 

 

 

.....

 

llllllllalllllllzlllllllmllIIJUIHIHIWIllHUlllNHll

megs 302074 0985

A» UBRARY
.{x \

6‘ 0—“) a Yuaibgd Stg‘hﬁ

Uni Draw

w.

 

 

 

 

This is to certify that the

dissertation entitled

3-D VOLUME RECONSTRUCTION OF THE DEVELOPMENT OF THE
EYE-TENTACLE COMPLEX IN PULMONATE GASTROPODS:
NEW APPROACHES TO AN EVOLUTIONARY QUESTION

presented by

Timothy Lancaster Murphy

has been accepted towards fulﬁllment
of the requirements for

PhD. Zoology

degree in

 

 

Major professor

Date /7ﬂlcem,é‘~’4 177?

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

 

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

 

DATE DUE

DATE DUE

DATE DUE

 

 

 

 

 

 

 

 

 

 

 

 

 

 

11/00 chIRC/Ddapres-p.“

 

 

3-D VOLUME RECONSTRUCTION OF THE DEVELOPMENT OF THE
EYE-TENTACLE COMPLEX IN PULMONATE GASTROPODS:
NEW APPROACHES TO AN EVOLUTIONARY QUESTION

By

Timothy Lancaster Murphy

A DISSERTATION

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

DOCTOR OF PHILOSOPHY

Department of Zoology

1999

ABSTRACT

3-D VOLUME RECONSTRUCTION OF THE DEVELOPMENT OF THE
EYE-TENTACLE COMPLEX IN PULMONATE GASTROPODS:
NEW APPROACHES TO AN EVOLUTIONARY QUESTION

By

Timothy Lancaster Murphy

The development of the eye-tentacle complex in a
basommatophoran snail, Helisoma anceps, and a stylommatophoran
snail, Anguispira alternata, was examined to determine whether the
differences in adult pattern can be attributed to early differences in
developmental position and / or possible tissue interactions. In order to
carry out the analysis it was necessary to develop a method that allows
for the precise comparison of position in all three dimensions of the
developing cerebral ganglia, eyes, and tentacles. The approach used in
this study was to combine a laser scanning microscope’s (Zeiss) ability
to create high resolution digital images, with the power of a Silicon
Graphics, Inc. workstation running the "volume rendering" software
program VoxelView/ E (Vitallmages, Inc.). The software program creates
three-dimensional "virtual" embryos that can be manipulated in ways

impossible to perform with the "real" embryos.

The results of this study found that the point of invagination of
the eye vesicles was in the same dorso-lateral position of the cephalic
plates for both species. However, in Anguispira altemata, the cerebral
ganglia were not found in close proximity to the cephalic plates in the
precise location of the eye vesicles either before or during invagination,
while in Helisoma anceps, the cerebral ganglia were found in contact
with the cephalic plates from the very beginning of eye-vesicle
invagination. The eye vesicle were found to maintain connection even
after the eye vesicles had separated from the overlying ectoderm. These
results are discussed in terms of the ability of the cerebral ganglia in
some adult basommatophoran snails to induce the formation of
supernumerary eyes and tentacles, thus suggesting that induction may
play an important role in the formation of eyes and tentacles during
embryogenesis.

The value of the method developed in carrying out this analysis,
laser scanning microscopy coupled with 3-D reconstruction by volume
rendering, is demonstrated in addressing these questions in snail
development/ evolution and its tremendous potential for other such

analyses is explored.

Copyright by
Timothy Lancaster Murphy
1 999

In memory of my brother Daniel:
May we never stop believing in ourselves and may we always remember to
dream good dreams.

ACKNOWLEDGMENTS

I take this opportunity to thank all the people who directly or
indirectly helped to make this dissertation project become a reality. With
gratitude, I thank my mentor, Dr. James W. Atkinson, for his support,
guidance and overall patience without which I would not have been able to
complete this work. I thank him for allowing me to share in his awe of
wonder for the natural world around us. But most importantly, I thank him
for believing in me. I also thank my committee members for all their time
and support: Drs. Martin Balaban, John I. Johnson, Donald 0. Straney, and
Joanne H. Whallon. I am grateful to Dr. Shirley A. Owens and fellow
graduate student Geoffrey Williams, for their support, friendship, and most
importantly humor. I thank fellow lab slugs, Alyssa L. Shearer and Miranda
A. Karson for showing just how interesting snails can be even after they
hatch. I especially thank Drs. Diana and Nestor Bello-DeOcampo for their
support, encouragement, and friendship. I am also grateful to my parents,
Donald H. and Jane B. Murphy, and my brother Dr. James B. Murphy for
their unconditional support and encouragement. Special thanks go to Dr.
Joanne Whallon and my father for their critical review of this manuscript.

Finally and most importantly, I extend my warmest appreciation to
Kelly J. Stevens for his great humor, love, and always being there when I

needed him most.

TABLE OF CONTENTS

 

 

 

 

 

 

 

 

 

 

LIST OF TABLES IX
LIST OF FIGURES X
LIST OF ABBREVIATIONS XII
INTRODUCTION 1
CHAPTER 1 7
A SNAIL’S VIEW 7
EVOLUTIONARY MODIFICATION OF THE EYE-TENTACLE COMPLEX: ADAPTATION T0 TERRESTRIAL
HABrrATs? ................................................................................................................................................. 8
DEVELOPMENT OF THE EYE-TENTACLE COMPLEX IN PULMONATE GASTROPODS ...................................... 9
CHAPTER 2 20
RESEARCH APPROACH 20
TRADTTIONAL METHODslAPPROACHES & THEIR LIMTTATIONS ................................................................ 20
USING LASER SCANNING MICROSCOPY & COMPUTER 3D “VOLUME” RECONSTRUCTION ....................... 22
CHAPTER 3 25
MATERIALS AND METHODS 25
SEIECTION OF SPECIES COMPARED .......................................................................................................... 25
Stylommatophoran snails ........................................................................................... 25
Basommatophoran snails ........................................................................................... 26
STAGING 0F EMBRYOS .............................................................................................................................. 28
Video tape recording of live embryos prior to ﬁxation .............................................. 28
MICROSCOPY PREPARATION ..................................................................................................................... 29
Fixation of embryos ................................................................................................... 29
Scanning Electron Microcopy .................................................................................... 29
Light Microscopy ....................................................................................................... 30
DIGITIZATION OP SECTION IMAGES ........................................................................................................... 30
Image format conversion and transfer from the LSM to the SGI graphics
workstation ................................................................................................................. 31
IMAGE PREPARATION ON THE 801 WORKSTATION ................................................................................... 32
Rectangular pixels to square pixels: Maintaining correct image aspect ratio .......... 32
Voxel-padding in VoxelView ...................................................................................... 33
The importance of the order of images ...................................................................... 33
Filling in the Z dimension: Interpolations ................................................................. 34
Reduction of interpolations: Sub-sectioning the sections .................................................. 35
Precise adjustment of section image-alignment using "Negative" interpolations ..... 37
IMAGE PROCESSING AND ANALYSIS .......................................................................................................... 38
Determination and delimitation of cell and tissue types ............................................ 38
3D visualization of 2D tracings: The "Geometry" Function ............................................. 38
Enhanced visualization ...................................................................................................... 38

vi

Recombining volumes Visually vs. Structurally ......................................................... 43

 

 

 

 

Visual merging: Multi-channel viewing ............................................................................ 43
Structural merging: Merge Volume (V oxelMath) ............................................................. 44
Image rendering ......................................................................................................... 44
Single Images and Animations .......................................................................................... 44
Annotations ........................................................................................................................ 45
CHAPTER 4 46
RESULTS 46
PART I: EFFICACY 0F 3D VOLUME RENDERING FOR THE STUDY OF SNAIL EMBRYOLOGY ....................... 46
Enhanced visualization and identiﬁcation of developing structures ......................... 46
Visualization of unidentified structures in 3D ................................................................... 46
Visualization of internal structures by digital "dissection" ................................................ 47
Visualization Of structures with increased contrast by digital "staining" .......................... 47
Visualization of Slices taken in planes independent of the original plane of sectioning by
"re-sectioning" ................................................................................................................... 48
Visualization of the whole embryo and any "sub-sets" of it, from all angles
simultaneously ................................................................................................................... 48
Ability to determine in all three dimensions the precise spatial-temporal position
and development of structures ................................................................................... 49
Ability to make direct quantitative measurements and comparisons ......................... 49
Measuring the accurate distance between structures from sectioned embryos .................. 49
Volumetric measurements of developing organ systems (morphometrics) ....................... 50
PART II: DEVELOPMENT OF THE EYE-TENTACLE COMPLEX IN THE AQUATIC BASOMMATOPHORAN SNAIL
HELISOMA ANCEPS AND THE TERRESTRIAL SNAIL ANGUISPIRA ALTERNATA ................................................. 52
General pattern of pulmonate development ............................................................... 52
Eye-tentacle complex development in the stylommatophoran snail Anguispira
altemata ..................................................................................................................... 55
Eye-tentacle complex development in the basommatophoran snail Helisoma
anceps ......................................................................................................................... 5 7
Comparisons .............................................................................................................. 58
PLATES ..................................................................................................................................................... 60
CHAPTER 5 115
DISCUSSION 115
TECHNIQUE ISSUES ................................................................................................................................. 1 15
Repeatability & Reliability ....................................................................................... 116
New Capabilities ...................................................................................................... 11 7
"Re-sectioning" of embryos ............................................................................................. 117
Digital "dissection" of embryos ....................................................................................... 120
Accuracy Of Visualizations ............................................................................................... 123
Quantitative Measurements ............................................................................................. 125
Accuracy of reconstructions ..................................................................................... 129
Left-right axis in reconstruction ...................................................................................... 129
Alignment of images ........................................................................................................ 131
Pixels to Voxels ........................................................................................................ I33
BIOLOGICAL ISSUES ................................................................................................................................ 138
Determination and delimitation of cells and tissue types ........................................ I38

vii

Gastropod phylogeny ............................................................................................... 140

 

 

 

 

 

 

Structure of the eye-tentacle complex in gastropods ............................................... 140
Sensory functions of gastropod tentacles ................................................................. 142
Homology of sensory structures ............................................................................... 143
Development ............................................................................................................. 145
Experimental evidence of possible inductive interactions between the cerebral ganglia
and overlying ectoderrn of the cephalic plates ................................................................. 148
Evolutionary modiﬁcations are modiﬁcations of development ................................ 151
CONCLUSION ........................................................................................................................................... 157
APPENDIX A 162
ASSORTED COMPUTER SYSTEM COMMANDS .......................................................................................... 162
APPENDIX B 164
VOXELMATH AND VOXELVIEW INFO ..................................................................................................... 164
APPENDIX C 169
DIMENSIONS FILE FOR VOLUMETRIC DATA SET: VOXELVIEw’s "_DIMENSIONs" FILE ............................. 169
APPENDIX I) 170
IMAGE ROTATION STEPS ......................................................................................................................... 170
APPENDIX E 171
MATERIALS & SUPPLIES ......................................................................................................................... 17 1
APPENDIX F 172
ASSORTED BIBLIOGRAPHIC REFERENCES ................................................................................................ 17 2
LITERATURE CITED 179

 

viii

LIST OF TABLES

Table 1 Calculation of Pixel Size (resolution) in the X-axis .............................................................. 36
Table 2 Calculation of Z-axis Interpolations ....................................................................................... 36
Table 3: Sub-dividing Histogram range: Mathematical operations ................................................. 42
Table 4 VoxelSeed Summary of Statistical Analysis ........................................................................... 51
Table 5 Developmental stages ................................................................................................................ 58

LIST OF FIGURES

Figure 1 Comparison of the eye-tentacle complex ............................................................................ 2
Figure 2 Patterns Of Gastropod Nervous Systems .......................................................................... 15
Figure 3 Dextral vs. sinistral spiral cleavage patterns ...................................................................... 17
Figure 4 Developmental Stages ........................................................................................................... 19
Figure 5 "Geometry" Function in VoxelViewZ E ............................................................................. 62
Figure 6 Digital "dissection" of the nervous system of snail embryo ........................................... 64

Figure 7 3D volume reconstruction of left cephalic plate Of Helisoma anceps embryo at time of
eye-vesicle invagination ............................................................................................................. 66

Figure 8 3D reconstructions the left cephalic plate of a stage B Anguimim altemata embryo .. 68

Figure 9 3D reconstruction of left and right cephalic plates in

 

Anguirpira altemata at time of eye-vesicle invagination ................................................... 70
Figure 10 3D reconstruction OfAnguimz'm altemata cephalic plates before and
during invagination of eye—vesicles .. - ...... . ......................................... 72
Figure 11 Comparison of 3D volume reconstructed embryo with SEM images ........................ 74
Figure 12 Digital re-sectioning Of embryo .......................................................................................... 76
Figure 13 Original 2D section image and 3D reconstruction showing nervous
system of H e/isoma anceps embryo ....................................................................................... 78
Figure 14 3D reconstruction and re-sectioned 2D Slices ................................................................. 80

Figure 15 More 3D reconstruction images Of left cephalic plate at time of eye-vesicle

invagination in H elzlroma anceps .................................................................................................. 82
Figure 16 Measuring distances in 3D using the "Trace" function .................................................. 84
Figure 17 Volume measurements using the "VoxelSeed" function .............................................. 86
Figure 18 SEM image of H. 4mm: and original 2D section of A. altemata with labels ............... 88
Figure 19 Light micrographs of cells of cerebral ganglion ............................................................. 90
Figure 20 Conversion of 2D sections to a 3D volume .................................................................... 92
Figure 21 Digital dissection: processing of images ............................................................................ 94

Figure 22 Shifting the range of voxel-values: Original image and histogram ............................... 96

Figure 23 Shifting the range of voxel-values: Image and histogram after multiplying the
values by 63 ................................................................................................................................ 98

Figure 24 Shifting the range of voxel-values: Image and histogram after multiplying the
values by 63 and dividing by 256 . ................ 100

 

Figure 25 Shifting the range of voxel-values: Image and histogram after multiplying the
values by 63 and dividing by 256 and shifting all non-zero voxel values by 64
("bias= 64') - - .......................................................... 102

 

Figure 26 Image “stack” order ............................................................................................................ 104

Figure 27 Comparison of SEM of whole embryo with partial 3D reconstruction:
Effects of correct and incorrect image stack order ...................................................... 106

Figure 28 Order of sections on the microscope slide vs. the order of digital images
needed to reconstruct correctly in 3D ............................................................................ 108

Figure 29 Digital realignment of sections of section images using “negative” interpolations 110

Figure 30 Computer generation of 3D voxels from 2D pixels ..................................................... 112
Figure 31 Maintaimng the correct image aspect ratio: rectangular to square pixels .................. 114

xi

LIST OF ABBREVIATIONS

ANG Angulsplra altemata

ASE Angulsplra altemata embryo

ANSI/IEEE 488-1978 (American National Standards Institute: Determines
hardware-related standards

CCD Charge-couple devise (used for digital cameras)

CD-ROM Compact Disk Read Only Memory (optical disk)

DOS Disk operating system

FAA Formaldehyde-alcohol-acetlc acid . A common fixative.

FTP File transfer protocol

GPIB AT (General Purpose Interface Bus for an AT class PC clone)

HAS Helisoma anceps

IBM PC Refers to a family of personal cmuters produced by IBM. The
term can also refer to computers that conform to set of loosely controlled standards.
These are also called IBM clones, IBM comgtib/es , or simply compatibles. These
terms are actually misnomers because many of the ﬁg produced by IBM do not
conform to industry standards. For example, IBM attempted to change the expansion
ms to MCA in its PS/2 line of PCs, but the industry did not follow suit.
(http://webopedia.intemet.com/TERM/l/lBM_PC.html)

IEEE 488 (Standard for GPIB established by Institute of Electrical and Electronics
Engineers)

LSM Laser scanning mlcroscopy

MB megabyte

N.A. Numerical aperture

nm nano-meter

PC Clone (IBM Clone; IBM Compatibles) Other companies adjusted to IBM’s
dominance by building IBM clones, computers that were internally almost the same as
the IBM PC, but that cost less. Because IBM clones used the same microprocessors
as IBM PCs, they were capable of running the same software.
(httpjlwebopedia.internet.com/TERM/p/personal_computer.html)

PC Short for personal computer or IBM PC. The first personal computer produced
by |_B_M was called the PC, and increasingly the term PC came to mean IBM or IBM-
compatible personal computers, to the exclusion of other types of personal computers,
such as Macintoshes.

In recent years, the term PC has become more and more difficult to pin down. In
general, though, it applies to any personal computer based on an Intel microprocessor,
or on an Intel-compatible microprocess__or. For nearly every other component, including
the operating system, there are several options, all of which fall under the rubric of PC.
[http://webopedia.internet.com/TERM/P/PC.html]

PERSONAL COMPUTER A small, relatively inexpensive computer designed for an
individual gel. In price, personal computers range anywhere from a few hundred
dollars to over five thousand dollars. All are based on the microprocessor technology
that enables manufacturers to put an entire it; on one grip. Businesses use personal
computers for word processing, accounting, desktop publishing, and for running
spreadsheet and database management applications.

Other companies adjusted to IBM’s dominance by building IBM clones, computers that
were internally almost the same as the IBM PC, but that cost less. Because IBM clones

 

xii

used the same microprocessors as IBM PCS, they were capable of running the same
software. (http://webopedia.intemet.com/TERM/p/personal_computer.html)

RAM Random access memory

SGI Silicon Graphics, Inc.

SP Standard play (VHS recording speed)

SVHS Super VHS

TAR Tape Archive (Unix system)

TCPIIP Transmission control protocol/Internet protocol

u(m) micro-meter (microns)

UNIX Pronounced yoo-niks, a popular multi-user, muﬂtskina operating system
developed at Bell Labs in the early 19703. Created by just a handful of programmers,
UNIX was designed to be a small, flexible system used exclusively by programmers.
Although it has matured considerably over the years, UNIX still betrays its origins by its
cryptic command names and its general lack of user-friendliness. This is changing,
however, with graphical user interfaces such as MOTIF.

UNIX was one of the first operating systems to be written in a high-level programming
language, namely Q. This meant that it could installed on virtually any computer for
which a C compiler existed. This natural portability combined with its low price made it
a popular choice among universities. (It was inexpensive because antitrust regulations
prohibited Bell Labs from marketing it as a full-scale product.)

Bell Labs distributed the operating system in its source language form, so anyone who
obtained a pppy could modify and customize it for his own purposes. By the end of the
1970s, dozens of different versions of UNIX were running at various sites.

After its breakup in 1982, AT&T began to market UNIX in earnest. It also began the
long and difficult process of defining a standard version of UNIX. To date, there are two
main dialects of UNIX; one produced by AT&T known as System Vand one developed
at Berkeley University and known as BSD4.x, x being a number from 1 to 3.

Due to its portability, flexibility, and power, UNIX has become the leading operating
system for workstations. Historically, it has been less popular in the personal computer
market, but the emergence of a new version called Linux is revitalizing UNIX across all
platforms. (http://webopedia.internet.com/'I'ERIVI/U/UNIX.htmI)

 

WORKSTATION 1) A type of computer used for engineering applications
(CAD/CAM), desktop publishing, software development, and other types of applications
that require a moderate amount of computing power and relatively high quality graphics
capabmﬁes.

Workstations generally come with a large, high-resolution graphics screen, at least 64
MB (megabﬂes) of RAM, built-in network support, and a graphical user interface. Most
workstations also have a mass storage device such as a disk drive, but a special type
of workstation, called a diskleﬁss workstation, comes without a disk drive. The most
common operating systems for workstations are UNIX and Windows NT.

In terms of computing power, workstations lie between personal computers and
minicomputers, although the line is fuzzy on both ends. High-end personal computers
are equivalent to low-end workstations. And high-end workstations are equivalent to
minicomputers.

Like personal computers, most workstations are single-M computers. However,
workstations are typically linked together to form a local-area network, although they
can also be used as stand-alone systems.

 

 

xiii

 

The leading manufacturers of workstations are Sun Mjcrosvstems, Hewlett-Packard
Company, Silicon Graphics Incorporated, and Compaq.

(2) In networking, workstation refers to any computer connected to a local-area
network. It could be a workstation or a personal computer.
[http://webopedia.internet.com/TERM/w/workstation.html]

xiv

INTRODUCTION

Evolutionary changes in morphology often occur when a group of
organisms invades a new habitat. Since changes in morphologl are
derived from changes in developmental processes and patterns of gene

expression, evolution can be defined as changes in developmental

patterns over time.

Within the gastropod group Pulrnonata (air-breathing snails and
slugs), one of the major differences between the "Basommatophora," a
prir11arily Mp group Of snails, and the "Stylommatophora," a
prlrnarily terrestrial group of snails and slugs, is in the structure and
pattern of their eye—tentacle complex. The names of these two groups
are based upon the position of the eyes relative to the tentacles. The
Water-dwelling basommatophoran snails have their eyes situated at the
base of a single pair of cephalic tentacles. The land-dwelling
Stylommatophoran snails and slugs have two sets of cephalic tentacles,
a Pair of smaller “anterior” tentacles situated just above the "labial
Palps" or upper lips, and a larger pair of “posterior” or "Optic" tentacles
lo(tested further back on top of the “head.” Their eyes are at the top of
this larger, posterior set of cephalic tentacles (Figure 1).

The evolution of the stylommatophoran eye-tentacle complex
reI3I‘esents a major restructuring of the main sensory system of these

amtnails. Included in this evolutionary modification was not only a

reorganization of parts, such as eyes moved to the top of the posterior
tentacles, but the creation Of novel features such as 1) the sensory pad

at: the tip of the tentacles, 2) the largest and most complex tentacular

 

{ ganglia of any gastropods

(Haszprunar, 1988), 3)

Stylommatophoran Basommatophoran
snail snail

and the ability to fully
retract the tentacles by

invagination.

 

It is not known

 

 

Fig ure 1 Eye-tentacle complex Comparison what, if any, the adaptive

(snails redrawn after Solem, 1974) signiﬁcance of these two

 

 

 

very different patterns of
the main sensory structures, which are so highly correlated with life in
different environments, may be. Although there is disagreement as to
the ancestry of these two groups (see below), there is little doubt that
the pattern of tentacles and eyes found in the Stylommatophora was an
eV<>lutionary novelty within the Gastropoda. How this restructuring Of
31—1011 important organ systems occurred evolutionarily is not lmown.
Since all evolutionary changes in morphology are based on changes in
the underlying developmental pattern, to fully understand the evolution
Of these structures requires knowledge as to where and how the

developmental pattern was altered.

While there have been descriptive studies of the general
development Of several species of each group, there have not been any

clear comparative studies of how the different patterns of the eye-

tentacle complex are established during development. DO the changes
occur early in development, with major reorganization and / or novel
developmental patterns? Alternatively, do the patterns of development
differ only later, with minor adjustments to existing developmental
pathways?

Snails are asymmetric animals, and during development, their
organs and tissues go through many different patterns of arrangement.
TO understand how one pattern develops into another, it is necessary to
be able to study the structure Of the embryo at different stages in all
three dimensions. Because of the small size of pulrnonate embryos,
traditional approaches to the study of their development have involved

COllecting embryos at different stages of development, ﬁxing them,
embedding them in some sort of supportive medium, sectioning, and
Staining the sections for observation under a microscope. While this
Yields very good two-dimensional images of these very small embryos,
the technique is limited in determining the three-dimensional
infOrrnation needed to understand and compare precise changes in
developmental patterns. A novel approach is needed that not only

Utilizes the spatial information available in the two-dimensional

sections, but extends that information allowing better determination of
spatial relationships in all three dimensions.

In this study the approach used to study snail development was
to combine a laser scanning microscope’s ability to create high
resolution digital images, with the power of sophisticated computer
“volume rendering” three-dimensional reconstruction software running
on a graphics workstation. While both of these technologies have been
used separately, as well as together, their combined complementary
capability had not been brought to bear on these types of questions of
development. This approach creates a “virtual embryo” on the computer
that can be studied and manipulated in ways impossible to perform on
“real” embryos. Changes in pattern in all three dimensions can be
examined quantitatively as well as qualitatively with great precision,
accuracy, and repeatability.

