Jim Diederich1, Renaud Fortuner2*, and Jack Milton1
1: Department of Mathematics, University of California, Davis, CA 95616, USA; 2: La Cure, 86420 Verrue, France
* Corresponding author, e-mail: fortuner@wanadoo.fr
Reproduced with permission from:
Nematology 2(8): 805-822.
Summary -- The need for a uniform method of describing structures (organs and sub-organs) of taxa prior to building a database for nematode identification and other computer applications is discussed. The classical representation of characters (entity -- attribute -- value) does not assure uniformity and guidelines are needed for a truly uniform representation of morphological-anatomical entities, i.e., the organs and sub-organs of nematodes or other biological groups. Views and other relationships are proposed to manage specific aspects of representation such as perspectives (e.g., front vs. lateral view), junction between two organs, structural overlaps, grouping of organs, sub-organs contained in several organs, and organs that could be arranged in more than one system. The value of the proposed character representation is discussed for identification, taxonomy (homologies), and other uses.
Résumé -- Une représentation uniforme du plan d'organisation des nematodes de l'ordre Tylenchida -- L'article considère la nécessité d'une méthode uniforme pour la description des structures des taxons dans le cadre de la construction d'une base de données pour l'identification assistée par ordinateur et d'autres applications informatiques. La représentation classique des caractères en entité - attribut - valeur ne suffit pas à elle seule à assurer l'uniformité et des directives sont nécessaires pour aboutir à une décomposition réellement uniforme. La représentation uniforme des 'entités' morpho-anatomiques est discutée, les entités en question étant les organes et sous-organes des nématodes et d'autres groupes biologiques. Des vues (au sens informatique) et d'autres relations sont proposées pour gérer des aspects spécifiques de la représentation tels que les différentes perspectives (vue de face et vue latérale par exemple), les jonctions entre deux organes, les recouvrements structurels, l'association de plusieurs organes, la présence d'un même sous-organe dans plusieurs organes et les organes qui appartiennent à plusieurs systèmes. La valeur de la représentation des caractères proposée dans cet article est discutée pour l'identification, la taxinomie (homologies) et d'autres utilisations informatiques en biologie.
Keywords -- anatomy, biological characters, biological database,
description,
homology, identification, morphology, organs, plan of organization,
representation,
schema design, structural decomposition, structure modeling,
systems,
uniformity
0. Introduction.
In two previous papers (Diederich, 1997; Diederich et al., 1997), we presented a set of design principles for developing character sets for morpho-anatomical databases within the Genisys (General Identification System) project (Diederich et al., 1998). We developed these principles to gain control over a large character set that at times seemed almost unmanageable due to its size and complexity. Although the resulting character set was much simpler, more compact, and more consistent than earlier versions, additional review has lead to refinements that can significantly improve and simplify the design of large character sets. Establishing a set of principles is an iterative process, alternating between discovery of representational problems and developing a consistent means of solving them.
Character sets that are not constructed in a systematic and consistent fashion may make design, use, and integration of related databases much more difficult to achieve, as illustrated in Diederich (1997) and Diederich et al. (1997). Without a methodology based on clear principles, it is extremely difficult to achieve a consistent representation either within a single large character set or when several character sets prepared by different authors need to be combined into a single database. Without consistency, integration of character sets for different families of nematodes (or other groups) becomes an arduous task and, even within a character set, effective use depends on consistency.
We have found that biologists are quite willing to comply with principles of database design provided they are able to incorporate the richness of their data and are not forced to oversimplify because of constraints set by others. This conflict between the uniformity required for effective use of the data and the full, rich, and varied representation demanded by biologists is one of the stumbling blocks faced in determining appropriate design principles for biological databases.
By ‘uniformity', we mean that a single method is used to represent a particular relationship every time it is found in a list of characters as part of the 'schema' of the database, i.e., the description of the structure of the data in the database such as the hierarchical decomposition of the biological systems and organs, their properties, and their states or values. A simple violation of uniformity would be to have a character such as in 'annuli, width' in one part of the schema, where a structure (annuli) and its property (width) are clearly delineated, and in another part of the schema have 'width of lumen in the procorpus', where the property is an amalgamation of a true property, the width, and structures, the lumen and the procorpus (Diederich et al., 1998).
By 'full, rich, and varied representation’, we mean that the representation of the characters is a natural one in the eyes of the biologist and it includes all the data that any biologist would like to record. In the scope of the present article, these data are all the structures (as defined below) that compose a nematode, even those that some authors would not consider to be 'useful' for identification, e.g., the spermatheca. All the various names used over the years by the authors for these structures must be included, both alternative names of a structure (e.g., oesophagus, esophagus, and pharynx) and invalid names that are traditionally used by some scientists (aerolations for areolations). All valid and invalid names of structures need to be included in the representation of the characters because i) everyone should be free to use any name (even invalid names because the right to make mistakes is one of the fundamental rights of humankind), and ii) authors do use these names, which means that a narrowly-defined system insisting on a single name per structure would have trouble coping with data entry using other names.
