2353? V???

INIHIHIHHIIIUI”Hill!!!“llHlIIIIIIHIHJIUIIUWI

293 00788 5993

LIBRARY
Michigan State
1 University

\ I

 

 

This is to certify that the

thesis entitled

A Nonconventional Morphometric Technique for
Measuring Ontogenetic Shape Changes
In Two Species of Centrarchid Fishes

presented by

James Edward Zablotny

has been accepted towards fulfillment
of the requirements for

M-S- degree in Zoology

 

ELgéCZ/‘L §L{t«,'~ve \1

Major professor

 

Date ”Ego ii“?

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

PLACE IN RETURN BOX to remove this checkout from your record.
To AVOID FINES return on or before date due.

DATE DUE DATE DUE DATE DUE

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

l

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

\J

lflfi

MSU Is An Affirmative Action/Equal Opportunlty Institution-
emu-Ann”.-

 

 

 

 

_—

A NONCONVENTIONAL MORPHOMETRIC TECHNIQUE FOR MEASURING
ONTOGENETIC SHAPE CHANGES
IN TWO SPECIES OF CENTRARCHID FISHES

by
James Edward Zablotny

A THESIS

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

SMASTER OF SCIENCE
Department of Zoology

1989

ABSTRACT
A NONCONVENTIONAL MORPHOMETRIC TECHNIQUE FOR MEASURING

ONTOGENETIC SHAPE CHANGES OF THE
FIFTH CERATOBRANCHIAL IN TWO SPECIES OF CENTRARCHID FISHES

by
James Edward Zablotny

The shape changes of the occlusive surface of the right
fifth ceratobranchial bone in pumpkinseed and bluegill sunfish
were investigated across an ontogenetic series of specimens.
The fifth ceratobranchials, from cleared and stained
specimens, were photographed and outlines of the right
occlusive surface were digitized to permit calculation of
medial axes. A series of lines normal to these medial axes
were drawn to the medial edge of the ceratobranchial and
recorded as normal length per relative length along the medial
axis. A geometric quintic equation was used to model the
curve of the medial edge of the occlusive surface.
Ontogenetic trajectories were calculated for the geometric
coefficients across the size range of specimens studied. All
geometric coefficients appear to be linearly related to
standard length. The ontogenetic trajectories for
ceratobranchial shape reveal that both species apparently
diverge in shape at sizes greater than 35 mm standard length.
For pumpkinseeds, the initiation of shape divergence from the
juvenile ceratobranchial shape appears well before the switch

in diet from cladocera and aquatic insects to snails.

ACKNOWLEDGMENTS

I wish to express my gratitude to the following people
who have provided me with countless assistance: Donald
Straney, guidance commitee chairman: Guy Bush and Robert
Anstey, guidance committee 1members. I ralso thank Fred
Bookstein for technical assistance with medial axis
implementation; Gerald Smith, J. Alan Holman and Edward Wiley
for specimens. Most of all, I thank my wife Jennifer for

helping me with the completion of the manuscript.

iii

TABLE OF CONTENTS

Page
LIST OF TABLES........................................ v
LIST OF FIGURES....................................... vi
INTRODUCTION.......................................... 1
MATERIALS AND METHODS................................. 7
RESULTS............................................... 18 ‘
DISCUSSION............................................ 45
APPENDIX A: DIGITIZED OUTLINES AND MEDIAL AXES....... 63
APPENDIX B: SOURCES AND LOCALITY DATA FOR SPECIMENS.. 76

BIBLIOGRAPHY...0......0..I.OOOOOOOOOOOOOOOOOOOOOOO0.0. 78

iv

Table

LIST OF TABLES

Page

Two Level Nested Anova Describing the
Measurement Error Produced During Normal
Line construction.OOOOOOOOOOOOOOOOOOO0.00....O 17

Coefficients of the Geometric Interpretation
of the Algebraic Quintic Polynomial: Lepgmis
mm.0.0C0.0.000000000000000.00.0000... 20

Coefficients of the Geometric Interpretation
of the Algebraic Quintic Polynomial: .Lepgmis

giprSUSOOOeoeoeeooooooeoooeooooooeooooooooeoo 21

Linear Regression Equations of the Geometric
Coefficients, tn, Versus Standard Length...... 30

Correlation Analysis for the Interaction
Between Geometric Coefficients with Size
Partialed out...0.000000000000000000000......O 31

Statistical Tests for Regression Line

Parallelism for Inter-Species Comparisons of

the Geometric Coefficients Versus Standard

Length Regressions............................ 38

Calculated Geometric Coefficients for Expected
Ceratobranchial Shapes Across Ontogeny........ 43

Figure

10

11

LIST OF FIGURES

Alignment Apparatus for Photographing
the Fifth Ceratobranchial Bones...............

Measurement Protocol for Extracting Normals
to themedial “is.0.00.0.0...OOOOOOOOOOOOOOO.

Blending Functions for the Quintic
PalynomiaIOOOOOOOOIOOOOOOOOOOOOOOOOOOOOOO .....

The Ontogenetic Trajectory of the Width of
the Posterior End of the Fifth Ceratobranchial
(b5) Versus Standard Length...................

The Ontogenetic Trajectory of the Width of
the Anterior End of the Fifth Ceratobranchial
(b1) Versus Standard lengthOOOOOOOOOOOOOOOOOOO

The Ontogenetic Trajectory of the Tangent of
the Outline of the Posterior End of the Fifth
Ceratobranchial (t5) Versus Standard Length...

The Ontogenetic Trajectory of the Tangent of
the Outline of the Anterior End of the Fifth
Ceratobranchial (t5)‘Versus Standard Length...

The Ontogenetic Trajectory of the Change in
Tangent of the Outline of the Anterior End

of the Fifth Ceratobranchial (b‘) Versus
Standard Length...............................

The Ontogenetic Trajectory of the Change in
Tangent of the Outline of the Posterior End
of the Fifth Ceratobranchial (b5) Versus
Standard Length...............................

Regression for b5 Versus Standard Length:
Natural Log-Natural Log Transformation........

Expected Curve Shape of the Medial Side of
the Fifth Ceratobranchial of Pumpkinseed

and Bluegill Sunfish. Calculated Curves
Contructed for Specimens Having Standard
Lengths of 31, 41, 51, 61, 71, and 111 mm.....

vi

Page

10

12

16

22

23

24

25

26

27

32

33

12

13

14

15

16

Growth Pattern of the Apophyseal Edge
in the Fifth Ceratobranchial of Pumpkinseed

sunfiShOO0......0..0..OOOOOOOOOOOOOOOOOOOOOOOO

Primitive Fifth Ceratobranchial Bone Shape

of the Warmouth Sunfish gheengpgygtng
WOOOOOOOOO..00...OOOCOOOOOOOOOOOOOOOOOOO

Similarity in Fifth Ceratobranchial Shape
between Small (<28 mm) Pumpkinseeds and
BluegillsOOOOOOI0.00000000000000000000.0.0....

Digitized Outlines and Medial Axes of Bluegill
Sunfish. Specimens are Ranked by Size from
Small to Large. Scale Distance is Equal to
One Millimeter................................

Digitized Outlines and Medial Axes of Pumpkin-
seed Sunfish. Specimens are Ranked by Size
from Small to Large. Scale Distance is Equal
to One Millimeter.............................

vii

41

51

52

63

69

INTRODUCTION

In general, bony fishes utilize the external jaws for
obtaining prey that is swallowed without mastication
(Alexander, 1974). Highly derived euteleostean fishes
(Greenwood, et. al., 1966) possess a set of pharyngeal "jaws"
for mechanical manipulation of food items within the pharynx
(Lauder 1983a). The euteleostean pharyngeal jaw functions as
a single unit. despite being' derived from. the posterior
viscerocranium elements (Harder, 1975). The upper
pharyngeal jaw consists of the second, third, and fourth
pharyngobranchial bones. The lower pharyngeal jaw element is
derived from the fifth ceratobranchial (Lauder, 1983). The
bones of the pharyngeal jaws are usually covered with teeth
on their occlusive surfaces. Extreme variability in shape of
the fifth ceratobranchial and its associated dentition appear
correlated with the prey type utilized by these fishes (Liem
and Osse, 1975). Pharyngeal jaws are simple within the
Centrarchidae and Serrannidae (the fifth ceratobranchials are
separate from each other), but specialized (fifth
ceratobranchials fused) in other higher euteleostean fishes

such as the Cyprinodontoidei, Pomacentridae, Anabantoidea,

1

2
Girellidae, Cichlidae, Embiotocidae, Labroidei, and Odacidae
(Liem and Greenwood, 1981).

Pharyngeal jaws can be used to assist in swallowing large
food items. When used in pharyngeal transport, the upper and
lower pharyngeal jaws are simultaneously adducted and
retracted.posterior1y (Lauder, 1983). By moving the upper and
lower pharyngeal elements differently, the jaws will function
instead as crushing surfaces masticating the food within the
pharynx. For this to happen, simultaneous contraction of the
pharyngeal jaw musculature adducts the upper and lower
pharyngeal jaws to produce the crushing forces needed to
masticate a ‘hard food items Based. upon the taxonomic
distribution and outgroup analysis, Lauder (1983) concluded
that the pharyngeal jaws were primitively used for food
transport. Pharyngeal trituration has arisen several times
as a further, more specialized function of pharyngeal jaws.
The sequences of’ this evolution from ‘transport, to food
processing roles can be studied most conveniently within the
North American centrarchid genus m. Both feeding styles

are found among the 11 species of Lewis sunfish. Two

species (L. gibbgsus and L. W) are pharyngeal
crushers, feeding extensively' on. snails. ILauder (1983)

concluded that these two species masticate snails with muscle
utilization patterns derived from the muscle patterns

associated with pharyngeal transport.

3

Lauder (1983) investigated the functional differences in
the pharyngeal jaw structure and feeding behavior in M
and emphasized the structural dissimilarities between species
that feed on snails and those that feed on zooplankton and
aquatic arthropods. Zooplankton feeders, like bluegill
sunfish (L. whims) , feature a narrow fifth
ceratobranchial covered with filiform teeth on the occlusive
surface. The two halves of the ceratobranchial are only
moderately apposed to each other along the midline of the
floor of the pharynx (J. Zablotny, pers. obs.). Lauder (1983)
noted that the snail specializing species have a broadly
widened tooth bearing occlusive surface of the fifth
ceratobranchial and a fairly strong region of apposition
between the right and left fifth ceratobranchials. The
mollusc crushers also display hypertrophy of the muscles
responsible for generating the powerful adduction of the upper
and lower pharyngeal jaws.

