 

 

6° 12° 20°

ACCLIMATION TEMPERATURE

Figure 14. Sequence of behaviors observed during critical thermal maximum tests of adult
redside dace ﬁom New York at three laboratory acclimation temperatures (means :1: 2
SE). OPS = opercular spasms, FLE = ﬁnal loss of equilibrium, SUR = surfacing, ESW =
eratic swimming, BSW = burst swimming. See Methods - Physiology - Thermal
tolerance for a description of each behavior.

56

Table 4. Sample sizes and standard lengths of redside and rosyside dace morphometric
specimens.

1! M L :t E

Wisconsin redside dace

Black River 27 63.9 :1: 2.1

Wisconsin River 18 62.1 i: 0.9
New York redside dace

Allegheny River 23 58.7 :1: 1.7

Susquehanna River 18 57.7 :1: 1.2
Michigan redside dace 28 59.1 i: 1.3

Rosyside dace 30 51.0 :1: 1.7

57
however, clear diﬁ‘erences between redside dace and the rosyside dace. As described by

Page and Burr (1991), the rosyside dace appeared to be a deeper bodied, stockier ﬁsh than
the redside dace. Examination of the scatter plots of landmark points and statistical
analyses conﬁrmed this observation. Overall differences in the mean shape conﬁguration
of rosyside dace compared to the redside dace were most prominent along the dorsal
surface of the form (Figure 15). The trend was for a heightening of the back and head and
a slight deepening of the caudal peduncle of rosyside dace.

A preliminary within group analysis of differences in mean landmark coordinates
(two separate MANOVA’S) was performed to test the decision to pool watersheds within
New York and within Wisconsin prior to an overall test for mean differences among
regions. The mean conﬁgurations of the two New York watershed groups were not
signiﬁcantly diﬂ‘erent (Wilk’s A, F30, 9 = 2.20, P > 0.10), and those groups were pooled for
further comparisons to the other groups of specimens. However, the two \Vrsconsin
watershed groups did prove to be signiﬁcantly different in mean landmark locations (F 30,14
= 6.91, P < 0.0002). Corresponding univariate tests of the individual landmark
coordinates suggested that the groups differed in the mean x-coordinate positions of
points 9, 10, 14, 15, and 16, indicating horizontal shifts in points along the dorsal and
ventral surfaces of the forms. Consequently, the Wisconsin redside dace were not pooled
for ﬁrrther comparison to the other redside dace groups.

Wilk’s test conﬁrmed a signiﬁcant group effect (P 120,435 = 12.18, P < 0.0001) in an
overall test of differences in mean shape conﬁgurations among Michigan, New York, and

the Wisconsin groups of redside dace and the rosyside dace, and corresponding univariate

58

.96 come 0238c e5 .93 :52 52085, - 588$? A8 cot/2 Moe—m - £3035»

56 etc.» 302 .96 sewing «concave Eamon BEBE wen—BE E0304 .2 was ~ 332. e333 8:33 a :o woman—toga wee
“co—«om soon 26: mos—«=0 .33 02938 28 22mm: Sou Sou eon—c 02mg. cow meouﬁawccoo omega Sconce moi—30 .2 oczwﬁ

2,

N“

2

36 62,3. Ease? 8% ensue 8% ensue 8%
a Becca? $er or; .62 5363 ounces

lllllll

59
tests detected signiﬁcant differences among groups for most x- and y-coordinates. There

were fewer signiﬁcant diﬂ‘erences among the redside dace when the analysis excluded
rosyside dace. The redside dace groups were signiﬁcantly different overall (1:90.240 = 6.79,
P < 0.0001) and in the x-coordinates ofpoints 6, 9, 10, 14, and 16 and in the y-
coordinates of points 2 - 7, 13, 15, and 16, despite substantial overlap among the groups
in the coordinate scatters of individual landmarks.

The thin-plate spline analyses provided insight into the scale and magnitude of
shape differences between groups and was useﬁrl in visualizing the directions of landmark
displacement. Linear transformation did not suggest signiﬁcant relationships among the
groups of redside dace. Uniform variation was virtually unrecognizable in the affine
mappings of the thin-plate spline transformations ﬁom the rosyside dace reference
conﬁguration to mean redside dace shape conﬁgurations. A scatter-plot of the uniform x-
and y-factors (Figure 16) showed that uniform variation was primarily in a negative
direction along the x- and y-axes. A principal components analysis of the uniform factors
(T abachnick and Bookstein, 1990) showed that uniform factor y loaded highest on the
ﬁrst of two principal components that described 58% and 42% of the total uniform
variation, respectively. The linear transformation was characterized by a slight tilting of
the thin-plate in which landmarks above and below the baseline displayed anterior, dorso-
ventral displacements that were proportional to their distance from the baseline. The
deformation corresponds to a homogeneous narrowing of the body and head of the

outgroup rosyside dace form in the transformation to the form of redside dace.

 

 

 

 

0.1
a
a A
0.05- x , o.
o O ‘a IO
0 K
>.. O i' 0 O 0 g :8 O O
a: ° 0 °
0 K a
E K ‘ ...“ K ° 8‘ a
<'0.05. ‘ . ’ 0. ‘ K
In. . c “or: It
0 It I a
E O O ‘ . x ‘ ‘x ‘ . O
o -01- . . f o
X K .03
E . I ,. ,
x ,5 ’
.015- ‘0 '. x ' °
K . x
0 ° 4:
-0.2t
o
-0.25 ‘ ‘ ‘ L ‘
-0.2 -0.15 -0.1 -0.05 0 0.05 0.1 0.15
UNIFORMFACI'ORX

Figure 16. Scatter-plot of the uniform factor x- and y-dimensions of the affine
transformation ﬁom the mean rosyside dace starting conﬁguration to individual redside
dace shape conﬁgurations. Symbols: ([1) rosyside dace at (0,0); (0) Michigan redside
dace; (x) New York redside dace; (0) Wisconsin - Black River redside dace; (A)
Wisconsin - Wisconsin River redside dace.

61
Non-uniform shape change had a much greater impact than uniform change on the

shape of the thin-plate splines. Non-uniform variation among the groups of redside dace
and the outgroup was observed as heterogeneous stretching of the thin-plate caused by the
independent movements of separate neighborhoods of landmark points. The Spline
analyses produced a set of 14 principal and partial warps that corresponded to non-
uniform components of shape change for each comparison of the rosyside dace starting
form to individual redside dace specimens. The partial warps were ordered by eigenvalue
in that the ﬁrst partial warp had the largest eigenvalue and corresponded to the smallest
scale shape change. Progressively smaller scale partial warps described variation not
already attributed to larger scale warps.

Overall, the thin-plate spline analysis identiﬁed several non-uniform differences in
shape that distinguished redside dace from rosyside dace, but few distinctive shape
differences were found among the groups of redside dace. Figure 17 and 18 illustrate the
thin-plate spline transformation from the rosyside dace starting form to each of the mean
redside dace shape conﬁgurations by mapping the deformation of a cartesian grid. Total
bending energies of the non-uniform transformation were comparable among all redside
dace groups (New York 0.0093, Michigan 0.0094, Wisconsin - Black River 0.0114,
“ﬁsconsin - Wisconsin River 0.0119) suggesting that the overall degree of bending was
similar.

The most profound difference between redside dace and rosyside dace was
regularly displayed in the largest scale partial warp 14 (Figure 19) though this warp was

not useﬁrl in distinguishing among the groups of redside dace. Partial warp 14 described a

62

deacon—«:3 000: 0330: are.» 302 05 8 :ouﬁammeg 000v 02¢.me

05 88m 8398829: 05. Am damage". 00% 0230: 59:32 05 8 Sundown—.3 000: 0233: 05 Bob :oweEommgb
0.3. 2 ewes—0 092a mo 3:38:80 88%—=15: e5 0:80 06 .«o mﬁobo “6:588 05 Bio: “means: 2E. .m:o38:wm:oo 2.0%
508 003 0330: 9 8395350 use—Sm 003 02932 05 .«o guns—comment 053m 03.— £5 05 mo masque:— 56050 .S 0.5me

m . .m 0 ".0.“ m. .u 0.“. ......... .<
u 12' ....... mum ........ o ...... n ...... e ...... . ...... a“ ...... "3.1..” ..... .
H. ...... f ..... .:.:...: . . ...: . u m m.“ m .::... .Nnn ..... i ...... .
m m: I . m m m m om m
: ...... . ...... . ...... . ...... . ...... . ...... M ...... . ...... M ...... . ...... . ...... m ...... Mrwﬁe ...... . ..... tire... ...... w
Na m A. m m . A. m .V O .n
a ...... e ...... e ...... e ...... e ...... e ...... m ...... o ...... .m ...... . ...... q ...... m ...... . ...... t...” .............. him .......
"a." m Om .u m A I .
....... Bate”-

63

dogﬁoo 0000 00:02 :02: 50:08.
5:088 2:5 . 05w . . . .3 - 20:03; 05 8 50.00% 00 00 >0
of. A“ .0w:M:=%oham—ﬁ MMN::0:¢:83030 02000: :32 0.005 - 20:32? 0.: 9 5083228 0000 020““: 05 Howﬁoaﬁﬂwmmm
5.0:. 0000 02000.. 2 5:83» o:8EwSHuM..=:.:o: 0:0 0:80 05 .3 300.220 00:20:80 05 «2:00 ”minus: 05. goaaswmgo 0:20
. . a: . 0000 02308 05 .«o 5005.889: 02:? 03: :0: 05 .«o 09:30:. 5550 .3 0 S». m

...... m. ..... ...... .. ...... Mus. ...... ..........! ...... . ...... . ...... . ...... . ............. . ...... . ..... . ....... .
m ...... .93... ...... ..... ............. m m 0H M m .. . .. m. m m
m ma. m . v0. . m E. ...... m ...... w ...... . ...... ...... . ...... ..:....“ ...... w ...... w... ..... .
m m m . . H m . . .. . .. ma om .. .. N. . ..