In the Genisys representation, each character is decomposed into three elements:
As we discuss in detail below, the hierarchical arrangement of the structures caused some difficulties that were not recognized in our earlier efforts. Our focus in this paper is to refine the specification of these structures. As we discovered additional problems in reviewing the character set, the challenge was to keep the design methodology simple. It is very easy to solve problems by framing a new concept to solve each successive difficulty, but this complicates matters for the designer. Therefore our goal was to limit the number of concepts and, within each concept, to identify major categories. For example, the concept of basic property was introduced by Diederich (1997) with four major categories, Appearance, Dimension, Location, and Quantity, each containing a few very fundamental basic properties for morpho-anatomical data. As a result, a great deal of the design effort can be focused on simply selecting the appropriate basic properties for a given structure, a considerable simplification of the design process.
Our aims in this article are the following:
1. To elucidate new problems encountered in arranging the
existing
structures in a uniform hierarchical manner
2. To explain the solutions by detailing the biological point
of view and the implementation concepts
3. To introduce a first list of structures (for Tylenchida, a
nematode order), which is available in Table 1 and on the Web at the
following
URL:
http://math.ucdavis.edu/~milton/genisys.html [This schema can
now be seen at: http://www.genisys.prd.fr/schema.html
(click on Tylenchida).]
The basic properties and states attached to each structure will be added to the Web list later as a final review of each is completed.
1. Representation
1.1 The plan of organization
Representation in the 'structure-basic property-state/value'
form is simple in principle. It follows the
entity-attribute-value
representation that is standard in database design (Date, 1995).
However, the complexity of biological characters presents challenges in
adhering to this design paradigm as discussed in some detail by
Diederich
et al. (1997).
As a guide for formulating this organization, we used what is traditionally called the 'plan of organization' of the various biological groups, generally at the phylum level, and specifically the nematode plan of organization. This plan is the description of the parts of an organism arranged by major physiological functions: protection, locomotion, digestion, excretion, circulation, control, and reproduction. These functions are provided by body envelopes, somatic muscles and skeleton, digestive system, excretory system, circulatory system, nervous system, and genital system, respectively.
The plan of organization of a biological group is well known to the biologists who specialize in this group and it is well documented: it is how this biological group is described in textbooks. As the plan of organization is based on physiological functions, it can be used in many domains besides taxonomy: physiology, ecology, molecular biology, etc.
The plan of organization of nematodes differs from the plan of other phyla not just by including different structures, but also by the absence of several systems, in particular the circulatory system. When the Genisys representation is used with other biological groups, the missing systems will be added.
1.2. Morphology and Anatomy
In textbooks, plans of organization of animal groups traditionally begin with a description of the external morphology, i.e., the description of the parts of an animal as seen from the outside. The whole body is divided into thorax, abdomen, tail, appendices, limbs, etc., depending on the biological group considered. For example, the body of a tylenchid is divided into a head, a neck, a body sensu stricto, and a tail.
In addition, many internal structures include openings to the outside, which are parts of the external morphology. In humans, the mouth and the anus (parts of the digestive system) or the eyes and ears (parts of the sensory system) are all visible externally. In the Genisys schema, to avoid duplication of structures, each structure is described within one of the internal systems, but a separate mechanism, called 'views' by database designers, is used to access them when looking at the external morphology of the organism. For example, labial sensillae are part of the sensory system, but they can be viewed with the rest of the external morphology of the head. Views have long been used in databases to tailor presentations of the data to particular types of user. A view in that context is composed of parts of database tables that are processed and put together in a new, different way that allows this user to view the data according to his specific needs. To the final user, it is not apparent that a view mechanism is being used. If a biologist calls for the external morphology of the head from a Genisys database, labial sensillae will be displayed without it being obvious that a view mechanism was used to create this display. Only the designer of the structures or others interested in the design will be concerned with this mechanism.
1.3. Hierarchical arrangement: ideal vs. reality
Based on the plan of organization, the list of structures is naturally arranged in a hierarchy, with every structure placed in one of the usual systems (digestive system, nervous system, etc., as listed above). Within each system, the structures are again arranged hierarchically with the major organs divided into several organ parts, which are structures of their own. Each structure can be seen as a part, i.e., a substructure, of a larger structure and at the same time, it is composed of several parts, which means it is the superstructure of smaller substructures. For example, in tylenchids, a major part of the digestive system is the oesophagus, but this structure is composed of several substructures: the procorpus, the median bulb, the isthmus, etc. (Fig. 1). Each of these substructures includes one or several parts. For example, the median bulb is a superstructure for a substructure such as the median bulb valve.
