The Inquiry Process: The Example

Animal behavior can be studied at several levels of analysis (Lehner, 1996, Hinde 1970).  The most fundamental objective must be the initial description and cataloging of the behaviors (making an ethogram).  Once a well defined ethogram has been established, quantitative and statistical techniques can be used to analyze the temporal sequences and frequency distribution of the behaviors.  Finally, the scientific method can be used to manipulate experimental variables to produce an understanding of the mechanisms controlling the expression of the behavior.

In this experiment the grooming behavior of the fly Drosophila melanogaster will be studied at three levels of analysis.  First, an ethogram of the behavior will be constructed through direct observation of the flies.  Second, statistical analysis including Markov Chain Analysis and Chi Square testing will be used to characterize the temporal sequences and patterning of  grooming behavior.  Finally, the use of surgical and genetic manipulations will be used to test a hypothesis concerning central nervous system control of this behavior.

The Model
Table of Contents

The grooming behavior of insects, including flies, has been studied by several authors (refs).  Grooming in these insects is a stereotypic behavior characterized by a predictable sequence of behaviors which often occur in cyclical bouts (ref).  Several authors have suggested that the patterns in stereotypic behavior such as grooming are strongly controlled by the nervous system.  External stimuli may initiate, terminate, or influence the intensity of the behavior, but not the overall patterns of the behavior (refs).  Therefore only experimental manipulations which effect the nervous system will alter the patterns and sequences of the behavior.

In this experiment, the grooming behavior of Drosophila melanogaster will be studied.  Comparisons will be made among three groups.  The groups will include normal wild type flies, normal wild type flies with their wings removed, and wingless flies with the mutation "vestigial wings".  The hypothesis is the patterns of grooming behavior will not be altered by removal of the wings since the neurons of the central nervous system controlling grooming remain intact.  An additional hypothesis is the patterns of grooming behavior will be altered by the vestigial mutation due to developmental consequences of this mutation.

Top of Page
The Model
Table of Contents

This example uses the fruit fly Drosophila melanogaster.  This species is easy to obtain and care for.  Its wide spread use in genetic studies opens the possibility of using various mutant strains in behavior experiments.  Materials need to maintain stocks of Drosophila are readily available from biological supply companies.  When using an animal as small as Drosophila, some consideration of the equipment needed for observing  the specimen and recording data.  In this case the authors used dissecting microscopes to observe the flies in small closed containers.  Data was recorded by the observer with a audio tape recorder as the fly groomed.  The behavior was also recorded using a VCR connected to the microscope via a "flex cam".

Simply finding the best methods for observing and recording the animal's behavior can be a valuable experience for the students.  Each species will present its own particular responses to handling.  In the case of Drosophila, the flies responded best when kept in containers which were at least a centimeter in diameter.  Flies in smaller containers spent most of their time attempting to escape.  Eliciting grooming behavior was also found to be enhanced when ether was used to disable the flies during handling.
 
At total of fifteen different grooming behaviors were observed.  Although these are presented as separate and distinct behaviors, there are instances where the fly appears to be producing intermediate forms of the behavior.  The process of deciding what behaviors constitute grooming and distinguishing among these various behaviors is an important lesson for students.  Producing consistent and reliable descriptions of behavior requires careful observation and discussion among the participating students. See Dawkins, 1976, and Armstrong, 1985 for similar studies.

• Front (F):  Using the two prothoracic (front) legs together in a back and forth direction.
• Head (H):  Using one or both (most common) prothoracic legs to groom parts of the head other than the mouth parts    (grooming the eyes and antennae).
• Tongue (T):  Using the two prothoracic legs together to groom the mouth parts.
• Rear (R):  Similar to Front except the two metathoracic (rear) legs are used.
• Abdomen (A):  Using one or both (most common) metathoracic legs to groom parts of the abdomen.
• Wing Under:  Using one (ULeft or URight) or both (UBoth) of the metathoracic legs to groom the ventral surface of the         wings. This was often accompanied by a depression of the abdomen.
• Wing Over (OLeft, ORight, OBoth):  Similar to Wing Under except the dorsal surface of the wing is groomed .
• Middle Leg with Front Legs:  The left (FLeft) or right (FRight) metathoracic (middle) leg groomed by the front legs.
• Middle Leg with Rear Legs (RLeft, RRight):  Similar to the previous behavior except the rear legs are used.

Fly Grooming Front Legs and Head.

Fly Grooming Abdomen, Wings, and Rear Legs.

Wingless Fly Showing Grooming Behavior of the Missing Wings.

Top of Page
The Model
Table of Contents

Once the ethogram was constructed, additional observations were taken to record the sequence in which these behaviors followed one an other.  A typical sequence for a normal wild type Drosophila is given below.  This is not a complete sequence.  A complete sequence can include several hundred observations if the fly is grooming intently.

F-H-F-H-F-H-F-H-F-H-F-H-F-H-(stop)F-H-F-(stop)-F-T-FR-F-(stop)-F-H-F-H-R-A-R-A-UL-R-UL-R-UL-R-UL-R-
UL-R-UL-R-UL-R-UL-R-OL-R-A-R--UL-R-UL-R-------
 
Sequences of data were recorded for two normal wild type flies.  A transition matrix was constructed for both using the following procedure.  The example given below is a matrix for the two normal flies combined.  This was done because the matrix for both flies were very similar, and pooling data improves the quality of the statistical analysis done later.
 

• Set up a table with as many rows and columns as there are separate behaviors. In this case the transition matrix has 15     rows and columns.
• Label the rows and columns with the corresponding behaviors. Label the top (columns) of the matrix as the preceding      behaviors and the side of the matrix (rows) as the following  behaviors.

