Telling apart <I>Felidae</I> and <I>Ursidae</I> from the distribution of nucleotides in mitochondrial DNA

Rank--frequency distributions of nucleotide sequences in mitochondrial DNA are defined in a way analogous to the linguistic approach, with the highest-frequent nucleobase serving as a whitespace. For such sequences, entropy and mean length are calculated. These parameters are shown to discriminate the species of the <I>Felidae</I> (cats) and <I>Ursidae</I> (bears) families. From purely numerical values we are able to see in particular that giant pandas are bears while koalas are not. The observed linear relation between the parameters is explained using a simple probabilistic model. The approach based on the nonadditive generalization of the Bose-distribution is used to analyze the frequency spectra of the nucleotide sequences. In this case, the separation of families is not very sharp. Nevertheless, the distributions for <I>Felidae</I> have on average longer tails comparing to <I>Ursidae</I>.