This hierarchical arrangement makes it easy to classify and retrieve
each structure. However, this ideal cannot always be
achieved.
In the early attempts to list Tylenchida structures, it was difficult
to
adhere to an ideal representation of the structures because of various
problems that were temporarily solved by creating artificial structures
such as:
- Junction between procorpus and median bulb (Fig. 1). Procorpus
and median bulb are real structures, but the junction between them is
an
imaginary plan, not a physical structure.
- Spermatheca in lateral view and Spermatheca in face view. There
is only one spermatheca per genital branch (Fig. 2) in a nematode and
there
should not be two separate structures to describe how it looks from
different
perspectives.
- Overlap of the subventral oesophageal glands (Fig. 1). The
part of the glands that overlaps the intestine is not differentiated
from
the rest of the glands and it should not be a different structure.
It may seem to be trifling to distinguish between real structures and artificial structures, but there are subtle effects in treating junctions, overlaps, etc., as structures: conceptualizing the schema is more difficult, ambiguities sometimes occur in new situations, and there is the temptation to create more artificial structures to handle them. Also, a system that explicitly supports the new concepts we introduce can aid the designer and the user.
New concepts that were created to handle these 'artificial'
structures
in a systematic manner will be described below.
2. Structural decomposition issues: structural relationships
In this section we present several new issues that arose in our review of the structures. We have attempted to satisfy several criteria in dealing with these issues while preserving the concept of structure in a strict sense. Morphological entities that are not actual structures are portrayed in a manner that is both more faithful in a biological sense and more consistent with database design practice. As a result, the hierarchy of structures is easier to handle and simpler.
In the future Genisys database, each concept will be represented in a text file using a specified syntax. Generally speaking, this will not be seen by any user. A designer will normally refer to the syntax only to see what can be represented by the concept. Both user and designer will normally interact with the schema via a tool for building and displaying the schema. This schema tool will also provide some automatic operations such as generating standard synonyms for a particular concept. Finally, each concept will be used within the database itself, often as the field name for the structure in question.
2.1. Perspective
Biological structures can be seen from various perspectives: face view, lateral view, posterior view, and cross-section after dissection. Some basic properties attached to a structure will differ depending on how the structure to be described is viewed. For example, a cylindroid organ such as the spermatheca of many species will be seen as a rectangle in lateral view but as a circle in cross section. The shape and dimensions of this structure are quite different in lateral and in face views.
In our previous versions of the list of structures, two separate structures were created in such cases, e.g., one structure called Spermatheca in lateral view and the other called Spermatheca in face view. In the final list, there is only one structure, the Spermatheca, with various perspectives. It is possible to attach perspectives to any structure so the values or states observed can be attached to the proper point of view. We called this concept 'perspective' instead of 'view' because the latter already has a specific meaning in database design.
Note that in the case of a cylindroid organ such as the spermatheca, the width in lateral view is equal to the diameter in cross section. The uniformity of representation we enforce makes it possible to enter this relationship only once, and make it valid for all cylindrical structures:
If structure shape = cylindrical
then structure lateral view, width = structure cross section, diameter.
This is a state-based relationship as discussed by Diederich (1997) and Diederich et al. (1998).
In other cases, the various points of view are specific for a particular organ. For example, the lateral fields are seen in lateral view as several parallel lines running along most of the body. In cross section, we see that these lines are in fact the edge of ridges or depressions of the cuticle (Fig. 3).
When a structure is described according to several perspectives, the
one the most frequently used by the authors (often the one with the
greater
number of basic properties attached to it) is called the primary
perspective
and is used by default for the structure. The secondary
perspective
is explicitly noted as a perspective. For example, in nematodes,
the lateral field is generally described in lateral view, and more
rarely
in cross section. The list of structures includes two entries:
Lateral field
and
'Perspective - Cross-section' Lateral field
Similarly, for the spermatheca we have:
Spermatheca
and
'Perspective - Cross-section' Spermatheca
In lateral view (default), the spermatheca is seen as a roundish or
a rectangular organ, depending on the species. When rectangular,
its length and its width can be entered simply as:
Spermatheca
shape rectangular
length (specific value)
width (specific value)
Most spermathecae seen as rectangular in lateral view are rounded
in cross section, which means that, in cross section, they do not have
a length and a width but a diameter. It is possible to enter a
value
for this diameter:
'Perspective - Cross-section' Spermatheca
shape circular
diameter (specific value)
A perspective consists of three components: the term
'Perspective',
its type, and the name of a structure. The textual representation
has the following form, i.e., syntax:
'Perspective - <type>'
<structure-name>
where <type> is one of {Face view = Anterior view,
Lateral
view, Cross-section, Posterior view} and <structure-name> is the
name
of the principal view, i.e., the structure. In the example above:
'Perspective -
Cross-section'
Spermatheca
'cross-section' is the type and 'Spermatheca' is the structure name.
