Proc. 11th Int. Cong. Genet. Pergamon Press, Oxford. 1963 (in press).
Genetics Today 2: 437-445 (1965)
The genetic-physiologic structure of species complexes in relation to environment.
William M. HIESEY
Dept. of Plant Biology. Carnegie Institution of Washington, Stanford, California, U.S.A.

The importance of genetic and cytologic studies in clarifying questions of taxonomic relationship is now firmly established. Yet relatively few researches have been directed toward revealing the genetic structure of species, especially their genetic composition as related to environment. Our inquiries in this area have reached the stage where we now regard a new principle—the principle of genetic coherence—as essential for interpreting plant relationships, and also for orienting quantitative physiological experiments designed to uncover mechanisms underlying natural selection.

Experimental evidence relating to genetic coherence has been derived mainly from the study of three species-complexes, each belonging b a diverse plant family. The three plant groups are: (1) North American members of the Achillea millefolium complex of the sunflower family: (2) North American members of the Potentilla glandulosa complex of the Drymocallis section of the rose family; and (3) members of the Mimulus cardinalis-lewisii complex of the Erythranthe section of the genus Mimulus, of the figwort family.


The central European members of this world-wide complex have been studied by Ehrendorfer (1952, 1959, 1959a) and Schneider (1958). On biosystematic evidence these investigators suggest that tetraploid and hexaploid species such as Achillea collina Becket and A. millefolium L. have evolved from diploid species having 2n=18 It chromosomes such as A. asplenifolia Vent, and A. setacea Waldst. and Kit. The polyploid species are more highly polymorphic than the diploids and often are not easily distinguished from one another. At all levels of ploidy the species and races of Achillea occupy ecologically distinct natural habitats.

In North America the Achillea millefolium complex is represented by many subspecies and ecological races within tetraploid A. lanulosa Natt. having 2n=36 chromosomes and hexaploid A. borealis Bong with 2n=54 (Lawrence, 1947; Ehrendorfer, 1952). A series of strikingly differentiated climatic races of these two occur along a west-east transect at 36° N. latitude in central California from sea-level to 3350 meters altitude (Clausen, Keck, and Hiesey, 1948).

FIG. 1. Above: Latitudinally contrasting ecological races of Achillea borealis originally from Kiska Island at 52° N. (P1), the San Joaquin Valley of California at 34° N. (P2), and their F1 hybrid. Below: Degrees of correlation between frequencies of nine parental characters segregating in a population of 300 F2 plants

Crossing experiments made within the North American Achilleas have extended our knowledge of their genetic structure. Hybrids between extreme and less extreme races within the tetraploid and the hexaploid groups have been studied. Crosses between the tetraploid lanulosa and hexaploid borealis yield pentaploid F1 progeny. In every instance the F2 progeny from such interracial hybrids segregate widely in an array of recombinations of the parental characters.

A particularly interesting F2 progeny will serve as an example (Fig. 1). It is a cross between contrasting latitudinal races of the hexaploid borealis, a race from Kiska Island in the Aleutian chain at 52° N latitude and another from the San Joaquin Valley in California at 34° N. The Kiska race is less than 10 cm high when grown at Stanford, and over a period of years has proven to be a non-survivor there. The contrasting San Joaquin Valley race grows to nearly 2 m in height at Stanford and is a vigorous survivor. The two races can be crossed reciprocally to produce fertile F1 hybrids, one of which is shown in Fig. 1.

The F2 progeny from this cross segregate between plants almost as dwarf as those of the Kiska parent and others as tall as the San Joaquin Valley parent (cf. Hiesey and Nobs, 1951, Plate 1). Plants in the experimental garden give the general impression of free random recombination of all the parental characters in the F2. Such an impression is superficial and untrue. When the characters of the individual F2 plants are measured for several years and the results are subjected to statistical analysis, it becomes clear that the parental characters are not inherited at random, but are partially linked together.

The polygon diagram in Fig. 1 illustrates these linkages in a much simplified form. Nine characters that distinguish the two parental races mere chosen for study. The lines connecting the possible combinations of paired characters indicate the degree to which any given pair of characters tends to be linked in their phenotypic expression. The degree of linkage is measured by the value of the correlation coefficient r computed from the frequency in F2 plants of particular combinations of phenotypes among a sample of 300 individuals.

