In 2011, Bakker et al.[1] published a new overview of the Dutch Rosa taxa in this journal. To order the previous chaos in Rosa sect. Caninae, the authors used a highly artificial typological species concept based on a priori determined set of characters to classify the observed variation in entities, which they subsequently treated as species. However, the discriminating characters used in this treatment are the position and persistence of the sepals and the diameter of the style opening, which are inherited solely from the pollen donor.[10] Other characters, like the presence or absence of hairs and glands and the structure of the epidermal wax on the leaves, are solely inherited from the mother plant. Since genetic recombination probably doesn’t occur by passing on genetic material to the next generation, I conclude that the ‘male’ characters cannot be used to delimitate taxa in sect. Caninae: these characters are stolen and are only dependent on the characters of the pollen donor; consequently they may differ between offspring of the same mother plant originating from different pollen donors and from generation to generation. It is recommended to develop a species concept which does justice to the evolutionary relations within the genus. If genetic recombination doesn’t take place, as is indicated by recent studies, species should be based on characters that are inherited in the maternal line, just like in apomict genera as Rubus and Hieracium. Since genetic recombination probably doesn’t occur, intermediate forms within section Caninae cannot be considered to be hybrides.