ﬂ. ...... . ...... ...... m ...... ... ...... . ...... . ...... m ...... ...... u . . . . .

.. o. H .m .. W M m .0... ....... ..... on ..... a... ..... H .0

..HH ...... ...... ...... ...... ...... ... ...... ............. m ...... w. .... m. .va ..

...... m ...... m ...... m. ..... ..:..». ...... w m m. . m n .... ..... ... ..... .. ......

m ..... . ............. . ............. ......w. ...... h ............ ......
....... 9
. . . . . m .0 .. . . m H . . . H. .. . ........... .
H. ...... 1...... ...... ...w..1........mH ..... . ...... . ..... Wm... ...... ...... .... .. .. 9.. . <
.mH. . H m .. m .. m . .. AH . ...... .. ...... ....N... ..... . ...... .
N... ...... . ...... . ...... ...... W ...... m ...... . ...... m. ...... . ...... . ...... . ..... .. ...... . ..... ...wm... . .. m . m
H. q. m .. . W . .. .. . W . . m ..... on. ..... .. . .. ...... 0
:3. ...... .. ...... m. ...... .. ...... m ...... m ...... . ...... . ...... .. ...... w ..... .m ...... m 0.. m.. a. o. H.
m m .. m m . m o. m m m m . .:......N............. ..... ...? ...... ._
....... ..:.-......v...-..:.......:.....333...a$33...-..rno...‘..uuscn¢cs:...-..:.-...uuac...:ua.su..o.u.w.uac..am...:.uaM m. .m

 

 

 

 

0.04
002,
o
8 K
a o ..0 ‘ ‘
3 Or A o & O ‘5: { g:
x <9 o 9:.“
”3 o ‘9- 0 Id
3 0 § (“‘35 .
>2 . 0 ' ooo‘ﬁ’itx 2 “"a
O ‘ O ‘ .
-o.02. x . 13%“. "
9“ K . " . o
x
o
x
3 o
A
-0. ' . . .
-0.04 -0.02 O 0.02 0.04
X-AXISLOADINGS

Figure 19. Distribution of loadings on the x- and y-dimensions of partial warp 14 for the
thin-plate spline transformation from the mean rosyside dace starting conﬁguration to
individual redside dace shape conﬁgurations. Symbols: (D) rosyside dace at (0,0); (0)
Michigan redside dace; (x) New York redside dace; (0) Wisconsin - Black River redside
dace; (A) Wisconsin - Wisconsin River redside dace.

65
transformation characterized by dorso-ventral depression of the trunk of rosyside dace

through downward shifts at points 9 and 16, and corresponding upward shifts at the ends
of the baseline, points 1 and 12. The displacement of the central landmarks relative to the
anterior and posterior ends of the baseline is a common feature of the largest scale
principal warp (Swiderski, 1993; Bookstein, 1991). Large-scale transformations
contributed most to the shape of the splines (high magnitude) but had low bending
energies (< 10% of the total) because their effects were distributed over broad regions
(low eigenvalues). Depression along the dorsal surface was consistent among the redside
dace and suggested an overall decrease in the height of the rosyside dace back relative to
redside dace, a difference apparent in the outlines of the mean shape conﬁgurations
(Figure 15) and in the total spline mappings (Figure 17 and 18).

Large-scale features of shape change were the most distinguishing shape
characteristic among the redside dace. Partial warp 11 (Figure 20) described primarily
dorso-ventral shifts along the ventral surface with high y-axis loadings on point 16, 14,
and 2. This transformation suggested that Wisconsin redside dace ﬁ'om both watersheds
had a shallower keel than New York or Michigan redside dace relative to the rosyside
dace starting form; however, the scale of these shape diﬂ‘erences is small and the scatters
of all redside dace groups overlap. Michigan redside dace were more similar to the
outgroup in this deformation than were the other redside dace groups.

Local changes in the positions of one or two points contributed the most bending
energy (> 7 5% of the total) to the thin-plate spline transformation from rosyside to redside
dace. Partial warp 4 (Figure 21) was consistently of highest bending energy (> 45% of the

total) and was of second highest magnitude aﬁer partial warp l4. Partial warp 4 was

 

 

 

 

0.02
0.015-
0.01 ..
a
x e
I ‘ ‘ 0°
8 0.005. ‘ o "
a ’ o o e o
x a A x‘ 6
a ‘ &: ea ‘95 .3 o 0
3 O - U. .3“ .
m ‘3 x a“ x: 3" ..
a K o ‘1‘. x f 0
>" 0.005- e o '5 a“ x
o x ‘
K o " x xx
'59 " o
-0.01 - o 0 0° It
0
-o_01_ . J . . . .
-0.02 -0.015 -0.01 -0.005 0 0.005 0.01 0.015
X-AXIS LOADINGS

Figure 20. Distribution of loadings on the x- and y-dimensions of partial warp 11 for the
thin-plate spline transformation from the mean rosyside dace starting conﬁguration to
individual redside dace shape conﬁgurations. Symbols: ([1) rosyside dace at (0,0); (0)
Michigan redside dace; (x) New York redside dace; (0) Wisconsin - Black River redside
dace; (A) Msconsin - Wisconsin River redside dace.

67

 

 

 

 

0.01
0.005, e
x
o 0 ’0" x
a :- .. :o
a O X K .‘ k 0 X ‘
ﬂ :3: - " .
3 OH '°"' “3% ogb"’i. a"
m x . x I o " x o '
a o X 2 ": K o
' e x0 O o ‘ .
>“ ‘ o A 0 g . “‘0
o x ‘ x
-0.005. e l
x
-0.01 . 1 .
0 0.005 0.01 0.015 0.02
X-AXISLOADINGS

Figure 21. Distribution of loadings on the x- and y-dimensions of partial warp 4 for the
thin-plate spline transformation ﬁ'om the mean rosyside dace starting conﬁguration to
individual redside dace shape conﬁgurations. Symbols: ([1) rosyside dace at (0,0); (0)
Michigan redside dace; (x) New York redside dace; (0) Wisconsin - Black River redside
dace; (A) Wisconsin - Wisconsin River redside dace.

68
driven by the horizontal, posterior movement of point 5, the posterior edge of the eye.

High loadings on the horizontal axis indicated that redside dace possessed a signiﬁcantly
narrower intraorbital diameter than rosyside dace, though no difference was indicated
among groups of redside dace specimens.

Rosyside dace were further differentiated ﬁom redside dace through a second,
small-scale transformation, partial warp 2. This warp described inward movements in
landmarks located along the operculum and base of the skull (points 6 and 8) of redside
dace relative to rosyside dace. Anterior displacement of these landmarks was most
pronounced for Michigan redside dace and least for Wisconsin - Wisconsin River redside
dace, though distributions of the two group’s loadings overlapped. Additional partial
warps depicted inward, posterior movement of the anterior naris (partial warp 1, point 3)
and the dorso-ventral displacement of landmarks identifying the caudal peduncle (warp 5
and 10, points 11, 13, and 9). Collectively, these shifts implied a decrease in relative head
dimensions and in the width of the caudal peduncle of the redside dace compared to the
rosyside dace. Redside dace did not differ appreciably in their loadings on these partial
warps.

Meristics. - There was no discernible pattern to minor variation in dorsal and anal
ﬁn ray counts. Redside dace from Michigan, New York, and Wisconsin consistently
possessed 8 separate dorsal ﬁn rays (Table 5); only 2 individuals ﬁ'om New York had 7
dorsal ﬁn rays. Counts of anal ﬁn rays were more variable (Table 6), though most

individuals (>86%) had 9 anal rays.

69

Table 5. Frequency distribution of dorsal ﬁn ray counts for redside dace morphometric
specimens from ﬁve locations.

 

Number of Dorsal Fin Rays
n 7 8

Wisconsin

Black R. 26 26

Wisconsin R. 22 22
New York

Allegheny R. 36 36

Susquehanna R. 17 2 15

Michigan 43 43

70

Table 6. Frequency distribution of anal ﬁn ray counts for redside dace morphometric
specimens ﬁ'om ﬁve locations.

 

Number of Anal Fin Rays
n 8 9 10

Wisconsin

Black R. 27 23 4

Wisconsin R. 19 18 1
New York

Allegheny R 42 34 2

Susquehanna R. 17 3 14

Michigan 43 3 37 3

71
Diﬂ‘erences among the redside dace were evident ﬁ'om lateral line scale counts

(Figure 22). Michigan redside dace had a lower median scale count (63) than either of the
New York watershed groups (Allegheny Susquehanna Rivers = 65) or “frsconsin groups
(Black River = 65; Wisconsin River = 66). A x2 contingency table analysis conﬁrmed that
Michigan had signiﬁcantly fewer lateral line scales (P < 0.001) than New York and

Wisconsin (watersheds pooled within state).

