European Journal of Neuroscience, Vol. 10, pp. 2964–2974, 1998 © European Neuroscience Association
Odour coding is bilaterally symmetrical in the antennal
lobes of honeybees (Apis mellifera)
C. Giovanni Galizia, Karl Na ¨gler, Bert Ho ¨ lldobler
1
and Randolf Menzel
Institut fu ¨ r Neurobiologie, Fachbereich Biologie, Freie Universita ¨ t Berlin, D-14195 Berlin, Germany
1
Theodor Boveri Institut der Universita ¨ t, Lehrstuhl fu ¨r Verhaltensphysiologie und Soziobiologie der Universita ¨ t (Zoologie II),
Am Hubland, D-97074 Wu ¨ rzburg, Germany
Keywords: development, insect, olfactory glomeruli, optical recording, statistical analysis
Abstract
The primary olfactory neuropil, the antennal lobe (AL) in insects, is organized in glomeruli. Glomerular activity
patterns are believed to represent the across-fibre pattern of the olfactory code. These patterns depend on an
organized innervation from the afferent receptor cells, and interconnections of local interneurons. It is unclear
how the complex organization of the AL is achieved ontogenetically. In this study, we measured the functional
activity patterns elicited by stimulation with odours in the right and the left AL of the same honeybee (Apis
mellifera) using optical imaging of the calcium-sensitive dye calcium green. We show here that these patterns
are bilaterally symmetrical (n = 25 bees). This symmetry holds true for all odours tested, irrespective of their role
as pheromones or as environmental odours, or whether they were pure substances or complex blends (n = 13
odours). Therefore, we exclude that activity dependent mechanisms local to one AL determine the functional
glomerular activity. This identity is genetically predetermined. Alternatively, if activity dependent processes are
involved, bilateral connections would have to shape symmetry, or, temporal constraints could lead to identical
patterns on both sides due to their common history of odour exposure.
Introduction
Information processing in the nervous system relies on a very complex
interaction between different neuron types. The connections that
underlie the computational architecture of the brain are partly genetic-
ally programmed, partly established in development via processes
such as activity dependent plasticity, and partly remain plastic during
the entire life span. Also activity dependent wiring can lead to highly
stereotyped functional maps. For example, in the mammalian lateral
geniculate nucleus (LGN), afferents from both eyes sort into separate
layers. This sorting is accomplished by local synaptic interactions in
the LGN, and is driven by spontaneous activity in the ganglion cells
of the retinae (Wong et al., 1993; Wong & Oakley, 1997). Nevertheless,
the resultant layering is highly predictable, and bilaterally symmetrical.
Further up the visual pathway, in mammals with binocular vision,
information from the left and right eye is organized in ocular
dominance columns, the layout of which is arranged in alternating
stripes. The spacing of these stripes is uniform, while the exact
pattern which they form is variable (Anderson et al., 1988). In fact,
modelling approaches rely on local, randomly distributed inhomogen-
eities to trigger the segregation into left-innervated and right-innerv-
ated areas for each particular map (Miller et al., 1989; Miller, 1994;
Obermayer et al., 1995). In other words, local, activity dependent
synaptic interactions lead to functional maps with predetermined
overall properties, but local variabilities. Therefore, these maps
are not bilaterally symmetrical. Nevertheless, in reversed occlusion
experiments in cats an intrinsic cortical component of the columnar
layout was found (Go ¨decke & Bonhoeffer, 1996).
The mechanisms involved in the interaction between genetically
Received 9 February 1998, revised 20 April 1998, accepted 21 April 1998
determined and/or activity modulated connections in sensory systems
can be well studied in the olfactory system and particularly precisely
in that of the bee brain. Olfactory receptor neurons project to the
olfactory antennal lobe (AL) in insects or the olfactory bulb in
mammals. These structures are organized in glomeruli, which are
believed to be the functional units in the first stage of central olfactory
processing. The glomerular activity patterns resulting from these
projections have been shown to be odour specific (Joerges et al.,
1997; Friedrich & Korsching, 1997). With the exception of locusts
(Hansson et al., 1996), single receptor neurons of adult animals
always branch in a single glomerulus (Boeckh & Tolbert, 1993).
Also, the projection patterns of receptor neurons all expressing the
same putative receptor protein have been shown to converge on a
particular glomerulus in mammals (Ressler et al., 1994; Vassar et al.,
1994). Consequently, it is the innervation pattern of receptor neurons
that creates a spatial map of the odour world accessible to the
individual animal (Mombaerts et al., 1996). However, it is still unclear
how this map is formed in development. Two extremes are possible:
each glomerulus could be predetermined for one (or more) particular
(i.e. genetically predetermined) receptor cell family, or an activity
dependent sorting mechanism could account for the specific innerva-
tion pattern.
An example of a genetically predetermined glomerular map is
seen in the macroglomerular complex devoted to sexual pheromone
processing in male lepidoptera (Hansson et al., 1992; Hildebrand,
1996). The separation between the main olfactory bulb and the
vomeronasal organ in mammals is also genetically determined (Dulac