The importance of accurately mapping the position of developing
structures and tissues is underscored when the role of tissue
interactions during embryogenesis is understood. Classic studies in
vertebrate embryologI revealed the phenomenon of "induction" in which
the fate of one tissue is determined by substances produced by adjacent
tissues (See Wolpert, 1998). Of special importance to the investigation
at hand is the well-studied phenomenon in which the developing
nervous system of the vertebrate embryo induces overlying ectoderm to

differentiate into the lens and cornea of the eye. Recent transplantation

studies involving adult basommatophoran snails indicate that the
nervous system of gastropods may have inductive properties also.
Cerebral ganglia transplanted into an adult host induced the formation
of supernumerary eyes and tentacles (Moffett and Austin, 1981).
Because Of their extremely small size, comparable transplantation
studies have not been attempted on the embryos of pulrnonate snails.

Nevertheless, the inducing ability of cerebral ganglia in adult
snails suggests that such induction may play an important role in the
formation of eyes and tentacles in the developing embryo. To determine
whether this is the case it is essential that the precise spatial
relationships among these tissues (developing ganglia, eyes, and
tentacles) be followed through development. Creation of a technique for
mapping these positions in three dimensions is of critical importance.

If the cerebral ganglia are responsible for inducing the formation
of the eyes and tentacles, one would expect to find a close positional
relationship between the cerebral ganglia and overlying ectoderm, in the
precise area where these structures will form. The differences in the
adult pattern of the eye-tentacle complex may be reﬂected in differences
in positional relationships in the embryo. Thus, it may be possible to
explain the difference in eye location between these two groups of snails
as based on a change in the position of the cerebral ganglia relative to

the overlying ectoderm during development.

This study set out:
1) to develop a method that allows for the precise comparison of
position in all three dimensions of the developing cerebral ganglia and
eyes, relative to the overlying ectoderm, between aquatic
basommatophoran and terrestrial stylommatophoran gastropods;
2) to determine the precise location within the cephalic plates of eye-
vesicle formation, and if the cerebral ganglia move into a position to be
able to interact with the overlying ectoderm in the exact area required
for their induction: and
3) to determine if there is a difference in the position of the cerebral
ganglia relative to the overlying ectoderm that may account for the
changes in the position of the eyes relative to the tentacles between

these two groups.

Cbapterl

A SNAIL’S VIEW

H ow to separate dice from dough - that is a mayor question

facing coo/uttering! biology today.
- Tyler Vol/e (1990)

The phylum Mollusca CUVIER 1797 is made up Of eight “classes” with
Class Gastropoda CUVIER 1797 comprising nearly 80% Of the phylum
(Solem, 1974). The class Gastropoda, which contains the snails and
slugs, is further divided into the subclasses Prosobranchia,
Opisthobranchia, and Pulmonata. The subclass Prosobranchia is the
oldest, and is believed to have given rise to the other two subclasses
(Schmekel, 1985: Solem, 1974, 1985; Hasprunar, 1988). Prosobranchs
are predominately marine, with some freshwater and terrestrial
members. Opisthobranchs are all marine. Pulmonates are mainly
terrestrial and freshwater.

Historically, gastropods have been divided into two groups based
on whether or not the pleural-visceral connectives of their nervous
system are crossed (“streptoneury”), forming a ﬁgure eight of the
visceral loop, or not crossed, or only slightly crossed peripheral nerves

(“euthyneury”) (Bullock and Horridge, 1965)1 (Figure 2). It is from an

euthyneurous group of the Prosobranchia that the Opisthobranchia and
the Pulrnonata are believed to have arisen (Hyman, 1967; Fretter and
Graham, 1962; Thompson, 1976). Although their relationship to each
other is still under debate, the Gastropoda are believed to form a

monophyletic group (Haszprunar, 1988).

EVOLUTIONARY MODIFICATION OF THE EYE-TENTACLE COMPLEX:

ADAPTATION TO TERRESTRIAL HABITATS?

As the ancestors of modern gastropods moved from marine to
freshwater and terrestrial habitats many physiological as well as
structural modifications occurred, e.g., changes involving water—
balance, reproduction, shell formation, etc. (See Fretter and Graham.
1962; Graham, 1985; Andrews, 1965; Andrews and Little, 1972; Creek,
1951; Solem, 1974, 1985). Along with obvious adaptive alterations in
structure and physiology, modiﬁcation of the major sensory system, the
eye-tentacle complex, has changed in the structure, function, number
of cephalic tentacles, and the position of the eyes relative to them, of
those animals that most successfully invaded land -- the
stylommatophoran puhnonates.

Aquatic snails have a single pair of cephalic tentacles, with their
eyes located at the base of the tentacles. The eyes are fused to the
base (as in the pulmonate basommatophoran snails, Lymnaea and
Helisoma), or located on short eyestalks located at the base (as in the

prosobranch Pomacea). The largest and most diverse group of terrestrial

gastropods is the Order Stylommatophora. All these animals (snails and
slugs), in contrast, have two pairs of cephalic tentacles, with their eyes
located at the tip of the larger posterior set of tentacles. (Figure 1) Each
Of the four tentacles has a large, complex, peripheral ganglion
innervating a chemo-sensory pad at the tip.

With the exception of the Limnaeids, in which the tentacles are
immobile spade-like structures, basommatophoran snails have long
slender solid tentacles that can only be shortened a small amount and
have very limited range of mobility. They move almost freely in the water
currents. On the other hand, stylommatophoran snails have tentacles
that can be retracted by invagination completely back into the body and
are highly mobile. The larger posterior "optic" tentacles can also be
lengthened and may be deployed in a variety of patterns and directions.

The success Of the Stylommatophora (over 30,000 species in
terrestrial habitats throughout the world —- which incidentally out-
numbers all terrestrial vertebrates (Barker, 1999; Solem, 1974, 1984)) --
and the universality among them of the same basic eye-tentacle-
cerebral ganglion pattern is strongly suggestive that the evolution of the

pattern has played an important role in their success.

DEVELOPMENT OF THE EYE-TENTACLE COMPLEX IN

PULMONATE GASTROPODS
The general development of gastropod molluscs has been worked

out for several species. Numerous cell lineage studies since the late

nineteenth century have established the general pattern of gastropod
development (Blochmann, 1882: Neritina; Heymons, 1893: Umbrella;
Kofoid, 1895: Lima): Meisenheimer, 1896; Limax; Conklin. 1897:
Crepidula, 1907: Phlgur, Holmes, 1900: Planorbis; Robert, 1902:
Trochus; Casteel, 1904: Fiona: Carazzi, 1905: Aplysia; Wierzeiski, 1905:
Physa; Delsman, 1912: Littorina). (For a more detailed review Of their
ﬁndings, see Raven, 1966.) Organization of the gastropod embryo is
based upon a ﬁxed relationship to the pattern of the blastomeres
(Raven, 1966; Verdonk and Biggelaar, 1983).

Cleavage in all molluscs, except the cephalopods which have
meroblastic discoidal cleavage (Raven, 1966). is a modiﬁed form Of the
radial type of holoblastic cleavage termed “spiral cleavage” by Wilson in
1892. The system of nomenclature used to label and thus follow the cell
lineage was developed by Conklin (1897) and modiﬁed slightly by
Wierzejski (1905). The ﬁrst cleavage divides the egg into two cells AB
and CD. Second cleavage divides these cells further in to the four
“quadrants” Of the eg, “A,” “B,” “C,” and “D”. Third cleavage separates
an animal “micromere” from the vegetative “macromere” of each
quadrant, thus forming the “ﬁrst quartette” of micromeres, designated
“ la-ld” and the macromeres now designated “ lA-lD” (Raven, 1966:
Verdonk, 1965).

Gastropods have what is termed “alternating spiral cleavage” with

the direction Of cleavage alternating between “dexiotropic” (“move to the

10

right” or in a clockwise direction), and “laeotropic” (“move to the left” or
in an anti-clockwise direction). When third cleavage2 is laeotropic
instead of dexiotropic, the cleavage pattern of the embryo becomes “a
mirror image” of the original pattern, and thus the morphology of the
adult animals become reversed (Raven, 1966; Verdonk and van den
Biggelaar, 1983). Crampton (1894) discovered the correlation between
the pattern of cleavage and direction of coiling of the shell (V erdonk and
van den Biggelaar, 1983). Some species of basommatophoran snails,
such as the Biomphalaria. Planorbis, Physa, and Helisoma, have
sinistral (or left-handed) coiling shells, all have the reversed cleavage
type and thus reversed body plans also (Verdonk and van den
Biggelaar, 1983). (See Figure 3)

The ectoderm is formed from the ﬁrst three quartettes of
micromeres. The ectoderm is divided into two regions by a band of
ciliated cells called the “prototroch”. The ectoderm anterior to this
prototroch is called “pretrochal ectoderm” and the ectoderm posterior to
the prototroch is called “posttrochal ectoderm” (Raven, 1966).

It is the pretrochal ectoderm, derived from the ﬁrst quartette of
micromeres (la- 1d), that gives rise to the head region. Within this head
region, two rounded areas develop from micromeres 1a and 1c, called
the “cephalic plates”. It is from these speciﬁc regions of pretrochal
ectoderm that the cerebral ganglia, cephalic eyes, tentacles, and any

tentacular ganglia are derived3 (Crofts, 1937; Moritz, 1939: Raven,

11

1966; Demian and Yousif, 1975; Verdonk and van den Biggelaar, 1983:
Jacob, 1984: Page, 1992).

In pulmonates, the cerebral ganglia are derived from the ectodenn
of the cephalic plates by cellular proliferation and delarnination, and
invagination. (Henchman, 1890: Limax maximus; Verdonk, 1965:
Limnaea stagnalis; Carney and Verdonk, 1970: Biomphalaria glabrata;
Morrill, 1982: Lymnaea palustris: Reviews: Raven, 1966; Moor, 1983.) A
pair of invaginations4 occurs with the innermost end forming a
thickening of the ectodermal wall with cells apparently migrating to the
cerebral ganglia, and developing into the “procerebrum” in the
Stylommatophora, or the “lateral lobes” in the Basommatophora
(Henchman, 1890; Raven, 1966; Moor, 1983).

The formation of the eyes in pulmonates occurs by invagination in
the lateral or dorsal lateral region of the cephalic plate forming a small,
single-layered vesicle (Henchman, 1890; Raven, 1966; Eakin and
Brandenburger, 1967: Moor, 1983). The eye vesicle then pinches Off
from the overlying ectoderm forming a closed ball Of cells. In the
prosobranch freshwater snail Marisa comuarietis (Demian and Yousif,
1975), the eyes form in the lateral region of the cephalic plates and then
migrate to a small protuberance that has formed in the lateral side of
each tentacle. The ectoderm above the eye thins out and becomes the
cornea. In the pulmonate stylommatophoran snail Helix aspersa (Eakin

and Brandenburger, 1967) the eye-vesicle is said to form in the tentacle

12

anlage, and then gets "carried up" with the developing tentacle. The
cephalic eyes are innervated by the cerebral ganglia in all gastropods.

At the time the eye-vesicles begin to form, the anlagen of the
tentacles start to sprout from the cephalic plates. The single pair of
cephalic tentacles of basommatophorans are solid and ﬁlled with
mesoderrn cells, while the two pairs of cephalic tentacles of
stylommatophorans are hollow (Henchman, 1890: Raven, 1966; Moor,
1983). The tentacles of stylommatophorans have a large, complex
tentacular ganglion located just below the tip Of each tentacle.

While this general pattern Of pulmonate eye-tentacle complex
development has been well established. these studies are not precise
enough to make direct comparisons Of developmental pattern of the eye-
tentacle complex between species. For example, these studies only refer
to where the eye-vesicles actually form in the cephalic plates as being
“dorsal” or “dorsal lateral" to the developing tentacles (Raven, 1966;
Eakin and Brandenburger, 1967; Moor, 1983). These descriptions,
While generally useful for gross comparisons, are not precise enough to
I'ule out or support the possibility for a change in the location of the
ey€=~vesicle formation that may account for the difference in the location
of the eyes in the adults. The lack of precision of these descriptions is
(1‘18 in part to the techniques used is these studies, i.e., sectioned and
WhOIe mounted embryos, which can make it difﬁcult to determine the

Precise position of very small structures in all three dimensions.

13

 

 

Figure 2 Patterns of Gastropod Nervous Systems

Structure and patterns of the nervous system from
the three sub-classes of gastropod molluscs. The visceral
loop of the Prosobranchia is crossed (streptoneury), while
the Opisthobranchia and Pulrnonata are not crossed
(euthyneury).

The ganglia believed to be homologous are shown
in the same color: Cerebral ganglia are in white; visceral
loop ganglia are in green; and pedal ganglia are in red.

(Based on Bullock, T. H., and Horridge, G. A.
(1965). Structure and Function in the Nervous Systems of
Invertebrates. Vol. 2. W. H. Freeman and Company, San
Francisco.)

 

14

 

Prosobranchia

cerebral

gangl ia\
visceral loop///'

ganglia

   

Pomatia Buccinum

Opisthobranchia
3 .
\)

   
 
 

pedal ganglia

Aplysia Coryphella

Pulmonata A

  

Lymnaea

Basommatophora Stylommatophora

 

 

Figure 3 Dextral vs. sinistral spiral cleavage patterns

Comparison of cleavage pattern of dextral (a-e)
verses sinistral (f-j) gastropods. Dextral snails have a
dexiotropic third cleavage with a clockwise shifting of the
resulting micromeres. Sinistral snails have a laeotropic
third cleavage with a counter-clockwise shifting of the
resulting micromeres.

Macromeres are labeled with uppercase letters and
micromeres are labeled with lowercase letters.

(Based on Verdonk and van den Biggelaar, 1983)

 

16

 

 

.0 if. .
e. an.” i@-§

.‘U " 33 at,
.. .b m.“ Q4 mm
ﬂ ‘9‘ Velma. ﬁle... .56.

 

 

 

Figure 4 Developmental Stages

Comparison of developmental stages:

Al-A3 Line drawing from cell lineage studies
(drawings based on Verdonk, 1965).

B1-B3 SEM images of Helisoma anceps.
C1-C3 Video images of Anguispira altemata.

Trochophore stage: A1, B1, Cl
Veliger stage: A2, B2, C2
Hippo stage: A3, B3, C3 (Late hippo stage D 8r E).

Ap, apical plate; cp, cephalic plate; hv, head vesicle; m, mouth;
p, prototroch; sh, shell(shell field); tent, tentacle.

 

18

 

 

trochophOre

  
  
  

,. {LEO
'r’r‘23:

-."’¢ 1.. t
' 2.3. ,2

 
    
   

     
    
  

 

      

-
‘.

' V
3") Q i
3"? Mfg ,‘L'jo n

‘3?” i 13;
.-Q‘ A . {312.9%
I , .u' P‘-ﬁ;. I
' w- .

  
 

 

    

    

 

C3

 

leChI

thrift

Objei

Com

€135:

Cbapter 2

RESEARCH APPROACH

‘ ‘T/Jeﬁrrt object of ”26’ painter i: to make a ﬂat plane appear a; a boa}! in
relief and projecting ﬂow that plane. "

- Leonardo da Vimi

If the change in position of the eyes in stylommatophoran snails is
achieved by a change in timing and / or position Of the cerebral ganglia
during development, one must be able to determine the precise spatial-
temporal changes that occur at the tissue level during development. The
technical challenges are:

to visualize, with high resolution, cell 8r tissue position in all
three dimensions in a manner that decreases researcher bias, increases
Objective interpretation by other researchers, and is highly reproducible:

to quantify changes during development by making direct
comparisons between specimens in a time sequence; and

t0 make quantitative comparisons of these developmental

Changes between species.

TRADITIONAL METHODS / APPROACHES & THEIR LIMITATIONS
The traditional method of drawing tracings of section images onto

glass plates and lining up these plates is an old (and inventive)

20

techrnc
SlI'UClL
tehnk
hinted

SU’UCI‘

image
task.r
must
SOHlei
resea
as 0;
01a.c
exist
{See
biolc
any
ﬂiel
rice

COIT.

Gide]

COmI

technique used to help visualize the three-dimensional (3D) position of
structures, as well as to convey the information to others. This
technique, (while useful in helping to visualize structures in 3D), is
limited in precision of positional information and its ability to help with
structural identiﬁcation.

Tracing techniques are based on the creation of “wire-frame”
images (called polygons on the computer) traced by hand, a laborious
task requiring days to months to create even one model. The researcher
must already have some idea of the identity and position (and
sometimes even shape) of structures to trace. Thus, there is the risk of
researcher-bias in drawing the structures as they are “believed” to be,
as opposed to how they "really are". The technique requires the creation
of a deﬁnitive “hard” boundary, whether or not such a boundary really
exists, me the specimen can be reconstructed and visualized in 3D.
(See Walter (1999) for a discussion on the use of “Fuzzy logic” in
biologi.) The ﬁnal images are only of the tracings themselves, without
any Of the rest of the data including the data used in determining where
the boundary was drawn. This limits the use of the resulting
reconstructions as a “feedback” system allowing reﬁnement and
correction of the 3D model.

Other techniques based on outlines or boundaries, such as the
Older clay modeling technique, or more recently, the use of the

computer to create “virtual” three-dimensional models, while adding to

21

the infor
"surface
based or
techniqt
enhance
efficient
deierrnii
possible

Changes

USIN

investig
tIaditioi
the higl
MM} n

re8111th

Cornput

the information Obtained are still subject to the same criticisms. The
"surface rendered” computer models, while a big improvement, are still
based on wire-frame (polygon) images. None of these methods allows
techniques such as digital ﬁltering and histogram manipulation to
enhance visualization of unknown structures. Thus, they are less
efﬁcient in differentiation and determination of structures, and in
determining the precise location of tissue positions needed to determine
possible sites of tissue interactions such as "induction" during early

changes in developing organ systems.

USING LASER SCANNING MICROSCOPY & COMPUTER 3D “VOLUME”
RECONSTRUCTION

A new approach was developed as part of the present
investigation to overcome the limitations and drawbacks Of these
traditional methods of 3D visualization. This new approach combines
the high resolution digitizing capabilities of a laser scanning microscope
(LSM) with the power Of computer three-dimensional “volume rendering”
software (e.g., VoxelVieWZE by Vital Images Inc., Fairﬁeld Iowa).

Once serially sectioned embryos are digitized by the LSM, the
resulting images can be visually enhanced and manipulated by the use
of techniques such as digital ﬁltering or deconvolution. The two-
dirnensional (2D) images of the series are then “stacked" by the

computer software to form a “virtual” three-dimensional whole embryo,

22

whi

COD

wh
b0]

l'CI

Th.
ear

l’is‘.

3D

COI

which is then manipulated as a single "volume" (or Object) by the
computer.

Unlike “surface rendering” techniques for 3D reconstruction,
which use only the “trace ﬁles” (or hand-drawn tracing around the
borders of structures) separated from the original data, "volume
rendering" maintains all the information from the original 2D images,
allowing for continual adjustment and manipulations of 3D renderings.
There is no need to create deﬁnitive “hard” boundaries (the computer
can display and manipulate gradations or “soft” boundaries) before
visualization. thus lessening the risk of researcher bias. Digital ﬁltering
and segmentation routines (to help differentiate Objects from their
background, needed to get quantitative information) can utilize all the
information in the data from all three dimensions.

After the section images of the embryos are digitized and "rebuilt,"
3D volume rendering programs can make direct point-tO-point
comparisons between embryos in a developmental time sequence,
allowing for accurate and precise tracking and analyses of changes
during development. It is now possible on the computer to “digitally
dissect” extremely small embryos, (i.e. to remove the structure(s) Of
interest from the rest of the embryo, or to remove structures obscuring
the structures Of interest). Thus creating an accurate 3D visualization of

structural detail and spatial-temporal information.

23

Combining laser scanning microscopy to create high resolution
digital images, with computer “volume rendering” techniques to create
virtual 3D models Of embryos, solves the challenges Of creating accurate
visualization of 3D structure, and allows precise, quantiﬁable
comparisons within and between species in a manner unmatched by

any other method.

24

lal

rel

Mi

gle

iOI
tn‘

den

Chapter 3

MATERIALS AND METHODS

"I t i: not enongb to believe w/Jatjon see. You must also

nnderrtand wbatjon see. "

Leonardo da Vinei

SELECTION OF SPECIES COMPARED

Stylommatophoran snails

The stylommatophoran snail used in this study was Anguispira
altemata SAY 1816. These animals breed and lay eggs readily in the
lab. Their eggs have a hard calcium crystalline shell, which allows
relatively easy removal of the embryos without damaging them. The
time to hatching is 28-34 days at 20-24 °C.

Adult snails were collected from Raynor Park woodlot, Mason,
Michigan. They were placed in one gallon, three gallon, or ten-gallon
glass bowls or tanks with 1-2 inches of commercial topsoil. Decaying
leaves and bark collected with the snails was added to the tanks. De-
ionized water was added every few days to maintain a moist
environment. Snails were fed carrots and lettuce every two or three

days, and any leftover food was removed at this time. White chalk was

25

added as a source Of calcium. The top quarter inch of soil was removed
every week and new clean topsoil was added to maintain soil depth.
The soil in breeding tanks was checked each day for new egg
batches, which were placed in a covered stender dish lined with a piece
of moist paper towel. Each batch was assigned a unique identiﬁcation
number with the date laid, tank number, and batch sequence. One or
more eggs in each batch were Opened to check stage of development.
Each batch contained seven to twenty-ﬁve eggs, and it takes up to 34
days to hatch. To collect representative embryos of each day of
development, the date the ﬁrst embryo from a given batch was collected
varied from day one to day ﬁfteen. This staggering of collection starting
dates allowed collection Of embryos at all stages of development. One A.
altemata embryo was sampled from each batch each day, thus embryos
from batches at different stages of development were collected each day.
Under a dissecting microscope, the embryos were removed from their
eggshells with dissecting forceps by cracking open the shell and pouring
the contents out onto a clean glass microscope slide in a glass petri
dish. The embryos were washed clean Of their capsule ﬂuid using a

Pasteur pipette to rinse them with distilled water.

Basommatophoran snails

The basommatophoran snail used in this study was Helisoma

anceps MENKE 1830. They reproduce readily in the lab and their eg

26

masses and capsules are transparent which makes it possible to view
developing embryos in vivo. The egg masses and outer casing are semi-
brittle, allowing removal of the embryos without damaging them. The
snails hatched in seven to eleven days at 21-24 °C. Adult snails were
collected from a pond in Vevay Township, Ingham County, Michigan.
They were placed in a ﬁve, ten, or ﬁfteen-gallon aquarium with under-
gravel ﬁltration, and accessory above-gravel ﬁlters. Aquarium heaters
were used to maintain temperatures between 21-24 0C. Food such as
lettuce, carrots, and TetraFin Basic Diet Goldﬁsh Food was added every
few days, and any left over food was removed. One-quarter of the water
was removed and replaced with ﬁltered pond water or de-ionized water
every two weeks.

Breeding tanks were checked each day for new egg masses. Egg
masses in this species are laid on the glass walls of the aquarium. Each
egg mass contained nine to twenty egg capsules. The egg masses were
removed by sliding a razor blade under them to separate them from the
aquarium walls and placed in a stender dish with de-ionized water. H.
anceps embryos were collected from one every four — six hours, to one a
day per case. Each egg mass was assigned a unique identiﬁcation
number with the date laid, tank, and case sequence number recorded.
Under a dissecting microscope, embryos were removed from their egg
capsule by gently teasing apart the transparent case and capsule and

pouring the contents out onto a clean glass microscope slide in a glass

27

petri dish. Embryos were washed clean of their capsule ﬂuid using a

Pasteur pipette to rinse them with distilled water.

STAGING OF EMBRYOS

To compare development of the embryos within and between the
two species Of snails studied, forty to ﬁfty embryos of each species were
analyzed. Since A. altemata takes twenty-eight to thirty-four days to
hatch at 20-240 Celsius, and H. anceps takes seven to eleven days at
21-240 Celsius, direct comparison based on time was not possible. A
rough staging system was created based on the presence and level of
development of structures such as the shell ﬁeld/ shell material,
mantle, stomodeum, radular sac/ radula, alimentary tract, heart,
kidney, foot/ podocyst (A. altemata), pedal ganglia, statocysts, etc. This
staging system followed the basic system of Morrill (1982), Cumin

(1972), and Raven (1966).