72

 

72

     

I|IIIIIHII

71

70

 

69

IIIIH‘H“

 

IIIIIIIll|HHIIIHIHIIIIIIHHH

    
  
 
  
 
    
    

IIIIIIIIIIlllllllllllllllIlIlIlIllllllllIIIIHIIIIIIIIIIllllllllllllllillll

/x/ /}//z 7, 71” 1,. 8

‘ ‘n’.

llllllllllllllllllllllll

 

////////// W
\O
—
I:.I.: I. IIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIHIIIIIIIIIIIIHIHIII
L}/’/,/////’//' a
.IIIIIIIIIIIIIIIIHHHIIIIIIIIIIIIIIIIIIIIIIIIII
11111 ///////////// 8

 

 

 

 

 

 

 

[
8
l l
\O <- N
113513003115

 

NUMBER OF SCALES

(16)

(23)

(32)

(27)

CJ MICHIGAN I NY, ALLEGHENY R. I NY, SUSQUEHANNA R. 3W1, BLACK R E WI, WISCONSIN R.
(14)

Figure 22. Frequency distribution of lateral line scale counts for redside dace morphometric specimens ﬁ'om ﬁve regions. Sample

sizes are given in parenthesis.

DISCUSSION

Behavior

Redside dace in Mchigan and New York used depth and current velocity
attributes of microhabitat in similar ways, but differed in their use of other microhabitat
attributes. Most individuals occupied mid-water positions in the deepest parts of pools.
From this position, redside dace struck repeatedly at items ﬂoating on or hovering above
the stream surface. They struck less frequently at objects suspended in the water column.
It was diﬁcult to generalize about the importance of substrate type or instream cover to
foraging redside dace because the substrates and cover types available in Seeley Drain and
the New York streams were different and redside dace use of these attributes differed
accordingly.

In Seeley Drain, instream visual cover was the only type of abundant instream
cover, and its use by redside dace was only slightly less than that expected ﬁ'om random
choice. Use of no cover was recorded for most observations. It is also possible that
depth was being used as a form of cover ﬁ'om avian piscivores by redside dace in Seeley
Drain and New York (Baltz and Moyle, 1984), but I was unable to test this hypothesis
with the data I collected. In the New York streams, boulders provided velocity shelters
that received a disproportionate amount of use by redside dace. Velocity shelters were the
only form of instream cover abundant in the New York streams. Structure overhanging

the stream was used in the New York streams, but the importance of overhanging

73

74
structure was less obvious in Seeley Drain where it was abundant over all of the pools

surveyed.

Diﬂ‘erences in the availability of habitat attributes have been shown to limit the
degree to which habitat use information may be reliably transferred from one site to
another (Barrett and Maughan, 1994). The importance of a habitat attribute may be
masked by its general abundance or attributes deemed important in one locality may not be
available in others. A comparative approach involving artiﬁcial manipulation of
microhabitat components would be useful to clarify the signiﬁcance of the patterns of
microhabitat use that I observed. It would be valuable to know if redside dace in Seeley
Drain would make use of velocity shelters if they were made available or would respond
to the selective removal of some overhanging cover. Velocity shelter did appear
important to rosyside dace that were observed to aggregate in eddies or depositional areas
next to fast water during high ﬂow conditions (Freeman, 1990). Redside dace did not
appear to make special use of areas adjacent to fast water (depositional areas at the heads
of pools), and there was no identiﬁable relationship between the distribution of redside
dace within pools and the proximity of fast water areas.

My observations of microhabitat use are in close agreement with the detailed
observations of Koster (193 9) who noted that redside dace primarily inhabited pools and
devoted the largest portion of their activities to foraging at or near the surface. It should
also be emphasized that the use and availability of microhabitat characteristics that I
measured are speciﬁc to mid-summer, daylight conditions and may differ at night and

during other parts of the year. Grossman and Freeman (1987) found that rosyside dace

75
occupied more benthic, low velocity positions during winter months. Also, the attributes

and signiﬁcance of spawning and overwintering microhabitats were not evaluated in this
study and may play a crucial role in determining redside dace distribution. Gravels
hypothesized to be necessary for spawning (Koster, 1939) may be limited in the sand-
dominated substrates of Seeley Drain, and increased siltation may pose an especially
serious threat to the reproductive success of redside dace.

The patterns of redside dace microhabitat use that I described appeared to
correspond closely to selective tradeoﬁs associated with the ﬁsh’s distinctive mode of
feeding, the activity that dominated nearly all observation periods. Natural selection
should favor individuals that maximize their ﬁtness through the selection of high quality
microhabitats and eﬁicient foraging strategies (Hill and Grossman, 1993). Through their
use of mid-water positions and exploitation of aerial, terrestrial prey, redside dace incurred
higher energetic costs associated with swimming at higher velocities (F acey and
Grossman, 1990) and perhaps increased their risk of predation by other aquatic and aerial
predators compared with deeper positions. However, at mid-water positions redside dace
could increase their net energy intake by reducing their distance ﬁom potential food items
and could increase their chances at detecting and capturing prey compared to deeper
positions (W ilzbach, 1985). In another sense, a mid-water position would reduce the risk
of predation by avian predators compared to a near-surface position that would be closer
to prey. Piscivorous ﬁshes may also present a threat to foraging redside dace; brook and
brown trout were observed at the New York study sites and brown trout are known to co-

occur with some redside dace populations in Michigan. Redside dace were observed to

76
form small aggregations when feeding, similar to rosyside dace (Freeman, 1990).

Aggregative social behaviors may oﬁ‘er a defense against other ﬁsh predators (Moyle and
Cech, 1988).

Redside dace may reduce direct competition for food with other ﬁshes by
displaying such a specialized foraging strategy (Daniels and “Wisniewski, 1994). A
correspondence between net energy intake and microhabitat use has been shown for other
ﬁshes inhabiting stream water-column microhabitats (Hill and Grossman, 1993; Hughes
and Dill, 1990). A bioenergetic model constructed by Hill and Grossman (1993) predicted
microhabitat use (focal velocities) for rosyside dace with 75% success and found that prey
capture success was the most important model component. They showed that the
energetic costs incurred by holding position at a given velocity was much less than the
energy available at that velocity, indicating that the beneﬁts of the position outweighed the
costs. In the same way, redside dace appear to balance the costs of swimming, predation
risk, and perhaps decreased food availability against the beneﬁts of increased prey capture
success and reduced competition. The redside dace foraging strategy appears to rely
heavily on visual accuity, a hypothesis supported by the uncommon observation of redside
dace in streams of low water clarity. Increased turbidities in the streams inhabited by

redside dace may pose a serious threat to redside dace foraging abilities and should be

guarded against by appropriate management efforts.

Physioloy
Metabolic rate. - Redside dace displayed a clear metabolic response to trial period

lengths and acclimation temperatures. Estimates of oxygen consumption rate became

77
more stable as the duration of the test increased. Presumably, test ﬁsh became more

accustomed to the change in water quality that accompanied the interruption of ﬂow. One
of the disadvantages of static respirometry methods that employ the sealed vessel
technique is the inability to maintain constant conditions during a trial (Winberg, 1956).
Subtle changes in temperature and in the concentrations of dissolved gases and ions in the
respirometer may elicit a physiological response in test ﬁsh. The activity levels of test ﬁsh
were also beyond control and varied to a minor extent among individuals. These
criticisms make the sealed vessel method less attractive as an estimator of standard
(resting) metabolic rate, though the procedure was a dependable method for measuring
resting routine rates for comparative purposes (Cech, 1990). The rates that I measured
were expected to overestimate standard metabolism and to include some variability
associated with spontaneous activity. Observation showed that such activity was minimal.
The rate of metabolism is a product of the environmental conditions that ﬁsh
encounter as well as their evolutionary history (Fry, 1971). The oxygen consumption
rates that I recorded for redside dace tended to be higher than rates determined for other
cyprinids using similar methodology (Figure 23). Central stoneroller and common shiner,
minnows that were abundant in the New York streams where I studied redside dace, had
similar oxygen consumption rates in a study by Clausen (1936). In addition, metabolic
rates measured for rosyside dace at 15°C (F acey, 1987) were identical to redside dace
tested at 14°C in the ﬁeld. Similar rates were also recorded for the plains minnow,
Hybognathus placitus, which has a more midwestern distribution but reaches similar

dimensions as the redside dace (Bryan et al., 1984). Minnows that had the lowest rates

78

.85: 25.53 326m Eu 83.382: maﬁa .23 32088 A v: ..m was 83. cage—598 5mg .mm 83E

.3338 o 3 £58 :oqumcoo 258.3 noun—2355 3 33:58 v.88 28:8!

