V. GRANT 47 



Criteria of Reproductive Isolation 



Relatively few groups of organisms have been studied from 

 enough points of view with sufficient thoroughness to reveal their 

 whole pattern of species-separating mechanisms. In most groups 

 the presence and distribution of isolating mechanisms will have 

 to be inferred from indirect evidence of various sorts. Among the 

 most important of these are the finding of discontinuities in the 

 variation pattern, marginal overlapping of distribution areas, com- 

 plete sympatry, or hybrid sterility in the laboratory or experi- 

 mental garden. 



If these criteria of reproductive isolation are consistent among 

 themselves in any given group, there will be no species problem. 

 It may happen, however, that one or more criteria will fail to 

 coincide with the rest. Under these circumstances an automatic 

 application of any single criterion in drawing species boundaries 

 might easily lead to a false picture of the natural species. The in- 

 consistency indicates that a critical appraisal of the criteria them- 

 selves is always needed. 



A reproductively isolated cross-fertilizing population should 

 reveal its genetic separation from other such populations by a 

 gap in the pattern of variation. Some plant species in every genus 

 do show the expected morphological discontinuity, but others for 

 various reasons do not. If applied uncritically, the criterion of 

 discontinuity would often result in treating as species populations 

 belonging to other categories. In a polytypic species with a dis- 

 junct distribution, the subspecies may be separated by morpho- 

 logical gaps, i.e., Gilia leptantha. The morphologically discrete 

 units in Agoseris seem to be the species groups; in Ceanothus 

 they frequently correspond to the sections; in the California oaks 

 one of the discrete entities is a subgenus; and in Allophyllum it 

 is the entire genus, which has, in fact, until recently been treated 

 as a single species. 



The overlapping range of variation of three biologically well 

 isolated species of Gilia is shown in Fig. 1. Each symbol repre- 

 sents a population. The coordinates give the measurements of the 

 two most important diagnostic characters for these three species, 