Video tape recording of live embryos prior to ﬁxation

To compare ﬁxed, sectioned, and 3D reconstructed embryos with
the morphologl of living embryos so as to determine the stage of
development, live snail embryos of both species were recorded for two or
three minutes on videotape. This was done after their release from the
egg capsule and prior to ﬁxation. In addition, some H. anceps embryos
were recorded while still in their egg masses and capsules (which, as

previously noted, are completely transparent). Videotaping was done

28

through a Wild M5-41059 dissecting microscope with a phototube
attachment (404892) connected to a Javelin Ultrichip CCTV camera
(JE-7442). This CCD camera was connected to a video recorder (VHS)
and images were captured on T- 120 VHS tape media recorded at

Standard Speed (SP).

MICROSCOPY PREPARATION

Fixation of embryos

The embryos were picked up with a glass Pasteur pipette (7.0 mm
OD) and transferred to l-DRAM glass vials, containing Formalin-Acetic
Acid-Alcohol (FAA) Galigher formulation (Galigher and Kozloff, 1971) for
ﬁxation of the embryos. The ﬁxative was washed out with, and the

embryos stored in, 70% ethyl alcohol.

Scanning Electron Microcopy

After ﬁxation, some H. anceps embryos were prepared for
scanning electron microscopy (SEM). They were dehydrated in a 70%-
95%- 100% ethyl alcohol series, with several changes Of absolute
alcohol. Specimens were critical-point dried in C02, sputter coated with
gold, and imaged in a JEOL JSM-6400V Scanning Electron Microscope
at 15kV accelerating voltage and a working distance of 39mm. Images
were stored as Tagged Image File Format (TIFF) images on a l-gigabyte

optical disk, as well as on loo-megabyte ZIPT" disks.

29

Light Microscopy

Embryos for light microscopy were dehydrated in 70%-95%- 100%
ethyl alcohol series, cleared with methyl benzoate, and inﬁltrated and
embedded in Paraplast (HRI 8889-501006) or Paraplast Plus (56—570C)
parafﬁn. Embryos were mechanically sectioned at a thickness of ﬁve or
ten microns. Sections were attached to glass slides using Mayer
albumen adhesive. Section ribbons were ﬂoated on water onto the
slides, warmed, and dried on a hot plate set to 37°C.

Slides were de-parafﬁnized in methyl benzoate. Sections were
stained following Heidenhain’s Iron Hematoxylin Method and counter-
stained with Orange G, as outlined in Galigher and Kozloff (1971).
Sections were cleared with HemO-D clearing ﬂuid and cover slips set

with canada balsam.

DIGITIZATION OF SECTION IMAGES

The sectioned embryos were imaged and digitized in transmitted
(brightﬁeld) (non-confocal) mode on a Zeiss 210 Laser Scanning
Confocal Microscope (LSM) (Carl Zeiss, Inc., Thornwood, New York)
using a 20X (0.5 NA.) dry or 25X (0.8 NA.) immersion Objective lenses
and the 633 nm line from the Helium-neon laser.

Images were collected starting with the middle section in the
series, which acted as a template or ﬁducial image for alignment of the

next two images. The last image aligned was used as the new ﬁducial

30

image for the next images. Aligning the images was done by hand using
the LSM’s “protect image” routine which allows the live image to be

overlaid on the protected image.

Image format conversion and transfer from the LSM to the SGI graphics

workstation

For volume rendering and image processing, the images created
on the LSM were transferred to a Silicon Graphics 4-D 30 Personal Iris
workstation running IRD( 4.0.5C (Silicon Graphics, Inc., Mountain View,
CA) in one of two ways:

1) Images were transferred from the LSM directly to the SGI via a
GPIB-AT (IEEE 488) (National Instruments Corp., Austin, TX)
connection, using the program VoxelScanZeiss. This program
automatically converts the native LSM picture ﬁle format "PIC" (not to
be confused with PICT or PIC used by Apple, Inc. brand computers) into
the native slice ﬁle format of the volume reconstruction program
VoxelView/ E.

2) Alternatively. images were transferred from the LSM to a Dell
Dimensions P90 computer (PC clone running DOS 6.22 and Windows
for Workgroups 3.11) via a GPIB-AT (IEEE 488) connection using the
software program LSM_NET (Carl Zeiss, Inc.). Images were transferred
in native LSM ﬁle format "PIC", and copies stored on loo-megabyte (MB)

ZIPTM disks and on CD-ROMs. The software program LSM_PC (Carl

31

Zeiss, Inc.) was then used to convert the images to TIFF 6.0 compliant
format (referred to as "TIFF", "TIF", "',tiff or "tif") (one of the formats
suitable for use in the software programs VoxelView/E 2.1.2 and
VoxelMath 2. l).

Unedited TIFF 6.0 image ﬁle copies were stored on 100MB ZIPTM
disks and on CD-ROMs. Images converted to TIFF format were
transferred to the Silicon Graphics, Inc., workstation. The SGI
workstation and DELL PC clone were connected via direct Ethernet
connection, facilitating image and ﬁle transfers between the computers

using FTP (File Transfer Protocol) client software over a TCP/IP network.

IMAGE PREPARATION ON THE SGI WORKSTATION

Rectangular pixels to square pixels: Maintaining correct image aspect
ratio

The LSM creates images of 512(x-axis) by 512(y-axis) pixels. The
individual pixels on the LSM have a 3:2 x/y aspect ratio, making them
rectangular. PC clones and workstations (such as the SGI) work with
1:1 x/y aspect ratio (square) pixels. Thus images created on the LSM
and displayed on a PC clone, MAC, or SGI, will appear "squeezed" along
the x-axis (see Discussion chapter).

This effect was corrected by “resampling” the number of pixels in

the y-axis from the original 512 to 341 pixels returning the images back

to their correct proportions. For the images transferred from the LSM

32

directly to the SGI via the GPIB AT connection, the program
VoxelScanZeiss performed the resampling from 512 x 512 pixels to 512
x 341 pixels automatically. The images transferred from the LSM to the
DELL PC clone then to the SGI were corrected using VoxelMath 2.1

routine Merge Operations> Resample XY.

Voxel-padding in VoxelView

The program VoxelView requires the dimensions to be evenly
divisible by four. Since the converted images are 512 x 341, the y-axis
dimension has three pixels added by VoxelView/E (although only when
viewing and saving) making the images 512 x 344 in memory. These
added pixels were removed when the images were “cropped” to remove
the LSM menu bar from the bottom of the images or to remove extra
"space" around the specimen or when creating a sub-volume Of the
specimen. The ﬁnal y-axis dimension was cropped to a number evenly

divisible by four.

The importance of the order of images

In this study, most of the images were collected from the middle
of the embryo back to the beginning and then to the end. This allowed
alignment Of the subsequent sections to the largest section with the
most structures present for registration (see Figure 28). Image ﬁles were
then renumbered individually or by the freeware program “Rname-IT'“

(by Steve H.).

33

The program VoxelView/ E reconstructs (or builds) the volume
from back-to-front by laying down whichever image has the ﬁle name
“1” (or "0") then stacking the ﬁle named “2” on top of it, and so on,
stacking the highest numbered (named) ﬁle on top (Figure 26 A). If this
is not the correct order for the images to be stacked, the resulting
embryo will have its left-right axis reversed. This is corrected by
renaming the ﬁles in reverse order, i.e., if there are ﬁfty images in the
set, image number “1” becomes image number “50”, and image number
“50” becomes image number 1, etc., as shown in Figure 28c. This was
done with the program VoxelMath 2.1 by “ﬂipping” the images along

their z-axis ("Flip-Z").

Filling in the Z dimension: Interpolations

The software program VoxelView/ E calculates the dimension of
the 3D voxels from the dimensions of the 2D pixels such that the
resulting voxels are iso-dimensional. Since the pixel resolution (i.e., x-
axis dimension or distance value of the specimen represented by the
pixel along the x-axis) is smaller than the distance between the images
or z-interval (in this study z-interval equaled the thickness of the
sections), and thus each voxel will only be as "thick" as the pixels are
wide, the resulting 3D reconstruction will appear "ﬂattened" in the z-
axis. To ﬁll out the space to better represent the true thickness of each

section "interpolated" images are calculated to ﬁll in the distance.

34

The number of interpolations needed was determined by dividing
the z-interval or step size (which in the present study since using
mechanically sectioned embryos equaled the section thickness) by the
"pixel resolution" (Table 1) or full ﬁeld view covered or represented by
each pixel in the x-axis (see Table 2). The size of the pixels was
determined by dividing the total x-axis distance of the sample in the
image (the screen width or "full ﬁeld width" of the LSM monitor) by the
number of pixels (512) (See Table 1). The full ﬁeld width of the image
was measured using the function “Measure” of the LSM 210. VoxelMath
2.1 Merge Operations > Resample Z was used to create interpolations

using non-integer resample values.

Reduction of interpolations: Sub-sectioning the seca'ons
The 20X and 25X objectives were used to create higher resolution

images Of just the developing eye—tentacle region of some of the
embryos. Since both the 20X and 25X objectives have a depth of focus
(depth of ﬁeld) smaller than the lO-um thick sections (see Table 1) two
images Of each section were created: one at the top of the section, and
one focused 5 microns into the sample. This allowed for a reduction by
half Of the number of interpolated images per “real” image required (see
Table 2) thus creating more accurate reconstructions (comparing with
whole mount, video taped images, or SEM images of embryos of the

same stage).

35

Table 1 Calculation of Pixel Size (resolution) in the X-axis.

A Dz(nm)= LSM Screen
>./N.A.2 Zoom width um Screen

633 28133.3 20 2838.8 512

633 7033.3 20 1419.4 512
633 2532.0 20 709.7 512
633 989.06 20 567.8 512

 

 

N. A. = numerical aperture A. = wavelength of light D2 = depth of field in focus

 

 

 

Table 2 Calculation of Z-axis Interpolations

 

Pixels to Voxel Interpolation;
LSM # of Pixel Size mm z-interval

  

 

 

 

 

screen (W71) x- resolution +

pixels axis. .. (mierval) pixel size
512 1.386 10 10+ 1.386
512 1.386 5 5 + 1.386
512 1.109 10 10+ 1.109
512 1.109 5 5+ 1.109

 

 

36

Precise adjustment of section image-alignment using "Negative"
interpolations

For very precise alignment [of sections], the program VoxelMath
2.1 was used to shift or rotate the problem slice the exact number Of
voxels needed to “best ﬁt” alignment. By using a negative integer for
interpolation in the "_dimensions" ﬁle of a volume data set, the
programs VoxelMath 2.1 and VoxelView/ E will display the 3D volume
with "empty" space between the sections instead of interpolated images.
By viewing the embryo with empty space between the "real" slice images
it is easier to determine which sections are out Of alignment with the
other sections.

Combining the use of negative interpolations with the ability of
VoxelMath 2.1 to view 2D images of the embryo "re-sectioned" in the X-Z
or Y-Z aids, and its ability to determine the X,Y, and Z coordinates of all
voxels by simply placing the cursor in the section image window on the
part Of the slice out of alignment and reading the voxel coordinates
displayed, it was possible to count the exact number of voxels an image
needed to be shifted to adjust its alignment with the rest of the embryo
(Figure 29). (It was very helpful to change all the scale values in the
"_dimensions" ﬁle to the value "1.0" so the coordinate readout equaled
the number Of voxels in each direction as Opposed to the "real world"

distance.)

37

Once the number of voxels a given slice image needed to be
shifted was determined, it was possible to use the "x-shift" or "y-shift"
math function in the Math Ops panel of VoxelMath 2.1 to correct its
alignment as compared with whole mount embryos, video images, or

SEM images of other embryos at the same stage Of development.

IMAGE PROCESSING AND ANALYSIS

Determination and delimitation of cell and tissue types

3D Visualizaa'on of 2D tracrhgs: The "Geometry" Funca'on

In early stages of embryo-genesis many structures have not
developed to the point of being readily recognizable (compared to the
adult or even older embryos). TO help make comparisons of structures
at different stages of development, and thus help identify unknown
structures in younger stage embryos the function "Geometry" in
VoxelView/E was used. This feature made it possible to trace known
and unknown features in the 2D section images and immediately
visualize their position in 3D relative to each other, and to the whole
reconstructed embryo. When the whole virtual embryo is rotated, the

image of traced structures in 3D rotates in synchrony (Figure 5).

Enhanced visualization
The LSM creates 8-bit images having 256 shades of gray. While

the developing nervous system and eyes are distinguishable by

38

morphological, positional, and in some cases cellular cues, the
grayscale value of each pixel/voxel (referred to as “voxel-value” in both
VoxelView/ E and VoxelMath 2.1) overlapped with other structures in the
embryo. This overlapping of voxel values limited the programs
VoxelView/E and VoxelMath 2.1’s ability to differentiate and delimit by
voxel value or voxel position the developing structures of the eye-
tentacle complex when three-dimensionally reconstructed.

To help visualize and differentiate the different structures two
techniques were employed to enhance their visualization (and thus
relative position) within the embryo: 1) “Digital Dissection” , and 2)
“Voxel-Value Range Partitioning” (assigning different ranges Of voxel

values to different structures).

Digital Dissegtiog
To enhance visibility Of the developing eye-tentacle complex, the

nervous system and eyes were “digitally dissected” from the rest Of the
embryo (i.e., the images of the cells forming the nervous system were
isolated from the rest of the embryo and made into their own separate
volume). The dissection of structures digitally is best performed before
adding interpolations if there are a large number of images.

Using the image manipulation program. Adobe Photoshop (version
3.0 or higher) or in some cases VoxelMath, each image was opened, its
histogram inverted, and the pixels of the image that represent non-

nervous system tissue were "painted over" with (i.e., converted to) the

39

same pixel color or value as that of the background of the image (i.e.,
turned black -- grayscale value=0) (Figure 21). The set of images created
was then saved as a new independent data set (volume), which
consisted Of just the developing nervous system while still in correct
spatial registration with the rest of the embryo. Adding Z-axis
interpolations for this new set of images creates a three-dimensional

volume of just the nervous system.

Voxel-Vgue Range Partitioning

To highlight the separate structures (e.g., the eyes, nervous
system, prototroch) following their digital dissection, each was assigned
to a different range of non-overlapping voxel values. The cells Of the
developing eye vesicles were assigned values in the range of 1-63 (0
being the value of the background), the cells Of the developing nervous
system in the range of 64-127 (Figures 22-25), and the rest Of the
embryo assigned values of 128-255. The assigning of structures to
separate ranges of voxel values was accomplished with the program
' VoxelMath 2. 1 in one of tvm ways:

1) Assigning a different range Of voxel values to different volumes; or
2) Assigning a different range Of voxel values to structures within the
same volume by using "selective application" of mathematical

operations using “Grease Pencil.”

40

Arugning diﬂerent range: of voxel value: to deﬂe’rent volume:

Each volume had its range of grayscale values (histogram range)
shifted accordingly by applying the mathematical operations as listed in

Table 3.

41

Table 3: Sub-dividing Histogram range: Mathematical operations

Structure/ Histogram Histogram range Histogram range

 

organ range 1-63 64-127 128-255
Eye anlagen * 63 / 255

Nervous * 63 / 255 bias=641
system

Rest of * 127 / 255 Bias = 128
embryo

 

 

 

 

 

Assigning dgﬁ'erent ranges of voxel values to structures within the same volume: Selective
apph'eation of mathematical operations using “Grease Pencil”

For younger embryos, which are smaller and thus comprised of
fewer images, assigning voxel values to given structures was
accomplished by ﬁrst shifting the whole volume into the range of voxel
values 128-255. Then, using the “Grease Pencil" function in VoxelMath
2.1 to apply the current Ops List only to those voxels selected, the
structures of the eyes, ganglia, prototroch etc., were shifted into their
own range of voxel values, as stated above.

To delineate the borders of the cephalic plates the function
Grease Pencil was used to convert the voxels representing the cells of
the prototroch (which are the ventral, medial to lateral borders) to voxel
value 1, and the last columnar cells of the dorsal most edge of the
cephalic plates to voxel value 2. Each had a unique color assigned using

the look-up table (LUT) features of the Histogram function of VoxelMath.

 

‘ “Bias” function adds the Argument value to each non-zero voxel value. This increases
equally all non-zero voxel values, thus shifting the histogram range from 0-63, to 64-
127.

42

Recombining volumes Visually vs. Structurally

Which software analysis functions will be applied (i.e., how the
3D reconstruction will be analyzed) determines what, if any, slice image
(2D) preprocessing is required. Some functions of the volume
reconstruction software, e.g., the ability to re-slice the virtual embryo in
any plane ("VoxelSlicer") and analyze the resulting images can only be
applied to a single volume in memory. Other functions allow for the
formation of images composed of several volumes in memory, i.e.,
"Multi-channel View." The latter function requires fewer processing
steps and so is faster, but limits the types Of analysis available. The
following functions cannot be used on a Multi-channel rendered
volume: VoxelS1icer, VoxelSeed, VoxelTrace, Histogram, any option in
Manage Data (Add Data set, Encode Volume, Save Volume, etc.), or the

add-on module VoxelAnimator.

Visual merging: M ulti-cbannel viewing

The individual volumes of the nervous system, eyes, and whole
embryo had separate color look-up tables and Opacity settings applied.
The eye vesicles were assigned a green spectrum. The developing
nervous system was assigned a red spectrum, and the volume of the
whole embryo was left in gray -scale. The prototroch was colored a solid
blue. and the dorsal border of the cephalic plates a solid magenta

within the whole embryo volume.

43

All the volumes were recombined visually as a single volume, with
their correct position maintained relative to each other, using the

function “Multi-channel View” in the program VoxelView/ E.

Structural merging: Merge Volume (Voxellllatb)

The separate volumes (see section Digital Dissection above) of the
eyes, ganglia, and the body of the embryo were recombined into a single
volume (one set Of images) using VoxelMath’s “Merge Volume> Option
“Montage” with Alternate Mode: Scalar=255. The Offsets in each
volumes “_dimensions” ﬁle was set to “0” creating 100% overlap. This

maintains the correct spatial registration of all structures to each other.

Image rendering

Single Images and Animations
3D images rendered with VoxelView/ E were either saved using

the Animations> Save Picture command, or 3D rotation images were
generated by using the Animations> Save Images to Disk command. In
the latter function, the computer generates multiple images "rotated"
around either the azimuth or elevation in steps of the number of
degrees speciﬁed (from 1 to 360). The combined set of images can be
viewed in quick secession as an "animation" thus aiding the visual 3D
effect. In both rendering functions the images are saved as SGI format

(.rgb) images.

44

 

Im ' in n I n P

Images Of the rendered virtual embryos which were saved in the
SGI image format RGB (. rgb) were converted to the TIFF image format
by the SGI program IMGCOPY before transfer to the PC. Since the SGI
workstation reads images from the lower left corner, and PC and MACS
read images from the upper left comer, the images transferred to the PC
needed to be ﬂipped vertically. This was done with the program Adobe

Photoshop (3.0 or above), or by other image processing programs.

Annotan'ons

The annotations (axis lines, measurement lines, and bounding
box) were created using the Measurements> Annotations function in
VoxelView/ E. The color of the lines can be selected, as well as distance
measurement values and calibration lines. The measurement values are
based on the values in the volume dimension ﬁle ("_dimensions") for the
axes scale values. The labels of the axis determined by the axes label
ﬁelds in the volume dimension ﬁle ("_dimensions"). The axes units value

is also determined by the text in the "units" ﬁeld of the dimensions ﬁle.

[IMAGES IN THIS DISSERTATION ARE PRESENTED IN COLOR]

45

Chapter 4

RESULTS

There is a single light of science, and to brighten it anywhere

is to heighten it evegyzvhere. "

-Isaae Asimov

PART I: EFFICACY OF 3D VOLUME RENDERING FOR THE STUDY OF

SNAIL EMBRYOLOGY
The use Of three-dimensional reconstruction, which combines LSM
digital imaging and the functions in the volume rendering software

program VoxelVieWZE 2.1.2 provided the following features for this

investigation.

Enhanced visualization and identiﬁcation of developing structures

Visualization of uniden riﬂed structures in 3D

The "Geometry" function of the software program made it possible
to trace known or unknown features in the 2D section images and
immediately visualize their position in 3D relative to each other, and to
the whole reconstructed embryo. When the whole virtual embryo is
rotated, the image of traced structures in 3D rotates in synchrony

(Figure 5). The 3D views of unknown structures were then able to be

46

compared with Older more developed embryos in 3D, thus aiding the

identiﬁcation of structures unknown in the younger stage embryos.

Visualization of internal structures by digital "dissection "

Embryos the size Of the pulmonate snail embryos (80-300um)
studied in this investigation are far to small to dissect in real life. Once
the 2D images are "rebuilt" into 3D virtual embryos, it is possible to
remove the outer tissues and organs leaving only the system Of interest,
in this case the developing nervous system (Figure 6 ). This made it
possible to determine the actual shape of the developing cerebral
ganglia and the trajectories of their "branches" or nerve tracks (Figures

7, 8. 9).

Visualization of structures with increased contrast by digital "staining "

Many structures of interest in this investigation were groups of
cells not yet developed to the point where they differentially stain. The
ability digitally to add or modify color for the purpose of improving
contrast enhanced the visualization of the structures, and aided in their
identiﬁcation (Figures 8 8t 9). A good example, the cells Of the
prototroch, are readily identiﬁable in a given 2D section image (Figure
10) as large clear ciliated cell(s). In 3D reconstructions and SEM
images (Figure 1 1) it clearly delineates the ventral and posterior-lateral
border of the cephalic plates (Figures 8 8t 10) Adding color to it to

differentiates it from the rest Of the ectoderm greatly enhancing its

47

visibility in the reconstructed embryo and thus improving its ability to

serve as a marker for the borders of the cephalic plates.

Visualization of slices taken in planes independent of the original
plane of sectioning by "re-sectioning "

The virtual embryos were "re-sectioned" in planes perpendicular
to those that the "real" embryos were physically cut. The small size of
the snail embryos in this investigation makes determining the
orientation of the embryos in mounting medium, and thus determining
the plane of sectioning, a major obstacle. Re-sectioning embryos makes
localization and identiﬁcation of structures easier. This capability is one
of the great advantages of digital imaging. "Matched" sets of slice images

in different planes can be viewed simultaneously (Figure 12).

Visualization of the Whole embryo and any "sub-sets " of it, ﬁ'om all
angles simultaneously

The ability to visualize the 3D shape of structures from multiple
angles simultaneously in separate viewing windows greatly aided in
their identiﬁcation (Figure 13 8: 14). Multiple views of the same embryo
or different embryos (or parts there of) side by side was very useful,
especially for side by side comparisons between embryos at different

stages of development.

48

Ability to determine in all three dimensions the precise spatial-temporal

position and development of structures

The ability to visualize the virtual embryo utilizing the features
and functions listed above (e. g., with overlying structures at various
degrees of being "dissected" away or rendered transparent and
structures of interest visually enhanced by the addition of color) allowed
the discovery of : 1) the exact position within the cephalic plates of the
invagination of the eye-vesicles (and distinguishing them from other
ingression of groups of cells), and in relation to the position of the
cerebral ganglia (and their branches) (Figures 8, 9, 10), and 2) the exact
position (and shape) of the cerebral ganglia before, during, and after

eye-vesicle invagination (Figures 7, 8, 9, 10, 13, 14, 15).

Ability to make direct quantitative measurements and comparisons

Measuring the accurate distance between structures ﬁ'om sectioned
embryos
One of the continuing problems in microscopy is the ability to

determine distances among three-dimensional structures. By using the
"Trace" function of the volume rendering software it is possible to select
any two points (voxels) within the 3D volume and have the computer
determine the distance between them in three dimensions (Figure 16).
Since structures are more complex than just two points in 3D space,

the Trace function allowed multiple distance measurements to be made

49

and the resulting values analyzed statistically. The results can be
formatted for importation into statistical analysis packages or
spreadsheet programs.

The distance of the cerebral ganglia from the buccal mass was
measured using the "lrace" function. The statistical analysis of the
results was automatically calculated (Figure 16). The distance from ﬁve
different locations from the edge of the buccal mass to ﬁve locations of
the cerebral ganglion (two from the top side, and three from the bottom)
were measured. The mean average distance of all the measurements
was 48.7um. The longest distance measured was 7 1.71mi (top most edge
of cerebral ganglion, thus furthest away). and the shortest distance
measured 33.8um. These distances compare well to the overall size of

the embryo.