 

«we ..:.: 53m 3o 93 ”no and v3 388 ”$535

82 ..1 so :80 NS 85 5c 8d 36 ”33.28885

32 .836 m3 3328 $33

$2 ..aagm “I. Rd ”:5 :.o §§a§~§8¢~

82 ...aaﬁoo 3... So 3.822% 3632.5»:

82 ..zaﬁﬁﬁooz 2.0 Ba 35 “.3239: 883 3.6
82 ...30 Baggage. «3 and N3 3&8 3.55
$2 as: Ed «no Ed e§a§§§e§5
#33885 85 3d 2... 5o assets» 2.32855

3.2 .3: as b..— NS 32% U
82 £38525ch Ed N3 86 .328 3.5.28
82 .5820 Ed 2:358 §.8§so
3 mm _ om _ om _ m g

5V oﬁsieﬁ 8€e=8<

79
compared to redside dace, Sacramento squawﬁsh Prychocheilus grandis (Cech et al.,

1990) and the tui chub Gila bicolor (McClanahan et al., 1986), have western distributions;
the squawﬁsh occupies warmer environments and is much larger than the redside dace.

Redside dace showed a signiﬁcant, steady increase in metabolic rate with
acclimation temperature in the second experiment. The upward trend in intensity
appeared to explain the differences between New York ﬁeld and lab results observed in
the ﬁrst experiment. Considered together, the results of the two experiments showed that
the mean resting routine metabolic rates of redside dace ﬁt a predictable exponential
function:

y = 0.0416e°-°“"‘ (r2 = 0.993)

where y = oxygen consumption rate (mg g‘” h") and x = acclimation temperature (°C;
Figure 24). Many researchers have demonstrated the inﬂuence of temperature on standard
metabolic rates (Ege and Krogh, 1914; Brett, 1964; Beamish, 1964b, and Mookherji,
1964; Cech et al., 1979, 1985, 1990; Kaufman and Wieser, 1992; Smirnov et al., 1987;
Konstantinov et al. 1990). The general relationship for temperate ﬁshes was synthesized
by Ege and Krogh (1914) who developed the “standar ” curve from static respirometry
data for temperature acclimated goldﬁsh, Carawius auratus. Winberg (1956) found that
the rate of standard metabolism of many ﬁshes followed a similar trend, though the
standard curve has been shown to oﬁ‘er a poor approximation of actual oxygen
consumption rates for some species (Fry, 1947, 1971; Brett and Groves, 1979). The
resting routine metabolic rates that I measured for redside dace showed close

correspondance to those predicted by the goldﬁsh curve (Figure 24). The standard curve

80

 

0.2 ‘

0.15 '

.0
...-s
l

0.05 '-

 

OXYGEN CONSUMPTION (mg g" 5‘ )

 

 

 

TEMPERATURE (C°)

Figure 24. Redside dace mean oxygen consumption rates in relation to Krogh’s “standard
curve” (dotted line) adapted from Winberg (1956). New York data are indicated by the
open circles and solid line, Michigan by the closed circle.

81
(Ege and Krogh, 1914) has a mean Q10 (increase in rate with a 10°C increase in

temperature) of 2.0 compared to a mean Q10 of 2.4 for redside dace. The metabolic rates
of redside dace were elevated above the standard curve but ﬁt within the 2.0 - 2.5 range
reported for ﬁ'eshwater ﬁshes (Fry, 1971; Brett and Groves, 1979).

Redside dace tested in the ﬁeld in Michigan showed a signiﬁcantly higher
metabolic rate compared to the value predicted for New York ﬁsh at the same
temperature (Figure 24). The Michigan mean consumption rate (0.13 mg g" h") was
beyond the 95% conﬁdence interval around the estimated rate of New York redside dace

at 10 °C (Y, = 0.09, 95% CI: 0.09 - 0.10). However, in comparing standard errors the

diﬂ‘erence is not great. I was restricted from testing more Michigan redside dace by
Michigan Department of Natural Resources policy regarding the handling of protected
species. Therefore, I did not have the opportunity to collect oxygen consumption data for
Michigan redside dace at other temperatures, so it is difﬁcult to generalize about
physiological differences in metabolism between redside dace in Michigan and New York.
The relatively higher rate for Michigan ﬁsh at 10 °C may indicate a true difference between
the groups, suggesting that redside dace in Seeley Drain are adapted to a different set of
environmental conditions than redside dace in New York or that redside dace in Seeley
Drain are stressed by an unmeasured environmental factor. It is unlikely that seasonal
inﬂuences separate ﬁom temperature were responsible for this difference because ﬁeld
experiments in Seeley Drain and New York were not more than one month apart and

occurred well outside of the reported spawning season of redside dace in either state.

82
Static respirometry methods worked well in the lab and in the ﬁeld. Although

metabolic rates have been estimated by other techniques at ﬁeld sites (F eldmeth and
Jenkins, 1973 ), few studies have utilized ﬁeld respirometry methods and equipment that
were nearly identical to those used in the lab (Mukhamedova, 1978). Field techniques
allowed the testing of naturally acclimatized ﬁsh in their native stream water, avoiding the
handling stress and lengthy acclimation periods necessary for ﬁsh transferred to lab
conditions. The method also eliminated the expense and eﬂ'ort required to maintain ﬁsh in
the lab. I was able to control most signiﬁcant external factors to the same degree as in lab
experiments (with the exception of a few rainy days), and the method required a minimum
of equipment and preparation. Field respirometry methods should be considered as a
credible alternative to lab approaches for simple experimental designs.

Thermal tolerance - Increasing acclimation temperatures produced a signiﬁcant
rise in redside dace CTM but at a lower rate than the increase observed in standard
metabolic rates. Mean CTM showed a 1.3 rate of increase from 6 - 20 °C acclimation
temperatures (1.2 for each 10 °C increase in acclimation temperature). It is not possible to
predict the relationship of Michigan redside dace CTM to the data measured for New
York redside dace without a comprehensive, detailed base of ambient monitoring data for
the New York and Michigan streams. I would expect redside dace from Michigan to have
higher thermal tolerances if environmental conditions in Seeley Drain proved to be more
variable than conditions in Rice Brook. Feminella and Matthews (1984) showed that
populations of orangethroat darter, Etheostoma spectabile, ﬁ'om streams with wide

temperature ﬂuctuations exhibited higher CTM than darters in spring-fed streams with

83
homogeneous temperature regimes. This prediction assumes that redside dace in Seeley

Drain are under no other forms of stress and are comparable in condition to redside dace
ﬁom New York.

The CTM of redside dace was similar though consistently higher than maxima
determined for other Cyprinidae (Figure 25), many of which are also found in the New
York streams studied. The fact that Clinostomus CTM ranged higher than the values
recorded for most minnows was in part explained by diﬁ‘erences in the experimental
endpoints chosen. Kowalski et al. (1987) used the onset of muscular spasms to mark
CTM, and Scott (1987) chose initial rather than ﬁnal loss of equilibrium, the endpoint that
I used. Muscle spasms and initial loss of equilibrium corresponded to the BSW behavior
that I recorded for redside dace and occurred at lower temperatures than the ﬁnal loss of
equilibrium (Figure 14). The two minnows that displayed a CTM higher than that of
redside dace, red shiner erinella lutrensis (Matthews, 1986) and tui chub Gila bicolor
(McClanahan et al., 1986), have western distributions and red shiner inhabit warmer
environments.

The measurement of thermal tolerances has been critical to understanding the
temperature requirements and preferences of different ﬁsh species and has been used to
model ﬁsh response to various environmental perturbations (Fry, 1971). Upper CTM is
regarded as an acute measure of thermal tolerance that has been shown to overestimate
the upper bounds of the “zone of tolerance,” deﬁned by Fry (1947) as the upper incipient
lethal temperature (UILT), where 50% mortality occurs. CTM places an upper bound on

the “zone of resistance,” where mortality is an unavoidable function of exposure time and

84

2.2 ...a a €382
$2 an a amass.
52 8%

£2 .a .0 amazes
$2 ..a .0 €352
$2 ..:. a £253.
$2 a? no 33333—
32 ..E a 8.15602
82 nausea:

33m 3305

EIlowaImlumoll>:._.

.monmm 2:296 1238 .89 mcouuciuoaov 8:838 3:55 182.5 .mm 033m

 

v. _ m oSoESoo 35:35::
a. _ m Seance SEEUEEN
man man mg 023M252» 35.8.:
e. _ m 338: aﬁeiuoﬁi
w. _ m 2533 wage?
edm zoMomESE 25982
a. _ m “.3358 §=§Q
Non eém mam Manges .8363 5.5
3m 2.39:5 §o=t§b
odm mSN 0mm “SEES? §Eoﬁo£~b
— _ p _ p b L
on 8 2 o 835

Aoov 2383th nom§==oo<

85
temperature. That mortality quickly follows CTM was evident in this study by the death

of three test ﬁsh immediately after showing signs of OPS.