Pairs of characters showing random segregation, without genetic correlation, are shown by thin broken lines, whereas r values abuse 0.11 are statistically significant at the 1 per cent level and are shown by solid lines. The heaviest lines indicate r values above 0.24.

Of the 36 possible combinations in Fig. 1, only 7 are inherited at random, and 29 pairs are significantly correlated at the 1 per cent or higher level. This shows an unmistakable tendency for the parental combinations of characters to be genetically linked, that is, to cohere.

Between 1952 and 1960 we conducted a long-term transplant experiment on the parental strains and the F1 and F2 progenies of the Kiska and San Joaquin Valley races of Achillea at our three altitudinal stations in California. The Stanford station is near the coast at 30m elevation, the Mather station is in the central Sierra Nevada at 1400m, and the Timberline station is near the Sierra Nevada crest at 3050 m. The climates at the three stations differ widely. Stanford has essentially a year-round growing season, Mather a 6 months growing period, and Timberline only about 2 months.

FIG. 2. Summary of results from transplanting cloned individuals of widely segregating F2 populations from contrasting interracial crosses of Achillea, Potentilla, and Mimulus to gardens at contrasting altitudes. See text.

Clones of 300 segregating F2 individuals of the cross Kiska San Joaquin Valley were planted in gardens at each station, and their survival, growth and development were followed year by year. Plants that died were repeatedly replaced, so that elimination by climatic factors as contrasted with accidental causes could be fairly accurately determined. The left-hand part of Fig. 2 summarizes in a highly simplified form some of the results from this experiment.

In Fig. 2 the plants of the F2 population were grouped into three categories: (1) those which tended to resemble the Kiska parent (P1), indicated by the solid black bars; (2) those which tended to resemble the Selma parent (P2), represented by the white columns; and (3) intermediate recombinations tending to resemble the F1 type, shown by the cross-hatched columns. The classification of the F2 plants into the three categories was based on a system of index numbers obtained by adding the values of 5 characters distinguishing the Kiska from the San Joaquin Valley parents when grown in the Stanford garden.

In the graph the data from Stanford are plotted at the bottom of the figure. those from Timberline at the top, and those from Mather in the renter. At all three stations the plants are grouped into two major categories, non-survivors and survivors. The survivors at each station are, in turn, separated into classes according to their average relative vigor as determined by over-all bulk measurements during a 5-year experimental period. The relative positions of the parents and of the F1 hybrids at each station are indicated by arrows and the symbols, P1, P2, and F1.

At Stanford, where the Kiska parent (P1) is a non-survivor and the San Joaquin Valley parent (P2) a vigorous survivor, the F1 is in an intermediate class with respect to vigor. The most vigorous F2 plants include the population fractions most resembling the San Joaqutn Valley parent and the F1 type. The weaker F2 plants at Stanford include mostly Kiska-like types and also a few F1-like recombinations. The fact that a few Kiska-like F2's are rather highly vigorous at Stanford is noteworthy.

At Mather both parents are non-survivors, but a substantial part of the F2 segregants of all three morphological classes show a surprisingly high degree of vigor in this environment. The F1 plants are also truly vigorous at this station, and nearly all of the wide array of F2 recombinations display growth far transcending that of either parent. This complementary genetic effect of recombining the genetic characteristics of the Kiska and San Joaquin Valley races is highly striking, and could scarcely have been predicted on the basis of the known characteristics of the parent races.

The most noteworthy result at Timberline is the high mortality of the P2, F1, and most of the F2 plants. The F1-like recomhinations are most frequent among the few surviving F2 Plant, although both Kiska-like and San Joaquin Valley-like plants are also represented.

The results from this experiment demonstrate three facts. First is the impressive array of recombination types that can he realized among the progeny of a single cross between two contrasting ecologic races of a species. Second, there are definite genetic limitations imposed upon the capacity of even such an array of F2 plants for survival when transplanted to the extreme environment ai Timberline. Finally, there is a tendency for F2 individuals morphologically resembling the parents to respond to transplanting more like the parent they resemble than like the opposite parent, a logical consequence of genetic coherence.