Volumem'c measurements of developing organ systems
(morphometrics)

The capability to easily determine volume differences and changes
of structures (within and between species) used in morphometric
analysis was demonstrated by using the "VoxelSeed" function. In
addition to displaying the results of statistical analysis, the program
changes the opacity settings of the voxels measured and those that were
not and displays the 3D volume showing which structures were
measured within the whole volume (or sub-volume) (Figure 17). Results

of the statistical analysis of the comparison of developing nervous

50

system (NS) and eyes in an H. anceps embryo are shown in the Table 4.
The measurement of the nervous system on the right also included
ganglia tissue besides the cerebral ganglion, thus accounting for its
much larger volume measurement of 144305.00 um3. The measurement
of the left cerebral ganglion only included the left cerebral ganglion
tissue and thus had a smaller measured volume of 45047 .00 mm. The
right and left eye-vesicles where found to have very similar measured
volumes

(14154.00 m3, and 14342.00 uma, respectively).

Table 4 VoxelSeed Summary of Statistical Analysis

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Morphometric Statistics of 4 Selected Volumes

Minimum Maximu Mean Median Mode Std. Dev
m

14154.00 1443050 54462.00 14342.0 14154.0 61629.96
0 0 0

Voxel - Metrics of Selected Points

Total Voxels Rendered 160154 Total Integral = 1.219689E

= +07

Minimum Maximu Mean Median Mode Std. Dev.
m

2 118 76.16 79 64 0.10

Morphometrlc Summary of 4 Individual Volumes ﬂim3)

Left Cerebral Ganglion 45047.00

Left Eye 14342.00

Right Eye 14154.00

Right NS 144305.00

 

 

 

 

 

 

 

 

 

 

 

 

51

PART II: DEVELOPMENT OF THE EYE-TENTACLE COMPLEX IN THE
AQUATIC BASOMMATOPHORAN SNAIL HELIsom ANCEPSAND THE

TERRESTRIAL SNAIL ANGUISPIRA ALIERNA TA

General pattern of pulmonate development

To compare development of the embryos within and between the two
species of snails, forty to ﬁfty embryos of each species were analyzed to
determine stages of development. The eye-tentacle formation occurs in
stages (late) A & B St C (Figure 4 A-C:2&3), which corresponds to
"trochophore," "veliger" and "adult-like form" of Morrill (1982),
"trochophore," "veliger" and "hippo" stage of Raven (1949) and ","E2 "E3"

to "E5" of Cumin (1972).

WW);

The embryo develops into a trochophore larva following
gastrulation. At this stage the following structures ﬁrst become visible:
the protonephridia (larval kidneys); the "prototroch", a band of ciliated
cells which delineates the dorsal and lateral sides of the stomodeum
and the ventral and lateral borders of two bilateral regions of numerous
small cells -- "cephalic plates"; a line of ciliated cells running
longitudinally along the ventral side of the foot anlage is present; and
the shell gland is visible as a concave or invaginated region of the

posterior ectoderm.

52

51mm
At this stage of development, the foot is visible as an outgrowth

below the stomodeum (Figure 4: A2, B2, C2). In Anguispira altemata the
foot is more developed than in Helisoma anceps mainly due to the
presence of the "podocyst" (an enlargement of the end of the foot found
in Stylommatophora) (Figure 4). The pedal ganglia are separate from the
ectoderm of the foot (Figure 18) and the statocysts are visible with
obvious lumina. The stomodeum is continuous with the esophagus, and
the radular sac is visible as a posterior outgrowth of the ﬂoor of the
stomodeum (Figure 13). There is shell material of larval shell or
"protoconch" visible covering the shell ﬁeld. The prototroch is more
distinct in H. anceps (Figure 4) and clearly delineates the ventral
borders of the now more distinct cephalic plates. The prototroch is not
as continuous in A. altemata.

At this stage of development the relatively few cells forming the
cerebral ganglia ( ~25-50) have separated from the overlying ectoderm.
These cells, along with the cells of the larger (more developed) pedal
ganglia, have larger nuclei than the cells of the surrounding mesoderrn
or ectoderm. Their cytoplasm appears lighter stained than the
ectodermal cells or surrounding mesoderrnal cells, with a single large
and dense staining nucleolus (Figures 18 8t 19), as has been commonly

reported for many other gastropods (Henchman, 1890; Raven, 1966).

53

In both species studied, under the developing tentacle (dorso-
posterior pair in A. altemata) of the latero-ventral region of the cephalic
plate, a large indentation has formed (Figures 10 8r 1 l, and 18a & 27a).
This large indentation appears in more developed embryos as a larger
tube-like pocket. From the 2D sections cells appear to be proliferating
at the inner-most end of the indentation/ tubes along-side the
developing cerebral ganglia. From the 3D reconstructions a large
branch of the cerebral ganglion can be seen projecting directly towards
this indentation (Figures 9 8r 1 1). (See discussion chapter for
consideration if these develop into what is described as "cephalic
tubes").

This stage corresponds with Morrill’s (1982) "Adult-like form" and
Cumin’s (1972) "E5" stage. At this stage, the shell covers a third or
more of the embryo (Figure 4:A3-C3). The foot, as well as the podocyst
of A. altemata. is much larger than in the previous stage, and the head
vesicle (a swelling of the head ectoderm, dorsal and posterior to the
cephalic plates) is more pronounced (Figure 4: C3, D, E). The anterior-
most region of the ventral surface of foot of H. anceps is ciliated.
Radular material. forming the radula, is now visible, and the nephridial
organs are more pronounced. The prototroch is no longer an
uninterrupted band in H. anceps, (Figure 4: B3) and is further reduced

in A. altemata.

54

Sdecdon
Thirty to forty embryos of each species were selected from these

stages. They were imaged on the LSM and further categorized as to
suitability for reconstruction (i.e., overall quality of sectioning, staining
properties, etc.), whether in whole or in part. Five to ten of each species
were reconstructed using "volume rendering" 3D reconstruction
software to determine the precise location of the points of invagination
of the eye-vesicles within the cephalic plates and the exact position of
the cerebral ganglia at that time. SEM images were also collected of H.
anceps (Figure 4; Figure 18:b; Figure 1 lzd-i); these were compared to
the 3D reconstructed embryos to help determine the level of
development, as well as the timing and positioning of points of

invagination of the eye-vesicles (Figures 7, 10, ll, 20).

Eye-tentacle complex development in the stylommatophoran snail

Anguispira altemata

In stage B embryos (early veliger) a very small ingression of cells
was observed in the dorso-medial region of the cephalic plates (Figure
8). This ingression consisted of 4-5 cells and was seen in two
contiguous sections. Thus it extended between only 10-20um. This
structure was not seen in any younger or older stage embryos.

3D reconstruction revealed that this ingression of cells lay just
above a group of cells forming a portion of the cerebral ganglion. It also

revealed that this dorso-medial ingression was relatively far removed

55

from the dorso-lateral location of the developing eye in later embryos.
(See Chapter 5 for discussion of the possible identity of this ingression
of cells.) A small line of cells "branching" off the cerebral ganglia as can
be seen directed at (almost in contact with) this small medial ingression
of cells (Figure 8). At this stage of development, there is no similar
branch or line of cells directed at the dorso-lateral point within the
cephalic plates where the invagination of the eye-vesicles will occur.

The earliest signs of the development of the eye-vesicles were
found in late stage B (veliger stage) to early stage C embryos. The eye-
vesicles were ﬁrst identiﬁable as small invaginations of the ectoderm
forming a rosette of cells which pinch off from the ectoderm (Figure 10).
The eye-vesicles were visible spanning three to ﬁve sections, thus about
30-50um in size. They consisted of approximately ﬁfty to seventy-ﬁve
cells each. There is a lumen formed which increases in size with age
after the vesicle has separated from the overlying ectoderm.
Pigmentation ﬁrst was visible in late stage C embryos. Later still, a lens
formed in the middle of the lumen.

The point of invagination of the eye-vesicles was in the dorso-
lateral edge of the cephalic plates (Figure 10). The location of where the
eye-vesicle formed was not on the tentacle anlage, but posterior to it (as
seen in Figures 10 8t 11). The cerebral ganglia were not found in contact
with or near the developing eyes. In over forty embryos analyzed in both

2D sections as well as in the 3D reconstructions(Figures 8, 9, l l),

56

spanning the stages of development from A to late C. neither the
cerebral ganglia, nor any of their parts (branches) were ever found in

contact with or even close to the eye vesicles, or the overlying ectoderm

from where they arose.

Eye-tentacle complex development in the basommatophoran snail
Helisoma anceps.

In H. anceps, the earliest signs of the development of the eye-
vesicles were found in late stage A embryos. The eye-vesicles were ﬁrst
identiﬁable as small invaginations of the ectoderm forming a rosette of
cells which pinch off from the ectoderm (Figure 7 b). The eye-vesicles
spanned two contiguous sections, thus about 10-20um in size, and
consisting of approximately ten cells. There is a lumen formed, which is
larger in size in older embryos. In stage B embryos the eye-vesicles are
lauger, composed of approximately twenty to thirty cells, and have
SEparated from the overlying ectoderm. In late stage C embryos the ﬁrst
Signs of pigmentation becomes visible. Later still, a lens forms in the
Inicidle of the lumen.

The point of invagination of the eye-vesicles was in the dorso-
lﬁlteral edge of the cephalic plates, slightly dorso-lateral and posterior to
the tentacle anlagen (Figure 1 lzd-f; Figure 15). Comparison of the SEM

kn'clges (Figure l lzd-ﬂ with the 3D reconstructed embryos and 2D

57

images of sections (Figure 7) conﬁrms this relatively early time of eye-
vesicle invagination compared with A. altemata.
At the time of invagination of the eye-vesicles the cerebral ganglia

were found were in direct contact (Figure 7). The cerebral ganglia were

 

found to be in contact with the eye-vesicles even after the eye-vesicles
had separated from the overlying ectoderm in older embryos (Figures

12.13, 14).

Table 5 Developmental stages

 
       

development ingression of Pigmentation
of cerebral unknown cells of eyes
anglia

  
  

 

eye-vesicle Radula teeth
invagination formed
tentacle
development
large indentation /
"cerebral tubes"

 

 

 

 

development tentacle Pigmentation
of cerebral development of eyes
ganglia
eye-vesicle Radula teeth
invaggiation formed

 

large indentation/
"cerebral tubes"

 

 

 

 

 

Comparisons

In both species studied the eye-vesicles developed from
LnVaginations of the ectoderm in the dorso-lateral region of the cephalic

Plates dorso-lateral to the tentacle anlagen. In Helisoma anceps this

58

invagination occurred early in development (trochophore - stage A) and
was laterally asynchronous -- the left side developing somewhat earlier
than the right. In Anguispira altemata the invagination occurred
relatively later in development (late veliger - stage B) and was laterally
synchronous -- both sides occurring at the same time.

In H. anceps the developing cerebral ganglion tissue was in
contact with the presumptive eye-tissue throughout the period of
development looked at during the present study. In contrast, the
presumptive eye—tissue of A. altemata was never in contact with
developing cerebral ganglion tissue.

In both species a large indentation of cells in the ventro-lateral
region of the cephalic plates (corresponding to the "cerebral tubes”)

appeared during the veliger stage (stage B).

59

PLATES

IMAGES IN THIS DISSERTATION ARE PRESENTED IN COLOR

60

 

Figure 5 "Geometry" Function in VoxelView/E

Screen shots of Geometry function. Figure 5a and
5b are matched views, as are So and 5d. The view of the
3D rendered embryo is the same view as its matched 3D
geometry view.

Individual structures are traced in the 2D section
images (5e), and their position in 3D relative to the other
structures is immediately visible, as is the view of the
whole embryo.

All structures are color coded between the 2D
section images (5e) and the 3D rendered images (5a and
5d).

 

 

61

 

Visualization of
relationship of
structures in 3D
which were traced
from 2D images of
sections

 

 

 

 

 

Figure 6 Digital "dissection“ of the nervous system of snail
embryo.

Images a-f show stages of dissection or separation
digitally of the developing nervous system and eye from
the rest of the embryo.

Images a) and b) show the whole embryo.

Image c) shows a subsection (sub-volume) of only
the region surround the nervous system and eye.

Image (1) shows the sub-volume from image c) with
the surrounding non-nervous tissue made transparent;
and images e) and i) show just the nervous system and
eye as its own independent volume.

Nervous system is red; eye is green; and the rest of the embryo
is gray.

 

63

 

Digital "dissection"

 

 

 

Figure 7 3D volume reconstruction of left cephalic plate of
Helisoma anceps embryo at time of eye-vesicle
invagination

Image a) is a 3D reconstruction showing part of the
cerebral ganglion connected to the eye-vesicle at the time
of its invagination (image b).

Image b) is a transmitted light micrograph used in
the reconstruction seen in images a, c, and d.

Image c) shows the location of the point of
invagination (green) of the eye-vesicle within the cephalic
plate. The cephalic plate is bordered on the ventral side
by the prototroch (in blue), with the dorsal border
highlighted in magenta.

Image (1) is a rotated view the same volume as seen
in c, (with the body tissue made transparent to allow
visualization of the internal cerebral ganglion (in red))
showing the cerebral ganglion in contact with the eye-
vesicle.

Cerebral ganglion in red, eye-vesicle in green, prototroch in
blue.

 

65

 

eye vesicle

/\t

   

 
 
   

eye—vesicle
invagination

66

 

 

Figure 8 3D reconstructions the left cephalic plate of a
stage B Anguispira altemata embryo

The 3D reconstruction this stage B A. altemata
embryo shows the left cephalic plate with its borders
highlighted. There is an ingression of a small group of
cells located dorso-medial within the cephalic plate seen
in images a, b, and c. Also the location of a large
indentation, which in later embryos appears as a larger
tube-like structure, can be seen.

In image b) the body tissue was made transparent
allowing the visualization of internal structures. The
cerebral ganglion has a small branch extending toward
the small ingression of cells. Image c shows a side view of
the left cephalic plate.

 

67

 

    

3D reconstructions
of
Anguispira altemata

 

 

Figure 9 3D reconstruction of left and right cephalic plates
in Anguispira altemata at time of eye-vesicle
invagination

These images show the position of the cerebral
ganglia at the time of eye-vesicle invagination. Image b
(bottom) shows the position of the cerebral ganglia
relative to the invagination of the eye-vesicles, and to the
cephalic plates (bordered by the prototroch).

There are four main branches of the cerebral
ganglia visible. The anterior dorsal branches form the
cerebral commissure; the anterior ventral branches are
the cerebral-pedal connectives. The largest ventral
posterior branch extends to the innermost end of the
large indentation. The smaller medial posterior branch
extends toward the point of invagination of cells seen in
Figure 8. As can be seen in Figure 9 a and Figure 9 b,
the branches of the cerebral ganglia neither extend nor
contact the eye-vesicles.

Structures in red are cerebral ganglia; Green are the eye-
vesicles, and blue the prototroch.

 

69

 

\IHI\.I|

141‘ "-
l

"W m m /I

lnguispiru allurnulu

'N
—
v
I
~
.—
.—
.—
’a
:2
no
—
—
f;
.-
.—
f:
a
C
.—'.
v
:.
—
T:
C
'a
P.
I
a.
C
.—
.—
—.
..
.-
f:
P.
C
A
—
_,
—
I:
C
/

 

 

 

Figure 10 3D reconstruction of Anguispira altemata cephalic
plates before and during invagination of eye-
vesicles

Image a) shows one of the original sections of the
embryo used for the reconstruction. The invagination of
the right eye-vesicle is visible as are cells of the
prototroch. The left side shows the left eye—vesicle. Pedal
ganglia are clearly visible, but no cells of the cerebral
ganglia are evident.

The location of the points of invagination of the
eye-vesicles within the cephalic plates is shown in
images b) and c). The points of invagination for the eye-
vesicles are clearly seen to be dorso-lateral to the large
indentations, and posterior / lateral to the tentacle
anlagen.

Images (1) and e) show the location of the
ingression point of the small group of cells dorso-medial
to the large indentation and tentacle anlagen compared
to the dorso-lateral location of the invagination of the
eye-vesicles.

 

71

 

 

eye i vaginati
{a

 
  
   
      

        
 

?

/

Eye-vesicle” /
c i

prof troch cell

     

  
 

  
 
  

   
  

Right side I '-' 180w:
i dyeioiln’g "—'
a . " ._.- left side

llllillll,lllvrll

‘.\ll|llll|(|ll.l| [ l

llrr‘lll.lllnll

14 II liltlllii |.lI|ll))
\\llll lllli'lll.l|

«r-ll lili>llll‘l.lll(vll

I)

 

 

Figure 11 Comparison of 3D volume reconstructed embryo with
SEM images

Images a—c are of the right cephalic plate at the
time of eye-vesicle invagination of Anguispira altemata
with different levels of opacity applied to the body tissue
revealing the internal cerebral ganglion. Notice the
branch of the cerebral ganglion that is directed at the
large indentation. No branch is directed at the eye-
vesicle.

Images d-f are SEM images of Helisoma anceps at
the time of eye-vesicle invagination. Comparison of the
SEM images H. anceps with the 3D reconstructions of A.
altemata show the location of the point of invagination of
the eye-vesicles are the same for both species.

 

73

 

 

Il‘llliﬂll‘
.lll|.l:‘_l‘

[illillll.lllilll
\\llil llllt'lll.li
It'll lli \I In

ulrlnwl ".lll"ill)|l

lr‘nlur Il'
.llllllt'l‘

llllll'lll.lllnll

\\|lil|l|1l’lll.li

(IHIW-‘j
(ill Ilt'\.l ll)

(I‘lt |)|.li thalliﬁllnll

left side
cephalic plate

stomodeum

d

Allr/I/Is/ul.) .l/Iumzlle developing
foot

Helisoma anceps

 

 

 

Figure 12 Digital re-sectioning of embryo

These images are of digitally re-sectioning of
Helisoma anceps embryo in planes perpendicular to the
plane in which the embryo was mechanically sectioned.

The eye-vesicles (in green) and the nervous system
(in red) are visible in cross section. While the resolution
of the images of re-sectioning is much less than the
original section images, they are still useful in
determining the positional relationship of structures.

 

75

 

cells of the
nervous
system

cells of the

eyc-Vt‘xtx In)

cells of the
nervous
system

b Re-sectioned in ZX plane

\"r~|lll‘:|| lmll'nl'

(liggilulli (‘lll ('mln‘yn

 

7f)

 

 

Figure 13 Original 2D section image and 3D reconstruction
showing nervous system of Helisoma anceps
embryo

Image a) is of one of the 2D images of the original
sections. The image had its grayscale inverted. Images b)
and c) are 3D reconstructions with different levels of
opacity (transparency) applied to all non-nervous tissues.

 

77

 

HSA94el 0

stmnodoum

Grayscale inverted

ll

 

Helisoma anceps stage C embryo

 

 

Figure 14 30 reconstruction and re-sectioned 2D slices

Image a) shows a 3D reconstruction with the body
of the embryo made transparent so as to visualize the
nervous system inside.

Image b) shows cross sections of same embryo
digitally re-sectioned in the Y-Z plane.

For images a) and b) cells of the nervous system are colored
red, and the eye-vesicles are colored green.

 

79

 

 

3D reconstruction and re-sectioning
of stage C Helisoma anceps embryo

30

 

 

Figure 15 More 3D reconstruction images of left cephalic
plate at time of eye-vesicle invagination in
Helisoma anceps

Images a) shows the location of eye-vesicle
invagination as dorso-lateral within the cephalic plate.
Images b) and 0) show the cerebral ganglion in contact
with the eye-vesicle.

Nervous system tissue is in red, eye-vesicle in green, and
prototroch in blue.

 

81

 

( 1)
L100 00
dorsal

Left cephalic plate

cerebral
ganghon

«hint .-
anterior ventral posterior

 

 

 

Figure 16 Measuring distances in 30 using the "Trace'
funcﬁon

This is a screen shot of VoxelView/E’s "Trace"
function. Any two (or more) section images are selected
and then a line (or multiple lines) are drawn from a
selected point in one section to a selected point in the
second section. Only the points of where the lines stared
in one section and ended in the other are visible in the
sections. The black arrows and arrow heads point to the
starting and ending points for some of the distances
measured.

In this example, ﬁve lines were drawn from the
buccal mass (lower image) to the cerebral ganglion
(upper image). Statistical analyses (morphometrics of the
lines, and distances) are displayed in the blue boxes.

 

83

 

Measuring distance between objects in
3-Dimensions using
VoxelMath 's T RA CE feature

.u... .L ,1/ "P! a ‘

l" —l:!

on: :35: 0'03 (in: tin: 603 a: do: 5:: do: '-

l)lll‘(‘;ll muss

LSM L5 5 E
LAB LA

 

 

 

Figure 17 Volume measurements using the “VoxelSeed“
function

Image a) shows the eye-vesicles (in green), and the
nervous system(in red) after being "seeded" (i.e., the
voxels that matched the criteria set in the green window
of image b) have their volume measured. The rest of the
embryo has its voxel opacity set to 50%, making the
structures that were measured more apparent.

Image b) shows the measurements determined by
VoxelSeed. The histogram and morphometrics of the
voxels counted are displayed in the blue windows.

 

85

 

 

 

 

M l—“llEll m T1

[brocnm (‘arplcrod 15037 voxels tam
We Cant 1433.3 vamis

5m: ”imam
wmmnwnas E3.“
mmmw -mm

a a

m n

H a

mm ”mm

Volumv I lixingram
Mm 10.111300 lvl.v:-: lll'l?2‘>w ul 1 ll
low " llll llvg ll 11‘: ill]
'l'm.»l‘l‘r_v"zrwrl ’1 Ni" Ulnlr’mu-

lunrml l ’l" lull lT‘rlrxll',‘

lllr

 

 

 

W

Volume Morphormlrirs Simistic;

vm —u

 

 

 

Figure 18 SEM image of H. anceps and original 2D section
of A. altemata with labels

Image a) shows an SEM image of a veliger stage
(stage B) H. anceps with labeled structures.

Image b) is one of the section images used for
reconstruction of veliger stage (stage B) A. altemata, with
structures labeled.

 

87

 

\ll ‘llll ‘tl’t'lllll

l‘iml

 

l ell cephalic plate

Pedal ganglion Prototroch

    

1

  
   
 
 
 
  
 
 
 
 
 
 

Buccal mass

I ' 1350
~ phagus
“ 0‘ ‘s - I. r
..s ,' ‘ Storage
. I
a. cells
0 ‘J
Podocyst Anterior ,
9
V 1 Right - ‘
cntra '
Left $
r .

  

Posterior

88

 

 

Figure 19 Light micrographs of cells of cerebral ganglion

Light micrographs showing visual and structural
differences of the cells forming a cerebral ganglion in
Anguispira altemata.

The cells of the developing cerebral ganglion have a
larger nucleus than the surrounding mesoderrn or
ectoderm cells, nucleus stains lightly with hemotoxylin
stains (more transparent), with a single large denser
staining nucleolus than the others cells around them.

Image b) is a higher magniﬁcation of the same section as
seen in image a).

 

89

 

l

7.. '

9 g " J.\
/
l
\"
k
I
"
Q
QUUH:4U l:3; (=49? :: 7 to

I

 

cerebral ganglion
"E" ' ~ r' "II-r
’ I“ ‘ _ 0

a .

."'
CO

' I

 

.- E=43 F0
b Anguispira altemata

9O

 

 

Figure 20 Conversion of 2D sections to a 3D volume

Images a-e show the change from a single 2D
section image (a), to a stack of 2D section images (b),
then to a stack of images with interpolated images
added(c), to 3D volume rendered reconstructions (d 8: e).

Image d) shows the position of eye-vesicle
invagination (arrow).

Image e) is the same data set with different
rendering settings applied. The prototroch is highlighted
in blue (arrow).

 

91

 

 

92

 

 

Figure 21 Digital dissection: processing of images

The images a-c show the steps used in "dissection'
or segmentation of the nervous system "by hand" (prior
to 3D reconstruction) using Photoshop 3.0.

The step of converting the non-nervous tissue to
black (pixel value "0") does not affect the registration of
the resulting images of the nervous system. The ﬁnal set
of images is saved as a separate data set and can be
visualize separately by 3D reconstruction or visualized
"merged" together with the whole embryo.

 

93

 

This is the original image
ol‘scclion 8.

It has been converted to a
TIF tile format and the
rectangular shaped pixels
have been com erted to
square pixels. Thus the
original SIZ x 5l2 pixels
are no“ 512 \ 34!.

 

 

HSA94e8 section #8

This is the same image
with the gray scale
inverted and the info
bar cropped out.
512x300

 

In this picture all but
the nervous

system has been
removed by hand
using Photoshop.