A schema tool should automatically generate a name and common
synonyms.
Synonyms are indicated after a sign '=' and they have the following
form:
<structure-name> <type> = <type> of the
<structure-name>
as in:
Spermatheca Cross-section = Cross-section of the Spermatheca.
The schema tool will by default show the hierarchical representation of the actual structures. Thus one would see Spermatheca in the hierarchy, but not its perspectives. However, the user or designer can opt to examine the perspectives of any structure and then the existing perspectives and their properties would be displayed. The designer can add perspectives to any structure, as needed to enter data.
Since perspectives have basic properties, within the database itself <structure-name> <type> will serve as its database name. From the Entity-Relationship model (Chen, 1976), both entities and relationships between entities are ultimately represented in the database as if relationships were also entities. The reason is that both entities and relationships have properties.
2.2. Junctions
In some cases, the place where two structures meet may have some properties, but it is not a real structure. For example, the median bulb is a spherical structure that lies between two cylindrical structures, the procorpus and the isthmus (Fig. 1). The junctions between these structures may have some characteristics that need to be described (e.g., bulb fused with procorpus vs. bulb clearly set-off by a constriction). A junction is a relationship between two structures. In database practice this is a relationship between entities.
Our original approach was to create an artificial substructure 'junction of procorpus and median bulb', but this proved unsatisfactory for the same reason as in perspectives. The final schema solves this problem by adding a new concept, 'junction', described as:
OesophagusProcorpus
'Junction' Procorpus / Median bulb
Median bulb
This arrangement clearly differentiates the real substructures,
procorpus and median bulb, from their junction.
The treatment of 'junction' seems straightforward, but, when applying it to the Tylenchida structure list, we discovered that it is not always easy to decide when a junction is needed as opposed to a structure of its own. For example, the vaginal sphincter is the place or junction where the vagina narrows before expanding into a vulva (Fig. 2). However, it can be described as a bona fide structure instead of a junction because it is a real structure composed of circular muscles. Generally speaking, a junction becomes a structure in its own right when it is a clearly distinct part, that is, when it would continue to exist if the adjacent structures were to be dissected out. Such 'junctions' are generally given a name of their own in the literature, e.g., Cardia is a true structure located at the junction between Oesophagus and Intestine. The concept of 'junction' is used only when the point where two organs meet is an imaginary line such as the junction between Procorpus and Median bulb, which has no specific name in the literature.
A junction consists of three components: the term 'Junction', and
the
names of two structures. The textual representation has the
following
form, i.e., syntax:
'Junction' <structure-name1>/<structure-name2>
as shown in the example:
'Junction' Procorpus / Median bulb
A schema tool should automatically generate a name and common
synonyms
of the following form:
= Junction of the <structure-name1> and the
<structure-name2>
= Junction of the <structure-name2> and <structure-name1>
= <structure-name1> <structure-name2> junction
= <structure-name2> <structure-name1> junction
which for the example above would automatically generate:
'Junction' Procorpus / Median bulb
= Junction of the Procorpus and the Median bulb
= Junction of the Median bulb and the Procorpus
= Procorpus Median bulb junction
= Median bulb Procorpus junction
As in the case of perspectives, junctions would appear in the schema
tool at the user's/designer's option when viewing the hierarchy of
structures.
Within the database, junctions would be treated similarly to
perspectives
since junctions have properties as well. The name of the field in
the database would be 'Junction of the <structure-name1> and the
<structure-name2>'
2.3. Overlap
In some nematodes, some organs continue past other organs and such 'overlaps' have properties of their own, in particular a length. However, it would be wrong to describe them as separate substructures, because they are not individualized organs: an overlap is only a part of a well-individualized superstructure and it is recognized (and described) only by virtue of this superstructure being longer than usual. For example, oesophageal glands can stop short of the beginning of the intestine, or they can be longer and extending over the intestine in a 'glandular overlap' (Fig. 1).
In other cases, the overlap is formed when an organ is folded over itself. For example, in some nematode species, the very long ovary is folded and its end overlaps its beginning.
To avoid creating artificial structures, these types of situation
are
described as 'overlaps', for example:
Subventral oesophageal glands
'Overlap' Oesophageal glands/Intestine = Glandular overlap
Intestine
Ovary
Flexure of ovary
'Overlap' Ovary /Ovary
As shown in the above examples, when the word 'overlap' is actually used by the authors in published species descriptions, e.g., glandular overlap, the traditional expression is added as a synonym.