Some Lepomis species also undergo an ontogenetic change
in both food habits and pharyngeal jaw action. Although young
centrarchids feed either on littoral or limnetic zooplankton
(Keast, 1980), the gastropod specialists tend to shift from
zooplankton to snails between 40 and 50 mm standard length.
Other minor shifts in diet may occur in M due to
competition based resource partitioning and niche shifting
(Werner and Hall, 1976: Laughlin and Werner, 1980: Mittelbach,

1984), and predation risk (Werner, et. al., 1983a: Mittelbach,

4
1984), but the functional change in fifth ceratobranchial use
appears only in those species which switch to feeding on
snails.

The ontogenetic shift in pharyngeal jaw function in some
Lepgmis offers an excellent opportunity to examine the basis
of morphological and functional evolution. The derived
crushing morphology of L. gippgsgg is achieved late enough in
ontogeny (standard length of circa 45 mm, Zablotny, pers.
obs.) that the morphogenetic transition can be studied in
whole-mount field caught fish. This makes a comparative study
of the ontogenetic basis of the evolution of pharyngeal jaw
morphology feasible, since the critical ontogenetic stages are
those that are most easily sampled. Because the use pattern
in some species shifts in free-living individuals (and not in
embryos), some questions about the process of morphological
divergence become tractable. We know very little about the
exact nature of the relationship between function and
structure during evolution. It is in cases of ontogenetic
switching in function that we can begin to assess, for
example, whether changes in function precede changes in
morphology (e.g. Mayr, 1958: 1960) or how important
environmental factors are in. precipitating' morphological
change (e.g. Smith-Gill, 1983).

I have examined the differences in ontogeny of the lower
pharyngeal jaw in two species of Lewis which differ in adult

feeding morphology. I have focused on the lower element

5

(fifth ceratobranchial) because the lower pharyngeal jaws have
been extensively used for taxonomic evaluation of species and
evolutionary relationships among fish taxa. The species
examined are phylogenetically closely related. Presently,
published phylogenies on the family Centrarchidae are based
on electrophoretic data (Avise and Smith, 1974: Avise, et.
al., 1977), and on the morphology of the acoustico-lateralis
system (Branson and Moore, 1962). Unfortunately, there are
no highly resolved phylogenetic hypotheses available for
evaluating the direct evolutionary relationships within the
genus Lepgmis (Humphries, pers. com.). Since Lauder (1983,
1983a) believes that pharyngeal transport is primitive to
snail crushing, I chose to represent the primitive morphology
(pharyngeal transporter) with the bluegill sunfish and the
derived (snail crusher) morphology with the pumpkinseed
sunfiSh. I am.interested in determining how the primitive
ontogenetic trajectory of bluegills is transformed into the
derived pumpkinseed condition. This study also examines how
the function of pharyngeal jaws (assessed by feeding
preference) is correlated with their morphology during the
transition from one feeding style to another in pumpkinseeds.

The ontogenetic shift in pharyngeal jaw function in
Lepgmis is a useful example of morphological change only to
the extent that it can be studied quantitativelyu The outline
of the fifth ceratobranchial in REGIME. for example is

smoothly curved without obvious landmarks. The ontogeny of

6
this bone in different Lepgmig species involves subtle shape
differences that simple measurements of linear dimensions
cannot describe well. Relatively few techniques are available
for quantifying shape change in landmark-free situations
(Oxnard, 1980: Bookstein, 1979, 1981a: Bookstein, et. al.
1985). In the present case, I have used the line skeleton
(Blum and Nagel, 1978: Blum, 1973: Oxnard, 1980: Bookstein
1981a, 1981c: Bookstein, et. al. 1985). The medial axis is
the locus of points inside an outline that are equidistant
from the outline. As a nonlinear axis of symmetry of an
outline, the medial axis is a useful summary of the outline
shape: the outline can be recovered by associating with each
medial axis point the distance from axis to outline. Because
I was interested in comparing how particular regions of the
fifth ceratobranchial changed with growth across species, I
could not effectively use techniques such as tangent angle
functions or Fourier analysis that are functions of outline
arc length. Instead, I have used the medial axis of each form
as a coordinate system within which to study complex shape

changes.

MATERIALS AND METHODS

An ontogenetic series of 25 bluegill (W W
Rafinesque) and 25 pumpkinseed (W W (Linnaeus))

sunfish, were selected from material loaned from the following
sources: The University of Michigan Museum of Zoology (UMMZ) :
The Museum, Michigan State University: The Field Museum of
Natural History (FMNH); and the University of Kansas Museum
of Natural History (KU). Specimens were chosen to provide a
reasonably complete coverage of standard lengths from 25 to
160 mm, spanning the size range of both species commonly found
in museum collections. Additional bluegill specimens were
captured by angling at Lake Ovid, Laingsburg, Michigan. Whole
specimens were cleared and stained using the technique of
Taylor (1967). The lower pharyngeal jaw (right and left
ceratobranchial V) were dissected from the cleared and stained
specimens and stored in 100% glycerine. A few grains of
thymol were added to inhibit mold growth.

To standardize specimen orientation for photography, each
pharyngeal lower pharyngeal jaw was positioned on a 4 mm layer
of plastic sponge foam with the occlusal surface uppermost.
A plastic microscope coverslip box (75x33x13 mm) served to
enclose the specimen and sponge foam, which were covered by

a solution of 3 parts glycerine and 7 parts 70% ethanol.

8

Two small sections of a transparent metric ruler were glued
to the underside of a large coverslip. The coverslip was
placed over the specimen with the ndllimeter scale of the
rulersi positioned along ‘the lateral edges of’ the lower
pharyngeal jaw. The sections of ruler provide a metric scale
at or near the plane of focus during photography (Figure 1).
Once placed on the microscope stage, two 100 g weights were
positioned on the coverslip to level the tooth surface of the
specimen to the plane of the camera lens and to minimize the
central rotation of the two halves of the ceratobranchials.
The resilience of the sponge foam to the force of the two
weights tended to push the teeth firmly against the glass
coverslip and stabilize specimens during photography.

A trinocular dissecting microscope (Wilde model M8, MPSSl
camera and MPS45 Photoautomat) and Kodak Panatomic X (ISO 32)
film were used for specimen photography. Only the right fifth
ceratobranchial and its millimeter scale were included in the
photograph. A Volpi Intralux 6000 fiber optics light source
provided even illumination. Negatives were enlarged and
printed on 8 1/2 x 11 in. Kodak high contrast resin coated
paper.

The outline of the specimen (including the cartilaginous
posterior and anterior apices) and millimeter scale were
traced from the photographic prints to transparent acetate
with a fine tipped permanent marking pen. Starting with the

anterior apex, each outline was digitized in a clockwise

9
direction on a Summagraphics Bit Pad-One digitizing tablet.
The number of digitized points used for shape analysis varied
from 135 to 220: more points were digitized on specimens
whose outline was more curved.

The medial axis (Bookstein, 1979) for each form was
computed using Bookstein's (1979) program via a Tektronix
T4014 graphics terminal. Hard copies of the graphics output
were obtained and inspected for digitizing errors. Those
forms having obvious digitizing errors were reanalyzed. The
90 degree crosshairs of a protractor were placed tangent to
points along the major medial axis component and were used to
draw normal lines perpendicular to the tangents of the medial
axis. Based on the degree of curvature found on the specimen,
32 to 65 normal lines were drawn per specimen. Branch points
on the medial axis were ignored: they were encountered only
on several pumpkinseed specimens. The ends of these normal
line segments at the outline boundary and at the medial axis
were digitized. The length of the normals was calculated and
serves as an estimate of the width function of the outline at
each medial axis point. Because the medial axis is an axis
of symmetry defined by the pair of "pseudonormal" lines for
each point on the medial axis, I chose to analyze only the
medial side of the fifth ceratobranchial. Since I used lines
drawn normal to the medial axis and not to the outline, the
width functions used to describe the form are an approximation

of this symmetry (Blum and Nagel, 1978).

10

Weights

mo. .1...

 

 

 

‘Coverslip

Sponge

  

Specimen

Plastic Microscope
Slide Box

Figure 1. Alignment Apparatus for Photographing the Fifth
Ceratobranchial Bones.

11

The medial axis and normals provide a coordinate system
for analyzing differences in outline shape. I set the origin
of this coordinate system at the posterior end of each
ceratobranchial and standardized the arc of the medial axis
to a length of one. Distance along the medial axis represents
an ordinate (Figure 2). A polynomial was fitted to the plot
of normal width versus relative medial axis arc length to
produce an equation expressing ceratobranchial width as a
function of position along the medial axis. I used a fifth
order polynomial,

Y = a0 + a1x + azx2 + a3x3 + a‘x‘ + asx5

to model the overall curve of the medial edge of the right
half of the lower pharyngeal jaw. Although the curve of the
pumpkinseed specimens are more complex than for the series of
bluegill specimens, comparison of the forms was simplified by
the use of the quintic model for both species. All of the
regression coefficients were significantly different from
zero.

To make the coefficients of the regression model
interpretable as shape descriptors of the estimated curve, the
fifth order polynomial was reparameterized and converted
algebraically to geometric form (Mortenson, 1985). This
parameterization decomposes the original quintic model into
six orthogonal component functions (F0 to F3) and a
corresponding set of six regression coefficients (bo to b5) of

the original quintic polynomial regression model.

12

_
_
@

omHX¢ Hfldcmx 0:“ OH mHflEOZ UCfiUUflHUXW HON HQUOUOHQ UCUEQHH—mflmz

m_x.< n_<_om_2 020...<

m02<._.m_o m>_._.<._mm o

 

 

SE CEEIEEEE is.

 

 

m.xe.e.poE o.
.coaceb

 

mc7._aEtoz

.N musmfik

Loug
SSNI'I "IVINHON
40 HLONE'I

13

Mortenson (1985) noted that these functions when used together
blend to produce a curve originally described by the algebraic
quintic equation. The values of the coefficients b0 and b1
correspond respectively to the posterior and anterior widths
of the medial half of the lower pharyngeal jaw. The next two
coefficients, b2 and b, are the slopes for the posterior and
anterior ends of the form. The acceleration or change in
slopes along the curve of the medial side of the fifth
ceratobranchial are b,‘ for the posterior and b5 for the
anterior ends. The change in magnitude and the sign of these
coefficients may be used to mathematically assess the overall
change in shape that is occurring during ontogeny.

The coefficients, bi's, of the geometric quintic were
calculated with an SPSS polynomial regression program.
Correlations and covariance among the geometric coefficients
for individual species were initially examined with a partial
correlation analysis (SPSS). Since most of the coefficients
were significantly correlated with standard length or size,
a partial correlation analysis was used to partial out the
contribution of size to the overall correlations among
coefficients (Table 5).