The registration ol‘the
NS is left unchanged.

 

94

 

 

Figure 22 Shifting the range of voxel-values:
Original image and histogram

Image a) is an image of the “digitally” dissected
nervous system (as shown in Figure 21 c).

Image b) shows the histogram and voxel value
statistics (lower right hand box in blue).

 

95

 

 

a Original image with unaltered histogram

 

3.13:; my.” Look-Up- Table Controls 2 z. s.
,_ n 111 u lll
. . p' '3'; ll

_' (*) 0111 Mill

Hill 0111

tomardrn, F9 m.

tart-i: _

l
.3
i

.4 ‘. d“

'. 1 r '7
.<' 115:5 9» giantess.

 

 

 

 

 

Figure 23 Shifting the range of voxel-values:
Image and histogram after multiplying the values
by 63

Image a) shows what the image looks like after
multiplying each voxel value by 63. Only 256 values are
shown at a time for 8bit data.

Image b) is the corresponding histogram for the
image above. Notice in the statistics, the computer keeps
tracks of all the voxel values, although only 256 values
appear in the image and histogram windows.

 

97

 

 

a Image after multiplying by 63 (only ﬁrst 256 values are shown)

 

Hill Illll I'illl

i'Illl ”III “III

Median%
Illll ”III “III

forwardFFT, F9 inverseFFl’, F10

‘ mesa; 'il «grail lair}

V v.1. ) 8.911353%! 5

Mask II 1 stogran

 

b Histogram of above image

98

 

 

Figure 24 Shifting the range of voxel-values:
Image and histogram after multiplying the values
by 63 and dividing by 256

Image a) shows the image of the nervous system
after being multiplied by 63 and divided by 256.

The histogram in image b) shows the voxel-value
range of the above image “squeezed” into one quarter of

its original range.

 

99

 

 

a Image after x63 and divided by 256

n "5 RGB-M look-Up-Tablc CORRDIS ;n,-_,
. ' win ”1“ HIH

_“ (*) rim Hill Hill

Medianx
um IIllI niii

 

iomardFFT, F9 m.

 

b Histogram after x63 and divided by 256

lf‘iel

 

 

Figure 25 Shifting the range of voxel-values:
Image and histogram after multiplying the values
by 63 and dividing by 256 and shifting all non-
zero voxel values by 64 ("bias=64")

The last mathematical manipulation of the nervous
system data set, adding 64 to all non-zero value voxels,
has the effect of shifting the range of voxel values for the
data set from 0-63 to 64- 127 . By using the “bias”
function the background of the image stays at voxel
value “0” (in this case the color lookup table in use has
“0” equals black as seen in image a). Thus, the program
can maintain the distinction between the “object”
(nervous system volume) and the “background”.

The range of voxel values has been “squeezed” into
one quarter of its original range, thus reducing the
volumes visual resolution to 1A: its original. This has no
effect though on the spatial resolution.

 

101

 

 

a Image after x 63 divided by 256 and "bias" 64

 

 

{ :illilltl RGB-M m-Up-leh Controls 25:.
II III II III
3‘. 7" t1

_‘ (*) um Illll

I‘llll l‘Illl

'i new.) '

' I ' . .- . V .
M19. 'F'r‘ursﬁ Finish" -

313-.i,,’;?id."_”..vig‘§'ﬁ 5A1;

.- s.‘

 

Mask Iiistogru

 

 

_
b Histogram after x 63 divided by 256 and "bias" 64

102

 

 

Figure 26 Image “stack” order

A) Images stacked in the proper order to maintain
the correct left-right axes. Image number “1” is
on the bottom, with images 2, 3, and 4 stacked
on top.

B) Images are in an improper stack order.

Redrawn from Czymmek et al., Exp. Mycol. 18:275-293.
1 994.

 

103

 

uh
l 3D volume reconstruction:
run Z-axls Image stock
\

 

 

Correct image stack order

m
l 3D volume reconstruction:
Yule Z-axls Image stock
\

 

        
   

r4». .
P at
Imagol‘l (mum)

Incorrect image stack order

(Modified drawing from Czymmek, et al., 1994)

104

 

 

Figure 27 Comparison of SEM of whole embryo with partial
3D reconstruction: Effects of correct and incorrect
image stack order

The SEM Images a) & b) show a snail embryo and
the region reconstructed in images c) & d), which is the
left cephalic plate and left side of the foot.

Image c) is a 3D reconstruction with the proper
order of images, thus the resulting reconstruction
appears correctly as the left side of the embryo.

Image (1) is a mirror image of image c) and shows
what the reconstruction would look like if done with the
reverse order of images. Thus, image d) appears to be the
right side of the embryo.

As can be seen by the above images, whether the
reconstruction looks like it is of the left or right side of
the embryo depends on the stack order of the images.

 

105

 

lllll lllll

 

d
Correct reconstruction Incorrect reconstruction

106

 

 

Figure 28 Order of sections on the microscope slide vs. the
order of digital images needed to reconstruct
correctly in 3D

As can be seen in this ﬁgure, the order of the
images collected from a microscope slide must be
reversed (i.e., the sequence of images reversed) to
maintain the correct left-right axis.

 

 

107

Correction for image
"stac " order needed
for 3D reconstruction

Section / Image # 1

l.
[11:33:

Section / Image # 50

 

 

 

”A \ Original image

5 ‘. w.- ‘ sequence
A}; ,

 

Image #1 Image #50

 

    

J , ,1 ~ Z-axis ﬂip 3’53.
F \x’ ' 1f“. reverses the sequence ,’ ,y. 9"“
" .- -t t 1 ~ oftmages ,1 3;“ if.
i . . , ;

 

Image #1 Image #50

108

 

 

Figure 29 Digital realignment of sections of section images
using “negative” interpolations

The use of "negative" interpolations to spread out
sections with empty space between them (white area
between sections), and changing the view to the X-Z axes
plane, allows one to see very clearly just how many
voxels a given slice maybe out of alignment with the
other shoes. This technique also makes it possible to
count the exact number of voxels any slice is out of
alignment, and thus know by how much and in which
axis to make any needed adjustments.

 

 

109

X-Z

X- Z view of' 'negative" interpolations

////,'///’/ ”///’ //¢666 // /181////
47/ Z/// ///4// / // // ////

 

16614 13 12
/////////// ///////////// 15
Slice 126 5X magniﬁcation
Slice 114
HSA94e10

 

 

Figure 30 Computer generation of 3D voxels from 2D pixels

The volume rendering program used in this study
(VoxelView/ E by Vital Images, Inc.) uses the 2D (x, y-
axes) pixel dimensions to calculate a virtual 3D (x, y, z-
axes) voxel. To render an image, the program stacks the
voxels of each image from back to front (towards the
screen).

(Redrawn from Vital Images, Inc. Fairfield, IA.)

 

111

 

Yaxis

 

 

 

 

Xaxis

to the screen

From 2D pixels
to 3D voxels

 

 

Figure 31 Maintaining the correct image aspect ratio:
rectangular to square pixels

Images created on the LSM are collected and
displayed with rectangular pixels having a 3:2 aspect
ratio. The images are collected with 512x 512 pixels (x,y).
When one of these images is displayed on a computer
using 1:1 aspect ratio (square) pixels, the image appears
stretched in the y-axis as shown in image a).

Image b) show correction for this by resampling
along the y-axis, converting from 512x 512 pixels to 512x
341 (x,y) thus, returning the image to its correct aspect
ratio.

 

113

 

512x
ECI

 

 

a) 512 x 512 pixels

l”06 33V

114

 

b) 512 x 341 pixels

Chapter 5

DISCUSSION

"A m'enlz'rt is not one 112/)0 can answer queriz'onr, but one

w/Jo can quertz'on annuerr. "

T/Jeodore Scbz'ck ]r., Skeptical Enquirer, 2 7 -2:3 9

TECHNIQUE ISSUES
The approach developed for this study, while replete with powerful new
tools for embryological investigations, is only as effective as the
techniques are free of error, and as the approach offers advantages
over, not just equal to, other approaches. At all stages in the creation of
knowledge errors can occur. One criterion of the competence of an
approach to the acquisition or creation of new knowledge is in its
robustness against the creation of artifacts, and its ability to help detect
or correct for self-generated error. The approach developed in this study

meets this challenge.

115

Repeatability & Reliability

The more traditional methods of embedding, sectioning, and
staining embryos for light microscopy were employed for creating the
serial sectioned embryos (as described in Chapter 3: Materials &
Methods), and thus are susceptible to most of the same criticisms.
Issues such as quality of ﬁxation, mechanical disruption (and thus
displacement of structures), shrinkage of tissues, and uncertainty of
staining properties are always present with these techniques.

The physical sections on glass slides will deteriorate from within
minutes to years, whereas digital data can (theoretically) be maintained
at extremely high coherence indeﬁnitely. Several copies were made of all
digital images, and stored on separate data medium: hard-drive of
computer, 100 MB Zip disks, QIC tape, and CD.

When working on digital data, unlike "the real thing", it is
possible to make multiple "exact" copies at each stage of data analysis.
Thus, manipulations to the data are functionally non-destructive,
reversible and repeatable. This allows multiple different analyses to be
performed on identical copies of data, reducing the wide variance of
results caused by unknown individual variation (when numerous
different original samples are used for each analysis). Since the copies
are identical, all individual variation originating from the specimen itself
(as well as those artifacts and changes created by physical processing)

are identical.

116

The original set of images, as well as intermediate sets, are always
available. This creates the ability go back a step or more and change
processing of the images from that point forward, without having to
start the whole process over from the beginning with a new sample.
Thus, the same sample can be analyzed and manipulated repeatedly, as
new algorithms become available. Working with true copies reduces the
number of specimens required and makes combining of multiple
analyses more accurate. It also means one can compare results of

different processing steps with their exact counterparts.

New Capabilities

Another criterion for critiquing an approach is how well it

compares in capabilities to other similar techniques.

"Re-sectiom'n " of emb as
g 17

The technique of "re-sectioning" the same specimen in different planes
is a powerful technique. The beneﬁts of being able to look at images of
structures cut through different planes have been utilized by people
using confocal microscopy to study small semi-transparent specimens,
as well as by people using imaging techniques such as CAT scans and
MRI for relatively large specimens (Baba, 1991).

Although digital re-sectioning provides additional spatial
information, there is a loss in spatial resolution. As with the other

imaging technologies (such as confocal microscopy, CAT scans, and

117

MRI), resolution in the z-axis, the long axis perpendicular to the plane
of scanning (x, y-axes) is the limiting factor (Pawley, 1990;
Radermacher, 1991). When re-sectioning the embryos in planes other
than parallel to the "real" section plane, the resolution is not as good
along the z-axis (compare Figure 12 with Figure 20 a).

Being able to put the slices "back-together" and to see the whole
specimen at once greatly aids in the interpretation and thus acquisition
of information from the 2D images. For example, if the specimen to be
examined is not able to be precisely oriented when either mechanically
sectioned or "optically" sectioned, the resulting 2D images can be
difﬁcult to interpret.

In the present study, by being able to look at slices of the
specimen, and tell where in the whole specimen they are from, allowed
for the discovery and exact placement of the small ingression of cells in
the stylommatophoran snail Anguispira altemata. and the
determination that they were not the earliest indication of the point of
invagination of the eye-vesicle. From the 2D section images of the
embryo, while showing an ingression of cells from the ectoderm, it was
not possible to deﬁnitively determine that they were not the beginning of
the development of an eye. By being able to precisely determine their
exact location within the cephalic plate, and compare the
reconstruction with later reconstructions showing the precise location

of the invagination of the eye-vesicles, it becomes evident that this

118

small ingression of cells could not be the anlage of the developing eye
(Figures 8 & 10).

In addition, while the 2D sections were needed to help identify
the several small groups of cells close to the above-mentioned
ingression of cells as cells resembling ( in staining characteristics and
morphology) the cells forming ganglia, it was not evident that these
ganglion-like cells formed a branch from the developing cerebral
ganglion until they were able to be viewed as a whole within the 3D
reconstructed embryo (Figures 8 & 9).

What the cells from this small point of ingression within the
cephalic plate become was not determined in this study. It is possible
that this small dorso-medial ingression corresponds to one of the
"cephalic tubes" referred to by Henchman (1890) and Raven (1966).
Henchman’s descriptions and drawings based on 2D sections of a large
invagination leading to the formation of a tube, as well as her
description of the position and types of cells involved correspond to the
large ventro-lateral invaginations found in the present study (Figures 8,
10, 11). Reports of a second invagination or cephalic tube occurring in
some gastropods (Henchman, 1890; Raven, 1966; Moffett, pers. comm.)
do not state from where within the cephalic plates they arise, or give a
detailed enough description to determine if the small dorso-medial

ingression of cells discovered in the present study corresponds to them.

119

Satisfactory identiﬁcation of these cells requires a ﬁner grained (more

embryos of intermediate stages) analysis than provided here.

Digital "dissection " of embryos
The snail embryos studied in this investigation are far too small

to be dissected by hand in any useful manner. The ability to dissect
digitally the virtual embryos proved to be a beneﬁcial method for
determining the shape, position and trajectory of branches (including
the connectives and commissure) of the cerebral ganglion. Moreover,
unlike dissection of the real specimen, digital dissection (since working
on an exact copy) does not destroy the rest of the embryo. The
structures dissected "out" can be compared directly to the rest of the
unaltered embryo as well as to the same structures from younger and
older stage embryos [Figures 6 & 13].

In the present study, there were two ways the structures of
interest were differentiated from the non-nervous tissues of the rest of
the virtual embryo depending on whether or not the result was to be a
single volume with all structures, or several separate volumes each
containing a different "dissected" structure. (See Material & Methods
chapter for discussion of differences of working with single versus

multiple volumes.)

A) Independent volumes of dig; ﬂy "dissected" structures

In the ﬁrst method, the resulting images of the "dissected" structures

were saved as a new set of images (Figure 21). The registration or

120

alignment relative to the rest of the embryo of the digitally separated
structures was not affected.

This method maintains the full visual resolution of the structures of
interest (as well as the spatial resolution), and thus functions, such as
Fourier Transformations or other ﬁltering functions, are able to analyze
ﬁner. details in the reconstructed volumes. Digitally dissected
structures, which were made into separate volumes in this manner,
were visualized within the unaltered original data set (in this case, the
whole embryo) using the function "Multi-channel View" in VoxelView.
This function allowed the simultaneous viewing of several volumes
"merged" together in the same 3D space (Figure 9), with the ability to

assign different ranges of colors and opacity to each volume.

B) ygxel-vglue pagiﬁgning
The second method was performed after the images of the

sectioned embryo were reconstructed. If done prior to interpolation, the
color ranges get partially mixed together during that process. If there
are relatively few images (and few interpolations), and one does not
want to sub-divide the volume into several separate volumes (e.g., so as
to use functions such as VoxelSlicer, etc.) the nervous system, eyes,
etc., can be assigned to a given sub-range of voxel values. Since the
workstation used in this study had only 64 megabytes of RAM memory

available, the original eight data bits per pixel were sub-divided. With

121

more memory, the volumes could be converted to 9-16 data bits per
pixels thereby maintaining full visual resolution.

Once the structures of interest were assigned a unique range (or
single) voxel-value, the software could then be used to differentiate the
structures by assigning a different "color lookup table" (LUT) to that
range of voxel-values (i.e., staining "digitally" the different structures,
see below) or by maldng the other structures more transparent (lower
opacity) allowing for the visualization of the internal developing nervous
system (Figures 7, 8, 1 1, 13, 14, 17). In this same manner, the rest of
the embryo could have its full range of voxel values made completely
transparent or "turned off" (i.e., not rendered) (Figure 6). Neither of
these techniques had any effect on the spatial alignment or registration
of the voxels being manipulated.

Each separate volume, along with the combined one, can be viewed
at the same time. Thus the cerebral ganglia were able to be viewed
independently from multiple angles, while simultaneously viewing them
in relation to other structures within the whole embryo (Figures 9 8L
1 l). The trajectory of the two branches (besides the connectives and
commissure) of the cerebral ganglia of A. altemata were able to be
determined and veriﬁed independently of the structures to which they

were directed.

122

3) Digital "staining" 9f structures

If the physical staining of the specimen differentiates even the
slightest amount, (or if there are differences in the thickness and optical
opacity in the sections when imaged and digitized), the computer can
enhance that difference and make it more visible and quantiﬁable (see
Figure 14). The intestine from the stomach to the anus on the right side
of the image is visible due solely to the developing gut being composed
of a we; concentration of cells.

For structures that were previously dissected digitally and made
into separate data sets or volumes, the application of separate LUTs can
be used to add color to increase contrast, thus making the different
structures recognizable when all the structures are visualized together.
The application of different colors can be applied to structures within a
single volume using the same "Grease Pencil" functions as described
above. This was used to make readily distinguishable developing

nervous system, prototroch, and eye-vesicles within the whole embryo.

Accuracy of w’sualiza dons

As stated above, the changing of voxel values for structures of
interest only affects visual resolution, not spatial resolution. Whether
changing the histogram of the structures of interest after separating
them from the rest of the embryo. or while still "inside", the effect on

their histogram and visualization can be seen in Figures 22-25 which

123

show the assignment of the developing ganglia to the histogram range
64- l 27.

Volume rendering allows one to work with "fuzzy" uncertain
boundaries among developing structures, in contrast to surface
rendering techniques, which require a prior! determination of distinct
hard edges of structures before any visualization of structures in 3D.
Enhancing the visualization of structures can be done using the Grease
Pencil function of VoxelMath 2. 1 to shift the range of voxel values while
maintaining full visual and spatial resolution. By "selective application"
of mathematical functions using Grease Pencil, the area of interest can
be shifted to its own range of voxel values (e.g., 256-512 for 9—16 bit
images) maintaining its full visual and spatial resolution. The
application of pseudo-color (via the use of LUTs) can then be applied to
highlight the area of interest visualized in 3D within the whole
specimen. This does not create hard boundaries of the structures of
interest, but allows one to selectively add color or contrast to a limited
region within the specimen and visualize in 3D within the whole
specimen helping to visually differentiate structures. Other functions
such as digital ﬁltering of data in 3D, can also be selectively applied
using Grease Pencil, as well as applied to the whole volume to help
differentiate structures using information from all three dimensions.

The accuracy of digital dissections or digital staining is based on

the ability to distinguish the structures to be manipulated. Unlike

124

surface rendered reconstructions (which are only of the outlines of
structures), more of the data is represented in the volume rendered
reconstruction. Adjustments can be made after all the separate volumes
are "merged" back together if it is determined after further study that

the digitally dissected structures are not what they were believed to be.

Quantitative Measurements

It is important to be able to measure structures and be able to
follow quantitatively their development or changes over time.
Quantitative measurements are needed to compare and determine
individual differences within and between species, thus allowing for the
creation of accurate "normal" tables of development. The determination
of intra-speciﬁc as well as inter-speciﬁc differences will help determine
changes in developmental processes of evolutionary signiﬁcance.

The ability to easily and accurately perform quantitative
measurements of microscopy data is an important capability. Recent
discussions on the Confocal Listserv
(CONFOCAL@LISTSEFWACSU.BUFFALO.EDU, 1999) attest to the amount of
interest (and confusion) on this subject. The problem is in trying to
make accurate measurements of distances among structures or to
determine their size (volume) from two-dimensional images of structures
that are three-dimensional. The use of three-dimensional reconstruction
by volume-rendering techniques allows for making easy, accurate, and

repeatable quantitative measurements.

125

Measuring structures from their images (image analysis) requires
distinguishing the objects to be measured from their background and
other structures. Algorithms for "mathematical morphology" (Gauthier,
et al., 1992) techniques including functions such as "dilation,"
"erosion," etc. are all available within VoxelMath 2.1 with the added
feature that unlike image analysis of single 2D images, volume
rendering can apply these algorithms to the whole volume, utilizing the
3D information as well. And unlike techniques, such as Stereologr
(Weibel, 1979) which are labor intensive, and more probabilistic than

voxel counting (Rigaut, et. al., 1992), morphometrics has become "point

and click."
A) Measuring dis. ﬂees beigeen Sguegges: the TRACE meg'en

Accurate measurement of distance cannot be made from a single
two-dimensional image ("x-y axes") (e.g., the computer screen or print)
of structures that are also separated by some distance in the third
dimension ("z-axis"). Modern confocal microscopes are able to produce
sets of 2D images at progressively deeper planes along the z-axis (2-
series) and computers can display these sets of images as if "stacked"
on top of each other. The opacity levels can be varied so as to render an
image very accurately as to the true structure. This gives very useful
"qualitative" data regarding the relative position of structures in the 3D.
The standard micron bar applied to the image is only useful for

structures that reside exactly in the same plane as that to which the

126

line of measurement is applied. By converting 2D pixels into 3D voxels
(Figure 30), the computer converts information of spatial position into a
format that is now quantiﬁable. Since the voxels have a positional value
assigned along a z-axis as well as the x and y-axes, each voxel is
locatable within a 3D Cartesian coordinate system. The change from 2D
to 3D increases the size of the data set. Massive increase in modern
computer power (memory, storage, and processor speed) allows
manipulation of this increase in size of the 3D data set.

To explore this capability, the lateral distance between the
developing cerebral ganglion and the buccal mass was measured. Since
the shape or surface of these developing structures were not smooth, it
was necessary to be able to take multiple measurements between them
from different regions and compare the results. The "Trace" function
allowed the ﬁve separate measurements to be made, and calculated the
results statistically (Figure 16). Thus allowing for a more conﬁdent

assessment of the "real world" distance between the structures.

B) Measuring the volmne gf structures: the SEED function

Measurements of volumes of 3D structures from 2D images is a
complex task. The accuracy of the 2D images, the number of
interpolations (if used), and the ability to distinguish structures from
one another and their background all can affect the accuracy of the
resulting values. One of the great advantages of using volume rendering

methods instead of methods such as Stereology (Weibel, 1979) is the

127

fact that volume rendering does not require the use of hard boundaries
(edges) to be pre-determined by the researcher.

The "SEED" function of VoxelView/E allows the researcher to
select the range of voxel values, gradient values (or even hexadecimal
values) combined with different levels of sensitivity to their "physical"
connection to the structure to be measured, as criteria. The program
will then display the selected voxels measured within the whole volume
thus making it possible see where the boundaries were determined and
if they agree with other evidence (i.e., by looking at the volume in 2D
sections or resections, and comparisons to other staining techniques
and specimens).

The program VoxelMath 2.1 can be used to make direct
quantitative (by values and position) comparisons of multiple "seeded"
volumes. Thus the effect of changes in criteria on volume
measurements can be directly compared and contrasted. To explore this
capability, the volume of the developing nervous system was measured
for one of the Helisoma anceps embryo (Figure 17; Table 4). The
measurement of the developing ganglia included the developing ganglia

besides the cerebral ganglion on the right side (Figure 17a), thus

accounting for the much larger volume of 144305.00 um3 measurement
(Table 4). The values for the right and left eye-vesicles are much closer

in value (14154.00 m3, and 14342.00 um3, respectively). All these

volume measures correspond well with estimates bases on the total size

128

of the embryo studied, and measurements correspond to what would be
expected from looking at the 3D reconstructions and 2D section images.
For measurements of both distance and volume, the program
automates the calculations, and allows multiple measurements to be
taken, performs the statistical analysis "on the ﬂy" and produces the
calculations in several formats and tables. These calculations can be
imported into statistical analysis programs or spread sheets. Multiple
measurements can thus quickly, reliably, and repeatedly be

determined.

Accuracy of reconstructions

The following is a discussion of several issues regarding spatial
resolution and accuracy, both quantitatively as well as qualitatively, of

3D reconstruction.

Leif-ﬁght axis in reconstruction

In reconstructing snail embryos, maintaining the correct left-right
axis in the 3D reconstructions is important because of the following:
First, though basically bilateral, snails are asymmetrical in their left-
right axis. Second, the species in this study differ in the directions their
bodies coil; A. altemata is “dextral” (coils to the right), and H. anceps is
“sinistral” (coils to the left). Dextral snails have the opening to their
respiratory, reproductive, and excretory systems on the right side of

their bodies, while sinistral snails have these structures and openings

129

on the left side of their bodies. Thus, the left-right axes are reversed
between dextral and sinistral snails. Third, there are also asymmetrical
differences in the timing of development of both left/ right paired
structures such as the nervous system, and unpaired organs such as
the reproductive organs. Finally, some snails go through “torsion”
during development, (a process in which their bodies twist 180 degrees
along their dorso-ventral axis. In the current study torsion was not an
issue because all the structures of interest are located in the

anterior/ dorsal part of the embryo unaffected by the twisting of the

more ventral/ posterior part of the embryo.)