An overlap consists of three or more components: the term 'Overlap',
the names of the overlapping structures including the same name
repeated
when a structure overlaps itself, the name of the overlap if it has one
in the literature, and synonyms of the name if they exist. The
form
is:
'Overlap'
<structure-name1>/<structure-name2>
= <name>
=
<synonym 1>
=
...
=
<synonym n>.
where the <name> and the synonyms are specified by the
schema
designer. Another example of the syntax (including an existing
synonym)
in the schema is:
'Overlap'
Intestine/Anus
= Post-rectal sac
The schema tool will also automatically generate the following
synonyms
as well.
Overlap of the
<structure-name2>
by the <structure-name1>
as in:
Overlap of the Anus by the
Intestine
Unlike what we saw with Junctions, the order of the two structures is important as it would be meaningless to create a synonym such as Overlap of the Intestine by the Anus.
If a particular name, such as Post-rectal sac, is specified and used by the authors for an overlap, this name will be used in the database to represent an entity with properties. If no name exists in the literature, then the first synonym would serve as the name used in the database. In terms of displaying overlaps in the hierarchy of structures, overlaps would be handled in the same way as junctions and perspectives.
2.4. Groupings or aggregations
The nematode body presents another type of difficulty. Traditionally, authors describe a 'head', often bulbous and offset from the 'neck', the part of the body that continues to the end of the oesophagus, and a 'tail', the part of the body posterior to the anus. The part between the end of oesophagus and the anus, i.e., between the neck and the tail, does not have a proper name and can be named the 'body proper' (Fig. 5).
Thus, the external morphology includes the following structures and
substructures:
Body
Head
'Junction' Head/Neck
Neck
Body proper
Tail
(Note that the junctions between neck and body proper and between body
proper and tail are not included because, so far, no basic properties
have
been described for these by the authors. It would be easy to add
them should the need arise.)
However, the description of species in one of the families of
Tylenchida,
Heteroderidae, often separates the neck (with the head) from the rest
of
the body, i.e., Body proper + Tail (Fig. 5). Basic properties
such
as length or shape are described for these two body parts and there
must
be a structure to which these basic properties can be attached in the
database.
We preferred to include such aggregation of structures as 'groupings':
'Grouping' Head + Neck
'Grouping' Body proper + Tail = Body behind the neck = Spherical part
of body
Other groupings may be used some day in future descriptions, which would avoid adding to the list of structures.
Syntax in the database:
'Grouping' <structure-name-1> + ... +
<structure-name-m>
= <name>
= <synonym1>
= ...
= <synonym n>
The name and synonyms must be specified, as none will automatically be generated by the schema tool. Since a grouping has basic properties with states or values, and since a grouping is a relationship between two or more structures, it is converted to the entity name in the database using the specified <name>.
2.5. Structure contained in several superstructures
Some structures are not entirely contained inside or are not part of a single superstructure but they continue over several structures. One example is the oesophageal lumen, a tube that starts at the base of the stylet, continues through the various parts of the oesophagus and opens into the intestinal lumen (Fig. 1). In our early efforts, we described several structures, called oesophageal lumen in the procorpus, oesophageal lumen in the median bulb, oesophageal lumen in the isthmus, and oesophageal lumen in the oesophageal glands, but this splits what is actually a single entity into several artificial structures. Following our philosophy of keeping as close to reality as possible, we created instead another relationship called 'within' to be used as follows with the single structure 'oesophageal lumen':
Oesophagus = Pharynx = (Esophagus)
Lumen = Oesophageal lumen
'Within' Procorpus
'Within' Median bulb
'Within' Isthmus
'Within' Oesophageal glands
A partial substructure consists of three or more components: the
name of a structure, the term 'Within', and the names of the partially
containing superstructures as in:
<structure-name
1>
'Within' <structure-name 2>
...
'Within' <structure-name
n>
where the first-named structure (structure-name 1) is within the
second-named
ones (structure-names 2 to n). Since properties can be attached
to
the structure as it relates to its containing
superstructure, field names in the database would have to be generated
from each combination, i.e., the name in the database would have the
form:
<structure-name-1> within <structure-name-i>, i
=
2, ..., n,
as in:
Lumen within the Procorpus
Lumen within the Median bulb
Lumen within the Isthmus
Lumen within the Oesophageal glands
2.6. Views and the 'Also in' relationship
2.6.1. Multiple systems for one structure
In our previous versions of the hierarchy of structures, a structure
could be placed in multiple locations within the hierarchy. While
this is important for a user when viewing the hierarchy, we found it to
be simpler for the designer to have each structure located in a single
location in the hierarchy. Usually there is little difficulty in
determining this for most structures. In other cases, the
situation
is more ambiguous. For example, the caudal alae are cuticular
folds
that are used as male secondary sexual organs (Fig. 4, C). The
proper
system for caudal alae is not obvious as this structure could be placed
by some authors in Body envelopes and by others in Genital
system.