To examine how the magnitude of the coefficients, bu
change through ontogeny for the sunfish, each coefficient was
regressed on standard length. I compared these regressions
between both species with a covariance analysis to test for

regression. parallelism (Zar, 1984). A. test for common

14
intercepts (Zar, 1984) was used. to further’ examine the
differences in y intercepts for regression lines having equal
slopes in both species. This procedure uses a modified t-test
to evaluate the regression elevations or intercepts as being
the same. Only one pair of regression lines, hm, was suited
for this statistical test.

Coefficients derived from the regression equations of
each coefficient versus standard length were used to calculate
expected pharyngeal jaw shapes across ontogeny. The expected
ceratobranchial shape was computed for 10 standard lengths
from 31 to 76 mm by 5 mm increments (Table 7). The bottom of
this size range was chosen at or near the standard length at
which the regression lines intersected one another: the
maximum size used was the maximum common to both data sets.
A standard length of 111 mm was also added to the chosen size
classes to represent the largest individual pumpkinseed size
class in the data set. At least 50 points representing a
relative arc length along the medial axis from 0 to 1 were
entered in the set of geometric quintic equations with the
proper coefficients for each size class. 'The resultant.points
were then plotted with a Hewlet Packard 7475A plotter (Figure
5).

To examine the sources of variation in measuring normal
line lengths, a minuten insect pin was inserted on the lateral
edge of one fifth ceratobranchial bone. The insect pin

provided a landmark for the distance measurement used in the

 

15
test procedure. The specimen was placed in the alignment
apparatus and photographed. This operation was repeated five
times. Each photograph was traced twice on transparent
acetate film. The distance of a normal line drawn from the
edge of the form opposing the pin placement to the point of
insertion of the pin was digitized five times for each
tracing. A nested analysis of variance (Sokal and Rohlf,
1969) was used to estimate the magnitude of construction.error
due to photographic and digitizing error on the length of the
normal line. F'tests for among photographs and among drawings
within photograph sources of variation were not significant
and accounted for 41.3 percent of the total variance (Table
1.). The remaining 58.7 percent of the variance was due to
errors in digitizing normal lines. The total measurement
error (s? = .00287) and coefficient of variation, V=3.25 is
actually quite small and does not contribute greatly to the

estimation of the curve studied here.

16

r3 . I - to):3 + st‘- 6:? F, . no»3 - 15x‘+ 6x5

 

 

 

 

O I O I

F2 8 y - 5,3 0 8:3- 315 F38 7x‘- 4x3- 3:5

o[\—I o/\I

 

 

 

 

If, 8 5(xz- 3x30 33’- x5) 5 ' 5X3' X4 ‘ 5K5

CONTRIBUTION TO SHAPE

 

 

 

 

/ \'/ \

O I O I

RELATIVE DISTANCE ALONG MEDIAL
AXIS

Figure 3. Blending Functions for the Quintic Polynomial.

17

 

 

osao. sun". av deuce

2.5m Sec. 58. name. 3 35:98 5.3:.
«$8633.— 553

6.3 «.38. 23.” A36. 38. n 3533a 9354
2.3 38. 23.." SS. 23. v manuauouona ocean
~n Have» a «n Am a: mu up scaveuuo>.uo oousom

.zumcoq on“: aneuoz unwashed: you
.n>o:< coumoz Hu>oq 038 .H manna

sawuewue> uo noousom one

RESULTS

The medial axis constructed from the digitized outlines
are depicted in Appendices A and B. The structure of the
medial axis is a simple smooth curve in bluegills (Appendix
A): the medial axis for pumpkinseeds (Appendix B) is more
elaborate and variable than for bluegills. Many of the medial
axis produced for large pumpkinseeds contain one or two triple
or branch points. I defined the major medial axis components
as those branches which represent the anterior-posterior axis
of the ceratobranchial. Although Oxnard (1980) and Bookstein
(1981a) acknowledge the usefulness of smaller branch segments
for quick visual evaluation of shape differences in similar
forms, I ignored them for the following reasons. Minor
branches of the medial axis at 45 to 50 percent along the
medial axis are consistently present for pumpkinseeds greater
than 50 mm standard length. Small pumpkinseeds and bluegills
did not feature any minor branches at that position along the
major medial axis component. These minor branch lengths were
not examined in this study. For certain cases, slight
deviations in digitizing produced extraneous branches that

were also excluded from the analysis.
18

19

A geometric quintic polynomial regression for normal
width versus relative distance along the medial axis
adequately models the curve of the right medial side of the
lower pharyngeal jaw. The polynomial regression produced r2
values for the geometric quintic equations ranging from 0.96
to 0.99 in bluegills and pumpkinseeds. The geometric quintic
model fit the set of bluegill curves slightly better than
pumpkinseed curves (Tables 2 and 3). Large pumpkinseeds were
least likely to be fit well by a fifth degree polynomial
curve. These were also 'the individuals with accessory
branches of the medial axis near the midpoint of the outline.

However, the difference in average r2

for the quintic model
fit to individuals with accessory branches (r630.989, n=13)
and without (r2=0.983, n=12) suggests that no discernible
imprecision was introduced by ignoring the information in
accessory branches.

Although the independent "variables", F}, of the
geometric version of the quintic polynomial are formally
orthogonal (Mortenson, 1985, pg. 49), the coefficients, by of
these functions need not be independent. With the exception
of coefficient b5, each coefficient was significantly
correlated with body size in both bluegills and pumpkinseeds
(Table 4). Partial correlations, removing the effect of size

were significant for coefficients b0 and b,, b‘ and b2, and b5

with b1 and b3 in both species (Table 5).

20

 

 

 

and amm.o evvv.hal dean.om ommh.¢n ~mna.mi mwen.o mmsw.o mv<=q
med omm.o Nooh.m| mumn.mo Noun.nl mumm.VI Noon.o momm.o meat:
mma omm.o wean.» Honn.mv nhmH.NI humm.~l wov~.c mmmn.o hv<za
«en nam.o mumm.w~ movm.o> Nbd~.~I Hnmu.MI v~h¢.o Name.o weds;
ooH Nam.o monu.ou hvvo.vm mmom.ou mooh.nl monv.c monv.o «(ta
no ooa.c unnm.o uumn.¢v nmnu.n: vmuc.nt ~mp~.o mumn.o meczq
on ham.o mmoo.~ numm.ve ncaN.nt mnvo.nt finmn.o noon.o «(Sq
an mam.o mhmn.v~| onuo.bm anhn.nt mhhn.nt m¢~H.o Nhnn.c NH¢za
we mmm.o mmnv.vu hvnu.mu vm-.fil mmmn.on muho.o mova.o scan
nv amm.o Omcb.bt cdmd.ed monn.dt undo.OI «$00.0 NanH.o «(In
an mmm.c uwnm.ml anvo.on Anom.dt mnmm.OI mmbo.o hood.o Duct:
an mmm.o oomv.ol uenm.n~ cvmm.c| «mam.ol nmho.c mm~H.o «(In
an mmm.o vcam.nl vaoo.mu o~hh.c| «moo.o ammo.o anuu.o >H¢=q
hm omm.c nman.fint vam.mH monv.dl homw.cl vnmo.o NQAH.O acid
mm Nam.o Heed." unvn.NH Nonn.OI mush.cl memo.o anH.o ondfiq
mm oam.o vma~.~l Hh-.mn onoo.OI nmn~.nt obso.c bmmu.o owdzq
on mmm.o mmvh.ul Hmnm.en mmmw.c| ncmm.OI moho.o oHnH.o mafia
On mmm.c num0.vt wewc.ba bmww.0I oucc.ou omvc.o when.o oudflq
on bmm.o ohno.mt anoo.mn nmmh.ol omno.nl cnvo.o mnem.o ads;
on bmm.o hmuw.nl nmmb.nd ammo.0t mHOQ.OI mvvo.o bNOH.o wmdtn
9N Nam.c vaH.vI behm.bd vHHB.OI mvmc.dl hnmc.o nNHH.o undid
mN vmm.c nvnm.o whmm.nd ownm.ct moco.~t ovmc.c VONH.O omdzq
hm mom.o nmmd.ml hnhh.hd Humo.cl bMHH.HI Hmno.o Oman.o mwdsa
mm mom.o owmo.~t «moo.mu bmom.0I emoc.at Homo.o awnu.o mudzq
mm nmm.o mmcw.¢l ncnm.wd Novw.OI bvoo.Hl momo.c NHO~.o Hmdzq
Am «A an .n ”a R .a an 553%
.mmmwmmmmmma mflammwfl “newfiozmaom Daucaso

ofionnomad ecu no sawumuwunuuusH Dwuuosoou ecu no musoauauuooo .N manna

21

 

 

 

HHH mmm.c ~mm~.m nume.o~n been.n| bmem.wt "665.6 mvvm.o mHHOq
ooH vmm.o nvmv.~nt mnhm.no «NnH.MI. cnvm.mt cmhv.c mmmn.° vnHQA
em cmm.o movm.bat emba.vm veo~.vt amuo.vl «bon.o m~mn.c «HHUQ
em mmm.o «mmm.n| emmv.~o neve.~t homo.nt hme~.c Nevm.c eNHuQ
ms mam.c wm-.h~ humm.ev memo.n| hmev.nt chum.o woun.c NmHuq
mu bum.o nmhh.mvt Nomm.ooa hemh.ml mmec.mt mnen.¢ enbn.o «anon
we omm.c cvmo.mt oom~.ow hmnm.~l nmvm.~t nth.c HON~.o eHua
he Hmm.o nnna.~t mmne.om hHHN.~I cmnn.nl onem.c Hbm~.o meuq
he nmm.o nmve.~ nmnh.hv mamv.~t nhnh.NI mmma.o mmm~.c mmHuq
«e mmm.o mehv.mul memo.me envv.nl Nmon.vl omm~.o mnmn.o demo:
co nmm.o «mnb.hl ovoo.~e mma0.nt cmov.nt camn.c emm~.o ENHOQ
cm «mm.o Hoeh.bt hnan.oe nhen.~I «mum.wl momd.o om-.o mnHQa
mm Nam.c mhnn.o~t ommn.nm cwmh.nt «hmm.nt ceNH.c uhNn.o VHHQA
om Nam.o mmmo.ea Humo.mn memn.o: ooHn.NI hmo~.c hmH~.o mNHQq
om mmm.o mmmh.el mmmn.vo NHNo.o| mmvo.a1 ovmu.o ammh.o cNHQq
we nmm.o mono.oa mnvm.n~ anna.n| nhem.ul nova.o hmn~.o .hnHQa
ve vmm.c ehb¢.ndt voo~.mn hamv.ml nmma.ml mnmc.o nhmH.o nHHuq
vv omm.o meho.ot «woo.en Nunh.al meem.ut oumo.o uwmn.o @NHOQ
ov Hmm.o och.HHI mnmn.an ehmm.ut mevb.ul mhho.o mmmu.c mnHUA
mm mmm.c hdhn.m hmca.hn Heme.OI vnmm.n| amho.o «unn.o mHHuq
mm vmm.o momv.n noon.mn mmna.ct wevn.ut nono.c hmho.c onHQq
mm mum.o Hhhm.m hubm.n~ Haem.63 amne.ut «chc.o eHnH.o mnHuq
mm mum.o momm.n oncm.ud emnH.OI brom.cl mmeo.o oomo.o anon
an «no.6 h~e¢.n hanw.c~ Hmon.o amnv.al oeoa.o NmNH.o emnHQq
hm umm.o neme.m vmvm.vu emwo.o nmmn.at o~oo.o omau.o nnHuq
Am «A an .n "a E .n on 553%
.mmwmmmwm mwammwm ”Heusocxuom educuso
eunuceoHd ecu mo cowunueumueucn Dauuesoeo ecu uo museueuumeoo .n eacns