Lethal temperature estimates, including CTM, may provide a reasonable
approximation of the optimum growth temperatures of temperate ﬁshes, though it is
acknowledged to be less accurate for this purpose than measures of ﬁnal preferenda
(Jobling, 1983; Reynolds and Casterlin, 1979). I applied the CTM value for redside dace
acclimated to 20 ’C (32.6) to two linear regression equations developed by Jobling (1983)
ﬁ'om a compilation of values reported in the literature that related lethal temperature to
growth optima (r2 = 0.86) and to ﬁnal preferenda (r2 = 0.88). The predicted growth
optimum for redside dace was 24.7 °C, almost identical to the predicted preferendum of
24.5 0C. These estimates are in the range of values measured directly for other minnows.
Cherry et al. (1977) reported ﬁnal preferenda of 26.2 °C for central stoneroller and 26°C
for rosyface shiner (Notropis rubellus). I have successfully acclimated a small number of
redside dace to temperatures approaching 23 °C with no visible signs of stress, though no
observations pertaining to tolerance at higher temperatures were made. The predicted
optimum growth temperature and ﬁnal preferendum are beyond the highest water
temperatures measured at the study sites in New York during July and were above
measurements made at the same sites in June by other invesitigators in previous years (R.
Daniels, unpublished data). However, redside dace are known to inhabit streams in
western New York that displayed August temperatures that were warmer than the
predicted growth optimum and ﬁnal preferendum. The predictions were close to the high

temperature recorded in Seeley Drain during the present study (24.4 °C, Table 2). This

86
was the highest temperature recorded in Seeley Drain during the past three years (T.

Coon, unpublished data). This suggests that redside dace in Seeley Drain and the New
York study streams were not challenged by extreme temperatures during the study period;
however, the temperature optima of many temperate ﬁshes have been shown to be near
the upper bounds of the zone of tolerance, the incipient lethal temperature (Fry, 1971).
Therefore, management eﬂ‘orts should be sensitive to changes in the stream ecosystem that
result in increased ambient temperatures, especially in streams where summer

temperatures may already surpass growth optima.

Morphology

Morphometric analyses employing thin-plate splines have been used to describe the
evolution of morphological structures among taxonomic groups (Swiderski, 1993;
Tabachnick and Bookstein, 1990) and to model the complex integration of ontogenetic
processes (Zelditch et al., 1992). The method was also useful in the present study to
clarify the morphological similarity of external body shapes among four groups of redside
dace in relation to a congener, the rosyside dace. Despite several signiﬁcant diﬁ‘erences
suggested by a MANOVA of shape coordinates, the thin-plate spline analysis suggested
only minor diﬂ‘erences in external body shape among groups of redside dace from
Michigan, New York, and Wisconsin; in general, variation in the geometric shape
descriptors was at least as great within as among groups. Diﬂ‘erences in mean body shape
that were identiﬁed among the redside dace were minute in comparison to differences in

shape between redside and rosyside dace and support only tentative conclusions about

87
patterns of morphological differentiation among geographically separate populations of

Clinostomus elongatus.

The redside dace differed from the rosyside dace in two respects. First, redside
dace demonstrated an overall body shape that was less robust than that of the rosyside
dace. This included both uniform and non-uniform dorso-ventral depression of the redside
dace form in the trunk and head, indicated by loadings on the uniform y-factor, on the
largest-scale partial warp, and on several intermediate and smaller-scale partial warps.
Second, a highly localized shape diﬁ‘erence showed that rosyside dace had a
proportionately wider intraorbital diameter than redside dace. This posterior widening of
the eye appeared to follow the outward shiﬁs observed in other regions of the head of
rosyside dace relative to redside dace.

The “ﬁsconsin redside dace specimens appeared to diverge from the other groups
and from rosyside dace in the largest number of morphometric characteristics. Wisconsin
- Black River redside dace differed slightly ﬁ'om Michigan redside dace in large and small-
scale partial warps (warps 10 and 2) while Vlfrsconsin River specimens diﬁ‘ered ﬁ'om
Michigan redside dace in partial warps 11 and 10. In two cases (warps 11 and 10), the
Michigan form deviated in the direction of the rosyside dace mean conﬁguration. Mean
shape outlines (Figure 15) also implied that shape similarities between Michigan redside
dace and rosyside dace were greater than were similarities between rosyside dace and
either Wisconsin group of specimens. New York redside dace were intermediate in most

respects. The spline mappings (Figure 17 and 18) suggested that the Michigan mean form

88
was deeper than either \Vrsconsin form; however, the high degree of intrasample variation

and intersample overlap even at this scale was demonstrated in Figure 19.

The ﬁn ray and lateral line scale counts that I made in this study fell within ranges
reported for the redside dace by Page and Burr (1991), Smith (1987), and Becker (1983).
Page and Burr (1991) described a range of 59 - 75 lateral line scales for redside dace
compared to a range of 60 - 72 counted in the present study. Counts of lateral line scales
also suggested a relationship in which Michigan specimens differed ﬁom the Wisconsin
redside dace. Michigan specimens had the fewest number of scales (median = 63), a
difference that, like body shape, was in the direction of scale counts presented for the
rosyside dace (43 - 57; Page and Burr, 1991). Numbers of dorsal and anal ﬁn rays were
not different among redside dace groups.

Collectively, the morphological evidence showed that differences among the
redside dace were rrrinute, but suggested that Michigan redside dace were most similar to
redside dace from New York and least similar to redside dace from Wisconsin, especially
ﬁ'om specimens representing the Black River drainage. Also, Michigan redside dace
exhibited a morphology that was more similar to that held by rosyside dace than were the
morphological characteristics exhibited by the other redside dace.

Results of the morphological comparisons described relationships among
Clinostomus that may have important evolutionary implications for the description of
post-glacial recolonization of the Great Lakes Basin by redside dace. In the absence of
parallel or convergent evolution, the results suggested that Michigan redside dace

displayed a more ancestral morphology than the New York or “fisconsin redside dace.

89
This hypothesis presumes that rosyside dace may be considered an approximation of the

ancestral form in that their external morphology has been inﬂuenced minimally by derived
characteristics. In addition, the hypothesis also assumes that the diﬂ‘erences in
morphology that I measured were genetic in origin and not a function of environmental
differences between the collection localities. Similarities among the groups may therefore
reﬂect a complex history of migration or local evolutionary diﬂ‘erentiation.

The Mississippi drainage is believed to have offered glacial reﬁrgia for numerous
cyprinids, including redside dace (Bailey and Smith, 1981). Migration routes into
deglaciated parts of the basin would have been aﬂ'orded by broad drainage connections as
early as 14,000 years ago in the upper Mississippi and the Illinois River drainages in
Wisconsin and Illinois, the Wabash - Maumee River connection in Michigan, and the Ohio
River drainage in New York. Michigan redside dace probably entered Detroit River
tributaries from the Wabash - Maumee River connection, a route that is believed to have
been open only brieﬂy after the retreat of glacial ice, closing approximately 13,600 years
ago (Bailey and Smith, 1981). Closure of the route would have isolated Michigan
populations of redside dace ﬁ'om eastern and western populations in New York and
Wisconsin. Consequent founder effects (Hartl and Clark, 1989) could have played a role
in maintaining ancestral morphological traits in the Michigan populations.

The biological and historical processes underlying the morphological variation that
I uncovered are unknown, and it would be inappropriate to invoke theories of adaptation
and natural selection or evolution through any other avenue to explain them. The
landmarks that I selected for analysis were arbitrary (Bookstein, 1990). Gross changes in

body dimensions as indicated in this study, for example, are not descriptive of the variation

90
in internal features that were likely responsible for the large-scale deformations. Because

of this, the true homology of the landmarks between forms is left uncertain as is the
structure or trait that might have been under selective pressure. The landmarks that I
selected conform to the deﬁnition of operational homology (Smith, 1990) that is simply a
correspondence of landmark positions from form to form. In this sense, the landmark
analysis at the scale I selected was less useful for phylogenetic or cladistic analyses
because it did not identify individual structures. Howeva, the method was valuable as a
tool for the preliminary testing of hypotheses of overall shape similarity among groups and
serves as a guide to the more detailed study of external and internal anatomical structures.
The morphological similarities described in this study should be compared to the results of
a molecular analysis to clarify the signiﬁcance of the subtle shape diﬂ‘erences among the

redside dace and to corroborate the relationships indicated by this study.

CONCLUSIONS

This study was designed to assess the phenotypic similarity of populations of
redside dace remaining in Michigan to populations in New York and Wisconsin, regional
centers of redside dace abundance. In doing so, this study provided information
concerning the species’ ecological requirements that can be used to guide management
decisions that will aﬂ‘ect the continued survival of redside dace in Michigan. It was not
known if the phenotypic traits selected for study were necessarily genetic or environmental
in origin. Research has shown that phenotypic variation may be attributed to either
inﬂuence (Shields and Underhill, 1993; Foster et al., 1992).

Redside dace in New York were similar to Michigan populations in some aspects
of microhabitat use, despite differences between Michigan and New York in the
availability of microhabitat attributes. In general, redside dace were found at low-velocity,
mid-water positions in the deepest parts of pools and were associated with locations that
possessed overhanging structures within 1 m of the stream surface. Redside dace at both
locations were observed to feed primarily at the surface, oﬁen leaping out of the water in
pursuit of ﬂying insects. Foraging dominated the activities of redside dace, and it was only
during competition for food items that redside dace appeared to interact directly with
other stream ﬁshes. The New York streams maintained a more diverse ﬁsh fauna, yet
observations of interspeciﬁc interactions with redside dace were rare. The Michigan and
New York streams were clear and offered well developed pools but differed markedly in

some aspects of water chemistry.