Maintaining the Qorrect Left-R_1g’ ht Axis:

Because of the way the VoxelView/E program reconstructs the
volume, there is potential for reversal of the z-axis of the volume that
would reverse whichever axis of the embryo is represented along this
dimension. VoxelView/ E uses the “compositing method” (or object
order) of volume reconstruction. The program reconstructs (or builds)
the volume from back-to-front by laying down whichever image has the
ﬁle name “1” (or "0") then stacking the ﬁle named “2” on top of it, and so
on, stacking the highest numbered (named) ﬁle on top (Figure 26 A). If
this is not the correct order for the images to be stacked, the resulting
embryo will have its left-right axis reversed. For example, the resulting

reconstruction will appear to be of the right side of the embryo (Figure

130

27 d), as opposed to being reconstructed as the correct left side of the
embryo (Figure 27 c).

Correction is accomplished by renaming the ﬁles in reverse order,
i.e., if there are ﬁfty images in the set, image number “1” becomes image
number “50”, and image number “50” becomes image number 1, etc.,
as shown in Figure 28 c. Renaming of ﬁles was carried out with the
program VoxelMath 2.1 by "ﬂipping" the images along their z-axis ("Flip-
2"), thus functionally reversing the order of the images. This was done
just before the volume interpolation ("Resample Z") step to decrease
processing time. Contrary to what is stated in the manual of
VoxelView/E for the “Position Volume” function, the “Flip Z-axis”
function in VoxelMath actually does reset the “origin” for the volume,
(i.e., the reference point 0,0,0 (x, y, z)), and thus changes the view of the
specimen in relation to the axes lines displayed when using the

“Annotations" option in VoxelView/E (Figures 6, 7, 8).

Alignment of images
Since one of the objectives of this study was to determine the

precise, three-dimensional location of the developing cerebral ganglia in
relation to where the eyes develop, it was important that structures
other than these be used to create or check the alignment of sections.
Points of reference in the images used for alignment were structures
other than the developing nervous system, e. g., the buccal mass,

radular sac, esophagus, shell ﬁeld, foot anlage etc. This decreases the

131

effects of expectations about the relative positions of the structures of
study (in this case the nervous system) on the alignment process. Since
the embryos were mechanically sectioned, alignment is still relative, not

absolute.

Digital Align—ment g Regi'stretign of the Image
The small size of the embryos studied not only limited the ability

to orient the embryos very accurately in parafﬁn (as mentioned
previously), it also interfered with embedding markers of some sort to
act as ﬁduciary marks (points visible in all sections that when aligned
correctly form a line(s) perpendicular to the plane of sectioning, and
allow accurate and objective alignment of the sections). Therefore, the
sections were aligned by hand on the LSM (as described in the Materials
and Methods chapter). Although there are commercial software
programs for alignment of images, these were unavailable for this study.
During the course of this study, it was discovered that two
features within the software programs used for the volume
reconstructions could be combined to create very precise, and more
accurate digital re-alignment of the images of sections. If the
interpolation value for a given volume is set to a negative integer in the
"_dimensions" ﬁle, VoxelView/ E and VoxelMath will display the volume
with empty space between the slice images (Figure 29). Using this
feature coupled with VoxelMath’s ability to view 2D slices of the volume

in the XZ and YZ views as well as the XY view, allows for the

132

determination of which slice is out of alignment with others, and byihow
many voxels ( Figure 29). VoxelMath 2. 1 can then be used to shift or
rotate the problem slice(s) the exact number of voxels needed to improve
its alignment with other sections, as compared with whole mounts,
video images, or SEM images of other embryos at the same stage of

development.

Pixels to Voxels

The computer program creates in essence, a three-dimensional
“voxel” from the two-dimensional pixels, and stacks them one on top of
each other to form a three-dimensional virtual model of the specimen,
called a “volume.” The voxels created are iso-dimensional in
appearance, that is the program uses the x and y pixel dimensions as
the z-axis dimension for calculation of the voxels, creating a three-
dirnensional volume (Figure 30).

Two issues aifecting the accuracy of the reconstructions are
speciﬁc to the creation of 3D voxels from 2D pixels: 1) changes in aspect
ratio of pixels; and 2) interpolation of voxels along the z-axis.

1) Changes in aspeet rang ef pixels

The ﬁrst issue arises from the fact that the digital images were
created on an LSM that creates rectangular pixels. The resulting pixels
need to be converted to square pixels for the 3D reconstruction software

to create proportionally correct 3D renderings. The LSM creates images

133

with 512x512 (x, y) pixels. The rectangular pixels can be converted to
square pixels maintaining the correct proportions of the specimen by
resampling the images either along the y-axis (512x341) or the x-axis
(768x512). While it can be argued it is better to resample along the x-
axis (thus not reducing the resolution of the image in either axis), doing
so requires much more computer resources and disk storage space.
Since the LSM scans the images along the x-ards, it can be argued that
the resolution in the y—axis is not as accurate and so the y—axis should
be the axis resarnpled. (See Pawley, 1990 and Hader, 1991, for
discussions of digital imaging in biology.) Resampling along the y-axis
creates images of 512x341. This is the default sampling method of
programs such as VoxelScanZeiss, and was used for the present study.

2)Ine latio fvoxls n th z-axi

In this study, the embryos were mechanically sectioned at ﬁve or
ten microns. Since all sections were digitized in sequence, the “section
thickness” of ﬁve or ten microns is also the “Z-Interval” (Figure 26). This
is the “real world" z-axis dimension/ resolution. Since the x and y-axes
dimensions are smaller than the z-axis dimension (Table 1) and the
program computes the z-axis dimension of each voxel equal to the x and
y-axes, when the program stacks the images (a single layer of voxels),
the resulting volume will look distorted (“squashed”) in the z dimension

(Figure 20 b) .

134

To correct for this squashing effect, images created by
mathematical interpolation are placed in between the original or “real"
images. The insertion of interpolated images between the real images
expands the volume in the z—axis dimension creating a more accurate
appearance of the specimen (Figure 20c). The computer program uses
linear interpolation of the real “slice” (2D) images to determine the voxel
values for the slice images to be inserted between them.

For example, some of the embryos used in this study were imaged
with a 20x objective with a “zoom” (or magniﬁcation by the LSM) of 20
(for a total on-screen magniﬁcation of 400). This creates a ﬁeld of view
(in x-axis) of 709.7um, scanned with 512 pixels making each pixel
represent 1.386um in the x-axis dimension (see Table 2). Since the
sections were mechanically cut at loum thick, (lOum divided by
1.386um=) 7.215 voxels would be needed to ﬁll (expand) the distance
from one image to the next. Without interpolations, the computer
“squeezes” all the information in the 10 pm thick section into 1.386um
and displays the volume as ﬂattened (Figure 20 b). Flattening of the
volume only occurs visually, since the computer will still measure the
distance as 10pm if that value was entered in the "dimensions ﬁle".

If the number of voxels needed to ﬁll in the “z-interval” is a whole
number (e. g., “three”), that number minus one (since the original image
thickness is included in the distance between the top of one image and

the top of the next), can be entered into the “interpolations” ﬁeld of the

135

"_dimensions" ﬁle for the volume. In this example the value “2” would
be entered (three voxels to ﬁll the space, minus one voxel of the original
image equals two interpolated voxels needed).

In the previous example which requires 7 .215 voxels (of 1.386 um)
the program VoxelMath 2.1 is used to create interpolations using non-
integer resample values ( Scalar value of 7.215) using the “Resample Z"
function in the Merge Window. Since it is impossible to have a partial
voxel (or pixel), the program calculates the number needed and resets
the distance value of each voxel so the distance is displayed as well as
measured with a "whole number" of voxels. If the "dimensions ﬁle" (or
image header) has the correct resolution values for the volume as
created, the function “Resample XYZ” will make all axis resolutions

equal.

Reduction of interpolations: Sub-sectioning t/Je sections
Interpolation of voxels, while improving the visualization of the

samples, does not add any real data, thus it is better to keep the
number of interpolations needed as low as possible to get more accurate
quantitative (and qualitative) information. If the z-axis or distance
between images is larger than the x-y axis resolution, then interpolated
voxels need to be created. Since mechanically sectioned embryos
stained with non-ﬂuorescent dyes were used in the present study, the
use of higher magniﬁcation objectives, could only increase the pixel

resolution (i.e., the smaller the area represented by a single pixel) in the

136

x and y—axes; the resolution in the z-axis of the created voxel stays the
same and thus _m_gge interpolations would be needed.

The 20X and 25X objects were used to create higher
magniﬁcation and resolution images of the developing eye-tentacle
region of the embryos. As shown in Table 2, the images created require
approximately 7—9 interpolations. This many interpolations create
volumes that have large “steps” between sections, degrading the quality
of the reconstructions. Both the 20X and 25X objectives have “depth of
focus” that are smaller (see Table 1) than the thickness of each section.
Two images of each section were created: one at the top of the section,
and one focused 5 microns down or into the sample. This was possible
due to the LSM having a mechanized scanning stage (i.e., a computer
controlled stage that can be moved in the z-axis in nano meter range
step-sizes). The level of accuracy, precision, and consistency of optically
sub-sectioning the mechanically sectioned images would not have been
possible using a standard microscope and digital camera.

Optically sub-sectioning created two images slightly different from
each other due to the fact that most of the cells of the sample are less
than 5 microns in diameter, so some cells appear in one of the images
in focus, and in the other image many are not visible at all. The cells
within the 10pm thick sections were spread back out more accurately.
This allowed for a reduction by half the number of interpolated images

per “real” image (although the total per section was the same) required

137

(see Table 2) thus creating a little more accurate reconstructions. With
the use of ﬂuorescent staining, it should be possible to use confocal
microscopy to "optically" resection the sections and if the z-axis step
size and optical depth can be made to match the x, y-axes resolution,

then no interpolations would be necessary.

BIOLOGICAL ISSUES

Determination and delimitation of cells and tissue types

Only a few species of pulmonate gastropods have ever been
studied with respect to development of their cerebral ganglia. Of
stylommatophorans, one of the most, if not the most, detailed study
was done on the slug Limax maximus by Henchman (1890). She
determined that the cerebral ganglia developed from ectodermal
thickenings of the cephalic plates, with a single large invagination
forming the "cephalic tubes." Whether there is one large invagination for
the cerebral tubes per side or two has been reported differently,
depending on which species were studied (See Henchman, 1890, and
Raven 1966, for reviews). Reports of differences (and disagreements) in
development demonstrates the problem with assigning "model" status
to a species to represent a large (and largely unstudied) group of
organisms.

Independent of their number, the cerebral tubes being such

obvious structures helped to determine which groups of cells were

138

forming the cerebral ganglia. Large proliferation of cells, which occurs
at the innermost ends of the cerebral tubes, end up next to (in contact
with) the groups of cells forming the cerebral ganglia.

The developing nervous system in these snail embryos is
composed of only a few cells at the time of eye development. The cells
that give rise to the eye-vesicles in the species studied do not show cyto-
differentiation until after they have separated from the overlying
ectoderm and formed a true vesicle.

The cells forming the ganglia did show some cellular differences at this
early stage of development using standard hematoxylin stains. When
stained with Heidenhain’s hematoxylin the cells forming the cerebral
ganglia showed nuclei that were larger, more transparent, and rounder
than the surrounding ectodermal cells or mesodermal cells (Figure 19).
The position of the cells forming the cerebral ganglia, (relative to other
developing structures such as the buccal mass, stomodeum, cephalic
plates) was also used in their identiﬁcation.

By looking at a series of embryos selected at different stages of
development, it is possible to extrapolate over time what a given group
of cells will become. The ability to determine the precise location in all
three-dimensions for a given group of cells frozen in time and to make
direct side by side comparisons with cells in the same location in other
embryos at later or earlier stages of development greatly facilitates the

determination and discovery of which cells develop into what

139

structures. The ability to study the developing embryo in all three
dimensions simultaneously, with the accuracy and precision that
volume rendering creates, is a great advancement for embryological

research.

Gastropod phylogeny

If we knew the relationship(s) of the Basommatophora to the
Stylommatophora, what the differences in development of the eye-
tentacle complex means in regards to the evolution of these two groups
would be more apparent. Pulrnonates are considered a grade of
structure, with its three orders (discussed below) reaching it
independently (Solem, 1985). Data from comparative morphology,
biochemistry, genetic structure, and biogeographic studies have not
been able to determine with any level of conﬁdence the relationships

among pulmonate orders (Solem, 1985).

Structure of the eye-tentacle complex in gastropods

Gastropod snails and slugs have a wide range of patterns of their
eye-tentacle complex. Prosobranch gastropods have a single pair of
cephalic tentacles, with eyes located on short eye-stalks situated at the
base of the cephalic tentacles. Alternatively, the eyes are located in
small bulges in the base of the tentacles, or embedded in the skin

around the base of the tentacles5.

140

Opisthobranch gastropods are the most diverse of all groups, a
fact that is also reﬂected in their sensory systems. Many
opisthobranchs have large cephalic sensory structures called
“rhinophores” which vary in structure from folded to linear, sheathed or
unsheathed in gross structure (Schmekel, 1985). Some opisthobranchs
also have cephalic tentacles, which are more or less outgrowths of the
labial palps or lips.

Pulrnonates are the most successful in freshwater and terrestrial
habitats. This subclass in composed of three orders all named by the
structure of their eye-tentacle complex; the Basommatophora,
Stylommatophora, and the very peculiar slug group
Systellommatophora. Aquatic snails of the order Basommatophora have
a single pair of cephalic tentacles, with eyes located at the base of the
tentacles. The eyes are either fused to the base (as in the
basommatophoran snails Helisoma), or embedded in the epithelium
between the tentacles as in Lymnaea.

The largest, and most successful group of terrestrial gastropods.
the Stylommatophora, have two pairs of cephalic tentacles, with eyes
located at the summit of the larger and posterior set. The tentacles of
stylommatophoran gastropods are completely retractable by
invagination, and can be manipulated with very precise control through

a large range of movement.

141

The last group, the Systellommatophora, is a small group of
terrestrial slugs that have, like the Stylommatophora, two sets of
tentacles with their eyes located at the tip of the larger and more
posterior set of tentacles. Where these two groups differ is in the fact
that the tentacles of the Systellommatophora are only contractible, not
retractable by invagination. They also have a uniform cerebral ganglion
structure, different enough for van M01 (1967) to keep them in their own

group (Solem, 1985).

Sensory functions of gastropod tentacles

While aquatic snails do appear to have chemo-sensory and tactile
functions of their tentacles (Boudko, et al., 1999; Chase and Tolloczko,
1985; Chia and Koss, 1982, 1984; Audesirk and Emery, 1978;
Audesirk, 197 5) most still have an organ for chemo-reception called an
“osphradium” associated with the ctenidium or gills contained in the
mantle cavity. Terrestrial snails and slugs have lost this structure, and
their tentacles (and lips) have taken over as the main site for chemo-
reception (Bullock and Horridge, 1965;Haszprunar, 1988).

At the tip of each tentacle of stylommatophoran gastropods is a
highly innervated chemo-sensory pad, supported by a complex and
large tentacular ganglion. The tentacular ganglia have been estimated to
contain 100,000 neurons each in the terrestrial snail Achatinafulica

(Chase and Tolloczko, 1993). This ganglion has a connection to the

142

cerebral ganglion that is independent of the innervation of the eye; the
optic nerve runs directly to the cephalic ganglion.

It is not known what, if any, the adaptive signiﬁcance of these two
very different patterns of the main sensory structures associated with
life in different environments may be. How this restructuring of such
important organ systems occurred evolutionarily is not known. The
adaptive value of some of the structural changes of the tentacles in
terrestrial gastropods are more easily explainable, given the lack of
surrounding water to help support the tentacles, as well as the
osphradium not being functional in a terrestrial environment. (Although
loss of the osphradium has also occurred in several orders of
opisthobranchs, such as the Nudibranchia, Pleurobranchomorpha, and
some of the Saccoglossa (Schmekel, 1985)). The evolution of a second
pair of cephalic tentacles, a large complex tentacular ganglia, and the

change in eye position are modiﬁcations which are harder to explain.

Homology of sensory structures

What, if any, are the homologous relationships between these
sensory systems? Are the cephalic tentacles of aquatic snails
homologous with the anterior or posterior pair of tentacles of terrestrial
snails? Or are the posterior (optic) tentacles of terrestrial snails

homologous with the "eye-stalks" of some prosobranch groups? Is the

143

anterior pair of tentacles of terrestrial snails derived from the labial
palps found on some aquatic snails?

While the question of homology of gastropod sensory systems,
such as the rhinophores, tentacles, and labial palps or lips, seems
straightforward, and has been treated as such in the literature, when
the actual structures and functions are compared among the great
diversity of gastropod groups, it becomes clear it is not such a simple
question.

The epithelium of gastropods is only one cell layer think, and
there are several cell types which require access to the surface, such as
the epidermal, ciliated, glandular, and neural sensory cells (Boudko, et
al., 1999; Hyman, 1967; Simkiss, 1985). There are two ways gastropods
have overcome this space limitation; one, create a second population of
sensory nerve cells with their cell bodies located below the basal lamina
of the epidermis with dendrites extended to the surface in-between the
other cells (Boudko, et al., 1999); and two, increase the surface area
that is innervated by growing out the epidermis into tentacles and
rhinophores.

The rhinophores are believed by some authors to have evolved
from a sensory organ called “Hancock’s Organ” and the dorsal lateral
edge of the cephalic shield of an Cephalaspidean opisthobranch. The
rhinophores of opisthobranchs and the cephalic tentacles of

prosobranchs (and thus also pulmonates) cannot be, therefore,

144

homologous (Schmekel, 1985). The Cephalaspidea do not have any
other cephalic tentacles, suggesting the tentacles of other
opisthobranchs may not be homologous with those of prosobranchs or
pulmonates either. This is quite interesting since the opisthobranchs
are believed to have arisen from prosobranchs with tentacles. The
diversity of pattern of the eye-tentacle complex attests to the
evolutionarily ﬂexible gastropod body plan, (another reason for the great
confusion of phylogenetic relationships).

More representative studies of species within the numerous
diverse orders and families of gastropods in regards to the development
of their sensory systems, including shared gene-complexes and gene
expression patterns, as well as more phylogenetic studies to determine
the relationships and thus direction of morphological changes are
needed. Until then, the questions as to which sensory structures are
really homologous and how they are modiﬁed developmentally and

evolutionarily can not be answered.

Development

The present study set out to determine: the precise location
within the cephalic plates of eye-vesicle formation; and if the cerebral
ganglia move into a position to be able to interact with the overlying
ectoderm in the exact area required for their induction; and to

determine if there is a difference in the position of the cerebral ganglia

145

relative to the overlying ectoderm that may account for the changes in
the position of the eyes relative to the tentacles between these two
groups.

In both species studied in this investigation the eyes were found
to develop by invagination of the ectoderm in the dorso-lateral region of
the cephalic plates; posterior and slightly dorsal to the tentacle anlagen.
The eye vesicles do not develop on the tentacle anlagen and thus the
location of the eyes at the tip of the tentacles cannot be explained by
the eye-vesicle anlage being on the tentacle anlage and thus being
carried up as the tentacle elongates.

The above ﬁnding suggests that in at least in this
stylommatophoran snail the change in development accounting for the
change in position of the eyes relative to the tentacle anlagen occurs for
some other reason later in development. Since the tentacles of
stylommatophoran gastropods are hollow, as opposed to the solid
tentacles of basommatophoran snails, the eye vesicles, once they
separated from the overlying ectoderm, may just move (“migrate”?) in to
the “open space” created by the elongation of the developing tentacles.
In addition, the present study did not ﬁnd any connection of the eye
vesicles with the cerebral ganglia during the stages of development
studied. Thus, unlike the basommatophoran snail, which was found to
have nearly continuous connection between the eye vesicles and

cerebral ganglia, the eye vesicles of the stylommatophoran snail are

146

“free” to move in to, and along with, the developing tentacles. How this
movement occurs, whether pushed by the development of the optic
nerve, or possibly pulled by a subsequent connection with the
epithelium of the developing tentacle, will not be known until further
study.

As mentioned previously, a large indentation of the cephalic
plate, ventro-lateral (under) to the developing tentacles, was seen in
young veliger stage (stage B) embryos of both snail species studied.
These indentations, which appear as “tube-like” structures in older
embryos, correspond with what has been reported in the literature as
the “cerebral tubes.” Although only one large indentation/ tube was
found in the present study, the literature reports two cerebral tubes for
some basommatophorans. A possible second tube was not readily
obvious in Helisoma anceps, but was not speciﬁcally looked for either.
In the stylommatophoran snail Anguispira altemata, the present study
did ﬁnd a small ingression of cells, dorso-medially within the cephalic
plate. Descriptions in the literature are insufﬁcient to determine if this
small ingression of cells corresponds with what has been reported as a
second cephalic tube in some other species.

While this study did ﬁnd that in the basommatophoran snail H.
anceps the cerebral ganglia were in the required spatial-temporal
position to be able to induce the overlying ectoderm to form the eyes, we

must not assume a causal connection (post hoc ergo propter hoe)6 i.e.,

147

that induction does therefore occur during normal development in H.
anceps. To deﬁnitively determine if induction by the cephalic ganglia
does occur, contact with the cephalic plate by a cerebral ganglion must
be experimentally blocked to see if eyes still form. However, we cannot
rule out induction, either, from this study’s ﬁndings.

On the other hand, since in the stylommatophoran snail
Anguispira altemata the cerebral ganglia were not found anywhere near
the point of invagination of the eye-vesicles in the cephalic plates, either
before or during invagination of the eye-vesicles, induction can be
considered unlikely. There is the possibility that the cerebral ganglia
either did have contact with the cephalic plates earlier in development
and then moved away, or that they had contact, but for such a short
time that they were missed in this study. Both of these alternatives are
of course possibilities that this approach can easily examine by looking
at earlier stages and specimens collected closer together in a

developmental time sequence.

Experimental evidence of possible inductive interactions between the
cerebral ganglia and overlying ectoderm of the cephalic plates
Two other lines of research support or suggest that the cerebral

ganglia may have inductive interactions with the overlying ectoderm to
form the eyes.
1) The ﬁrst line of research includes studies on prosobranchs and

basommatophorans and include deletion of macromere and micromere

148

cells studies and experimental disruption of development by treatments
such as a) exposure to ions such as lithium, b) heat-shock, and c)
centrifugation. All these methods cause very characteristic
malformations of the developing embryos (Verdonk, 1965; Cather, et al.,
1976; Morrill, 1982). (See Raven, 1966, for review.)

What is so intriguing about these studies in regards to the
question of the possible role of induction by the cerebral ganglia in the
formation of the eyes, is that is all these studies report the same order
of disruption of development -- "eyes>tentacles>cerebral ganglia"
(Raven, 1966). Most of these studies looked at whole embryos, and thus
could not report on the presence of cerebral ganglia, and the eyes were
only reported if they had reached the point of pigmentation (Verdonk,
1965). The studies that did look at sectioned material, such as in
studies on Limnaea stagnalis, in the sections one clearly sees a cerebral
ganglion in contact with the eye (Raven, 1966). Even more telling is that
where there are "supernumerary" eyes there are also supemumerary
cerebral ganglia connected (Raven and Beenakkers, 1955). Where the
eyes develop close together, there the cerebral ganglia also are closer
together (shortened commissure) (Raven and Bennakkers, 1955; Raven,
1966). All these ﬁndings are in accord with the idea that the cerebral
ganglia may in fact perform an inductive role in development. Although
no comparable studies have been carried out on stylommatophoran

gastropods.

149

2) The second line of research involves studies of transplantation
of cerebral ganglia from one host to another in Pulrnonate snails. The
most interesting studies are those of Moffett and Austin (1981). They
found that after transplantation of a cerebral ganglion into an adult
host of the basommatophoran snail Melampus bidentatus (Say)
"supernumerary" eyes and tentacles developed. One eye develops on the
transplanted ganglion itself, and one at the base of the new extra
tentacle, located just above the implanted cerebral ganglion. Thus,
clearly in at least this species and in the adult, the cerebral ganglia are
capable of inducing (either directly or indirectly) the formation of eyes
and tentacles.

