ISOLATING MECHANISMS AND SPECIES FORMATION 



especial interest. Hollingshead, using the extensive materials of Babcock's 

 laboratory, has made many interspecific crosses in the genus Crepis, a 

 weed of cosmopolitan distribution. When C. capillaris is crossed to C. tec- 

 torum, the outcome depends upon the strain of the latter which is used. 

 The Fi from some strains is fully viable; from other strains, the Fi includes 

 fully viable plants and plants which die in the cotyledon stage in a ratio 

 of 1 : 1; while from some strains all of the progeny die in the cotyledon 

 stage. It appears, then, that tectorum has a gene which, in crosses to 

 capillaris, behaves as a dominant lethal. Every effort to find a phenotypic 

 effect of this gene in pure tectorum has failed, but the gene has also been 

 found to behave as a lethal in crosses with several other species of Crepis. 

 This gene evidently functions harmoniously with the tectorum genome, 

 although in an unknown way. But, in combination with the capillaris 

 genome, it is so disharmonious that lethality results. 



A similar, and perhaps more thoroughly understood, situation has been 

 discovered in tropical fishes by Bellamy. In the moonfish, Xiphophorus 

 maculatus, there is a dominant, sex-linked gene IV which causes an in- 

 crease in the amount of black pigment, with NN being darker than Isln 

 fishes. Only the recessive allele of this gene is found in the closely related 

 swordtail, X. helleri. The two species can, however, be crossed, and the 

 Fi is also fertile. If NN moonfish are used for this cross, then the resulting 

 hybrids will all be Nn, and will in all other respects combine one genome 

 from each of the parent species. In these Nn hybrids, the amount of black 

 pigment exceeds that of the NN moonfish parent. If the Fi hybrids are 

 now backcrossed to pure nn swordtails, the backcross generation should 

 have approximately three-fourths of its chromosomes derived from sword- 

 tail, and some of these will also be Nn. In these, the pigmented tissue 

 becomes tumorous, and so the gene may be regarded as a lethal. The gene 

 N, then, produces a normal color variant in pure moonfish; in the hybrids, 

 it produces a more extreme expression of the same character; but in a 

 genotype closer to pure swordtail, this same gene N produces a more 

 extreme effect, melanotic tumors, and has thus become an interspecific 

 lethal. In other words, a gene which is beneficial, or at least harmless, in 

 its proper genetic background may become lethal when placed in a dif- 

 ferent genetic situation. It is not known whether these data have any 

 relationship to the fact that pigmented growths of mammals are some- 

 times precursors of cancer. 



The problems of hybrid sterility are not fundamentally different from 

 those of interspecific sterility. In fact, hybrid sterility could be regarded 

 as a special form of interspecific sterility in which the defect is simply 

 delayed for one generation. But this delay makes it possible to get some- 

 what more insight into the genetic and cytological mechanisms which are 

 operative. When sterile hybrids are studied cytologically, much the most 

 common anomaly found is that the chromosomes fail to synapse. The 

 unsynapsed chromosomes will typically be distributed without division 

 to the two daughter cells at one of the maturation divisions, with the 

 result that the distribution of the chromosomes is random. All of the 

 chromosomes might go to one pole and none to the other, or equal num- 



279 