91

92
Resting routine metabolic rates and critical thermal maxima were shown to

increase for New York redside dace at several acclimation temperatures and followed a
trend that is well established for other minnows and temperate ﬁshes. Metabolic rates
measured at a single temperature for Michigan redside dace were slightly higher than
expected ﬁ'om the relationship determined for New York ﬁsh; however, because we were
prohibited from testing metabolic rates of Michigan redside dace at more than one
temperature, the overall signiﬁcance of this diﬂ‘erence was not clear. The critical thermal
maxima of redside dace in New York suggested an optimum grth temperature that is
near the maximum stream temperature recorded during this study in Seeley Drain.
Redside dace in Seeley Drain could be threatened by minor, prolonged temperature
elevations because optimum grth temperatures and ﬁnal preferenda are near the upper
incipient lethal tolerance limit of most temperate ﬁshes (Fry, 1971).

Only minor diﬂ‘erences in morphology were found among groups of redside dace
from Michigan, New York, and Wisconsin. The diﬂ‘erences suggested that Michigan
specimens were more similar to New York than to Msconsin specimens in body depth,
though overall differences between New York and Msconsin redside dace were extremely
subtle as were differences between watersheds within states. Michigan redside dace
specimens appeared to differ most from specimens collected in the Msconsin - Black
River watershed, the western-most group included in the analysis. The ecological
signiﬁcance of these differences was not ascertained. The scale and number of shape
differences was much greater between redside dace and rosyside dace than among the

groups of redside dace. Among the redside dace, Michigan specimens bore the closest

93
resemblance to rosyside dace in external body shape and in lateral line scale counts. The

apparent retention of some ancestral traits by Michigan redside dace relative to the other
groups may be related to the early, post-glacial isolation of the southeastern Michigan
watersheds following the closure of a major southern dispersal route.

There are several avenues of future study that should be considered. Artiﬁcial
manipulation of instream microhabitat characteristics would be helpful in clarifying
ambiguous results due to differences in habitat availability within and between study sites.
Some habitat attributes that were available in the New York streams were not available at
the Michigan study site, impeding a direct comparison of habitat use between sites for
those attributes. It was also diﬂicult to assess the importance of habitat attributes when
use followed availability. In that case, it was not possible to separate selection of an
abundant habitat attribute from random choice. Cover types could be experimently added
and removed ﬁom pools in Michigan and New York streams to assess redside dace
response to structure that is not normally available. Another set of research needs
includes the direct determination of a more detailed array of physiological parameters
including measurement of active metabolism, ﬁnal preferenda, growth optima, and
deﬁning the zone of temperature tolerance for redside dace. This information would allow
more accurate prediction of the species’ physiological tolerances and suggest conditions
that would maximize (or threaten) ﬁsh condition. Finally, a phylogenetic analysis that
includes representatives ﬁ'om more regions across the species’ range and integrates

detailed features of external and internal anatomy with measures of molecular similarity

94
would help to resolve the historical relationships of the isolated populations of redside

dace included in this study.

The results of this project emphasize that efforts to enhance the prospects for
redside dace survival in Michigan will be strongly tied to the maintenance of beneﬁcial
aspects of the stream corridor and to ensuring a level of high water quality in the stream
reaches that redside dace occupy. This study suggested that redside dace require cool,
clear pools that are bordered by well-developed riparian vegetation. A suitable riparian
corridor including dense, low vegetation would promote the production and concentration
of terrestrial insects that serve as the primary food source of redside dace. Riparian
vegetation would also provide shade to encourage cooler stream temperatures and serve
to reduce erosional inputs that add sediment to pools, decreasing depth and water clarity.
Though not evaluated in this study, gravel riﬂles are reported to provide necessary
spawning habitat for redside dace suggesting that a pool - riﬂle conﬁguration should also

be protected.

APPENDDC

APPENDIX

Museum Specimen List

All specimens used in the morphometric and meristic analyses of this study were obtained
from the University of Michigan Museum of Zoology, Ann Arbor, MI.

C Iinostomus elongatus:

Middle River Rouge, Detroit River drainage, Northville, Wayne County, Michigan.
UMMZ 061357, 066545.

Upper River Rouge, Detroit River drainage, Farmington, Oakland County,
Michigan. UMMZ 136878.

Mud Creek, Allegheny River drainage, Cherry Creek, Chautauqua County, New
York. UMMZ 180797, 180960.

Cold Spring Creek, Allegheny River drainage, Randolph, Cattaraugus County,
New York. UMMZ 180823.

Dunderburg Creek, Susquehanna River drainage, Gilbertsville, Otsego County,
New York. UMMZ 109773.

Unnamed stream, Black River drainage, Medford, Taylor County, “ﬁsconsin.
UMMZ 078727, 078740.

95

96

Big Eau Pleine River, \Vrsconsin River drainage, Colby, Marathon County,
Wisconsin. UMMZ 075738.

Clinosromusﬁmduloides:

Tributary of Hyco Creek, Roanoke drainage, Leasburg, Caswell County, North
Carolina. UMMZ 177051.

Dry Creek, Tennessee River drainage, Hardin County, Tennessee. UMMZ
200924.

BIBLIOGRAPHY

BIBLIOGRAPHY

Altmann, J. 1974. Observational study of behavior". sampling methods. Behaviour 49:226-267.

American Public Health Association, American Water Works Association, Water Pollution
Control Federation 1989. Standard methods for the examination of water and
wastewater. 17th Edition. APHA, Washington, DC.

Bailey, R. 1951. A revised list ofthe ﬁshes orowa, with keys for identiﬁcation, pp. 315-365 i_r_r
J. Harlan and E. Speaker, Iowa ﬁsh and ﬁshing. Iowa Conservation Commission, Des
Moirres, IA

Bailey, RM. and GR Smith 1981. Origin and geography ofthe ﬁsh fauna ofthe Laurartian
Great Lakes basin. Canadian Journal of Fisheries and Aquatic Sciences 38: 1539-1561.

Baltz, D.M. 1990. Auteeology, pp. 585-607 i_r_r Schreck, CB. and PB. Moyle, eds.
Methods for ﬁsh biology. American Fisheries Society, Bethesda, 1W).

Baltz, D.M, P.B. Moyle, and NJ. Knight. 1982. Competitive interactions between bentlric
stream ﬁshes, riﬂe sculpin, Corals gulasus, and speckled dace, Rhinichtlnls osculus.
Canadian Journal ofFisheries and Aquatic Sciences 39:1502-1511.

Baltz, D.M, P.B. Moyle. 1984. Segregation by species and size class of rainbow trout (Salmo
gairdren') and the Sacranremo sucker (Catastomrrs occidentalis) in three California
streams. Enviromnental Biology of Fishes 10:101-110.

Barrett, PI. and CE. Mauglm 1994. Habitat preferences ofintroduced smallmouth bass in a
central Arizona stream. North American Journal of Fisheries Management 14:112-118.

97

98

Beamish, F.W.H 1964a Seasonal changes in the standard rate of oxygen consumption of
ﬁshes. Canadian Joumal of Zoology 42: 189-194.

Beamish, F.W.H_ 1964b. Respiration of ﬁshes with special emphasis on standard oxygen
consumption 11. Inﬂuence of weight and temperature on respiration of several species.
Canadian Journal onoology 42:176-188.

Beamish, F .W.H. and PS. Mookherjii. 1964. Respiration of ﬁshes with special emphasis on

standard oxygen consumption I. Inﬂuence of weight and temperature on respiration of
goldﬁsh, Carassius auratus L. Canadian Journal of Zoology 42:161-175.

Becker, CD. and RG. Genoway. 1979. Evaluation of the critical thermal maximum for
determining thermal tolerance of freshwater ﬁsh Environmental Biology of Frsha
4(3):245-256.

Becker, G. 1983. Fishes of Wisconsin. University of Wisconsin Press, Madison.

Bookstein, FL. 1992. Morphometric tools for landmark data: geometry and biology.
Cambridge Univ. Press, New York, NY.

Bookstein, FL. 1990. Higher-order features of shape change for landmark data pp. 237-250
E Rohlf F.J. and F .L. Bookstein, eds. Proceedings of the Michigan morphometrics
workshop, Special Publication No. 2. Museum of Zoology, University of Michigan,
Ann Arbor, MI.

Bookstein, F .L. 1989. Principal warps: thin-plate splines and the decomposition of
deformations. IEEE Transactions on Pattern Analysis and Machine Intelligence
1 1(6):567-585.

Bookstein, FL. 1986. Size and shape space for landmark data in two dimensions. Statistical
Science 1:181-242.

Bovee, KD. 1986. Development and evaluation of habitat suitability criteria for use in the
instream ﬂow incremental methodology. Instream Flow Information Paper 21. US.
Fish and Wildlife Service Biological Report 86(7).

99

Bovee, KD. and RT. Milhous. 1978. Hydraulic simulation in instream ﬂow studies: theory
and technique. Instream Flow Paper No. 5, US. Frsh and erdlife Service/Oﬁoe of
Biological Services - 78/33.

Brett, JR 1964. The respiratory metabolism and swimming performance ofyoung sockeye
salmon Journal ofthe Fisheries ResearchBoard ofCanada 21: 1183-1226.

Brett, JR and TDD. Groves. 1979. Physiological energetics, pp. 279-3 52 i_r_r Hoar,W.S. and
DJ. Randall, eds. Fish Physiology Vol. VI. Academic Press, New York, NY.

Bryan, J.D., LG. Hill, and WH Neill. 1984. Interdependence of acute temperature preference
in the plains minnow. Transactions of the American Fisheries Society 113:557-562.