22

1.00-

0.75 -

0.50 -‘

b0

0.25 -

 

 

 

oco . , . . . , . ,
20 60 100 140 180
Standard Length (mm)

o--- Lepomts glbbosue
e — Lepomle machrochlrus

Figure 4. The Ontogenetic Trajectory of the Width of the
Posterior End of the Fifth Ceratobranchial (b0)
Versus Standard Length.

b1

Figure 5.

23

 

 

 

I 1
140 180

I
20 60 100

Standard Length (mm)

o--- Lepomie glbbosus
e — Lepomle machrochlrus

The Ontogenetic Trajectory of the Width of the
Anterior End of the Fifth Ceratobranchial (b1)
Versus Standard Length.

24

 

 

 

‘8 T I ' 7
20 60 100

140 . 180
Standard Length (mm)

o--- Lepomls glbboeus
o — Lepomls mechrochlrue

Figure 6. The Ontogenetic Trajectory of the Tangent of the
Outline of the Posterior End of the Fifth
Ceratobranchial (b2) Versus Standard Length.

b3

25

 

 

Figure 7.

 

Standard Length (mm)

om Lepomls glbbosus
e — Lepomls machrochlrus

The Ontogenetic Trajectory of the Tangent of the
Outline of the Anterior End of the Fifth
Ceratobranchial (:5) Versus Standard Length.

26

125
-r o /

u 0
100 /

b4

 

 

 

T T m
140 180

T T
20 60 100

Standard Length (mm)

o--- Lepornle glbbosue
e — Lepomts machrochlrue

Figure 8. The Ontogenetic Trajectory of the Change in
Tangent of the Outline of the Anterior End of the
Fifth Ceratobranchial 0%) Versus Standard Length.

27

 

 

 

 

301
O O

4
20"

I' O
10“ O O C o

O
03%; ° '. __
.0. \_ 1‘ O
‘W\
In 10‘ :8 OOo\\ .
c ' ”a \
\
‘ O \ .
O \

-20-' O \\

a C

\\
-30“
O

.40-

4
~50 fl I v I for ' I I I ' I T I I I
20 4O 60 80 100 120 140 160 180

Standard Length (mm)

em Lepomls glbboeue
e — Lepomle machrochlrue

Figure 9. The Ontogenetic Trajectory of the Change in Tangent
of the Outline of the Posterior End of the Fifth
Ceratobranchial (:5) Versus Standard Length.

28

Because the coefficients within these three groups behave
differently during ontogeny, it seems that these correlations
have few implications for the main. purpose of this study.
However, because these correlations have a complex basis
(including possibly both biological and computational origins)
and a difficult interpretation, further study of correlations
among parameters of polynomial equations in similar situations
is needed.

Ontogenetic trajectories for the coefficients of the
geometric quintic are presented in Figures 4 through 9. To
a good first approximation, the value of each coefficient
appears to be linearly related to standard length. Over
ontogeny, the coefficient bo is most highly correlated with
standard length (Figure 4; for bluegills r2=0.94, for
pumpkinseeds rz=0.85) . The quality of the fit of coefficients
b1 through b, varies from good to fair. Coefficient b1
correlated well with standard length (Figure 5: for bluegills
r2=0.83, for pumpkinseeds r2=0.69). Coefficient b2 also
correlated reasonably well with standard length (Figure 6:
rz=0.88 for bluegills, rz=0.58 for pumpkinseeds). For
coefficient b3 (Figure 7), both species show fair correlations
with size (r2=0.61 for pumpkinseeds, rz=0.58 for bluegills).
Coefficient b, (Figure 8) correlated well with standard length
for bluegills (rz=0.90) and pumpkinseeds (r2=0.77).
Coefficient b5 does not appear to correlate with body size for

the two species (Figure 9: bluegills r220.04, pumpkinseeds

29

rz=0.18) . The variance inherent to b5 tends to increase with
fish body size. Because a log transformation can frequently
improve the fit in situations where variances increase with
size, I also regressed the log transforms of b5 and standard
length. This transformation resulted in a worse fit of the
model (Figure 10; bluegills, 380.0063: pumpkinseeds,
1¥=0.0856). I conclude that there is no clear association
between bs value and standard length.

Coefficients bo describes the width of the posterior end
of the ceratobranchial (the point of insertion of the fifth
branchial adductor). The slope of the ontogenetic change of
bo *with size appeared to differ significantly between
bluegills and pumpkinseeds. Inspection of residuals, though,
indicated that one pumpkinseed, labeled LGIZO, has an
unusually wide posterior end for its size (Figure 4).
Examination of the photograph of this ceratobranchial revealed
that its posterior region was slightly occluded by overlying
soft tissue and the thickness of the posterior edge was
overestimated. Therefore, this specimen was excluded from
analysis of b0. Digitization of the remainder of the form
appeared reasonable and the rest of the coefficients for
specimen LGIZO were not removed from the analysis” The slopes
of the ontogenetic trajectories of bo in fact do not differ
significantly between species (Table 6: standard.length by
species interaction is not significant). Likewise, there is

statistically no detectable difference between intercepts for

3O

 

 

 

 

 

Table 4. Linear Regression of the Geometric Coefficients
versus Standard Length.
ggpgmig macrochirus Rafinesque
Coefficient n r2 Y - mx + b
b0 25 0.94 Y - .0044x - .02435
b1 25 0.83 Y - .00295x - .02847
b2 25 0.88 Y - -.03273x + .12152
b3 25 0.61 Y - -.01926x - .15357
b4 25 0.90 Y - .48884x + .87434
b5 25 0.04 Y - .0447lx - 5.7218
‘Egpgmig gibbosus (Linnaeus)
coefficient n r2 Y’- mm + b
b0 25 0.41 Y - .004lx + .03397
b0 24 0.85 y - .0044x - .00425*
b1 25 0.69 Y - .0061x - .13597
b2 25 0.58 Y - -.0403x - .26624
b3 25 0.58 .Y - -.0523x + .98594
b4 25 0.77 Y - 1.0646x - 14.9917
b5 25 0.18 Y - -.2312x + 5.9002

* Specimen LGIZO was excluded from this regression.

 

 

31

Table 5. Correlation Analysis for the Interaction between
Geometric Coefficients with Size Partialed Out.

Islamic 911229.525 (Linnaeus)

 

 

 

b0 b1 b2 b3 bl. b5
b0 - ns ns ns ns ns
b1 - ns 0 . 66'" ns 0 . 68'" .-..
b2 - -o . 46" -0. 85"” ns
b3 - -o . 63"" o . 88'"
b4 - -o . 49"
b5 "

Lgpgmig EQEIQQDLIES Rafinesque

b0 )01 102 b3 b4 lbs
100 - 0 . 46" -o . 77' ns ns ns
b1 - -o.43' 0.75"" ns 0.63""
b2 - ns -0 . 81'" ns
b3 - -0 . 46* 0 . 86""
b, - -o . 46"
b5 ‘

 

*

' p < 0-5. " p < 0.05, ” p < 0.005

 

32

 

 

 

‘

Lengthof Normal Lines (mm)
ID

d

Figure 11.

33

Lepomls glbbosus

 

 

 

 

Lepomls macrochlrus

 

 

 

 

 

 

.2 .4 .6 .8 1
Relative Distance Along Medial Axis

Expected Curve Shape of the Medial Side of the
Fifth Ceratobranchial of Pumpkinseed and Bluegill
Sunfish. Calculated Curves Constructed for
Specimens Having Standard Lengths of 31, 41, 51,
61, 71, and 111 mm.

 

 

34

these two trajectories (Table 7). Although the posterior end
of the ceratobranchial in pumpkinseeds averages slightly wider
than in bluegills (Figure 4), the difference in regression
intercepts is well within the range of variation seen within
species. Both species appear to share a common pattern of
growth of the posterior width of the fifth ceratobranchial,
best described by the equation y-0.0044x-0.0143.

Coefficient b, describes the width of the anterior end of
the ceratobranchial. The anterior portion of the lower
pharyngeal jaws of adult pumpkinseeds is usually broader than
in adult bluegills (Figure 5). The slope of the regression
of coefficient b1 on standard length is significantly greater
(Table 6, p<0.001) in pumpkinseeds than in bluegills (Table
4: bluegills, slope= 0.00295: pumpkinseeds slope= 0.0061).
The two regression lines intersect at a standard length of 34
mm and the data points at smaller sizes do not differ
appreciably between species. The two species seem to diverge
in anterior ceratobranchial width relatively early in
ontogeny. Faster growth in this feature may provide
pumpkinseeds with greater surface area for stronger
articulation with the anterior pharyngeal elements to better
support the large forces generated in crushing snails.

Divergence between species of the trajectory of :5 is
less marked than for b1, b3, or b‘. Coefficient b2 is the
tangent at the posterior end of the outline curve. 'The values

of b2 influences the shape of the geometric regression on the

35

posterior third of the curve (Figure 6). Slopes of the
trajectories of bz are marginally distinguishable (p=.033:
Table 6). Values of b2 are consistently smaller in
pumpkinseeds than in bluegills, even in the range of sizes
(<40 m) where the two species become indistinguishable in
coefficients b1, b3, and b‘. Because the slopes of the
trajectories differ between species, the trajectories must
intersect. Whether they do at some ontogenetic stage before
the earliest ones included here, or whether the
ceratobranchials of the two species begin development with
different values of b2, can not be resolved with the data at
hand. Within the ontogenetic range studied here, the
contribution of b2 to the shape of the posterior arm of the
ceratobranchial decreases with size. The decrease is slightly
faster in pumpkinseeds. Overall, the aspects of shape indexed
by b2 seems of minor importance in determining differences
between bluegills and pumpkinseeds.