Count the number of times each behavior in the preceding columns was followed by the remaining behaviors in the      following rows.  For example, count the number of times Front was followed by Head and enter that frequency in column F row H (340 transitions).
•  Traditionally this type of analysis does not include the possibility of a behavior following itself (i.e.. F-F).  This is       considered to be a single occurrence of F.

• The individual behaviors do not follow one another randomly.  Some behaviors form common pairs in sequences. (e.g. F and H, R and A)
• Grooming involving the front legs only (F,H, T, FL, FR) are almost exclusively independent of behaviors involving the rear legs (R ,A, wing grooming elements).  There are very few transitions from the front group to the rear parts of the body.
• Grooming the wings and grooming the rear legs and abdomen separate into clusters, but are connected by frequent transitions (i.e. Rear to Under wing and Abdomen to Under wing).

The transition matrix table above shows that all of the cells have a significant Chi-Square value.  Closer examination reveals that the Front Legs transition to Front Legs much more than expected and much less to the Wings or Rear Legs than expected.  The Wings transition significantly more than expected to Wings and Rear Legs, and significantly less to Front Legs.  Similarly, the Rear Legs have significantly high transitions to Rear Legs and Wings, but a significantly reduced transition to Front Legs.

In other words, when a fly is grooming the front part of its body, it tends to alternate among the various behaviors associated with the front of the animal (F, H, T, FL, FR) and avoids shifting grooming to the wings or rear part of the body.  Grooming of the remaining parts of the body is more complex.  When grooming the wings, the fly will tend to continue grooming the wings or it will shift to the rear elements.  In the same way, when grooming the rear elements, the fly will tend to continue grooming the rear or shift to the wings.

If a Chi-Square analysis is made over the entire transition matrix the following patterns emerge:

• Grooming the Front legs is usually followed by grooming the Head and to a lesser degree by the Tongue or Middle legs. This can be represented by the code: F --- H (T, FL, FR).
• H --- F (T, FL, FR) In other words, the fly spends time grooming the front part of its body by alternating between the front legs and the head and occasionally grooming the tongue or the middle legs.
• R --- A or Under wing (Over wing, RL, RR)
• A --- R or Under wing (Over wing, RL, RR) In other words the fly spends time grooming the rear part of its body by alternating between the rear legs and the abdomen or the underside of the wings.
• Wings --- R (other wing, RL, RR, A)

Top of Page
The Model
Table of Contents

 There are several experimental approaches that can be used with Drosophila.  Students familiar with this animal from their study of genetics may suggest mutants which have altered body parts involved in grooming (Vestigial wing, Apterous wing, Antennapedia).  Two experimental manipulations were done in this exercise.  The first was to remove the wings from normal wild type flies and observe the effect on grooming behavior.  The hypothesis was the grooming behavior would not change since the nervous system controlling the behavior was still intact.

The results of the removing wings supported the hypothesis.  These flies showed a transition matrix very similar to normal flies.  The behaviors associated with grooming the wings (FR, FL, RR, and RL) were still present.  The quantity of data was too small to make efficient use of the Chi-Square analysis and therefore a statistical analysis is not included with these results.

The second experiment used flies with the mutation Vestigial wings.  These flies have a very small non-functional wing.  The hypothesis was the behaviors associated with grooming the wings would be absent due to possible developmental effects on the nervous system controlling the behavior.  Unfortunately the behavior of these flies was dramatically different from the wild type flies.  They appeared to be in a state of hyper-activity and showed very little grooming behavior.  Some isolated wing grooming behaviors may have been observed, but they were so infrequent reliable conclusions could not be made.

Top of Page
The Model
Table of Contents

There are many different avenues which can be explored as alternatives or extensions to this exercise.  A few are listed here.
 
Using Drosophila:

• Using different mutants in Drosophila such a antennapedia, aperterous, etc.
• Surgically removing different body parts such as the legs.
• Observing flies in different positions.  Drosophila will often walk upside down on the cover of the observation container. The effect of gravity in this case may have interesting consequences for grooming since the fly must take one or more legs off the surface.
• Looking at individual variations in grooming behavior.  In this experiment, the two normal flies had a strong tendency to groom the left wing more than the right.  Other variations may be observed.

• There are many other arthropods which have different leg morphology (grass hoppers, spiders, mosquitos, etc.) which may show alterations in grooming to accommodate differences in anatomy.
• There are arthropods which live in different habitats which may require changes in grooming behavior (water striders, spiders on webs, water beetles)

• Most students will realize their pet cats, dogs, and bird groom themselves.  This could be an ideal "at home" laboratory exercise for students.
• Farm animals and zoo animals are an opportunity to send students out to observe these animals more closely than they might otherwise be inclined to do.

• There will be a lot of interest in this possibility.  Using humans as subjects can be accomplished in several settings.
• The cafeteria
• The mall
• The prom
• Some discussion concerning the ethics and etiquette of doing behavior observations on humans should be included in any project such as this.

Grooming behavior is only one of several interesting behavior that can be studied with Drosophila, other arthropods,        vertebrates, and humans.  Feeding and drinking behavior, general circadian activity, and reproductive behaviors will be among those students will be interested in discussing.  The analysis tools used in this experiment can be used to study social behaviors as well.  Rather than recording preceding and following behaviors, record the behavior of one animal and the response of a second animal.

Top of Page
The Model
Table of Contents

• Standard laboratory report at the end of the exercise.
• While students are working their individual participation will be assessed according to rubrics given to students.
• Students will be required to find another animal and behavior to study and produce an ethogram for the behavior.
• Students will be required to read and critique a book  or article of a animal behaviorist (Von Frisch, Tinbergen, Hinde, Lorenz) and relate their work with Drosophila to the book/article.