Cain, T.1993. Beyond dollars and sense: debatingthevalue ofnongame ﬁsh.
Fisheries l8(7):20-21.

Cech, J.J. Jr. 1990. Respirometry, pp. 335-362 in Schreck, CB. and PB. Moyle, eds.
Methods for ﬁsh biology. American Fisheries Society, Bethesda, MD.

Cech, 1.]. Jr., C.G.Campagna, and SJ. Mitchell. 1979. Respiratory responses of largemouth
bass(Micropterus salmoides) to environmental changes in temperature and dissolved
oxygen. Transactions of the American Fisheries Society 108:166-171.

Cech,J.J. Jr., MJ. MassingilLB. VondracekandAL. Linden 1985. Respiratorymetabolism

of mosquitoﬁsh, mm ﬁrms: eﬂ‘ects of temperature, dissolved oxygen, and sex
diﬂ‘erence. Environmental Biology of Frshes 13(4): 297-307.

Cech, J]. Jr., S]. Mitchell, D.T. Castelberry, and M. McEnroe. 1990. Distn'bution of
California stream ﬁshes: inﬂuence of environmental temperature and hypoxia
Environmental Biology of Fishes 29295-105.

Cherry, D.S., KL. Dickson, and J. Caims Jr. 1977. Preferred, avoided, and lethal temperatures
of ﬁsh during rising temperature conditions. Journal of the Fisheries Research Board of
Canada 34:239-246.

100

Clausen, RG. 1936. Oxygen consumption in ﬁesh water ﬁshes. Ecology 17:216-226.

Coon, T. 1993.Projectedunpactofwastewatadisdnrgeonredsidedacegm
dwinSeeleyDrain, Michigan UnpublishedreporttoGiﬂ‘els—WebsterEngineers

C00per, EL. 1983. ﬁshes of Pennsylvania and the northeastern United States. Pennsylvania
State University Press, University Park, PA.

Cowles, RB. andC.M Bogert. 1944. Apreliminarystudyofthethennalrequirementsof
desert reptiles. Bulletin of the Museum of Natural History 83 2265-296.

Cox, D.K. 1974. Effects of three heating rates on the critical thermal maximum of bluegill, in
Gibbons, J .W. and RR Sharitz, eds. Thermal Ecology. USAEC Technical Information
Center, Oak Ridge, TN.

Daniels, RA and SJ. \Vrsniewski. 1994. Feeding ecology of redside dace, Clinostomus
elongatus. Ecology of Freshwater Fish 3: 176-183.

Eddy, S. and J. Underhill. 1974. Northern ﬁshes. University of Minnesota Press, Minneapolis,
MN.

Ege, R and A Krogh. 1914. On the relation between temperature and the respiratory
exchange in ﬁshes. Intern. Rev. Ges. Hydrobiol. Hydrog. 7:48-55.

Ehrlich, PR and J. Roughgarden. 1987. The science of ecology. Macmillan Publishing Co.,
New York, NY.

Elliot, J.M. and W. Davison. 1975. Energy equivalents of oxygen consumption in animal
energetics. Oecologia 19: 195-201.

Endler, JA 1986. Natural selection in the wild. Princeton University, Princeton, NJ.

101

Evans, H and E. Deubler Jr. 1955. Pharyngeal tooth replacement in Sematilwratrmraculanu
and Clinastormrs elangatus, two species ofcyprinid ﬁshes. Copeia 1955: 31-41.

Facey,DE.1987.Maabolkmnsuahnsarﬂnuaolnbimsdecﬁmmfomareamﬁshes
DissataﬁonUrﬁvasityofGeorgiaAﬂmsGA

Facey, DE and GD. Grossman 1990. The metabolic cost of marntarnrng position for four
NorthAmericanstreamﬁshes: eﬂ‘ectsofseasonandvelodty. PhysiologicalZoology
63(4): 757-776.

Feldmeth, CR and TM Jenkim Jr. 1973. An estimate ofenergy expenditure by rainbow trout
(Sabnogar'rdrenﬁnasmallmountain stream. Journal oftheFrsheriesResearchBoard
of Canada 30:1755-1759.

Felley, J .D. 1984. Multivariate identiﬁcation of morphological-environmental relationships
within the Cyprinidae (Pisces). Copeia 1984(2): 442-455.

Feminella J.W., and W.J. Matthews. 1984. Intraspeciﬁc diﬂ‘erences in thermal tolerances of
Etheostama .spectabile (Agassiz) in constant versus ﬂuctuating environments. Journal
of Fish Biology 25:455-461.

Foster, SA, JA Baker, and MA Bell. 1992. Phenotypic integration of life history and
morphology: an example ﬁom three-spined stickleback, Gasterosteus aculearus L.
Journal of Fish Biology 41(Supplement B): 21-35.

Freeman, MC. 1990. Foraging behavior of the rosyside dace, Climsoanusﬂmduloides: the
importance of social interactions. Dissertation. University of Georgia, Athens, GA

Fry, F.E.J. 1971. The eﬂ‘ect of environmental factors on the physiology of ﬁsh, pp. 1-98 i_r_r
W.S. Hoar and DJ. Randall, eds. Fish Physiology Vol. VI. Academic Press, New
York, NY.

Fry, F.E.J. 1947. Effects of the environment on animal activity. University of Toronto Studies
in Biology Series 5521-62.

102

Fry, F.E.J. and 1.8. Hart. 1948. The relation of temperature to oxygen consumption in the
goldﬁsh Biological Bulletin 94:66-77.

Gatz, A,Jr. 1979. Ecological morphology of freshwater stream ﬁshes. Tulane Studies in
Zoology and Botany, 21: 91-124.

Gilbert, C. 1980. Redsidedacemm p. 148inD. Lee, C. Gilbert, C.
Hocutt, R Jenkins, D. McAllister and J. Stauﬂ‘er, Jr., eds. Atlas of North American
ﬁeshwater ﬁshes. North Carolina State Museum of Natural History, Raleigh

Green, C. 1935. The distribution of Wisconsin ﬁshes. Wisconsin State Conservation

Grossman, GD. and M Freeman 1987. Microhabitat use in a stream ﬁsh assemblage. Journal
ofthe Zoological Society ofLondon 212: 151-176.

Hankinson, TL. 1920. Report on investigations of the ﬁsh of the Galien River, Berrien County,
Michigan. Occasional Papers of the Museum of Zoology Univ. of Michigan 89: 1-25.

Harlan, J .R and EB. Speaker. 1987 . Iowa ﬁsh and ﬁshing. Iowa Conservation Commision,
Des Moines, IA

Hartl, D.L. and AG. Clark 1989. Principles of population genetics. Sinauer and Associates,
Inc. Sunderland, MA

Hayes, DB. 1993. A statistical method for evaluating differences between age-length keys with

application to Georges Bank haddock W m. Fishery Bulletin
91(3):sso-557.

Hill, J. and GD. Grossman. 1987. Home range estimates for three North American stream
ﬁshes. Copeia 1987(2): 376-380.

Hill, I. and GD. Grossman. 1993. An energetic model of microhabitat use for rainbow trout
and rosyside dace. Ecology 74(3):685-698.

103

Hubbs,C.andKLagler.1964.FrshesoftheGreatI.akesregionUniversityofMichigan
Press,AnnArbor.

Hughes, N.F. andL.M. Dill. 1990. Position choiceby drift-feeding salrnonids: model andtest
forucﬁcgrayﬁngﬂhymallusarcﬁarsﬁrsubarcﬁcmoumainstreams, interiorAlaska
Canadian Journal of Fisheries and Aquatic Sciences 47: 2039-2048.

Hutchison, V.H. 1961. Critical thermal maxima in salamanders. Physiological Zoology 34:92-
125.

Hutchison, VH and JD. Maness. 1979. The role ofbehavior in temperature acclimation and
tolerance in ectotherrns. American Zoologist 19:367-384.

Jobling, M 1981. Ternperaturetoleranceandtheﬁnal preferendurn-rapidmethodsforthe
assessment of optimum growth temperatures. Journal of Fish Biology 19:439-455.

Karr, JR, L.A Toth, and DR Dudley. 1985. Fish communities of midwestern rivers: a
history of degradation. Bioscience 35(2):90-95.

Kaufman, R and W. Weiser. 1992. Inﬂuence oftemperature and ambient oxygen on the
swimming energetics of cyprinid larvae and juveniles. Environmental Biology of Frshes
33: 87-95.

Kerr, SR and RA Ryder. 1977. Niche theory and percid community structure. Journal of the
Frsheries Research Board of Canada 34: 1952 - 1958.

Konstarrtinov, AS., V.V. Zdanovich and D.G. Tikhomirov. 1990. The effect of temperature
ﬂuctuations on metabolic rate and energetics of juvenile ﬁsh. Journal of Ichthyology
30(1):38-47.

Korovin, VA 1976. Metabolic rate of the underyearling of the carp, Cygnus mic, adapted
to different water temperatures. Journal of Ichthyology 16(1): 168-172.

104

Koster, W. 1939. Some phases ofthe life history and relationships ofthe cyprinid, W
m (Kirtland). Copeia 1939:201-208.