Coefficient b, is the tangent of the outline of the curve
at the anterior end and influences the shape of the regression
in the anterior third of the curve (Figure 7). As with
coefficient b2, coefficient b3 decreases as body size increases
(Figure 7) . The decrease is faster in pumpkinseeds (Table 4:
slope=-0.052) than in bluegills (Table 4: slope=-0.019) . The
two ontogenetic trajectories for coefficient b3 intersect at
34 mm standard length (Figure 7) as do the trajectories for

b1. The ontogenetic trajectories for coefficient b3 appear

36
to be negatively related to the ontogenetic trajectories of
In, the width function of the anterior end of the
ceratobranchial (Figure 5 and 7). However, the high degree
of association between these two variables is accountable to
high correlations with size (Table 5).

Coefficient b, represents how rapidly the tangent of the
outline changes at the posterior end of the curve. Its value
affects curve shape most in the posterior part of the middle
third of the curve (Figure 8). Pumpkinseeds change
significantly faster in this coefficient with size than do
bluegills (Table 4: Bluegills slope=0.489: pumpkinseeds
slope=1.065; Table 6: p<0.001) . The ontogenetic trajectories
of coefficient bu diverge earlier during ontogeny (at 28 mm
standard length) than for the other coefficients (Figure 8).
How the tangent of the outline changes near the posterior end
of the ceratobranchial strongly influences the shape for the
edge where the fifth ceratobranchials articulate. The more
rapid. growth of the tooth. covered occlusive surface in
pumpkinseeds produces the flared posterior edge of the fifth
ceratobranchial (Figure 12). Examination of the ontogenetic
trajectory of coefficient b, for other euteleost fishes may be
useful in understanding’ the evolution. of a fused fifth
ceratobranchial in these pharyngognathous fishes.

The slopes of the ontogenetic trajectories for bluegills
and pumpkinseeds were statistically indistinguiShable. in

coefficients bs (Figure 9). This coefficient is the analog

37

of b‘ (Figure 8) . Bluegill ceratobranchials change relatively
little in coefficient b, over ontogeny (slope=0.045) but
pumpkinseeds show a general decrease with size (slope=-0.23).
However, the variance within species is considerable,
preventing the test of parallelism of slopes from detecting
differences between species. ‘Unlike other coefficients, data
for bs suggest that within species variation in this
coefficient may increase with size. The variation in shape
indexed by b5 can not be explained with a size-based model.

The shapes of the fifth ceratobranchial in both.sunfishes
was adequately modeled by the geometric quintic regression.
Figure 11 displays the expected medial outlines of this bone
in pumpkinseeds and bluegills at different sizes (along a
"straightened" medial axis abscissa). This figure summarizes
visually the growth trajectories illustrated numerically in
Figures 4 through 9. ‘The posterior end of the ceratobranchial
developed similarly in the two species. Although similar at
young stages, the width of the anterior tip of the
ceratobranchial grows faster in pumpkinseeds than in bluegills
yielding an adult pumpkinseed ceratobranchial with a wider
anterior end. The slopes or tangents of the posterior and
anterior regions of the ceratobranchial are negative and

become more steeply

t1

 

38
Table 6. Statistical Tests for Regression Line Parallelism

for Inter-Species Comparisons of the Geometric
Coefficients versus Standard Length Regressions.

Coefficient b

 

0
Source of Sun of df Mean F Sig. of
variation Squares Square F
Wi mini-Residual . 09285 45 . 00206
SL 1.14551 1 1.14551 555.148 0
Species .00354 1 .00354 1.715 .197
SL by Species .00003 1 .00003 .015 .904ns
(model) 1.14908 3 .38303 185.626 0
(total) 1.24194 48 .02587

ns - not significant

Coefficient b1

 

Source of Sun of df Mean F Sig. of
variation Squares Square F
Within+Residual .30503 46 .00663

SL .7949? 1 .79497 119.888 0
Species .06994 1 .06994 10.548 .00218
SL by Species .10610 1 .10610 16.000 .00023***
(model) .97101 3 .32367 48.812 0
(total) 1.27604 49 .02604

*** p <.001

Table 6 (cont'd.).

39

Coefficient b2

 

 

Source of Sun of df Mean F Sig. of
variation Squares Square F
‘Within+Residual 21.86753 46 .47538
SL 76.02410 1 76.02410 159.922 0
Species 10.23439 1 10.23439 21.529 .00003
SL by Species 2.28600 1 2.28600 4.809 .03341*
(model) 88.54449 3 29.51483 62.087 0
(total) 110.41202 49 2.25331

* p (.05

Coefficient b3

Source of Sum of df Mean F Sig. of
variation Squares Square F
Within+Residual 38.77362 46 .84290
St 42.18086 1 42.18086 50.042 0
Species 7.47786 1 7.47786 8.872 .00461
SL by Species 11.68191 1 11.68191 13.859 .00054***
(model) 61.34064 3 20.44688 24.258 0
(total) 100.11426 49 2.04315

*** p <.001

Table 6 (cont'd.).

40

Coefficient b

 

 

4
Source of Sum of df Mean F Sig. of
variation Squares Square F
Within+Residual 5851.50169 46 127.20656
SL 22544.35223 1 22544.35223 177.226 0
Species 3824.36815 1 3824.36815 30.064 1.718-6
SL by Species 3552.78130 1 3552.78130 27.929 3.368-6***
(model) 29921.50168 3 9973.83389 78.407 0
(total) 35773.00337 49 730.06129
*** p <.001

Coefficient b5
Source of Sun of df Mean F Sig. of
variation Squares Square F
Within+Residual 7360.20445 46 160.00444
SL 6.62088 1 6.62088 .041 .83970
Species 19.00690 1 19.00690 .119 .73192
SL by Species 611.66709 1 611.66709 3.822 .05665ns
(model) 637.29488 3 212.43163 1.328 .27690
(total) 7997.49932 49 163.21427

* p (.05

 

a.28mm
9
A t

Posterior

b.100mn1
Appophysealarea

Figure 12. Growth Pattern of the Appophyseal Edge in the
Fifth Ceratobranchial of Pumpkinseed Sunfish.

 

Anterior

42
negative with increased body size. The slope of the outline
at each end was always more steeply negative in pumpkinseeds
than in bluegills. The slope of the outline at the posterior
end changes faster in pumpkinseeds than in bluegills. This
influences the shape of the tooth covered occlusive shelf of
pumpkinseeds which forms the edge along which the two
ceratobranchials articulate. The ontogenetic trajectory for
the influence of anterior curve acceleration was quite
variable among individuals and this aspect of ceratobranchial
shape is probably explained better by non-ontogenetic

variables.

43

 

nop.msn oasH.noA emso.vu mans.eu nssem.o mavo¢.o ”an
use.asr ossm.me moom.~r caan.nu nopmn.c macnn.o on
mam.oar msmm.oo ssns.~r amma.nr nasa~.o mason.o A»
amn.ar aas~.mm mmes.~r oe~m.~r neee~.o muoe~.e we
no~.er mesa.me ¢¢o~.~u me~s.~n naen~.o maeo«.o He
546.5: mmue.e¢ mmem.sr enam.~r nemo~.o ma~e~.c em
Hmm.mr amen.an came.sr mamn.~u names.o mac-.c Hm
mne.¢r mmsa.nn mmav.au oc~A.~r "was”.o mamma.o we
mam.nr aeme.o~ vama.sr mesm.Ar nH¢HH.c maesa.o as
n~4.~u annn.n~ meas.or opas.fiu nonoo.o mavma.o on
so~.Ar moao.oa vane.cr mmam.ur mammo.o manna.o an
cueceq
an .n ma ~n .n on eueeeaum

“unenccuqv mmwmmmflm mflflmmwfl

.hsevouco umou0< menecm Hewcuceucoueueo
oeueenxm you nucewuueoo Ouuuefioeo oeumasoneo

.b OHQMB

44

 

omms.or «Ans.mm osm~.~r msam.nr omom~.o movev.o AHA
mnmn.~- ase~.mn made.an omen.~r nsmm~.o mooan.o on
«ssm.~- omam.mn oanm.au n~o~.~r emomA.o moae~.o as
mosa.~r mens.nn seme.flr sano.~r numea.o mooe~.o we
memm.~u enme.on vamn.au omso.ar scama.o mccv~.o «a
oma~.n- vm¢~.a~ H~n~.H- «HHF.AI neenH.o mo-~.c em
maes.nr ~moa.m~ amnA.Ar ssem.nn amANH.o mooc~.o Hm
Hmee.n- osen.n~ mano.su scon.ar nmsca.c moosa.o as
smma.nr mesm.o~ nmnco.r vo-.su ovmao.o mccmA.o He
-HH.¢- oase.ma noeqo.u aemo.sr nuaso.c mcqnfl.o on
mmmn.e- sumo.ea neoma.- Hnma.or ommeo.o mamas.o an
caucus
an .n as ~n Pa on eueeeaum
maem0:uumm mmuacmmucuuermqammuq

.c.e.ueoo. u wanna

DISCUSSION

The morphometric technique presented here uses the medial
axis in a nonconventional way to describe the ontogenetic
shape changes of the fifth ceratobranchial in pumpkinseed and
bluegill sunfish. Original applications of the medial axis
were used to visually represent complex shapes (Oxnard, 1980)
or to provide landmarks for measuring lengths and angles of
medial axis arcs (Bookstein et. al., 1985). The technique
used for this study is nonconventional to these other medial
axis applications in that the medial axis is used as a
landmark for setting a coordinate system for numerically
analyzing the curvature of the ceratobranchial bone.
Numerical approaches are easier than purely visual descriptive
methods for discerning subtle changes in shape and.provide the
means for statistical evaluation of shape changes. For
centrarchid ceratobranchials, the medial axis is ideal for
testing whether this technique can provide a means to
mathematically model curvature of this simple landmark free
form. Limitations of the technique are few and may be
particular to this data set. Most difficulties encountered
usually were involved with digitizing the ceratobranchial
outline. These situations were easily avoided by tracing the

outline of the form on clear acetate.

