American Naturalist 65: 144-148 (1931)
Internal Factors Affecting Discontinuity Between Species
Dr. Edgar Anderson

Washington University
Missouri Botanical Garden
1Read at the joint discussion on ''The Species Concept'' at the Fifth International Botanical Congress, Cambridge. August, 1930.

MAY I preface my remarks by repeating my conviction that the species problem is a problem and that as one of the fundamental problems of biology it is worthy of study as an end in itself and not as a mere corollary to work in ecology, systematics or genetics? Not until a fairly large number of species have been carefully and methodically studied over their entire range shall we be able to discuss the problem intelligently.

Aside from the work on cultivated plants by Vavilov and his associates, there are not even complete morphological surveys of any of the higher plants. For what species can the following simple questions be answered: What is its actual distribution? Does it grow in different situations in different parts of its range? What, for some easily measured character such as height or flower size, are the largest and smallest values for the species, what the average for the species as a whole? Do the values for different localities depart significantly from the mean of the species? Over what part of its range is it the most variable, over what part the least? Are there distinctive color forms and if so what is their comparative frequency in different localities! Are there any geographical regularities in the distribution of such conspicuous characters as hairiness, prostrate habit, etc., and if so, are they paralleled in other species having approximately the same distribution!

For the past six years I have been attempting to gather data which would answer just such questions. I have been studying a few species of Iris, the genus Aquilegia and several species of Aster. While the work is far from complete it has progressed far enough to convince me that species (though they are fundamental biological units) are of quite a different nature in these three genera. By this I mean that the morphological resemblances and genetic connection between the individuals which make up a species are widely different in these three groups of plants.

Lacking the precise information which would answer the question, what differences between species in these three genera may we logically expect on purely a priori assumptions, admitting for the purpose of the argument only that species are recognizable groups, the conclusion is unavoidable that they will, as units, be greatly affected by any factor which affects the degree of their isolation. Roughly we can classify these factors as internal and external. There must be many factors common to all Irises or to all Aquilegias which affect the nature and the degree of the isolation between species in each genus.

In the case of the genus Iris I think we may profitably consider not only that genus but practically the whole of the Liliiflorae (the Iris, Amaryllis and Lily families) since they are a natural group with many features in common. In the first place, polyploid series have been reported for a number of genera in these three families. That is to say that within many genera in this group (and sometimes even within the same species) forms occur whose chromosome numbers are simple multiple series. This will have a profound effect upon the relations between species.

To take a hypothetical example which differs in no essential detail from cases already reported in cytological literature, let XX and YY represent two diploid species. ZZZZ a tetraploid species, one letter being used to represent a haploid chromosome complement. If XX and YY, for instance, were species with twenty-four chromosomes (2n), ZZZZ would have forty-eight and each letter would represent a group of twelve chromosomes. XX and ZZZZ hybridizing produce a sterile triploid XZZ, which by doubling gives rise to the hexaploid XXZZZZ. By analogy with such cases as are known we would expect the hexaploid to be fertile and true-breeding. These relationships are represented in Fig. 1, A. the broken lines indicating the genetic relationships of the species involved.

Fig. 1. Interspecific relationships in A (the Liliiflorae) and B (the genus Aquilegia).

It is so commonly assumed in monographic work that morphological resemblance is a true guide to phylogenetic origin that it may be worth while to call attention to their differences in this particular example. The broad bands indicate the morphological resemblance to be expected on the assumption that it would be proportional to the number of sets of chromosomes in common. The example is a relatively simple one. As a matter of fact much more intricate interspecific relationships might very conceivably occur. Phylogenetic relationships and morphological resemblances in such groups will be reticular rather than dendritic. That is to say that the course of evolution, instead of being represented as a tree with diverging branches, may be more aptly likened to a complex and irregular web, with threads of varying thickness.

In the second place, the Liliiflorae are distinguished by highly developed vegetative propagation, as for instance in such genera as Tulipa, Lilium, Narcissus and Iris. Hybrids so sterile that in other groups of plants they could not survive for more than a season or two may, by vegetative propagation, live for centuries and even spread across continents. There is therefore, among the, Lilliflorae, the opportunity, on the one hand, for the production of complex, intergrading forms, and, on the other. for their survival and propagation. Newton and Darlington (1929) have shown that certain common tulips are sterile polyploid clones. Many of the bearded irises (Pogoniris) are sterile, and cytological examination proves them to be complex polyploids. Hybrids between I. virginica and I. versicolor are practically sterile, yet in certain localities, by vegetative propagation, they occur to the exclusion of the parent species.

In Aquilegia, on the other hand, there is little or no asexual propagation. It is difficult, even in an experimental garden, to keep a particular clone alive for more than four or five years. Sterile hybrids will therefore be of no importance. There is furthermore no poiyploidy in the group, either among the wild species or the numerous garden forms which have been investigated. True-breeding hybrids can not therefore be produced. and only relatively homozygous hybrid segregates will perpetuate their kind.

Fig. 1, B, continuing the symbolism of the previous example, represents the results of a species cross in Aquilegia. The first generation hybrid, XY, being practically fully fertile, a flood of segregating forms will be produced in the second generation which will exhibit various recombinations of the characters of species XX and XY.

An actual case of such a cross between two species of Aquilegia is that of A. formosa of the Pacific Coast of North America and A. flavescens of the Rocky Mountains. Pavson (1918) in his detailed monograph of the North American Aquilegias says:

The greatest development of flavescens occurs in the higher mountains of Montana and adjacent Wyoming and Canada. As we go westward from this region we find the species apparently merging into formosa: the sepals become salmon-colored or pink, the laminae shorter, and the spurs straight. This transitional area is often characterized by the lack of typical plants of either species, and in the mountains of Custer County, Idaho, the author has seen great patches of a variety with beautiful salmon-colored flowers entirely replacing the red formosa and the yellow flavescens. Since in the centers of their ranges formosa and flavescens are amply distinct, the author is very loath to treat one plant as a subspecies of the other. It would seem best to retain each as a species, never forgetting, however, that in certain regions the two actually merge.

There are furthermore in Aquilegia only slight physiological barriers to crosses between species, such as occur in practically all other genera. The most diverse Asiatic, American and European species cross readily and give fertile hybrids. Even Aquilegia ecalcarata Maxim, can be hybridized with A. vulgaris, though it is so distinct that Drummond and Hutchinon once (1920) placed it in Makino's genus Semiaquilegia. Species of Aquilegia will therefore be distinct only as long as that isolation persists. Whereas in Iris we may find several species growing together (as, for instance, Iris versicolor and I. prismatica in New England salt marshes) this will seldom or never occur in Aquilegia.

I have taken asexual propagation, polyploid series and physiological isolation as representatives of the internal factors which affect specific isolation and which whole genera or even families of plants may have in common. There must be many other such factors. May we not therefore logically expect that, even though species prove to be actual biological units, their relationships with each other and the relationships of individuals within species will vary from genus to genus and from family to family.