While there have been transplantation studies of the cerebral
ganglia in the stylommatophoran snail Helix aspersa (Gomot, et al.,
1990) and the stylommatophoran slug Limax maximus (Sokolove, et al.,
1983) neither reported any development of supernumerary eyes or
tentacles. Both studies were looking to determine if the cerebral ganglia
could survive over time, with the study on H. aspersa lasting up to a
year post implantation. The study on Limax maximus had the cerebral
ganglion transplanted in the foot region, and even the
basommatophoran snail Melampus bidentatus does not develop eyes or
tentacles when the cerebral ganglia are transplanted there (Moffett,

pers. comm.) )

150

These transplantation studies reﬂect what was found in the
present investigation -- that the cerebral ganglia may in fact play a role
by inducing the eyes and tentacles in the basommatophorans and that

this inductive role may have changed in stylommatophoran snails.

Evolutionary modiﬁcations are modiﬁcations of development

Developmental biologists have known for a long time that the
same mechanisms of development operate in all groups of animals.
More recently, biologists studying the molecular basis of development
have discovered the same developmental/ genetic control mechanisms
underlying some of these developmental mechanisms, for example,
homeobox genes, HOX, PAC6, etc. (Raff and Popodi, 1996). The fact that
developmental/ genetic control mechanisms are so highly conserved
across all groups of animals (and some plants) may explain long-term
evolutionary trends, as well as the speciﬁc types of morphological
changes that have occurred (Raff, et a1. 1990).

During the last couple of decades, biologists have begun re-
evaluating the role of development in evolutionary change (Atkinson,
1992). They have come to realize that not only does morphological
evolution require changes in the underlying developmental program,
but the mechanisms for evolutionary innovation are developmental and

genetic control mechanisms (Maderson, 1975; Raff and Kauffman,

151

1983; Buss, 1987; Raﬁ, et al. 1990,Gilbert, et al., 1996; Sommer and
Stemberg, 1996). Developmental mechanisms are beginning to be
thought of as not just "passive reﬂections" of changes in the underlying
genes. (See Gerd Mfrller’s Developmental Mechanisms at the Origin of
Morphological Novelty: A Side-Effect Hypothesis ( 1990), for a discussion
of non-genetic control mechanisms of development.) But, seen as an
active participant not only in determining which types of morphological
changes can occur, but also as the source of evolutionary novelty --
creation of new body plans and new structures.

This view is a departure from previous thoughts on the
relationship of development to evolution, i.e., that only phyletic
information resides in the ontogeny of individuals (Gould, 1977). The
view that development plays an active role in evolution redeﬁnes
evolutionary changes in terms of how they occur, rules that govern
them, evidence of their occurrence, and changes the "level of study" of
them (Raﬁ, et al. 1990). In recent books, such as "Embryos, Genes, and
Evolution" (Raﬁ & Kaufman, 1983), "The Evolution of Individuality"
(Leo Buss, 1987), "Evolutionary Innovations" (ed. Matthew Nitecki,
1990), Raff‘s most recent book, "The Shape of Life" (1996), and "The
Origin and Evolution of Larval Forms" (Eds. Brian K. Hall and Marvalee
H. Wake, 1999), the authors all argue for an integrated approach in the
study of development and evolution. A more robust evolutionary and

developmental synthesis, superseding the “Modern Synthesis” of the

152

1950’s, is developing. Consequently, evolutionary questions are starting
to be studied in terms of the "active" role development plays in
specifying evolutionary modiﬁcation, as well as the evolution of novel
structures.

There have been two main lines of approach for studies seeking to
understand the role of development in shaping evolutionary change.
The ﬁrst approach (based on theoretical considerations) presupposes
that changes in morphologr in response to selection are not truly
random, but limited or channeled (directed?) by the nature of the
complex interaction of underlying genetic and developmental systems
(Raﬁ, et al. 1990). This is referred to as "developmental constraint"
(Waddington, 1941) on evolution and may account for some long-term
evolutionary trends (Raﬁ, et al. 1990). The idea is that underlying
developmental processes and patterns of gene expression may set
directional rules governing evolution.

In this approach, one looks for developmental/ genetic control
mechanisms that may be able to act as buﬁers to selection and in
which systems these act. C. H. Waddington came up with the concept of
"canalization" of developmental pathways. He showed that developing
systems "...show a certain degree of stability, in the sense that it is
quite diﬁlcult to persuade the developing system not to ﬁnish up by
producing its normal end result." (Waddington, 1966). Recent work on

the heat-shock protein (Hsp90) (Rutherford and Lindquist, 1998) may

153

have discovered the ﬁrst molecular mechanism for buﬁering of
development.

The second approach is a more general one looking to discover
changes in the developmental programs of the organisms being
compared. In this approach one compares diﬁerences in underlying
developmental programs of two or more groups of animals, and tries to
determine which developmental / genetic control mechanism can help
explain the speciﬁc type / direction of change that occurred. This
approach is based on the concept of “dissociability” of developmental

processes (Raff, et al. 1990), and was described by Needham, (1933) as:

Developmental processes which happen together or sequentially in tirne are
not necessarihl tightly coupled mechanisticalhl and may he shiﬂecl relative to each
other in evolution without disngbting development.

There have been two main processes suggested which are of
evolutionary signiﬁcance: heterochrony and induction. The ﬁrst
process, heterochrony, is relative changes in timing of developmental
process, usually in reference to somatic versus gonadal or reproductive
structures. In the book Embryos and Ancestors (1958) de Beer
"generalized" the term "to mean a change in developmental timing of a
feature relative to the equivalent feature in its ancestor" (Raﬁ, 1996).
Heterochrony has been generally regarded as one of, if not the most,
universal and important modes of how development evolves (de Beer,
1958; Gould, 1977; Buss, 1987). Recently though, researchers argue it

is heterochronic pattern that is common and not heterochronic

154

processes leading to evolutionary changes. "There has been surprisingly
little (essentially none!) in the way of critical tests of the basic
assumption that observed heterochronies are dissociations of processes
in time." (Raﬁ, 1996). After setting out to write a "review of cases in
which heterochronic processes had an important role in larval
evolution..." Hart and Wray,(1999) had came to this conclusion:
"Heterochrony has been a potent general organizing principle in the
history of evolutionary biology...However, it seems to us that
heterochrony as a process has been relatively unimportant in both the
evolution of larval forms and the revival of interest in them...."

The second process, induction, has been studied since the 1920s
for its important role in organizing the diﬁerentiation of diﬁerent regions
of the embryo (Raﬁ, 1996). Has Spemann and Hilde Mangold were the
ﬁrst to study the "organizer" in amphibian embryos, and determined
"that activation of the genetic material resulted from the epigenetic
interactions among inducing and induced tissues" (Raﬁ, 1996).
Induction may thus act as a constraint to changes in developmental
patterns or process. Changes in inductive interaction, such as
dissociation of previously required inductive events, may allow for
speciﬁc evolutionary modiﬁcation (Raﬁ and Kaufman, 1983).

Whether the dissociation of a previously required inductive
interaction has allowed for the major restructuring of the

stylommatophoran eye-tentacle complex will only be known after more

155

species from each of these pulmonate groups are studied as well as the
other species of prosobranch and Opisthobranch gastropods. How
representative the species looked at in this investigation are for their
respective groups is not known, and will not be known till more
comparative studies, on a more representative survey of members of
each group, are performed. In the course of the investigation a powerful
approach has been developed for making the kinds of comparisons,
qualitative and quantitative, that are needed to answer these types of

developmental and evolutionary questions.

156

CONCLUSION
In the stylommatophoran snail Anguispira altemata, this study
did gt ﬁnd the cerebral ganglia in close proximity to the cephalic plates
in the precise location of the eye vesicles either before or during
invagination.

In the basommatophoran snail Helisoma anceps, this study did

ﬂ

ﬁnd the cerebral ganglia to be in contact with the cephalic plates from
the very beginning of eye-vesicle invagination, and to maintain
connection even after the eye vesicles had separated from the overlying
ectoderm.

As for the question of whether there is a diﬁerence in the location
of the point Of invagination of the eyes that could account for the
diﬁerence in the pattern of the eye-tentacle complex in the adults, the
answer is no. The point of invagination of the eye vesicles was in the
same dorso-lateral position of the cephalic plates for both species.

The use of laser scanning microscopy combined with three-
dirnensional reconstruction using volume rendering techniques was
found be a powerful tool for this type of embryological study. It was
eﬁective and eﬁicient for determination of the precise position of the
cerebral ganglia and the trajectory of its "branches" within the embryo,
before, during, and after eye-vesicle formation. The techniques and

functions available by this approach also allowed the precise

157

measurement of distance between the developing cerebral ganglion and
buccal mass to be determined, and to compare the volumes of the
developing eye-vesicles, as well as the cerebral ganglia. The ability to
analyze the same embryo from multiple angles, re-section the same
embryo in diﬁerent planes, and "dissect out" digitally the structures of
interest, meant this study needed far fewer real embryos at each stage
of development.

The ability to rotate the whole reconstructed embryo, or "sub-

 

volumes" of it, and to enhance visualization by such techniques as
digital "dissection," digital "staining" , etc., allowed the unequivocal
determination that none of the "branches" of the developing cerebral
ganglia in the stylommatophoran A. altemata connect with or even
point to the place in the overlying cephalic plates where the eye-vesicles
form (Figure 8 & 9).

The evolution of scientiﬁc beliefs, practices and approaches is as
interesting a subject matter as that to which the philosophical approach
of the "scientiﬁc method" has been, and continues to be, applied. It can
be argued that the part of the scientiﬁc enterprise that has developed
the fastest, and brought with it a far more sophisticated understanding
of the world within and around us, is that of scientiﬁc instrumentation.

In 1890, Annie P. Henchman made the argument for the use of
the modern technique of serial sectioning to study the development of

gastropod embryos in her paper on "The Origin and development of the

158

Central Nervous System in Limax maximus." Referring to the confusion
and contradiction in the ﬁndings of other researchers studying the
development Of the nervous systems in gastropods, she writes:

Since the observations of the earlier writers, down to about 18 74, were carried
on without the aid of sections, their conclusions do not merit that degree of
conﬁdence which is to he accorded those who have availed themselves of this

means of stuajt.

In 1992, Louse Page, in trying to make sense of the conﬂicting
and confusing results of studies on the development of the nervous
system in Opisthobranch gastropods, takes this one step further
making the argument that serial thin sections are needed to study the
developing nervous system in gastropods.

All previous accounts of sequential neurodevelopment in nudihranchs have been
histological, yet this method cannot reveal with certainyr all neuronal ingression
sites from ectodermalproliﬁration phzcodes or trajectories of earbr connectives and

axon tracts.

In 1999, the present investigation, while utilizing just a few of the
abundant number of sophisticated tools that are made available only by
the technique of three-dimensional reconstruction by "volume
rendering," has demonstrated the power and richness of this approach

to overcome problems in the study of embryogenesis. A hundred years

159

from now, what will researchers say about what we consider "cutting
edge" and "sophisticated" research methods? It will be exciting for
future generations to see what new technologies were created. Even
more exciting will be what has been learned about the many varied and

wonderful organisms on this planet and how they develop.

160

APPENDICES

161

APPENDIX A

ASSORTED COMPUTER SYSTEM COMMANDS

File manipulation commands for the SGI (Unix) workstation

Case Sensitivity
The UNIX operating system is “case sensitive” which means the

system recognizes uppercase (or capital) characters and their lowercase
version as diﬁerent characters. For example, the letters “a” and “A” are
treated as diﬁerent characters. The DOS operating system is not case
sensitive, so the letters “a” and “A” are treated as the same character.
TAR (Tape Archive) commands

1) To copy ﬁles to the ISO-megabyte QIC tape drive in the SGI:

IRIS 1% tar cv {full or relative ﬁle or directory path name)
E . g, to copy the directory “embryo_volume_1” in the user directory
“murphy” the command would look like:
“IRIS 1% tar cv /usr/people/murphy/embryo_volume_l”

2) To retrieve ﬁles from the ISO-megabyte QIC tape drive in the SGI:

IRIS 2% tar xv {full ﬁle or directory path name}
E.g, . to retrieve the single ﬁle “image_ﬁ1e_1.tif” the command would look

like:

“IRIS 1% tar cv
u sr/ people /murphy/ embryo_volume_1 / image_ﬁle__l .tif'

162

3) To print out a list in the console window the complete contents of a

tape in the drive:
“IRIS 3% tar t”

4) To store all the ﬁles in a directory in a single TAR ﬁle located on the

harddrive:

IRIS tar cf {name of tar ﬁle to store all ﬁles in} {path name of
directory to store in tar ﬁle}

“IRIS 1% tar cf /usr/peop1e/murphy/embryo_1.tar

usr/ people / murphy/ embryo_ l ”

The resulting ﬁle (icon looks like a “safe”) can be transferred to the

DELL PC via FTP. If the .tar ﬁle is smaller than 100MB it can be stored

on a ZIPTM Disk.

5) To retrieve individual ﬁles or directories from the tar archive ﬁle:
IRIS tar cf {name of tar ﬁle) {path name of directory or ﬁle
retrieve}

“IRIS 22% tar va /usr/people/murphy/embryo_l.tar

usr/ people / murphy/ embryo_ 1 /image_ 1 .tif’

163

APPENDIX B

VOXELMATH AND VOXELVIEW INFO

l)V@meHﬂedrhhynogzdﬁphg

IRIS 22%

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

Welcome to VOxelMath:

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

/usr/peOple/murphy/Committee_Meeting_Stuff/all

Physical memory Status:

Free ........ 37.89 Mbytes
Available ... 67.54 Mbytes
Total ....... 72.00 Mbytes

Volume 1 dimensions: x[512] y[300] z[36] (8-bit) Voxels

Memory requested for THIS volume: 5.27 Mbytes
Memory requested for ALL volumes: 5.27 Mbytes
Memory needed, including buffers: 16.77 Mbytes
Memory needed, including program: 23.93 Mbytes
Maximum physical memory available: 65.79 Mbytes
Maximum virtual memory available: 77.80 Mbytes

164

2) Reading _a Tiff image ﬁle “header”

[This volume had the "color" button on the LSM on when the
images were collected. Thus the LSM_PC program saved the
color information in RGB/colormap in the image file.

(tiff version on our SGI IRIX 4.0.5C is something like
2.4. VoxelView and VOxelMath need version 6.0 Of tiff to
read.)]

IRIS 6% VoxelLoader -h ./ANG96/ASE901—F/ase901-F

II

II

/usr/people/murphy/disk2/ANG96/ASE901—F/ase901—F

000.tif 001.tif 002.tif
003.tif 004.tif 005.tif
006.tif 007.tif 008.tif
009.tif OlO.tif Oll.tif
012.tif 013.tif 014.tif
015.tif 016.tif 017.tif
018.tif Ol9.tif 020.tif
021.tif 022.tif 023.tif
024.tif 025.tif 026.tif
027.tif 028.tif 029.tif
030.tif 031.tif 032.tif
033.tif 034.tif O35.tif
036.tif 037.tif 038.tif
039.tif 040.tif 041.tif
O42.tif 043.tif O44.tif
O45.tif O46.tif 047.tif
O48.tif O49.tif

TIFF Directory at Offset 0x8

II

Subfile Type: (O = 0x0)

Image Width: 512 Image Length: 512

Bits/Sample: 8

Compression Scheme: none

Photometric Interpretation: palette color (RGB from

colormap)
Software: "ZIF 1.81 MAR-93"

165

Make: "Carl Zeiss, Oberkochen, Germany"
Model: "Laser Scan Microscope"
Samples/Pixel: l

Planar Configuration: single image plane
Color Map: (present)

Volume header:

magic: 43970
volume_form= l
size= 38797312
pathname=
/usr/people/murphy/disk2/ANG96/ASE901-F/ase901-F
data bits= 8
normal bits: 0
gradient bits: 0
bits/voxel: 8
orig. xsize= 512
orig. ysize= 512
orig. zsize= 50
xsize= 512
ysize= 512
zsize= 148
interpolations: 2
xoffsetz 0.00
yOffset= 0.00
zoffset= 0.00
voffset= 0.00
xscale= 1.3860
ysca1e= 0.9240
zscale= 5.0000
vscale= 1.0000
xunits=
yunits=
zunits=
vunits=
xlabel= X
ylabel= Y
zlabel= Z
v1abel= Voxel_Value

166

3) TIFF Header Comparison TIFF
A) Saved in Photoshop 3.0-5.5, converted to MODE grayscale.

This method keeps the barscale from the LSM.
/ usr/ people / murphy / disk2 / AN G96 / 0H02 / Right_eye_tentacle / 3
6/gl.tif: Warning, unknown ﬁeld with tag 34377 (0x8649)

ignored.

/usr/peop1e/murphy/disk2/ANG96/0H02/Right_eye_tentacle/
36g1.tif
/usr/people/murphy/disk2/ANG96/0H02/Right_eye_tentacle/36/
gl.tif: Warning, unknown field with tag 34377 (0x8649)
ignored.
TIFF Directory at Offset 0x8

Subfile Type: (0 = 0x0)

Image Width: 512 Image Length: 512

Resolution: 72, 72 pixels/inch
Bits/Sample: 8

Compression Scheme: none

Photometric Interpretation: min-is—black
Samples/Pixel: 1

Rows/Strip: 512

Planar Configuration: single image plane

Volume header:

magic: 43970

volume_form= 1

size= 262144

pathname=
/usr/peOp1e/murphy/disk2/ANG96/0H02/Right_eye_tentacle/36

data bits: 8

normal bits: 0

gradient bits: 0

bits/voxel: 8

167

B) TIFF converted and saved in LSM_PC: MODE RGB (Index color in
Photoshop 3.0-5.5?)
This method loses the barscale from LSM.

IRIS 6% VoxelLoader -h
/usr/people/murphy/disk2/ANG96/0H02/Right_eye_tentacle/36

I,

/usr/people/murphy/disk2/ANG96/0H02/Right_eye_tentacle/36

036.tif
TIFF Directory at Offset 0x8
Subfile Type: (0 = 0x0)

Image Width: 512 Image Length: 512

Bits/Sample: 8

Compression Scheme: none

Photometric Interpretation: palette color (RGB from
colormap)

Software: "ZIF 1.81 MAR-93"

Make: "Carl Zeiss, Oberkochen, Germany"

Model: "Laser Scan Microscope"

Samples/Pixel: 1

Planar Configuration: single image plane

Color Map: (present)

Volume header:

magic: 43970

volume_form= 1

size= 262144

pathname=

/usr/people/murphy/disk2/ANG96/0H02/Right_eye_tentacle/36

data bits: 8

normal bits: 0

gradient bits: 0

bits/voxel: 8

168

APPENDIX C

DIMENSIONS FILE FOR VOLUMETRIC DATA SET:
VOXELVIEW'S "_DIMENSIONS" FILE

*VoxelView Data Set Descriptor File (ﬁeld format shown):

*ttt‘ttttﬁ

'- [zsize] [xsize] [ysize] [interps] [bits/voxel] *
* [zoﬁset] [xoﬁset] [yoﬁset] [voﬁset] *
* [zscale] [xscale] Lyscale] [vscale] *
* [zunits] [xunits] [yunits] [vunits] *
* [zlabel] [xlabel] (ylabell [vlabel] *

*This header must also remain in the ﬁle VERBATIMI! *“*********
37 412 300 0 8

0.000000 499.042969 499.042969 0.000000 0

10.00000 1.380000 1.380000 1 .000000
microns microns microns greyscale
z__axis x_axis y_axis voxelvalue

169

APPENDIX D

IMAGE ROTATION STEPS

The correction can be done in a couple of ways: a) the digital image on
the computer can be “ﬂipped” horizontally (left to right) and then ﬂipped
vertically (top to bottom); or b) the image can be “rotated” 180 degrees.
The program VoxelMath allows one to rotate or ﬂip all the images that
comprise a sample in a single step.

A) 1. Flip all images in x-axis: In the Merge Volume window,
under section “Unary (V 1) Operations” selection “Flip X-Axis”: This
reverses the voxel order along the x-axis, thus functionally “ﬂipping”
the volume left-to-right. This creates a second volume data set in
memory (only). 2. Repeat procedure on newly created volume data set
and select Option “Flip-Y Axis” to reverse the voxel order along the y-
axis, thus functionally ﬂipping the volume top-to-bottom.

B) Rotate the volume: function “TWist” in the “Operations
Window: Ops Panel: Math Buttons.” “Argument value” set to 180. This
rotates (twists) the image 180 degrees. “Do Opts on Vol” will rotate the
whole stack of images 180 degrees. New volume saved. (The original

non-rotated volume is left intact.)

170

APPENDIX E

MATERIALS & SUPPLIES

Chemicals

Mayer Albumen Adhesive (04261 1 Carolina Biological Supply
Company)

Alpha brand white chalk No. 314005 by Weber Costello, or
Hygieia brand by Dixon

Video Eguipment

Mitsubishi SVHS video recorder (HS-U62) and a Quasar VHS
video recorder (VH200)

Video images were captured on either Maxell XL-HIFI T-120 or

Polaroid Supercolor Plus T-120 VHS tape media

Computer Equipment

Dell Dimensions P90 Computer with Six gigabyte hard drive and
sixty-four megabytes of RAM

Silicon Graphics, Inc. Personal his running IRIX 4..0.5.C with 72
Megs RAM, 780 internal Hard Drive and 2 GB external SCSI Hard

Drive with 500MB swap ﬁle.

171

APPENDIX F

ASSORTED BIBLIOGRAPHIC REFERENCES

Page , L. R. citations

Page, Dr. Louise Ft. Bickell; Teaching: Developmental Biology; Reasearch:
Developmental Biology: My research investigates how evolutionary innovations and
patterns of phylogeny might be understood in terms of changes to developmental
programmes. I am currently studying comparative patterns of development in
gastropods to test long-standing hypotheses about the early evolution and phylogeny
of this group. These studies focus on morphogenesis of the nervous system, shell
musculature, and shell form and employ electron microscopical and immuno-
cytochemical methods. Molluscs are useful models for this research because
hypotheses about historical changes in shell form and shell muscles can be tested
against the fossil record; Dept. Biology. Univ. Victoria, Victoria, B.C., V8W Canada
(http://www.oz.netlcgl-blnlhgslsearch.pllnervcusl&lsysteml)

Bickell Page, LR. 1989. Autotomy of cerata by the nudibranch mollusc Melibe leonina:
a morphological and neurophysiological inquiry. American Zoologist
28(4):68A.[N]

Bickell Page, L.R. 1989a. Autotomy of cerata by the nudibranch Melibe leonina
(Mollusca): ultrastructure of the autotomy plane and neural correlate of the
behaviour. Philosoph. Trans. Roy. Soc. London, series B, Biological Sciences
324(1222):149-172.[N]

Bickell Page, LR. 1991. Repugnatorial glands with associated striated muscle and
sensory cells in Melibe leonina (Mollusca, Nudibranchia). Zoomorphology
110(5):281-291.[N]

Bickell Page, L.Fl., see also: Page, LR.

Page, LR. 1992. New interpretation of a nudibranch central nervous system based on
ultrastructural analysis of neurodevelopment in Melibe leonina. I. Cerebral and
visceral loop ganglia. Biol Bull 182:348-365

Page, LR. 1992. New interpretation of a nudibranch central nervous system based on
ultrastructural analysis of neurodevelopment in Melibe leonina. ll. Pedal,
pleural, and labial ganglia. Biol Bull 182:366-381

172

 

Page, LR. 1992. Thoughts on the Developing Opisthobranch Nervous System
Comparisons to Monoplacophorans. American Zoologist 32(5):118A.

Page, LR. 1992. New interpretation of a nudibranch central nervous system based on
ultrastructural analysis of neurodevelopment in Melibe leonina. l. Cerebral and
visceral loop ganglia. Biol. Bull. 182(3):348-365.[N]

Page, L.R. 1992a. New interpretation of a nudibranch central nervous system based on
ultrastructural analysis of neurodevelopment in Melibe leonina. ll. Pedal,
pleural, and labial ganglia. Biol. Bull. 182(3):366-381.[N]

Page, LR. 1993. Development of behavior in juveniles of Melibe leonina (Gastropoda;
Nudibranchia). Mar Behav Physiol 22:141-161

ﬂ‘n-m.4 "1
A .