45

46

The medial axis for centrarchid fifth ceratobranchial
bones features a long arc which represents the anterior-
posterior axis. This major axis of the ceratobranchial was
always present in both bluegill and pumpkinseed sunfishes.
The major axis of the fifth ceratobranchial is the
"operational homology" used for comparative purposes in this
studyu Branch.points were usually found.only at the endpoints
of the medial axis in bluegill ceratobranchials and were
present there as well as elsewhere along the medial axis in
pumpkinseed ceratobranchials. Because of the noise in
digitizing the posterior end of the ceratobranchial (J.E.
Zablotny, pers. obs.), branch lengths at the posterior end of
the fifth ceratobranchial were not included within the sample
of normal lines. The side branches at 45 to 58% of the
distance along the major medial axis were only found in large
individual pumpkinseeds and in only two of the larger bluegill
specimens.

The individual ontogenetic 'trajectories. of ‘the shape
descriptors of the fifth ceratobranchials in bluegill and
pumpkinseed sunfish are linearly related to size. This is a
nontrivial result; inspection of an age series of
ceratobranchials would not easily suggest a linear shape
change over ontogeny because of the curvature of the bone.
This seems contradictory to the expected since shape change
is usually modeled as a nonlinear ontogenetic process (e.g.

Alberch et. al., 1979). The curvilinear relationship of shape

47

with time should asymptote to a target adult morphology.
Perhaps, the strictly linear change of shape descriptors with
size found here is due to bias in sampling fishes within 25
and 155 mm standard length. This size range is only a subset
of the possible size range (both species can exceed 350 mm
standard length: Lee et. a1. 1980) and especially excludes
large male bluegills. However, very large specimens of
bluegill and pumpkinseed sunfish are uncommon in nature and
none greater than 150 mm were available for this study. The
bias towards small individuals reflects the availability of
specimens from museums. It is likely that large specimens
greater than 150 mm may display different ontogenetic
trajectories for shape descriptors of the fifth
ceratobranchial than the values reported here.

Size also has some drawbacks as an index of age in these
centrarchids. Growth rates in fishes are directly related to
available resources and overcrowding tends to reduce resource
availability and.to decrease growth rates. Dominey (1980) has
shown that size related alternate breeding strategies for
centrarchids occur in naturural populations as well. Dominey
(1980) discovered subtle phenotypic differences between
"sneaker" and territorial breeding male bluegill. Rapid
sexual development in "sneaker" males may indirectly affect
the development of other structures. The sexual condition of
the specimens used in this study was not ascertained and this

phenomena.may have contributed to the variance observed in the

48
shape descriptors that were correlated with size. However,
the relatively tight associations between shape descriptors
and size suggests that these possible drawbacks pose little
problem for this study; if anything, identification of these
effects would probably increase the goodness of fit of the
linear relationships.

The specializations in the shape of the fifth
ceratobranchial in adult pumpkinseeds are presumably derived
from an ancestral shape within Lepomis. Phylogenetic
understanding of the genus Lepomis is limited at this time.
The phylogenetic arguments for ‘the evolution of trophic
specialization in Lepomis presented here are generalizations
based on current knowledge of the group. Humphries (pers.
com.) suspects that the warmouth sunfish, Chaenobryttus
gulosus, is the outgroup for Lepomis. Initial changes in the
shape of the fifth ceratobranchial probably were derived from
an ancestral sunfish possessing similar pharyngeal jaw
morphology to the pharyngeal jaw type found in Chaenobryttus.
As adults, species of Lepomis have lower pharyngeal jaws
featuring various degrees of apposition between right and left
fifth ceratobranchial bones (see Trautman, 1981, p. 585,
Fig.14l, #11; p.589, Fig. 142, #9; p.580, Fig. 140, #9; and
p. 599, Fig. 145, #10). By visual inspection, bluegill fifth
ceratobranchials are intermediate in the extent of apposition
and I base the changes in morphology of pumpkinseed

ceratobranchials in terms of deviation from bluegill

Em
C

 

49
pharyngeal jaw shape. Small individuals of species of Legging
appear to have apposition sites similar to the apposition site
in warmouth sunfish lower pharyngeal jaws.

The primitive ceratobranchial shape features a rather
narrow occlusive shelf filled with fine, filiform teeth.
Additionally, primitive fifth ceratobranchials possess a
narrow apophysis or articulating surface located anteriorly
between the right and left fifth ceratobranchials elements
(Figure 8) . The ceratobranchial shapes mostly associated with
snail crushing include: a broad crushing surface studded with
molariform teeth, a broad region apposition of the left and
right halves of the lower jaw and an increase in the lateral
depth of the ceratobranchial. Ideally, the processes of
morphological specialization in the fifth ceratobranchial need
be addressed to understand the type and roles of mechanisms
which have occurred in the evolution of the genus Lepomis.

The process for morphological differentiation in trophic
structures of the Centrarchidae have not been addressed, Fink
(1982) mentions that in nature "changes in developmental
timing and their epigenetic consequences are suspected of
being instrumental in acquisition of evolutionary novelties,
including those often major changes associated with large
scale cladal diversity." The gastropod crushing type of
pharyngeal jaw develops from a precursor shape common to
bluegill sunfish. All of the shape descriptors within this

study are similar between bluegills and pumpkinseeds at some

50

point in development, and some descriptors have identical
trajectories in both species. This means that both species
have common shapes at some time in development (Figure 9).
For body sizes less than 35 mm standard length, pumpkinseeds
and bluegills have lower pharyngeal jaws of very similar
shape. The ceratobranchial elements of the viscerocranium are
endochondral (Harder, 1975) and do not begin to ossify until
fish exceed 20 mm standard length. The data from this study
do not address species differences in growth or shape of the
cartilaginous model of the fifth ceratobranchial. However,
the data strongly suggest that divergence in shape occurs at
sizes greater than 34 mm in standard length. Since the
cartilaginous model is nearly (but not completely) ossified
in fishes of about 28 mm standard length (J. E. Zablotny,
pers. obs.), the ontogeny of the bony fifth ceratobranchial
begins at a very similar shape in both species.

The shape descriptors exhibiting wide differences in
ontogenetic slopes between pumpkinseeds and bluegills always
have greater absolute values of slopes in pumpkinseeds. The
morphological differences in the two species of centrarchid
lower pharyngeal jaws may be easily explained in terms of a
common hetrochronic change. Increased ossification rates of
the occlusive tooth bearing surface of the pharyngeal jaw in
pumpkinseeds accelerate the development of width in
pumpkinseed pharyngeal jaws as compared to pharyngeal jaw

development in bluegills.

51

Posterior

 

Anterior

Figure 13. Primitive Fifth Ceratobranchial Shapez.
WW

52

.maaumeasm
use moeemcwcnssm .cumceu puepceun as mmvv Hanan
seesuem enecm Hewcoceucoueueo cuuum cw auwueawawm

 

 

.vH eusmwm

'

l
av<24
020..
Houueusc nowueueom

I i

if
DN<24 .104

 

 

 

 

 

53
DEVELOPMENTAL RELATIONSHIP BETWEEN MORPHOLOGY AND BEHAVIOR

Mayr (1960, 1976) proposed that behavior always precedes
morphological change during evolution. Although Bock (1976)
mentions that the study of structures by themselves may not
be adequate to evaluate adaptation, the shape analysis
techniques employed in this study, and published ecological
data on sunfish feeding, provide the means for examining
ontogenetic shape change as a set of continuous values and
permit hypotheses construction in terms of ontogenetic change
and ecological correlation. The appearance of new behaviors
may invoke changes in the selection of structures that are
used in the behavior pattern (Mayr, 1960).

The data from this study reveals that pumpkinseed fifth
ceratobranchial shape is well differentiated from bluegill
ceratobranchial shape at approximately 50 mm standard length.
The evidence I gathered suggests that the initial shaping of
the fifth ceratobranchial appears approximately from 28 to 34
mm standard length, well before the trophic switch to snails
by pumpkinseeds. However, Mittlebach (1984) notes small
pumpkinseeds do include a small portion of pond snails in the
diet whenever small hatchling snails are present. Young
pumpkinseeds may be severely constrained by gape size which
prevents them from harvesting larger snails. Smaller
pumpkinseeds are also known.tolcrush small snailsrduring times
of high availability (Mittlebach, 1984). However, these'

snails make up only a small percentage of the total diet and

54
do not reflect a major change in diet for small pumpkinseeds
(Mittlebach, 1984). This may reflect the lack of
morphological specialization in small pumpkinseeds which is
necessary for the fifth ceratobranchial to structurally
accommodate the compressive forces produced by pumpkinseeds
during bouts of snail crushing.

Morphological constraints may negatively affect the
energetic reward for switching to other'prey types in response
to interspecific competition. Behavioral plasticity in
switching behavior appears to occur in nature. For localities
in South Michigan, Mittlebach (1984) noted that pumpkinseeds
begin switching to snails from 45 mm to 50 mm in standard
length. Pumpkinseeds greater than 70 mm standard length
readily take snails as the major dietary component.
Ontogenetic changes in the shape of the fifth ceratobranchial
with body size in pumpkinseeds may be associated with a
decrease in handling times for processing pond snails as body
size increases in pumpkinseeds (Mittlebach, 1983). Osenberg
and Mittlebach (1988) note that large pumpkinseeds have a
greater probability of crushing a snail shell than small
pumpkinseeds. Although small pumpkinseeds are able to crush
snails, these individuals require longer handling times and
tend to reject more snails that are unsuitable for crushing
(Osenberg, 1988). The behavioral switch to snails in
pumpkinseeds may be consistent for regions where other

sunfishes are in sympatry. Ontogenetic shape constraints for

55
pumpkinseeds may not provide competitive parity with other
centrarchid species within the community.

Likewise, pumpkinseeds are usually the only
representative Lepgmis and mollusc crushing fish species in
the northern part of its range. At the time of Pleistocene
glaciation, Bailey and Smith (1981) noted that L. W was
the only Lepgmis to occupy the Atlantic drainage refugia.
Scarola (1979) mentioned that pumpkinseeds are the most common
sunfish in the Northeast. The feeding habits for allopatric
populations of pumpkinseeds differ from those found in
sympatric with other centrarchids. Confer and Blades (1975)
remarked that pumpkinseeds from New England ponds utilize a
stereotypical behavior when capturing suspended zooplankton,
a preferred prey item found in open water habitat. Apparently
pumpkinseeds can be encountered in littoral zone habitat in
some of these Northeastern lakes (C. Folt, pers. com.). In
New Brunswick, Reid (1930) stated that dragonfly nymphs were
the preferred prey item taken by pumpkinseeds. The causal
agents for this behavior pattern and the associated pharyngeal
jaw morphology of these populations have yet to be studied.
The appearance of crushing behavior in pumpkinseeds may be
initiated by the presence of superior competitors that prey
on zooplankton.

