Ant in a Petri Dish

In a laboratory experiment I recorded the spatial movements of an ant. The recording needed to be very effective with the smallest amount of technology possible. The following setup was then created (figure 6.3. The ant was put on the backside of a petri dish. To prevent the ant from escaping the petri dish was put in the middle of a tank filled with water, so that the surface of the petri dish and the water were at the same level. The water also prevented the heat from the overhead projector used to affect the ant. The projector made it possible to digitize the ant's movements directly on a digitizer tablet, to which the installation was projected.

Figure 6.3: Instrumental setup for the data collection of the ant. 1. overhead projector, 2. tank filled with water to keep the heat from the animal and to prevent the ant from escaping, 3. petri dish, 4. ant, 5. digitizer tablet, 6. digitizer mouse.

With this setup, four runs of five minutes each with pauses of five minutes in between were recorded. Locations were recorded every second. In the first two runs, only the animal was situated on the petri dish. After another pause of five minutes, a small piece of a rubber cannula was put onto the petri dish, and immediately after that the third run lasting another five minutes was recorded. After the third pause, a piece of sugar was offered to the ant and left on the petri dish, immediately followed by the recording of the last run. After this procedure the ant was immediately returned to the place of its capture. The observations after its release gave no indications of abnormal behavior.

In figure 6.4 a T-y plot is shown for the first run. Three things can be noted. The first is that the animal was almost always moving. The second thing is that the animal made a longer rest after about three thirds of the time, marked with a green circle in figure 6.4. The third and probably most interesting thing noticeable are the turn-points of the ant, i.e. the time-points at which the animal was changing its direction by a sharp turn. They are indicated by red bullets in the figure. More precisely it can be seen that in the first half of the run there was a tendency to show more turn-points than in the second half.

Figure 6.4: T-y plot for the ant in phase 1. Turn-points are marked with red bullets, the other event mentioned in the text is marked green.

Figure 6.5: T-$\nu$ plot for the ant in phase 1.

In figure 6.5 a T-$\nu$ plot indicates the speed at which the animal was running in phase 1. The speed of the ant is decreasing during time (Regression values corrected for temporal autocorrelation (data when the ant was standing still excluded): r²=0.23, p=0.041).

As a last T-plot with a single time axis a T-$\sigma$ plot is illustrated in figure 6.6.

Figure 6.6: T-$\sigma$ plot for the ant in in the middle of the observation period in phase 1. Time-frame width is set to 1 minute and 15 seconds.

In this plot the standard deviation for the locations within a moving temporal data frame of 1 minute and 15 seconds is displayed. The period in which the ant stayed for some time at one location can be easily depicted in the last quarter of the plot. Aside from that several times there are periods with a minimum in the standard deviation. They coincide with the turn-points discovered in figure 6.4 with few exceptions. These exceptions are probably additional turn-points which have not been identified before due to fact that the T-x plot was used as a static plot. When used in the dynamic form as a T-r plot, in which the direction of the spatial axis for which the plot is drawn can be changed interactively, they would have been easily recognised. This is again a restriction based on the static characteristics that printed graphics are confronted with. It is not an ideal way to illustrate dynamic interactive processes.

After these relatively simple plots I will now provide the TT-$\delta$ and TT-$\pi$ plots for the four runs. The plots are illustrated in figures 6.7-6.14.

Figure 6.7: TT-$\delta$ plot for the ant in the petri dish phase 1 (undisturbed). Short distances are indicated in blue, medium ones in green and large distances have a yellow to red color.

Figure 6.8: TT-$\delta$ plot for the ant in the petri dish phase 2 (undisturbed).

Figure 6.9: TT-$\delta$ plot for the ant in the petri dish phase 3 (rubber cannula).

Figure 6.10: TT-$\delta$ plot for the ant in the petri dish phase 4 (sugar).