(The question here is not to decide who is right or wrong, but to
design
a system that can handle any entry by authors or users, even erroneous
entries.) In Table 1, this structure is placed under
Secondary
male sexual organs, in the male genital system (because its primary
physiological
function is reproduction), but some users may not realize this and look
for this structure under Cuticle, in Body envelopes. We then use
the relationship 'Also in' to indicate this secondary location:
Caudal alae ('Also in' Cuticle)
2.6.2. Muscular and glandular systems
The basic list of structures in our design does not include all of the systems generally considered in a plan of organization. For example, it does not have a glandular system, because glands are part of various physiological systems, particularly the digestive and genital systems. Also, the muscular system in the Genisys representation includes only the muscles that are connected with locomotion, i.e., the somatic muscles. There are other muscles (stylet muscles, digestive sphincters, vaginal and spicule muscles, etc.), but they belong to other systems (digestive system, genital system).
Still, users may want to look at all the glands or all the muscles that exist in nematodes. This can be achieved in a set of characters such as the Genisys representation by keeping each muscle and each gland within the system associated with its major physiological function (e.g., Vagina sphincter with Female genital system) while using the schema tool and database 'views' to display on demand the lists of structures comprising the glandular system or the muscular system. This avoids duplicating structures, which would occur if the list included, e.g., two copies of vagina sphincter, one in the female genital system and the other in the muscular system.
2.6.3. External morphology
As explained above, some structures in various internal systems open
to the outside and these openings must appear in the external
morphology,
under the corresponding body parts. 'Also in' relationships are
used
to support the 'external morphology' view. For example, the
phasmid
is a chemosensory organ that opens in the lateral field, generally on
the
tail or somewhere along the body (Fig. 4, B). As its primary
function
is sensory, it is placed in the nervous system, but a 'Also in'
relationship
links it to the lateral field. Interestingly, lateral fields also
are part of an internal system (under Cuticle in Body envelopes), but a
second 'Also in' makes them viewable with the external morphology:
Phasmids
Phasmid opening ('Also in' Lateral fields)
Lateral field ('Also in' Body)
'Also in' is different from the other relationships in the sense that it is just used by the schema tool to display the structure in another location(s) in the schema, but not to create virtual structure names such as Junction of the Procorpus and the Median bulb.
2.7. Conclusion
The various database relationships introduced here, perspectives, junction, within, etc., were derived from problems we encountered with a nematode database. Some of them will probably be useful with other biological groups. For instance, the structures of any living being can be viewed from different perspectives. Others may solve nematode-specific problems only. On the contrary, we can expect to find other types of problems when we deal with other biological groups. No matter what these problems will prove to be, we believe that they should never be solved by adding artificial structures to the lists, but by defining new types of relationships in the database. In other words, any biological problem is better solved in a systematic way using principles developed for biological databases, as we have done here and elsewhere.
3. Discussion
3.1. A uniform and representative decomposition
3.1.1. Uniformity
Uniformity was obtained in part by separating structures from basic properties and from states and values in our previous work (Diederich, 1997; Diederich et al, 1997; 1998). It was further enhanced here by representing artificial structures as relationships between real structures such as junctions and overlaps. There is very little uniformity in many existing character lists and there is very little guidance in achieving uniformity in systems that aid in the construction of character lists. As pointed out by Diederich (1997) it is not unusual to find non-uniform representations within character lists. Indeed, we are far from blameless in this respect, as can be seen in the character list proposed at the beginning of the Nemisys project (Fortuner, 1989, pages 353 seq.). Actually, it was the errors we noticed in this original schema that led to the development of the principles set forth in our other papers cited here. Comparison of the list in Fortuner (1989) with the latest version (Table 1) can be very instructive.
3.1.2. Representation
The list of structures proposed (Table 1) and discussed here is representative in the sense that every structure that is traditionally described in a nematode is present in the list, and it is listed as a structure in only one place, in what appears to be a natural position within the list. (The natural position of a structure is the position that is linked with its primary physiological function). All known structures have one preferred name (for the sake of uniformity), but all synonyms are included as well (to preserve freedom of the authors). We hope people will use the standard names but they are free to use any of the terms that have developed over decades in the thousands of descriptions that have been published. These terms are present in the list because all synonyms, even invalid ones (included parenthetically in the list), are supported.