Pumpkinseeds from different lake communities may be under
different selection pressures for rates of development in the

pharyngeal jaws. I hypothesize that pumpkinseeds from high

56

competitive environments should express greater developmental
acceleration for producing the crushing type morphology than
those from noncompetitive lake systems. Behavioral plasticity
in switching behavior may indirectly provide the natural
selection for increased rates of growth and shape modification
of the pharyngeal jaws in pumpkinseeds in competitive
environments.

Resource partitioning appears to be a major mechanism for
alleviating competitive effects in centrarchid communities
(Werner and Hall, 1976) and may factor heavily in the natural
selection of trophic morphology. Keast (1980) remarks that
resource partitioning is commonly used by larval and post
larval fish communities at peak periods of utilization while
these young fishes are relatively undifferentiated regarding
trophic structures. Similarly, there are no significant
morphological differences between bluegills and pumpkinseeds
less than 34 mm standard length that might serve to alleviate
interspecific competition. For specimens less than 35 mm
standard length, t-tests show no significant differences
between bluegills (n=9) and pumpkinseeds (n=6) for the shape
descriptors b0, b1, b4, low significance for be, and b3 and high
significance for b5. However, this high significant
difference between pumpkinseeds and.bluegills for coefficient
In is suspect due to the low correlation with standard length.

Resource partitioning in larger centrarchids has also

been noticed for competition based centrarchid communities.

 

 

57
Werner and Hall (1976) demonstrate decreased growth rates in
addition to niche shifting for coexisting pumpkinseed,
bluegill, and green sunfishes. Werner and Hall (1976) believe
that the niche shifting into suboptimal diets are results of
interspecific competition in centrarchid fish communities.
High predation risk by largemouth bass also forces small
bluegills to switch to vegetation based prey from open water
prey (Werner et. al 1983). Unlike pumpkinseeds, the switch
in habitat type and prey is not correlated with morphological
changes in the pharyngeal jaw of bluegills. The shape
analysis of the fifth ceratobranchial hints that a
morphological bottleneck may prevent young pumpkinseeds from
effectively using feeding refuges distinct from the diets of
its competitors while at small body sizes. iExamination of the
plots of the expected shapes (Figure 5) indicates strong
similarity in shape of pumpkinseed ceratobranchials to
bluegill ceratobranchials for the smaller size classes.
MORPHOLOGICAL PLASTICITY

Epigenetic modification (Waddington 1953, 1956a, 1956b)
or morphological plasticity (Smith-Gill, 1983) occurs in the
trophic structures of fishes. Meyer (1987) , Sage and Selander
(1975), and.Kornfield and Taylor (1983) all noted the presence
of phenotypic plasticity in trophic morphology of certain
species of cichlid fishes. Meyer (1987) strongly suggests
different feeding behaviors as the principal influence on

trophic morphology in Cighlasoma managuense. Rubin and Lanyon

58
(1984) mention that minor alterations in loading on bone can
drastically remodel bone.

Within the Centrarchidae, Ehlinger and Wilson (1987)
found intraspecific morphometric differences between bluegills
occupying open water and vegetation habitat types. The
pumpkinseed sunfish may be the best species for testing
hypotheses of environmental influences on the shaping of
trophic morphology. The wealth of ecological and
morphological data on pumpkinseeds provide adequate background
for understanding the sources of morphological plasticity
observed in nature. Unlike redear sunfish, pumpkinseeds
switch feeding modes between crushing and pharyngeal transport
when presented hard and soft food types (Lauder, 1983).
Pumpkinseeds may experience force loading during snail
crushing that vastly differ from the forces produced by
pharyngeal transport which may cause new bone remodeling
patterns in the ceratobranchial. Detailed captive breeding
studies may also help to separate any epigenetic effects of
behavioral plasticity that may confound hypotheses concerned
with the genetic basis off structural shapes. I suggest
further studies be undertaken to evaluate the plasticity of
diet switching with regards to community structure in
pumpkinseeds and to correlate the impact of switching behavior
to the ontogenetic trajectory of the morphogenesis of the
fifth ceratobranchial bone.

INFLUENCES OF BEHAVIOR ON MORPHOLOGICAL DIVERSIFICATION

59

Mayr (1976) noted that behavioral changes and the change
in selection pressure usually cause rampant structural
reorganization in the organism. In the centrarchids, no major
repatterning of muscle homologies are observed in the
pharyngeal region. Lauder (1983) tested for differences in
muscle activity and morphology between snail crushing and
zooplankton feeding species of centrarchids. According to
Lauder (1983) the bluegill and pumpkinseed share the derived
condition ("thickened dorsal intermuscular aponeurosis, left
and right obliquus dorsalis 2 and anterior transervsus
dorsalis do not form a continuous sheet of muscle at the
midline") over the primitive condition ("obliquus dorsalis 4
passes anteromedially beneath the posterior transverse
dorsalis anterior fibers") found in.n1gzgpteru§ and the green
sunfish Lepomis cyanellus. This structural repatterning found
in the upper pharyngeal jaw structure is not correlated with
the behavior shift in pharyngeal jaw manipulation and appears
to have arisen independently of the behavior modification in
the centrarchids. Greater cross sectional areas have been
found in the fifth.branchial adductor, levatores externi.3 and
4, pharyngocleithralis externus and internus, and
pharyngohyoideus muscles for pumpkinseeds as compared with
bluegill pharyngeal musculature (Lauder 1983).

Lauder (1983) noted most changes in the musculature of
the pharyngeal jaws involve an increase in area of the

physiological cross section of pharyngeal musculature between

6O

zooplankton feeders and gastropod crushers. Major differences
in function result from changes in muscle firing patterns with
modification in fifth ceratobranchial shape rather than
through the evolution of new muscle groups. For Lemme
species feeding on zooplankton or soft bodied prey, pharyngeal
transport involves an overlapping activity pattern of the
pharyngocleithralis internus with the retractor dorsalis.
Both the upper and lower' pharyngeal jaws retract
simultaneously, moving the food items into the esophagus
(Lauder, 1983) . Coactivation of the pharyngeal muscles adduct
the upper and lower pharyngeal jaws during stereotypical
crushing mode (Lauder, 1983). The redear, pumpkinseed, and
green sunfish depend on this behavioral pattern for crushing
snail shells (Lauder, 1983).

Changes in the shape of the lower pharyngeal jaw
correlate well with the appearance of ecological and species
diversification in fishes. Unlike the more derived
pharyngognath teleost fishes like the Cichlidae, Embiotocidae,
Labridae, Odacidae, and Scaridae (Liem and Greenwood, 1981),
the centrarchidae all share the primitive condition of
possessing an unfused lower pharyngeal jaw. The medial sides
of the fifth ceratobranchials closely appose each other in
pumpkinseed specimens larger than 75 mm standard length (J.
E. Zablotny, pers. obs.). I hypothesize that the broadly
apposing medial edges of the paired fifth ceratobranchial in

large pumpkinseeds serves to reduce the degrees of freedom of

61

movement of the lower pharyngeal jaw while undergoing snail
crushing. To promote crushing efficiency, this may prevent
the halves of the ceratobranchials from yielding and folding
around a gastropod shell during adduction of the pharyngeal
jaws. The nonspecialized centrarchidae feature a narrow
region of close apposition in the lower pharyngeal jaw as
opposed to the broad zone of contact found in pharyngognath
gastropod crushers like the pumpkinseed and redear and
pumpkinseed sunfish.

The results of this study reveal that.a geometric quintic
is reliable for modeling the ontogenetic shape changes in the
fifth ceratobranchial bone in pumpkinseed and bluegill
sunfish. Pumpkinseeds appear to diverge in ceratobranchial
shape from bluegills approximately 35 mm standard length, well
before the ontogenetic diet shift to snails. It is suggested
that the plasticity in switching behavior may influence
selection for more rapidly developing crushing morphology in
pumpkinseeds from highly competitive habitats than those from
less competitive habitats. Epigenetic influences of switching
behavior on fifth ceratobranchial morphology may also occur.
Comparisons of populations of pumpkinseeds allopatric and
sympatric with other centrarchid species may be of use in
testing for epigenetic and. evolutionary changes in
morphological shape of the fifth ceratobranchial. Finally,
the centrarchids of the genus Legging may provide insight into

understanding the evolution of more ecologically complex,

62

speciose assemblages of teleost fishes.

APPENDICES

63
APPENDIX A

DIGITIZED OUTLINES AND MEDIAL AXES

 

 

 

 

 

 

 

 

 

 

 

Q::::::T”’r LMA25
—

LMA21
'—

J \LMAZ4
—

LMA26

 

/

—

 

($7

Figure 15. Digitized Outlines and Medial Axes of Bluegill
Sunfish. Specimens are Ranked.by Size from Small
to Large. Scale Distance is Equal to One
Millimeter.

64
Figure 15. (cont'd.).

LMA19

 

 

\

‘

 

LMA23

 

 

LMA18

 

 

 

/
/___
V ——

 

 

 

 

E ; \LMM

“

LMA6

 

 

 

 

 

Figure 15. (cont'd.).

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

E / LMA20
LMA16
5;; W“
——t H.
_ LMA17
h :._

 

66
Figure 15. (cont'd.).

 

 

 

 

 

 

 

 

 

LMA2
’> _____
\
LMA9
LMA7

 

 

 

 

 

 

 

 

 

 

 

 

Figure 15. (cont'd.).
N12
—_
LMAS
—
LMA45
\ /
E i
LMA4

 

68
Figure 15. (cont'd.).

F— ANN
/ _ LMA46
E /\

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

69

 

 

\. LGl1

 

 

 

 

LGlas

LGI35
\

Figure 16. Digitized Outlines and.Media1 Axes of Pumpkinseed
Sunfish.

Specimens are Ranked by Size from Small
to Large. Scale Distance is Equal to One
Millimeter.

70

Figure 16. (cont'd.).

LGI82a

 

 

 

 

 

 

 

 

 

 

 

 

 

 

l

71

Figure 16. (cont'd.).

 

 

 

 

 

 

LGI13
—

LGl37

LGIZO

LGI25

 

 

72
Figure 16. (cont'd.).

LGI14

 

 

 

 

 

 

 

 

 

 

 

 

 

LGI39
LGI27
.I-;:::r______—‘,a""""—aa//’
LG|41
7““-___

 

 

73
Figure 16. (cont'd.).

LGI22

 

LGl2e

 

LG|12

 

 

 

 

 

Figure 16.

(cont'd.).