Page, LR. 1993. Developmental analysis reveals labila and subradular ganglia and the
primary framework of the nervous system in nudibranch gastropods. J
Neurobiol 24(1 1 ):1443-1459

Page, LR. 1993. Developmental analysis reveals labial and subradular ganglia and the
primary framework of the nervous system in nudibranch gastropods. J.
Neurobiol. 24(11):1443- 1459.[N]

Page, L.R. 1993a. Development of behaviour in juveniles of Melibe leonina
(Gastropoda, Nudibranchia). Marine Behaviour & Physiology 22(3):141-161.[N]

Page, LR. 1994. The ancestral gastropod larval form is best approximated by
hatching-stage opisthobranch larvae: evidence from comparative development
studies, pp. 206-223. In: W. H. Wilson, S. A. Stricker, & G. L. Shinn. (Eds.).
Reproduction and development of marine invertebrates; symposium, Friday
Harbor, Washington, USA, June 9-11, 1992. ix+325 pp. Johns Hopkins
University Press.[N]

Page, LR. 1995. Similarities in Form and Developmental Sequence for Three Larval
Shell Muscles in Nudibranch Gastropods. Acta Zool. 76(3):177-191.

Page, L.R., see also: Bickell-Page, L. R.[N]

Gas 0 Ph 1 n :O rReference

Bandel, K. (1993) Caenogastropoda during Mesozoic times. Scripta Geologia,
Special Issue 2: 7-56.

173

Adamkewicz, S.L., Harasewych, MG. (1994) Use of random amplified polymorphic
DNA (RAPD) markers to assess relationships among beach clams of the genus
Donax. The Nautilus, Supplement 2: 51-60.

Bandel, K., Riedel, F. (1994) Classification of fossil and Recent Calyptraeoidea
(Caenogastropoda) with a discussion on neomesogastropod phylogeny. Berl.
Geowls. Abh. (E) 13: 329-367.

Bieler, R. (1992) Gastropod phylogeny and systematics. Annual Review of
Ecological Systems, 23: 311-338.

van den Biggelaar, J.A.M. (1993) Cleavage pattern in embryos of Haliotis tuberculata
(Archaeogastropoda) and gastropod phylogeny. Journal of Morphology, 216:
121-139.

van den Biggelaar, J.A.M. (1996) The significance of the early cleavage pattern for the
reconstruction of gastropod phylogeny. In Origin and Evolutlonary Radiation
of the Mollusca, Taylor, J.D. (ed), Oxford University Press, Oxford. Pp. 155-
160.

Freeman, G., Lundelius, J.W. (1992) Evolutionary implications of the mode of D
quadrant specification in coelomates with spiral cleavage. Journal of
Evolutionary Blology, 5: 205-248.

Fretter, V., Graham, A. (1962) British Prosobranch Molluscs, Ray Society, London.

Golikov, A.N., Starobogatov, Y.l. (1 975) Systematics of prosobranch gastropods.
Malacologla, 15: 185-232.

Haszprunar, G. (1985a) The fine morphology of the osphradial sense organs of the
Mollusca. Part 1: Gastropoda-Prosobranchia. Philosophical Transactions of
the Royal Society of London, 8 307: 457-496.

Haszprunar, G. (1985b) The Heterobranchia- a new concept of the phylogeny and
evolution of the higher Gastropoda. Zeit. Zool. System.Evol.-forsch. 23: 15-37.

Haszprunar, G. (1988) On the orgin and evolution of major gastropod groups, with
special reference to the Streptoneura (Mollusca). Journal of Molluscan
Studies, 54: 367-441.

174

Haszprunar, G. (1993) Sententia. The Archaeogastropoda: a clade, a grade or what
else? American Malacologlcal Unlon Bulletin, 10: 165-177.

Healy, J.M. (1988) Sperm morphology and its systematic importance in the
Gastropoda. In: Prosobranch Phylogeny, Malacologlcal Review
Supplement 4: 251 -266. Ponder, W.F. (ed).

Healy, J.M. (1996) Molluscan sperm ultrastructure: correlation with taxonomic units
within the Gastropoda, Cephalopoda and Bivalvia. In: Origin and Evolutionary
Radiation of the Mollusca. Taylor, J.D. (ed), Oxford University Press, Oxford.
Pp. 99-1 13.

Kantor, Y.l. (1996) Phylogeny and relationships of Neogastropoda. In: Origin and
Evolutionary Radiation of the Mollusca. Taylor, J.D. (ed), Oxford University
Press, Oxford. Pp. 221-230.

Lindberg, DR. (1988) The Patellogastropoda. Malacology Review, Supplement 4:
35-63.

Lindberg, D.R., Ponder, W.F. (1991) A phylogenetic analysis of the Gastropoda.
Geological Society of America. Abstracts, 1991 annual meeting, 23 (5):
A160-A161.

Lindberg, D.R., Ponder, W.F. (1992) Discovering key innovations in the phylogeny of
gastropod molluscs. Program Abstracts, XI Hennlg Society Meeting, 1992:
48.

Lindberg, D.R., Ponder, W.F. (1996) An evolutionary tree for the Mollusca: Branches
or roots? In: Origin and Evolutlonary Radiation of the Mollusca, , Taylor,
J.D. (ed), p. 67-75.

Ponder, W.F. (I970) Some aspects of the morphology of four species of the
neogastropod family Marginellidae with a discussion on the evolution of the
toxoglossan poison gland. Journal of the Malacologlcal Society of Australia,
2(I): 55-8I.

Ponder, W.F. (I973) The origin and evolution of the Neogastropoda. Malacologla,
I2(2): 295-338. Ponder, W.F. (ed) (1988) Prosobranch phylogeny.
Malacologlcal Review, Supplement 4, 346 pp.

Ponder, W.F. (1991) Marine valvatoidean gastropods - implications for early
heterobranch phylogeny. Journal of Molluscan Studies, 57: 21-32.

175

Ponder, W.F., Lindberg, D.R. (1996a) Gastropod phylogeny - challenges for the ’90s.
In: Origin and Evolutionary Radiation of the Mollusca, Taylor, J.D. (ed), p.
135-154.

Ponder, W.F., Lindberg, D.R. (1996b) (In press). Towards a phylogeny of gastropod
molluscs - a preliminary analysis using morphological characters. Zoological
Journal of the Llnnean Society.

Ponder, W.F., Waren, A. (1988) Classification of the Caenogastropoda and
Heterostropha - a list of the family-group and higher category names. In:
Prosobranch Phylogeny. Malacology Review, Supplement 4: 288-326. ,
Ponder, W.F. (ed).

an... a
tip-LA

Reid, 0.6. (1989) The comparative morphology, phylogeny and evolution of the
gastropod family Littorinidae. Philosophllcal Transactions of the Royal ,
Society of London, Series B, 324: 1-110.

Riedel, F. (1995) An outline of cassoidean phylogeny (Mollusca, Gastropoda). Contr.
Tertiary and Quaternary Geology, 32: 97-132.

Rosenberg, G., Kuncio, G.S., Davis, G.M., Harasewych, MG. (1994) Preliminary
ribosomal RNA phylogeny of gastropod and unionoidean bivalve mollusks.
Nautilus, Supplement 2: 111-121.

Salvini-Plawen, L.V., Haszprunar, G. (1987) The Vetigastropoda and the systematics
of streptoneurous Gastropoda (Mollusca. Journal of Zoology, London. 211:
747-770.

Taylor, J.D. ( 1996) (ed.). Origin and evolutionary radiation of the Mollusca. Oxford
University Press, Oxford.

Taylor, J.D., Kantor, Y.l., Sysoev, A.V. (1993) Foregut anatomy, feeding mechanisms,
relationships and classification of the Conoidea (=Toxoglossa) (Gastropoda).
Bulletin of the Natural History Museum of London. (Zoology) 59:125-170.

Taylor, J.D., Morris, NJ. (1988) Relationships of neogastropods. Malacology Review,
Supplement 4: 167-179.

Taylor, J.D., Morris, N.J., Taylor, ON. (1980) Food specialization and the evolution of
predatory prosobranch gastropods. Palaenotology, 23: 375-409.

176

Thiele, J. (1929-31) Handbuch der Systematischen Welchtierkunde. Vol. 1. Jena.

Tillier, S., Masselot, M., Herve, P., Tillier, A. (1992) Phylogénie moléculaire des
Gastropoda (Mollusca) fondée sur Ie séquencage partiel de l'ARN ribosomique
28 S. Comp. Rend. Acad. Sci. (Paris) 134 (ser. 3): 79-85.

Tillier, S., Masselot, M., Guerdoux, J., Tillier, A. (1994) Monophyly of major gastropod
taxa tested from partial 28$ rRNA sequences, with emphasis on Euthyneura
and hot-vent limpets Peltospiroidea. Nautilus, Supplement 2: 122-140.

Tillier, S., Masselot, M., Tillier, A. (1996) Phylogenetic relationships of the pulmonate
gastropods from rRNA seqences, and tempo and age of the stylommatophoran
radiation. In: Origin and Evolutionary Radiation of the Mollusca. Taylor, J.D.
(ed), Oxford University Press, Oxford. Pp. 267-284.

Wenz, W. (1938-1944) Gastropoda, Teil 1: Allgemeiner Teil und Prosobranchia. In:
Handbuch der Palaozoologle, 6. Schindewolf, O.H. (ed), Berlin. vi+1639 pp.

Snail Image citation: http://www.nearctica.comlnathistlmolIuskslgastrop.htm

177

BIBLIOGRAPHY

178

LITERATURE CITED

"Ongz'na/iy i: tbe art of concealinyour Journey. "

- Unknown

Andrews, E. B. (1965). The functional anatomy of the mantle cavity, kidney and blood
system of some pilid gastropods (Prosobranchia). J. Zool. 146: 70-94.

Andrews, E. B., and Little, C. (1972). Structure and function in the excretory system of
some prosobranch snails (Cyclophoridae). J. Zool. 168: 395-422.

Atkinson, J. W. (1992). Conceptual issues in the reunion of development and evolution.
Synthese. 91 : 93-1 1 0.

Audesirk, T. E. (1975). Chemoreception in Aplysia califomica. I. Behavioral localization
of distance Chemoreception used in food-finding. Behav. Biol. 15: 45-55.

Audesirk, T. E., and Emery, D. G. (1978). Sensory cells in Aplysia. J. Neurobio. Vol. 9.
NO. 2’ pp- 173'179.

Baba, N. (1991). Computer-aided three-dimensional reconstruction from serial section
images. In “Image Analysis in Biology," ( Donat-P. Hader, Ed.) pp. 251-270.
CRC Press, Boca Raton.

Barker, G. M. (1999). Naturalised terrestrial Stylommatophora (Mollusca: Gastropoda).
Fauna of New Zealand 38, 253 pages.

Blochmann, F. (1882). Ueber die Entwicklung der Neritina fluviatilis MGII. Z. wiss. Zool.
36: 125.

Boettger, C. R. (1954). Die Systematik der euthyneuren Schnecken. Zool. Anz., Suppl.
17: 253-279.

Boudko, D. Y., Switzer-Dunlap, M., Hadfield, M. G. (1999). Cellular and Subcellular
Structure of Anterior Sensory Pathways in Phestilla sibogae (Gastropoda,
Nudibranchia). J. Comp. Neurol. 403: 39-52.

179

Bullock, T. H., and Horridge, G. A. (1965). Structure and Function in the Nervous
Systems of Invertebrates. Vol. 2. W. H. Freeman and Company, San Francisco.

Burch, J.B. 1982. Freshwater snails (Mollusca: Gastropoda) of North America. Bull.
Environ. Protec. Agency, EPA-600/3-82-026: i-vi, 1-294.

Carney, T., and Verdonk, N. H. (1970). The early development of the snail
Biomphalan’a glabrata (Say) and the origin of the head organs. Neth. J. Zool.
20: 93-1 21 .

Carazzi, D. (1905-6). L’embriologia dell’ Aplysia e i problemi fondamentali dell’
embriologia compara. Arch. Ital. Anat. Embriol. 4: 231, 459; 5: 667.

Casteel, D. B. (1904). The cell-lineage and early larval development of Fiona marina, a
nudibranch mollusc. Proc. Acad. Nat. Sci. Philadelphia 56: 325.

Chase, R., and Tolloczko, B. (1985). Secretory glands of the snail tentacle and their
relation to the olfactory organ (Mollusca, Gastropoda). Zoomorphology 105: 60-
67.

Chase, R., and Tolloczko, B. (1993). Tracing neural pathways in snail olfaction: From
the tip of the tentacles to the brain and beyond. Microsc Res Tech 24: 214-230.

Chia, F.-S., and Koss, R. (1982) Fine structure of the larval rhinophores of the
nudibranch, Rostanga pulchra, with emphasis on the sensory receptor cells.
Cell and Tissue Research. 225: 235-248.

Chia, F.-S., and Koss, R. (1984) Fine structure of the cephalic sensory organ in the
larva of the nudibranch Rostanga pulchra (Mollusca, Opisthobranchia,
Nudibranchia). Zoomorphology. 104: 131-139.

Conklin, E. G. (1897). The embryology of Crepidula. J. Morph. 13: 1.

-------- (1907). The embryology of Fulgur. A study of the influence of yolk on
development. Proc. Acad. Nat. Sci. Philadelphia 1907: 320.

Creek, G. A. (1951). The reproductive system and embryology of the snail Pomatias
elegans (Muller). Proc. Zool. Soc. London122: 599-640.

180

Crofts, D. R. (1937). The development of Haliotis tuberculata, with special reference to
the organogenesis during torsion. Phil. Trans. R. Soc. Lond. 8228: 219-268.

Delsman, H. C. (1912). Ontwikkelingsgeschiedenis van Littorina obtusata. Diss.
Amsterdam 1912. --Cf. Entwicklungsgeschichte von Littorina obtusata.
'ﬁjdschr. Ned. Dierk. Ver. (2) 13: 170 (1914).

Eakin, R. M., and Brandenburger, J. L. (1967): Differentiation in the eye of a pulmonate
snail Helix aspersa. J. Ultrastruc. Res. 18: 391 -421.

Fretter, V., and Graham, A. (1962). British Prosobranch Molluscs. Ray Soc., London.

Gilbert, S. F., Opitz, J. M., Raff, RA. (1986). Resynthesizing Evolutionary and
Developmental Biology. Developmental Biology. 173: 357-372.

Gomot, P., Gomot, L., Marchand, C. -R., and Colard, C. (1990). Extirpation and
transplantation of the brain of the snail Helix aspersa: a study of the survival of
the animal and the implant. Can. J. Zool. 68: 1505-1512.

Graham, A. (1985). Evolution within the Gastropoda: Prosobranchia. In “The Mollusca,”
(E. R. Trueman and M. R. Clarke, eds.), Vol.10, pp. 151-186. Academic Press,
Inc., New York.

Hart, M. W., and Wray, G. A. (1999). Heterochrony. In ”The Origin and Evolution of
Larval Forms,” (B. K. Hall and M. H. Wake, eds.). Pp. 159-165. Academic
Press, Inc. San Diego.

Haszprunar, G. (1988). On the origin and evolution of major gastropod groups, with
special reference to the streptoneura. J. Moll. Stud. 54: 367-441.

Henchman, A. (1890). The origin and development of the central nervous system in
Limax maximus. Bull. Mus. Comp. 200/. Harvard College. Vol. 20. No. 7.

Heymons, R. (1893). Zur EnhNicklungsgeschichte von Umbrella mediterranea Lam. Z.
wiss. Zool. 56: 245.

Holmes, S. J. (1900). The early development of Planorbis. J. Morph. 16: 369.

Hyman, L. H. (1967). The Invertebrates. Volume VI. Mollusca I. McGraw-Hill Book
Company. New York.

181

Jacob, M. H. (1984). Neurogenesis in Aplysia califomica resembles nervous system
formation in vertebrates. J Neurosci. Vol. 4, No. 5: 1225-1239.

Kotoid, C. A. (1895). On the early development of Limax. Bull. Mus. Comp. 200/.
Harvard Call. 27:35.

Maderson, P. F. A. (1975). Embryonic tissue interactions as the basis for
morphological change in evolution. Amer. Zool. 15: 315-327.

Meisenheimer, J. (1896). Entwicklungsgeschichte von Limax maximus L. I. Furchung
und Keimblatterbildung. Z. wiss. Zool. 62: 415.

Milne-Edwards, H. (1846). Sur la classification naturelle des mollusques gastéropodes.
Institut 14(1), 295-296. As stated in: Schmekel, L. (1985). Aspects of Evolution
within the Opisthobranchs. In "T he Mollusca,” (E. R. Trueman and M. R. Clarke,
eds.), Vol.10, Evolution, pp. 221-267. Academic Press, Inc., New York.

Moffett, S. and Austin, D. R. (1981 ). Implanted cerebral ganglia produce
supernumerary eyes and tentacles in host snails. J. Exp. Zool. 216: 321 -325.

Moor, B. (1983). Organogenesis. In “The Mollusca,” (K. M. Wilbur, ed.), Vol. 3:
Development, N. H. Verdonk, J. A. van den Biggelaar, and A. S. Tompa, eds.
Academic Press, Inc., New York.

Moritz, C. E. (1939). Organogenesis in the gastropod Crepidula adunca Sowerby. Univ.
Calif. Berkeley Publ. Zool. 43: 217-248.

Morrill, J. B. (1982). Development of the Pulrnonate Gastropod, Lymnaea. In
"Developmental Biology of Freshwater Invertebrates.” Pp. 399-483. Alan R.
Liss, Inc. New York.

Miller, G. B. (1990). Developmental Mechanisms at the Origin of Morphological
Novelty: A Side-Effect Hypothesis. In ”Evolutionary Innovations.“ (Ed. Matthew
H. Nitecki).pp. 99-130. University of Chicago Press. Chicago.

Page, L. (1992): New interpretation of a nudibranch central nervous system based on
ultrastructural analysis of neurodevelopment in Melibe leonina. I. cerebral and
visceral loop ganglia. Biol. Bull. 182: 348-365.

Pawley, J. B. (Ed.)(1990). Handbook of Biological Confocal Microscopy. Plenum Press.
New York.

182

Raff, R. A. (1996). The Shape of Life: Genes, Development, and the Evolution of
Animal Farm. The University of Chicago press. Chicago.

Raff, R. A., and Popodi (1996). Evolutionary Approaches to Analyzing Development. In
"Molecular Zoology: Advances, Strategies, and Protocols.” (Eds. J. D. Ferraris
and S. R. Palumbi). Pp. 245-265. Wiley-Liss, Inc. New York.

Raven, C. P. (1966). Morphogenesis: the Analysis of Molluscan Development.
Pergamon Press, Oxford.

Raven, C. P., and Beenakkers, A. M. Th., (1955) On the nature of head malformations
obtained by centrifuging the eggs of Limnaea stagnalis. J. Embr. Exp. Morph. 3,
286.

Rigaut, J. P., Carvajal-Gonzalez, S., Vassy, J. Confocal image cytometry - Quantitative
analysis of three-dimensional images obtained by confocal scanning
microscopy. In “Image Analysis in Biology,“ (Donat-P. Hader, Ed.) pp. 109-133.
CRC Press. Boca Raton.

Robert, A. (1902). Recherches sur le développement des Troques. Arch. Zool. Exp.
Gen. (3) 10: 269.

Rutherford, S. L., and Lindquist, S. (1998). Hsp90 as a capacitor for morphological
evolution. Nature 396: 336-342.

Schmekel, L. (1985). Aspects of Evolution within the Opisthobranchs. In “T he
Mollusca,” (E. R. Trueman and M. R. Clarke, eds.), Vol.10, Evolution, pp. 221-
267. Academic Press, Inc., New York.

Simkiss, K. (1985). Molluscan Skin (excluding Cephalopods). In “The Mollusca,” (E. R.
Trueman and M. R. Clarke, eds.), Vol.11, pp. 11-36. Academic Press, Inc.,
New York.

Sokolove, P. G., Melrose, G. R., Gordon, T. M., O’Neill, M. C. (1983). Stimulation of
Spermatogonial DNA Synthesis in Slug Gonad by a Factor Released from
Cerebral Ganglia under the Influence of Long Days. General and Comparative
Endocrinology. 50: 95-104.

Solem, A. (1974). The Shell Makers: Introducing Mollusks. Wiley, New York.

183

Solem, A. (1985). Origin and Diversification of Pulrnonate Land Snails. In "T he
Mollusca,” (E. R. Trueman and M. R. Clarke, eds.), Vol.10, Evolution, pp. 269-
293. Academic Press, Inc., New York.

Thompson, T. E. (1976). “Biology of Opisthobranch Molluscs.” Ray Soc., London. As
stated in: Schmekel, L. (1985). Aspects of Evolution within the Opisthobranchs.
In "The Mollusca,” (E. R. Trueman and M. R. Clarke, eds.), Vol.10, Evolution,
pp. 221-267. Academic Press, Inc., New York.

van Mol, J. (1967). Etude morphologique et phylogénétique du ganglion cérébroi'des
des Gastréropodes Pulmonés (Mollusques). Mem. CI. Sci, Acad. R. Belg.
Collect. 80 37: 1-168.

Verdonk, N. H. (1965). "Morphogenesis of the Head Region in Limnaea stagnalis L.“
Thesis. Utrecht.

Verdonk, N. H., and van den Biggelaar, J. A. M. (1983). Early Development and the
Formation of the Germ Layers. In “The Mollusca,” (Verdonk, N. H., and van den
Biggelaar, J. A. M., eds.), Vol. 3, Development, pp. 91-122. Academic Press,
Inc., New York.

Waddington, C. H. (1941). Evolution of developmental systems. Nature 147:108-110.

 

(1957). The Strategy of the Genes. Allen & Unwin. London.

 

(1962). New Patterns in Genetics and Development. Columbia
Univeristy Press. New York.

(1966). Principles of Development and Differentiation. The Macmillan
Company. New york.

 

Walter, U. (1999). When fuzzy thinking is a good thing. [Online] HMS Beagle. Issue 60.
Available August 06, 1999. httgJ/biomednet.com/hmsbeagle/60/viewgts/op gd.

Weibel, E. R. (1979) Stereo/ogical Methods. Academic Press, London.

Wierzejski, A. (1905). Embryologie von Physa fontinalis L. Z. wiss. Zool. 83: 502.

Wilson, E. B. (1892). The cell lineage of Nereis. J. Morphol. 6: 361-481.

184

Wolpert, L. (1998). Principles of Development. Oxford University Press, New York.

Zilch, A. (1959). Gastropoda. Teil 2. Euthyneura. In “Handbuch der Paléozoologie” (W.
Wenz, ed.), Vol. 6, pp. 1-200. Gebr. Borntraeger, Berlin.

RNAME-IT 3.11 by Kent H. Copyright © 1998
Homepage: http://www.angelfire.com/ca/kent/ -- the gateway to my internet
empire
If you have problem accessing my homepage then try my other sites...
http://trailerpark.com/phase1/next/fileutil.html -- all my programs (freeware) are
here

 

185

ENDNOTES

 

l for arguments against this division see Boettger, 1954; Zilch, 1959.

2 Although, contrary to what Verdonk and van den Biggelaar (1983) wrote, it is the third

cleavage and not the second cleavage that determines the ‘cleavage attem” for the
embryo, and thus the handedness, “dextral” or right-handed coiling, or ‘ sinistral” or left-
handed coiling, of the shell. See Raven, 1966, for examples showing this to be true.

3 While all published work ees on the pretrochal origin of the cerebral ganglia, Louise P
épers. comm.) believes s e has discovered migraungﬁorttmcbal cells contributing to :E:
ormation of the cerebral ganglia in the opisthobranch elibe leonina.

4 Whether one or two occur on each side is uncertain. See Henchman (1890) and Raven
(1966) for review.

5 One of the most striking exce tions are the Strombidae which have what apgears to be
exceedingly long eye-s s, wi just a small distal branch of tentacle (Hyman, 1 67).

6 Assumed Causal Connection

In its classical form, this fallacy is called the post hoc ergo ropter hoc, which is Latin for
"After this, therefore as a result of this." Known also as 'falie cause," this fallacy involves a
confusion of the time of an event or action and the results of that event or action. Sim 1y
because something happens at a particular time, and something else happens after it, e
two need not be related in any way but temporally.
(httpz/ /www.rhetor.com/compete/fallacieshtm, 1999)

186

IES

"Ijljjjjljjljjlllllllllll"