Is there only one possible place for a structure? Is it in the proper system? Yes, if the structure has only one major physiological function and we know it. For example, the intestine can only be placed in the digestive system. If fifty nematologists were asked to place this structure in the hierarchy, they would find its correct place without prompting. The proper location of other structures such as Caudal alae is less clear, but the 'Also in' relationship can be used whenever there is some question on the proper location of a structure: it can be placed in one system and said to be 'also in' another system. We have seen above how this can be used for external morphology and for particular structures such as muscles and glands.
Naturally, we cannot claim that it is always clear where the best place within a hierarchy might be or to determine whether something is a true structure or not. We do not claim that our approach makes the task of creating a hierarchical list of structures simple. We only claim that it presents a set of concepts and principles by which much of the hierarchy can easily be formulated and difficult cases can easily be identified and be properly, though not necessarily easily, resolved.
As for new structures, in most cases they can be simply added to the schema. In some cases, some additional effort will be necessary to determine how to handle them, occasionally requiring developing a new conceptual basis for doing so, as we have done with junctions and the other concepts presented here.
3.2. Homology
In the present study, we propose a way to store existing knowledge in a database according to a sound set of principles. This process should start from the data proposed by the authors, either in past publications or, as will be possible in the future, by direct entry into a database when one is built according to the Genisys principles. However, in biology, facts are not always what they seem to be. This is particularly true with structures, whose very nature is sometime obscured by problems of lack of homology.
Homology is defined as similarity in the structures of two or more taxa that is due to inheritance from a common ancestor rather than due to independent acquisition (homoplasy). In the case of homoplasy, similar structures evolve either in unrelated organisms from different antecedent structures -- this is called convergence, e.g., the fins of whales and fish -- or independently in related taxa along the same development pathways -- this is called parallelism, e.g., the reduced post-uterine sac of females of various nematode families.
As defined, homology is not 'basic data', that is, data that comes straight from the observation of specimens. Conclusions about homology (or lack of it) are derived from other types of studies (embryology, development, histology, etc). The end product of such studies (homology vs. homoplasy) is better described as a relationship between basic data.
The most frequent and obvious case of homology is that which exists between two structures with the same name in two different species. Species 1 has stylet knobs; species 2 has stylet knobs; we will assume that stylet knobs in 1 are homologous to stylet knobs in 2. This relation is the default situation and it will not need to be represented explicitly. Complications arise in two cases:
1- When two structures of the same name are NOT homologous in separate taxa.
For example, the word 'mucro' is used to describe both a short projection at the tail end of many nematodes (Fig. 4, B) and a pointed structure in the oesophagus of some Xiphinema (a genus of the family Longidoridae), where it represents the relict of the replacement stylet (Fig. 4, D). The two structures have no relation whatsoever with one another and it is clear that they are not homologous.
2 - When two homologous structures have different names in different species.
The stylet is moved by muscles that are attached to a specialized part called 'swellings' in the sub-order Aphelenchina and 'knobs' in the sub-order Tylenchina (Fig. 1). Swellings and knobs are probably homologous.
Of course, the information about homologous and non-homologous characters must come from taxonomists, who are also the users of such data. The primary Genisys mission is the representation of existing data in a uniform manner that will make it easier to use them. At this moment, all we can say is that homology (or lack of homology) has the flavor of state-based relationships (Diederich, 1997), but, until we know how the system will use homology in the important contexts and how it would work in queries and updates in a Biological DBMS (Database Management System), we cannot be sure that this concept falls indeed in this category. This will be the subject of future studies within the Genisys project. For such studies, we would welcome collaboration with scientists who are familiar with the various uses of homology.
3.3. The future
The first attempt to establish a complete list of nematode structures for database purposes was prepared for the Nemisys project and published in Fortuner (1989) with about a hundred structures and several hundred characters (including properties and various possible states/values). During the course of the project, this list was expanded until it included several hundred structures and almost a thousand characters when properties were added to each structure. The present list includes less than half of the expanded set of structures and each structure is described by one or several out of 20 basic properties (Diederich, 1997). The list of characters had to grow tremendously before we could discern patterns and make some sense out of its complexity, which resulted in a list that is both more representative and more uniform. We are now reviewing the basic properties and states for Tylenchida and will add them to the list on our web page.
Although we used tylenchid nematodes as an example schema because the biologist of the Genisys team (RF) is a plant nematologist, the Genisys representation and the solutions we present here are usable for any biological group. We would welcome collaboration with specialists of other biological groups to develop new schemas outside nematology. Similar lists of structures can be created for other biological groups, both animal and vegetal. The process would parallel what was done in nematology, starting from the traditional plan of organization of the group and listing the structures according to the present guidelines.