 

74

 

 

 

 

 

LGI4

LGI‘IB

 

75
Figure 16. (cont'd.).

LGI16

 

 

76

APPENDIX B

SOURCES AND LOCALITY DATA FOR SPECIMENS

Emilee—1w“:

university of Michigan Museum of Zoology, UMMz
103258- M1., Macomb Co., Missmoris Gravel Pit,
24 April 1935, G. F. Cooper:
1.6113 44m: L61l4 58m: LGIZO 50m: LGIlS 29mm;
16125 58am: 1.6126 64m; L6132A 28am; 1.6133 27mm;
16135 28m;
103263— M1., Maccmb Co., Huron River,
24 April 1935, G. F. Cooper:
1.6128 84m; L614 68m; L6128 671110; 1.6129 67mm;
196846- M1., Huron Co., Pigeon River,
27 July 1908, A. S. Seathers:
L611 28mm: 06130 28mm:
2103258- L6126 44mm;

Field Museum of Natural History, FMM-t
NY., North Rose, Sodus Creek,
21 October 1925, A. C. Weed:
13426- L6134 100m;
13431- 16116 11111111;
13435- L6112 9411111:
13443- 1.6118 75mm; L6122 79mm;

42342- 1LL., Skokie Lagoon,
30 July 1939, L. P. Woods
(.6139 60mm: 06141 64mm;
43300- [.6136 40m; 1.6137 46m;

Lepomis macrochirus:

university of Michigan Museum of Zoology, UMMz
71380- M1., Newaygo Co., Long Lake,
9 July 1926, Langloise and
LMAZ 38m, LMA7 46am, LMA8 37m;:LMA10 39m;
107943- M1., Jackson Co., watkins Lake,
29 September 1934, 1.F.R. Staff:
LMAB 88m; LMA4 100m; LMAG 30m:
113172- M1., Mason Co., Gooseneck Lake,
4 September 1936, E. R. Kuhne:
LMAZO 250m: LMA23 28mm; LMA24 28mm; LMAZS 27mm;
LMA26 28m:
210030- FLA” Lake Co., N. Shore Wildcat Lake,
R. M. and S. Bailey:
LMA9 43m;

77

Field Museum of Natural History, FMNH
4332- LMA20 35mm;
43327- LMAl 30m; LMA16 35m; LMAl? 38m; LMAla 30mm;
LMA19 26010;
43998- LMAlZ 79m;

Personal Collectionp J. E. Zablotny
M1., Eaton Co., Lake Ovid
October 1986: '

-LMA45 93mm; LMA46 144mm: LMA47 128mm: LMA48 148m;
LMA49 158nm

 

BIBLIOGRAPHY

BIBLIOGRAPHY

.Alberch, P., Gould, S. J., Oster, G. F., and D. 8. wake. 1979. Size
and shape in ontogeny and phylogeny. Faleobiology. S(3):269-3l7.

Alexander, R. M. 1974. Functional Design in Fishes. Hutchinson & Co.
LTD, London 160 pages.

Avise, J. C., and M. H. Smith. 1974. Biochemical genetics of
sunfish 1. Geographic variation and subspecific intergradation in the
bluegill Lepcmis machrochirus. Evolution 28:42-56.

Avise, J. C., Straney, D. 0., and M. H. Smith. 1977. Biochemical
genetics of sunfish 1v. Relationships of centrarchid genera. Copeia
2:250—258.

Bailey, R. M., and G. R. Smith. 1981. Origin and geography of the fish
fauna of the Laurentian Great Lakes basin. Can J. Fish. Aquat. Sci.
38:1539-1561

Blum, H. 1973. Biological shape and visual science (Part 1). Journal
of Theoretical Biology 38(2):205-287.

Blum, H., and R. N. Nagel. 1978. Shape description using weighted
symmetric axis features. Pattern Recognition 10:167-180.

Bock, Wk J., and G. von Wahlert. 1965. Adaptation and the form-
function complex. Evolution. 19(3):269-299.

Bookstein, F. L. 1979. The line-skeleton. Computer Graphics and Image
Processing 11:123-137.

Bookstein, F. L. 1981a. Looking at mandibular growth: some new
geometric methods. Pp. 83-103 in: D. S. Carlson (ed.), Craniofascial
biology. university of Michigan Center for Human Growth and
Development, Ann Arbor.

Bookstein, F. L. 1981c. Foundations of morphometrics. Annual Review of
Ecology and Systematics 13:451-470.

Bookstein, F., Chernoff, 8., Elder, R., Humphries, J., Smith, 6., and
R. Strauss. 1985. Morphometrics in evolutionary biology, The geometry
of size and shape change, with examples from fishes.

Special Publication 15, The.Academy of Natural Sciences of
Philadelphia. 277 p.

78

79

Branson, B. A., and 6..A. Moore 1962. The lateralis components of the
acoustico lateralis system.in the sunfish family Centrarchidae. Copeia
l962:1:108.

Confer J. L., and P. 1. Blades. 1975. Omnivorous zooplankton and
planktivorous fish. Limnol. Ocean. 20:571-579.

Dominey, W. J. 1980. Female mimicry in male bluegill sunfish- a
genetic polymorphism? Nature284:546-548.

Ehlinger, T. J., Wilson, D. S. 1987. Complex foraging polymorphism in
bluegill sunfish. Proc. Natl. Acad. Sci. USA, In Press

Fink,‘Wz L. 1982. The conceptual relationship between ontogeny and
phylogeny. Paleobiology. 8(3):254-264.

Harder,‘W; 1975..Anatomy of fishes. Part 1 text. E. Schwizerbart'sche
verlagsbuchhandlung (Ngele u. 0bermdller)., Stuttgart. 612 p.

Keast,.A. 1980. Food and feeding relationships of young fish in the
first weeks after the beginning of exogenous feeding in Lake Opinicon,
Ontario. Env. Biol. Fish. 5(4):305-314.

Kornfield, 1. L., and J. N. Taylor, 1983. A.new species of polymorphic
fish, Cichlasoma munckleyi from.Cuatro Cienegas, Mexico (Teleostei:
Cichlidae). Proc. Biol. Soc. wash. 96:253-269.

Lauder, G. V. 1983a. Functional design and evolution of the pharyngeal
jaw apparatus in euteleostean fishes. 2001. J. Linnaean Soc. 77:1-38.

Lauder, G. v. 1983. Functional and morphological bases of trophic
specialization in sunfishes (Teleostei, Centrarchidae). Jou. Morph.
178:1-21. -

Laughlin, D. R., and E. E. Werner. 1980. Resource partitioning in two
coexisting Sunfish: pumpkinseed (Lepomis gibbosus) and northern
longear sunfish (Lepomis megalotis peltastes). Can. J. Fish. Aquat.
Sci. 37:1411-1420.

Lee, D. 8., Gilbert, C. H., Hocutt, C. H., Jenkins, R. 8., McAllister,
D. E., and J. R. Stauffer. 1980-et seq. Atlas of north american
freshwater fishes. Publication #1980-12 of the Narth Carolina
Biological Survey. 867 p.

Lien, R. F., Greenwood, P. H. 1981. A.functional approach to the
phylogeny of the pharyngognath teleosts. Amer. Zool. 21:83-101.

Liem, K. F., and J. Wt Osse. 1975. Biological versatility, evolution,
and food resource exploitation in African cichlid fishes. Amer. Zool.
15:427-454. '

Mayr, E. 1958. Behavior and systematics. In A. Roe and G. 6. Simpson
(eds): Behavior and Evolution. Yale university Press New Haven, pp.
341-362.

80

Mayr, E. 1960. The emergence of evolutionary novelties. In: The
evolution of life. Ed. S. Tax. Chicago Uhiv. Chicago Press, 349-380.

Mayr, E. 1976. Evolution and the diversity of life, selected essays.
The Belknap Press of Harvard university Press, Cambridge, Mass. 721 p.

Meyer, A" 1987. Fhenotypic plasticity and heterochroney in Cichlasoma
managuense (Pisces, Cichlidae) and their implications for speciation
in cichlid fishes.Evolution, 41(6):1357-1369.

Mittelbach, Gt 6. 1983. Optimal foraging and growth in bluegills.
Oecologia 59:157-162.

Mittelbach, Ge 6. 1984. Predation and resource partitioning in two
sunfishes (Centrarchidae). Ecology 65:499-513.

Mortenson, M. E. 1985. Geometric modeling. John Wiley & Sons, Inc.,
New Yark. 763 p.

Osenberg, C.‘W;, G. G. Mittelbach. 1988. The effects of body size on
the predator-prey interaction between pumpkinseed sunfish (Lepgmis
gibbosus) and gastropods. submitted to Ecological Monographs.

Oxnard, E. C. 1980. The analysis of form: Without measurement and
without computers. .Amer. Zool., 20:695-705.

Reid, H. 1930. A.study of Eu tis ibbosus (L.) as ocurring in
Chamcook lakes., N.B. Can. 8 o . F155. 5(165:457-466.

Rubin, C. T., L. E. Lanyon. 1984. Regulation of bone formation by
applied dynamic loads. J. Bone Joint Surg. 66A:397-402.

Sage R. D. and R. K. Selander. 1975. Trophic radiation through
polymorphism.in cichlid fishes. Proc Nat. Acad. Sci. USA 72:4669-4673.

Scarola, R. 1973. Freshwater Fishes of New Hampshire. New Hampshire
Fish and Game Dept, Div. of Inland Marine Fisheries.

Smith-Gill, S. J. 1983. Developmental.plasticity: Developmental
conversion versus phenotypic modulation. Amer. Zool. 23:47-55.

Sokal, R. R., and F. J. Rohlf. 1969. Biametry.‘W; H. Freeman and
Company, San Francisco. 776 p.

Taylor, wu R. 1967. An enzyme method for clearing and staining small
vertebrates. Proc. U.S. Nat. Mus. 122:1-17.

Trautman, M. B. 1981. The Fishes of Ohio. Ohio State university Press,
Columbus, Ohio.

Waddingtcn, C. H. 1953. Genetic assimilation of an acquired character.
Evolution 7:118-126.

weddington, C. H. 1956a. Principles of embryology. Allen and thin,
London.

81

waddington, C. H. 1956b. Genetic assimilation of the bithorax
phenotype. Evolution 10:1-13.

wetner, E. E., and D. J. Hall. 1976. Niche shifts in sunfishes:
experimental evidence and significance. Science 191:404-406.

wetner, E. E., J. F. Gilliam, D. J. Hall, and 6. G. Mittelbach. 1983
An experimental test of the effects of predation risk on habitat use
in fish. Ecolggy 64:1540-1548.

Zar, J. H. 1984. Biostatistical Analysis. Prentice-Hall, INC.
Englewood Cliffs, N.J. 718 p.

MICHIGAN STATE UNIV. LIBRARIES
lllllllllllllllllllllHlllllllllllllllllllllllllllHIWHNHI
31293007885993