Kowalski, KT, J.P. Schubauer, C.L. Scott, and JR Spotila 1978. Intraspeciﬁc and seasonal
diﬂ‘erencesinthetemperaurretoleranceofstrearnﬁsh. Journal ofThermalBiology
3: 105-108.

Larning, PR, C.W. Fumton, D. Roberts, and MJ. Armstrong. 1982. Behavioural,

physiological and morphological adaptations of the shanny (Blemrr‘uspholrls) to the
intertidal habitat. Journal of Marine Biology 62: 329 - 338.

Matthews, W.J. 1986. Geographic variation in thermal tolerance of a widespread minnow
Notropis Iurrensis of the North American mid-west. Journal of Fish Biology 28:407-
417.

McClanahan, L.L., CR Feldrneth, J. Jones, and D.L. Soltz 1986. Energetics, salinity and
temperature tolerance in the mohave tui chub, Gila bicolor mohavenszls'. Copeia
l986(1):45-52.

McKee, P. and B. Parker. 1982. The distn'bution, biology and status ofthe ﬁshes CW
anomalm mastering Messrs. Emir 1111919211115, and ammo seams in
Canada Canadian Journal of Zoology 60: 1347-1358.

Meade, L., D. McNeely, L. Komman, and A Surmont. 1986. The redside dace, Clinostomus
elongatus (Kirtland) in Kentucky, with comments about its habitat requirements.
Transactions of the Kentucky Academy of Science 47: 121-125.

Miller, RR, JD. Williams, and IE Williams. 1989. Extinctions of North American ﬁshes
during the past century. Fisheries 14(6):22-38.

Monalran, J. 1991. Development ofhabitat suitability data for smallmouth bass and rock bass in
the Huron River, Michigan thesis. Michigan State University, East Lansing, MI.

Moyle, PB. and JJ. Cech, Jr. 1988. Frshes: an introduction to ichthyology. Prentice Hall, Inc,
Englewood Cliffs, NJ.

105

Mukhamedova, AF. 1978. The level of standard metabolism of young silver carp,
amendments mm Journal of Ichthyology 17(2):292-299.

Page, L.M and BM Burr. 1991. Aﬁeld guideto ﬁeshwaterﬁshes. HoughtonMiminCo,
Boston, MA

Paladino, F..,V JR Spotlla,JP Schubauer, andK..T Kowalski.1980.'1hecriticalthermal

maximum: atechmqueusedtoeluddatephysidogicalstressandadaptaﬁoninﬁshes.
Rev. CanBiol. 39(2): 115-122.

Parker, R, P. McKee, and R Campbell. 1988. Status of the redside dace, W
m in Canada CanadianField-Nanrralist 102: 163-169.

Patterson, C., D.M. Williams, and CJ. Hurnphries. 1993. Congruence between molecular and
morphological phylogenies. Annual Review of Ecology and Systematics 24: 153-188.

Potvin, C. and DA Roﬂ‘. 1993. Distnbuiton-free and robust statistical methods: viable
alternatives to parametric statistics? Ecology 74(6): 1617-1628.

Reinthal, P. 1990. Morphological analyses of the neurocranium of a group of rock-dwelling
cichlid ﬁshes (Ciclﬂidae: Perciformes) ﬁom Lake Malawi, Aﬁica Zoological Journal of
the Linnean Society 98: 123-139.

Reighard,J. 1910. Methodsofshrdyirrgthehabitsofﬁslreawiﬂranaccoumofthebreeding
habits ofthehorneddace. Bulletin ofthe US. Bureau ofFishes

Reynolds, W.W. and ME. Casterlin. 1979. Behavioral thermoregulation and the “ﬁnal
preferendum” paradigm. American Zoologist 19:211-224.

Robins, CR, RM Bailey, C.E. Bond, JR Brooker, EA Lachner, RN. Lea, and W.B. Scott.
1991. A list ofcomrnon and scientiﬁc names ofﬁshes ﬁ'om the United States and
Canada American Fisheries Society, Special Publication No. 12.

106

Rohlf; F.J. and FL. Bookstein 1990. Proceedings ofthe Michigan morphometrics workshop.
Special Publication No. 2, University of Michigan Museum of Zoology, Ann Arbor,
MI.

Rudstam,L. G. andan Magnuson.1985.Predictingtheverticaldistnbutionofﬁshpopulaions:
analysis of cisco, Coregonus arterb'r’, and yellow perch, Percaﬂawxens. Canadian
JomnalofFisheriesanquuaticSciences42: 1178- 1188.

SAS Institute Inc. 1988. SAS/STAT user’s guide. Release 6.03 Edition SAS Institute Inc.,
Cary, NC.

Schwartz,F andJ.Norvell.1958.Food,growthandsexualdimorphismoftheredsidedace

Logging m (Kirtland) 1n Linesville Creek Crawford County, Pennsylvania.
The OhioJoumal ofScience 58: 311-316.

Scott, N.L. 1987. Seasonal variation ofcritical thermal maximum inthe redbelly dace,
Phgxinus EEG—111M (Cyprinidae). Southwestern Naturalist 32(4):43 5-438.

Scott, W.B. and EJ. Crossrnan. 1973. Freshwater ﬁshes of Canada Fishery Research Board
of Canada Bulletin 184.

Shields, BA and JG. Underhill. 1993. Phenotypic plasticity of a transplanted population of
dwarf cisco, Coregonus w'tedii. Environmental Biology of Fishes 37:9-23.

Smimov, P.B., IA Barybina, and LB. Klyashtorin 1987. Relationships between standard
metabolism and temperature injuvenile paciﬁc salmon Journal ofIchthyology
27(2): 15-22.

Smith, C. 1987. The inland ﬁshes of New York State. Cornell University Press, Ithaca, NY.

Smith, GR 1990. Homology in morphometrics and phylogenetics, in Rohlf; F.J. and FL.
Bookstein eds, Proceedings of the Michigan morphometrics workshop. Special
Publication No. 2, University of Michigan Museum of Zoology, Ann Arbor, MI.

107

Smith, GR, J. Taylor and T. Grimshaw. 1981. Ecological survey ofﬁshes intheRaisinRiver
drainage, Michigan Michigan Academy of Sciences, Arts, and Letters 13: 275-305.

Sokal, RR and F.J. Rohlf. 1981. Biometry. W.H Freeman, New York NY.

Strauss, RE. 1979. Reliability estirmtes forlvlev’s Why inng the forage ratio, and a
proposed linearindexoffood selection TransactionsoftheAmericanFisheries
Society 108:344-352.

Strauss, RE. and FL. Bookstein 1982. The truss: body form reconstructioru in
morphometrics. Systematic Zoology 31(2):113-135.

Swiderski, D.L. 1993. Morphological evolution of the scapula in tree squirrels, chipmunks, and
ground squirrels (Sciuridae): an analysis using thin-plate splines. Evolution 47(6): 1854-
1873.

Tabachnick, R and FL. Bookstein. 1990. Resolving factors of landmark deformation:
Miocene Globorotalia, DSDP site 593, pp. 269-281 in Rohlf, F .J . and FL. Bookstein
eds, Proceedings of the Michigan morphometrics workshop. Special Publication No.
2, University of Michigan Museum of Zoology, Ann Arbor, MI.

Trautman, M 1981. The ﬁshes of Ohio. Ohio State University Press, Columbus, OH.

Underhill, J. 1986. TheﬁshfaunaofthelaurentianGreatIakes, the St. LawrenceLowlands,
Newfoundland and Labrador, pp. 105-136 'm C. Hocutt and E. erey, eds. The
zoogeography of North American ﬁeshwater ﬁshes. J. Wiley & Sons, New York NY.

Warren, M.L. Jr. and BM. Burr. 1994. Status of freshwater ﬁshes of the United States:
overview of an imperiled fauna. Fisheries 19(1):6-18.

Webb, P.W. 1971. The swimming energetics of trout. I. Thnrst and power output at
cruising speeds. Journal of Experimental Biology 55:489-520.

White, M.M. 1988. Genetic variation and population structuring in the rosyside dace,
Clinostomusﬂmduloides, in Ohio. Ohio Journal of Science 88(3): 1 14-1 16.

108

Wilzbach, MA 1985. Relative roles of food abundance and cover in determining the
habitat distribution of stream-dwelling cutthroat trout (Salmo clarla'). Canadian
Journal of Fisheries and Aquatic Sciences 42:1668-1672.

Wrnberg, 6.6. 1956. Rate of metabolism and food requirements of ﬁshes. Fisheries
Research Board of Canada Translation Series 194(1960).

Yezerinac, S.M., S.C. Lougheed, and P.Handford. 1992. Measurement error and
morphometric studies: statistical power and observer experience. Systematic
Zoology 41(4):471-482.

Zar, 1H. 1984. Biostatistical analysis. Prentice Hall, Englewood Cliﬂ‘s, NJ.

Zelditch, M.L., F.L. Bookstein, and BL. Lundrigan. 1992. Ontogeny of integrated skull
grth in the cotton rat Sigrnodon ﬁrlviventer. Evolution 46(4):]164-1180.

Zelditch, M.L., W.L. Pink and D.L. Swiderski. 1995. Morphometrics, homology and
phylogenetics: quantiﬁed characters as synapomorphies. Systematic Biology
44: 179-189.

":lrlrlrr11.111111“