Extensive transplant and genetic Studies on altitudinal races of members of the Potentilla glandulosa complex have been described (Clausen, Keck, and Hiesey, 1940; Clausen and Hiesey, 1958) and will not be reviewed here. Only two major points pertinent to our present topic need to be mentioned.

The first relates to the clear evidence in Potentilla of genetic coherence between characters that distinguish the climatic races. As in Achillea, coherence can be expressed through degrees of correlation between pairs of characters in segregating F2 Populations (cf. Clausen and Hiesey, 1958, pp. 112-122, 1959). In the diploid species Potentilla glandulosa having n=7 chromosomes the ecologic races are distinguished by good marker characters that can be used in genetic investigations. Analyses of F2 and F3 progenies from such interracial crosses make clear that in Potentilla the inheritance of most of such characters is governed by multiple gene systems. The multiple genes governing the expression of each character are inherited independently, but are partially linked with genes regulating other characters through their location on the same chromosomes.

A second major result from the experiment on Potentilla glandulosa that relates to our present topic is snmmarized in simplified form in the graphs at the center of Fig. 2. Like the corresponding series of graphs for Achillea in the same figure at the left. the Potentilla data summarize the results of a corresponding long-term transplant experiment in which 575 cloned F2 plants of a cross between an alpine and a foothill race of Potentilla glandulosa were transplanted at Stanford, Mather, and Timberline to study their survival and growth response-patterns in comparison with their original parents.

The graphs for Potentilla and those of Achillea reveal striking similarities, and also marked differences. At Stanford the similarities are emphasized: in both groups one of the parents is a vigorous survivor, the other is not, and the F1 is a moderately vigorous survivor. At Timberline the contrast between Achillea and Potentilla is striking: the survival among the F2's of Potentilla is much greater than among the F2's of Achillea. This situation suggests that the genes of the foothill race complement those of the subalpine Potentilla parent in enhancing the survival of their F2 progeny at the alpine station. On the other hand, the genes of the Kiska and of the San Joaquin Valley races of Achillea did not work together to transmit this capacity to their offspring.

In both Achillea and in Potentilla the range of segregation among contrasting interracial F2 hybrids is spectacularly wide. There is at the same time an unmistakable tendency in both groups for the F2 progeny to respond to transplanting more like the parent they resemble than like the opposite parent. This tendency is clearly a manifestation of genetic coherence as illustrated in the correlation diagrams (cf. Clausen and Hiesey, 1960).


We have felt the need to supplement the study of the genetic structure of ecological races and their natural selection with investigations of specific physiological mechanisms that underlie natural selection. Neither Achillea nor Potentilla are as well suited as Mimulus for this kind of study.

Vickery (1951) crossed a lowland red-flowered form of M. cardinalis from near the coast of California with a lavender-flowered alpine form of M. lewisii. The F1 hybrid was fertile, and the F2 segregated widely in all manner of recombinations. This experiment led us to further studies along the lines already described for Achillea and Potentilla, but conducted in a more detailed manner. The members of the M. lewisii-lewisii group, like those of Potentilla glandulosa, are diploid and have n=8 chromosomes. Red-flowered M. lewisii consists of a series of ecological races native to habitats ranging from sea-level to upper middle attitudes, and covers latitudes from about 33° to 50° N. Mimulus cardinalis likewise is composed of many races, but they occupy a higher altitudinal segment than M. lewisii, from 1800 to 3300 m attitude; lewisii also occurs farther north, from about 37° to 60°.

The ecologic races of M. cardinalis are clearly distinguished from those of M. lewisii by vegetative and floral characters that also serve as excellent markers in genetic experiments. In this respect the races of Mimulus are superior to those of Potentilla for genetic studies although both groups are diploid. In range of diversity of form the races of Mimulus are comparable with the members of the Achillea millefolium and the Potentilla glandulosa complexes. It s interesting to compare what we know about the genetic structure of Mimulus with that of Achillea and Potentilla.

The F2 progeny of a hybrid between of M. cardinalis from near sea-level and an alpine race of M. lewisii from 3350 m elevation consisting of 300 plants were cloned and studied by the methods described above for Achillea and Potentilla. The results are summarized at the right of Fig. 2.

