First published online August 17, 2006
Journal of Experimental Biology 209, 3336-3344 (2006)
Published by The Company of Biologists 2006
doi: 10.1242/jeb.02364
Visual and tactile learning of ground structures in desert ants
Tobias Seidl and
Rüdiger Wehner*
University of Zurich, Institute of Zoology/Neurobiology,
Winterthurerstrasse 190, 8057 Zurich, Switzerland
View larger version (11K):
[in a new window]
|
Fig. 1. Schematic view of the experimental setup. Training took place in a 10 m
channel with an exit hole on the side, where ants were able to enter and leave
the setup and forage to a feeder 9 m down the channel. The coloured
rectangular area denotes the position of the ground landmark during training
(PI-position). For tests, the ants were transferred into another channel 18 m
long and aligned parallel to the training channel. Within this test channel
ants were presented with a ground landmark at varying positions (LM-position)
relative to the position indicated by their path integrator (PI-position,
experiment 1), or with ground landmarks differing from the training landmark
in visual or tactile properties (experiment 2). The first six U-turns of the
ant's search behaviour were recorded. Drawings are not to scale.
|
|
View larger version (13K):
[in a new window]
|
Fig. 2. Schematic representation of the position of the nest (N) as defined by the
path integrator (PI-position, open arrowhead) and as defined by the ground
landmark (LM-position, filled arrow) in the different sub-sets of experiment
1. During training both positions coincided, but in the test situations the
LM-position was usually shifted away from the PI-position towards the point of
release (R). This decoupling of PI-position and LM-position ensured that
within the test channel the ants encounter the landmark before they have run
off their home vector, i.e. before they have reached the PI-position. F and
open square, feeder in the training channel; filled square, point of release
in the test channel; N and filled circle, nest in the training channel; heavy
bar, ground landmark.
|
|
View larger version (21K):
[in a new window]
|
Fig. 3. Search density distribution of the ants' nest search behaviour exhibited
under different training and test conditions. Control 1: ants were trained and
tested without any landmark. Control 2: ants were trained with a landmark
located at the nest entrance but tested without one. Test 1A: ants were
trained and tested with a landmark at the nest entrance. Black square,
position of landmark; dotted line, position of path-integrator-defined
nest.
|
|
View larger version (25K):
[in a new window]
|
Fig. 4. (A) Search density distribution and (B) differential search density
distribution of the ants' nest search behaviour exhibited under different test
conditions. All ants were trained with a landmark (black square) directly by
the nest and tested with a displaced landmark (coloured square). Test 1A:
LM-position was identical to the one in the training situation: 9 m (black
line and square). Test 1B: LM-position, 7.5 m (red line and square). Test 1C:
LM-position, 6 m (blue line and square). Test 1D: LM-position, 2 m (green line
and square). Control 2: no landmark present during test. Note: in B the data
of control 2 (in A) were subtracted from each of the other data sets for
display reasons.
|
|
View larger version (15K):
[in a new window]
|
Fig. 5. Search density distribution of ants that faced a landmark for the first
time (test 1E). Ants were trained without a landmark, but were presented with
a landmark (LM-position: 7.5 m) in the test situation.
|
|
View larger version (16K):
[in a new window]
|
Fig. 6. Visual properties of the landmarks used during experiment 2. The optical
properties have been determined following Kollmeier
(Kollmeier, 2005 ) by measuring
the remission properties under natural (sunlight) conditions at the specific
wavelengths of the ant's light receptors (absorption maxima: green 500 nm,
bandwidth 90 nm; UV 350 nm, bandwidth, 60 nm). The optical properties of the
landmarks are predominantly defined by the paint used and not by the surface
roughness.
|
|
View larger version (30K):
[in a new window]
|
Fig. 7. Search density distribution of ants that faced different types of landmark
stimuli. Ants were trained with a black and rough landmark at the nest
entrance (black square) and later tested with a landmark of different
properties defining the nest entrance at 7.5 m (red square, LM-position). Test
2A (shaded area): Control experiment with a black and rough landmark (same
experimental situation as test 1B, but new data set). Test 2B (thin solid
line): a grey and rough landmark deprived the ant of the visual contrast. Test
2C (broken line): a black and smooth landmark changed the surface roughness,
but left the visual contrast intact (as compared to the training situation).
Test 2D (dotted line): the lower hemispheres of the ants' eyes were covered
with lighttight paint depriving the ant from any visual cues from below.
|
|
View larger version (30K):
[in a new window]
|
Fig. 8. Ants perceive a ground landmark only from a short distance. The angle of
vision under which the ground landmark (length, 1 m; width, 0.07 m) appears in
the ant's visual field depends strongly on the distance of the ant from the
landmark (shown here for distances <0.5 m). The height of the eye above
ground (2-10 mm) has a minor effect. (Inset) Vertical expansion of the ground
landmark used in the current experiments within the ant's visual field at
different distances of the ant from the landmark (eye 4 mm above ground). The
landmark remains extremely small (<1°) up to an approach of about 20
cm. Then it rapidly expands covering a large part of the ant's ventral visual
field.
|
|
View larger version (34K):
[in a new window]
|
Fig. 9. (A) Typical surface profiles of the structures used in experiment 2,
determined using a contact-profilometer. The roughness is characterized by
Ra, defined as the arithmetic mean of the deviations from the base
line measured over an evaluation length (DIN EN ISO 4287). The mean diameter
of the grains deposited on the abrasive paper used for experiments 2A, 2B and
2D (compare Table 1) is 270
µm, the sand glued to the channel ground had a mean diameter of 200 µm.
In contrast to the abrasive paper the sand grains are deposited next to each
other without gaps. (B) Dorsal view of a tarsus of Cataglyphis fortis
(ar, arolium; cl, tarsal claws). Tarsal claws are separated from each other by
approximately 320 µm. (SEM micrograph courtesy of Andrew Martin.)
|
|
© The Company of Biologists Ltd 2006
