Behavioral vectors, pooled over protagonists of the same experiment. Ordering of orbits is as in the template vector (see text): from short orbits (left) towards long orbits (right). Only occurrence/nonoccurrence of orbits was considered (maximal five entries, corresponding to five experimental protagonists). (a) Mated female in the experiment with normal males (histogram display). (b) Females in immature, mature, mated states. Counts are coded from white (0) to black (5 counts). Bars corresponding to particular vector components/orbits have equal widths.
Similarity matrix evaluated for all individuals involved (light shadings indicate highly similar behaviors; diagonal values were set to 0.5). The plot shows strongly individual normal male behavior, and lack of individual behavior in all experiments involving fruitless mutations. Emerging patterns strongly correlate with group boundaries indicated by thin lines. They demonstrate strong group coherence of individual behaviors. Dark subdiagonal group-squares indicate enhanced behavioral distance between male/female protagonists of the same experiment.
Similarity matrix (density plot over , diagonal elements set to 0.5). Lighter shading indicates higher similarities; the self-similarities were set to zero. Circles indicate similarity maxima (red) and minima (blue). For the meaning of the indices, see Table II .
Shannon entropies calculated for the five courtship experiments. Each data point corresponds to the average obtained from five different Drosophila individuals. Involved protagonists, from left to right, bottom/top: (1) mature females/normal males; (2) immature females/normal males; (3) normal males/fruitless males; (4) mated females/normal males; (5) mature females/fruitless males. In experiment 3, the value obtained for normal males has been connected with the female data points, and that of fruitless males with those of the normal males. This avoids crossing of the male/female behavioral categories.
Behavioral discrimination potential determined by a mean-difference test (light shading indicates low discriminability), showing again largely individual behavioral variability on top of clear group coherence. Majority voting can be used to determine whether one class should be distinguishable from another.
Encoding scheme: 37 fundamental acts found in the time series are encoded into numbers, some of which are sex specific. For illustrations, see Ref. 17 .
Correspondence between indices and Drosophila behavioral vectors (see text).
Article metrics loading...
Full text loading...