In the example schema, we focused on taxonomic data because the team biologist specializes in nematode taxonomy and because taxonomic data, and particularly the list of structures presented here, can be used in other fields, as discussed below. These choices make the present list (Table 1) very easy to use 'as is' by nematode taxonomists. Authors of new nematode species can pick up from the list the structures to be described, add to each structure one or several basic properties (see list in Diederich, 1997) and describe the states or the values observed in the specimens of the species being described. The structures of tylenchid nematodes are quite well known by the specialists in this field, but so far they have never been properly organized according to a comprehensive and consistent set of design principles, one goal of our efforts. We hope that the present list is almost complete, at least for the structures typically described by taxonomists.
In case an author wants to describe a structure that is not in the list, this person is asked to contact the present authors for updating the list and web page. This situation is most likely to occur when specialists of other domains, such as physiologists or morphologists, start using the list for their own purposes. For example, a study on the ultrastructure of Hoplolaiminae (Mounport et al., 1993) needed structures (granular matrix; electron-dense ovoid structures) that are not on the list because they appear at a level below the usual level of taxonomic descriptions. (The reason is that such studies require special observation techniques, such as SEM, TEM or staining, and taxonomists usually don't go into such detail). The Genisys representation was created with this possible extension in mind.
We welcome collaboration with nematode physiologists/histologists to
enlarge the list of structures to the tissue/cell/cell component
level.
The current guidelines can be used for such an extension. Any new
situation that proves to be not amenable by the exiting guidelines will
be studied and new guidelines will be proposed. To be usable in
physiology
for example, the extended list of structures needs to be provided with
physiology-specific basic properties and states/values. The
implementation
of a list of physiological basic properties will require an entirely
new
effort, almost certainly including physiologists. It is still too
early to indicate the possible outcome of such an effort. We can
only predict that the present list of structures will be used.
References
Chen, P. (1976). The entity-relationship model: towards a unified view of data. ACM TODS 1: 9-36.
Date, C. J. (1995). An introduction to database systems (6th ed.), Reading, MA, USA, Addison-Wesley Pub. Co., xxiii + 839 pp.
Diederich, J. (1997). Basic properties for biological databases: character development and support. Journal of Mathematical and Computer Modeling 25 (10): 109-127.
Diederich, J., Fortuner, R. & Milton, J. (1997). Construction and integration of large character sets for nematode morpho-anatomical data. Fundamental and Applied Nematology 20 (5): 409-424.
Diederich, J., Fortuner, R. & Milton, J. (1998). A general structure for biological databases. In: Information Technology, Plant Pathology and Biodiversity (Ed. by Bridge, P., Jeffries, P. Morse, D.R. and Scott, P.R.), CAB International, Wallingford, UK: 47-58
Diederich, J. & Milton, J. (1989). NEMISYS, an expert system for nematode identification. In: Fortuner, R. (Ed.). Nematode identification and expert-system technology. New York, NY, USA, Plenum Publishing Corp., pp. 45-63.
Fortuner, R., 1989. Nematode identification and expert-system technology. New York, NY, USA, Plenum Publishing Corp., ix + 386 pp.
Mounport, D., Baujard, P. & Martiny, B. (1993). Observations on
the cuticle ultrastructure in the Hoplolaiminae (Nemata:
Hoplolaimidae).
Nematologica 39, 240-249.
List of illustrations (to be scanned and added later to this page)
Figure 1 — The anterior part of the digestive system: oesophagus and stylet. (Hoplolaimus clarissimus, Fortuner, 1974, courtesy Cahiers ORSTOM, série Biologie).
Figure 2 — The genital system. Ditylenchus myceliophagus in Fortuner, 1982, courtesy Revue de Nématologie.
Figure 3 — The lateral fields in lateral view (left) and in cross section (right). Pratylenchus zeae, Fortuner, 1976, courtesy of CAB International, CIH Descriptions of Plant-parasitic Nematodes.
Figure 4 — A: Tail of H. clarissimus female with post-rectal sac. B: Tail of Helicotylenchus dihystera female with terminal mucro. C: Tail of H. clarissimus male with caudal alae. D: The oesophageal mucro of Xiphinema longicaudatum. A, C, from Fortuner, 1974; B: from Fortuner, Merny & Roux, 1981; D : from Swart & Quénéhervé, 1998 ; courtesy of Cahiers ORSTOM, sér. Biologie, Revue de Nématologie and Fundamental and Applied Nematology.
Figure 5 — Body and body parts of a typical nematode (A: Helicotylenchus
dihystera) and five Meloidogyne females (B: Meloidogyne
mexicana).
A from Fortuner, 1981, courtesy Revue de Nématologie; B
from
Cid del Prado, 1991, courtesy Fundamental and Applied Nematology.