GENERIC CRITERIA IN FLOWERING PLANTS 



321 



view of genetics and evolution but, by itself, is no argument for demolishing genera. It is 

 as if we said that unless one particular type of morphological change occurred in the 

 course of evolution we would not regard any other differences, however great, as justifying 

 the recognition of genera. 



The taxonomist can and should make use of genetical evidence to improve his classifi- 

 cation, bearing in mind that it is a general purpose one and not primari'ly a geneticist's 

 one. Hybridisation experiments in the Hordeeae support the separation of Hordeiymus 

 and throw light on the genus Agropyron, to take two examples of interest to British botanists. 

 Agropyron, as represented in this country includes two non-rhizomatous species and three 

 or perhaps four rhizomatous species. These two groups of species have been placed in 

 separate genera, Roegneria and Elytrigia, the latter differing from the former in having 

 spikelets which fall entire at maturity, instead of the glumes remaining attached to the 

 rhachis, larger anthers, deeply grooved caryopses and long rhizomes. Interspecific hybrids 

 though usually, if not invariably, sterile occur commonly within these genera, and species 

 of Elytrigia are known to cross with Elymus and Hordeum. There appear, however, to be 

 no records of hybrids between the British species of Roegneria and Elytrigia. These facts 

 suggest that the relationships between the two genera, though they are morphologically 

 similar, may not be particularly close. In this particular group inter-generic sterility appears 

 to be a character of positive value and brings the number of correlated characters available 

 for separating Roegneria and Elytrigia up to at least five. 



On the other hand when the intergeneric hybrid between Festuca pratensis and 

 Lolium perenne (known as X Festulolium loliaceum) and its parents are re-examined it is 

 found that the generic characters are all associated with the form of the inflorescence. In 

 Festuca this is a more or less branched panicle and the spikelets have two glumes. In 

 Lolium the inflorescence is typically a spike and, except in the terminal spikelet, one 

 glume is suppressed; however forms of L. perenne with branched inflorescences occur 

 not infrequently, and here the spikelets have two glumes. Such plants are difficult to 

 distinguish from Festuca except by the rather more strongly compressed spikelets. Here 

 the genetical evidence for a close relationship between Festuca pratensis and Lolium 

 perenne is supported by the morphological evidence, and the probability is that the generic 

 limits have been wrongly drawn. There seems to me to be no doubt that Lolium perenne 

 should be included in Festuca, leaving Lolium temulentum and allied annual species in 

 Lolium, though the whole problem requires a thorough re-investigation. 



The case of Ammophila and Calamagrostis is even clearer. Though the two genera 

 look different, the difference is due to the fact that Ammophila has short panicle branches 

 and large spikelets while Calamagrostis has long panicle branches and small spikelets ; the 

 details of spikelet structure are extraordinarily alike in both. There is nothing but a 

 well-established tradition to prevent us from following the genetical indications and 

 returning Ammophila to its old place as a species of Calamagrostis, and it is time we broke 

 from tradition here. 



To sum up, I would suggest that genera should be delimited by as many different 

 but correlated characters as possible and that careful consideration should be given to 

 the validity of genera where the number of characters involved is small, even though these 

 be reproductive and therefore traditionally respectable ones. In fact genera should be 

 recognisable at sight, at least in the great majority of cases. 



The characters which can be used in addition to those of the flower and fruit are 

 cytological, genetical and anatomical ones: geographical distribution and the duration of 

 the plant may also be useful in certain groups. These criteria are not of equal importance 

 in different families and an attempt should be made to assess their validity for the genera 

 being investigated. 