Figure 6.11: TT-$\pi$ plot for the ant in the petri dish phase 1 (undisturbed). Blue indicates parallel, green intermediate, red anti-parallel.

Figure 6.12: TT-$\pi$ plot for the ant in the petri dish phase 2 (undisturbed).

Figure 6.13: TT-$\pi$ plot for the ant in the petri dish phase 3 (rubber cannula).

Figure 6.14: TT-$\pi$ plot for the ant in the petri dish phase 4 (sugar).

Let us first have a look at the TT-$\delta$ plots for the first two 'undisturbed' runs (figures 6.7 and 6.8). The first characteristic we can see is that almost all of the plot is covered with a more or less regular pattern of parallel and antiparallel blue lines in respect to the base diagonal line. They indicate that the ant is continuously running on the same path (in this case along the edge of the petri dish), sometimes changing the direction in which the path is run. This is a very stereotype movement pattern. The second thing noted is the blue square in the upper right quarter of the plot. This shows a rest taken by the animal. The animal has passed this location several times before, and passed it several times afterwards, as it is indicated by the numerous blue horizontal lines to the left and right (and of course below and above) the blue square. In the second run (figure 6.8) these regularities persist, but there are at least two things that are different from the first run. The first is that in the middle of the run the animal continued running in the same direction for a long time. After that a second difference can be noted when the animal changed its behavior. It then ran for five sequences in a row three times in one direction, and then three times in the other direction (see table 5.2g).

When a rubber cannula was put onto the petri dish at the beginning of run three (figure 6.9) this 'pattern' changed a little. It can now be seen that the ant stayed at several occasions at a location. By examining the TT-$\delta$ plot in more detail, it can be found that there were six places where the animal was spending a longer time. It was expected that the animal stayed for longer times at the location where the rubber cannula was placed. This became true, but additionaly the ant also stayed at other locations more frequently than in the previous runs. As the rubber cannula was placed in the center of the petri dish and the 'long runs' of the ant were taken place at the edge of the petri dish, it can easily be determined, which rests were located at the cannula place. They are the blue squares that have no blue bands to the left or right in the plot, as they were not passed by the animal while running around the edges of the petri dish. Astonishingly the longest two resting periods have not taken place at the cannula, but somewhere on the edge of the petri dish, both of them at the same location.

In the last run, a piece of sugar was put onto the petri dish, at some distance to the rubber cannula and the edge of the petri dish. As can be seen in figure 6.10, the movements of the ant changed dramatically. Although the regularities known from the previous runs are still visible in parts of the TT-plot, the animal used a significant time spending on one location, the piece of sugar and at other locations. The most prominent feature in this plot is the large blue square located in the center. I would like to examine that period to greater detail using the data of the TT-$\pi$ plot shown in figure 6.14. First I generated the histogram of the values representing the parallelity of the movements (figure 6.15 for the whole TT-$\pi$ plot. These values seem to be distributed evenly over the whole range from 0 to $\pi$. In a second step I created a histogram for the values in the middle of the TT-$\pi$ plot (the large blue square in figure 6.10). This is shown in figure 6.16. Now a very distinct regularity appears. Five angles of parallelity show high values. These are: 0, $\pi/4$, $\pi/2$, $3/4\pi$ and $\pi$. Three of these values can be easily explained by the edges of the piece of sugar, which had the approximate shape of a square. The other two indicate either that the animal could see the edge lying on the diagonal of the sugar, or that it had constructed a mental map of the place knowing that it needed to run in that specific direction to get to the other corner. This could be further investigated by further subdividing this period in smaller time-frames. It is the intent of this thesis to develop new methods and not to produce biological results on animal behavior. Therefore, I will now leave this example of an ant's (techniquely limited) spatial movements.

Figure 6.15: Histogram of parallelity values for the TT-$\pi$ plot for the whole phase 4.

Figure 6.16: Histogram of parallelity values for the TT-$\pi$ plot for the time with little movement in the middle of phase 4.