At Stanford the F2 progenies of Mimulus, Achillea and Potentilla show a marked parallelism of response. In all three groups one of the parents is a non-survivor, the other parent a vigorous survivor, and the F1 hybrid a moderately vigorous survivor. At Mather the performance of the F2 progeny of Mimulus contrasts markedly with that of either Achillea or Potentilla. Neither the alpine parent of Mimulus (P2) nor the coastal parent (P1) survives at Mather, but the F1 not only survives but displays considerable vigor. Relatively few individuals of the F2 progeny survive, and very few display marked growth. Genes contributed in equal amount by the alpine and lowland parent apparently result in a complementary balance in the F1 favorable for survival at Mather, but when recombined among the F2's this balance is apparently disturbed so that only rare recombinations survive there successfully.

FIG. 3. Degrees of correlation between frequencies of thirteen parental characters, segregating in an F2 population of 300 Mimulus plants of a cross between an alpine farm of M. lewisii and a lowland form of M. cardinalis. Drawings by Malcolm A. Nobs.

At Timberline the alpine M. lewisii parent survives and the influence of the non-surviving lowland M. cardinalis parent is shown through an enhanced survival and vigor of the F1 and many of the F2 progeny in this extreme climate.

The graphs for Mimulus at Stanford and Timberline indicate, as for Achillea and Potentilla, that the fractions of F2 progeny most resembling the parental forms tend to respond like the parent they resemble. At Mather, where neither parent survives, such an effect of genetic coherence is not expressed.

The striking effect of the genic composition of the parental races on the transplant responses of their F2 progeny is clearly evident in alI the three groups, Achillea , Potentilla. and Mimulus.

The same F2 progeny of Mimulus used in the transplant experiments were tested for genetic coherence using 13 characters, in the same manner as in Achillea and Potentilla. The results are summarized in Fig. 3. Of the 78 possible combinations of the 13 characters scored, 72 combinations, or 92 per cent are significantly correlated at the 1 per cent level. Only 8 per cent of the paired combinations have values of r between 0.00 and 0.11, indicating random distribution. The evidence showing genetic coherence at Mimulus is thus even stronger than in Achillea or Potentilla (cf. Nobs, Hiesey, and Milner, 1963). As in the two other genera, the direction of the linkages in Mimulus favors a higher frequency of F2 progeny resembling one or other of the parental forms than would he expected on the basis of random recombination.


Our main conclusions can be summarized as a statement of principles that appear to apply generally to higher plants.

  1. The inheritance of characters distinguishing ecological races is mostly governed by multiple genes. Simple Mendelian segregation being rare;
  2. Systems of genetic coherence characterize ecologic races; when two races from different environments are crossed, the F2 tends to segregate with a higher frequency of parental types than would be predicted on the basis of free random recombination.
  3. Such coherence systems do not preclude the production of striking recombinations which provide rich potentials of genetic variation for further natural selection; genetic coherence is probably the basis for the differentiation of ecologic races, subspecies and species in higher plants.



H. B. KRIEBLE (U.S.A.): Because the grouping of the F2's into three classes is an arbitrary arrangement, would it not be better to use a continuous type of analysis such as regression or correlation? It seems to me that there is some danger of arbitrarily setting up ecological races that do not really exist as separate entities.

W. M. HIESEY: I fully agree that the setting up of these three classes is highly arbitrary and is a gross over-simplification of a complex situation made rather necessary by this very brief presentation. In the full analysis of these F2 populations we need to go into much more detail than that which we have presented today. However, the over-all picture as we now interpret our data is essentially as represented by these necessarily crude diagrams.

J. L. CROSBY (England): Have F2 families a high survival rate? If not, character correlation might be a selection phenomenon rather than an effect of linkage.

W. M. HIESEY: Fertilities are very high and germination is excellent the F populations. The populations used in these studies were random samples.

F. EHRENDORFER (Austria): What happens at the natural zone of contact between Mimulus cardinalis and M. lewisii?

W. M. HIESEY: There is only one known instance where we have found M. cardinalis and M. lewisii growing together. Here we were unable to find any evidence of natural hybridization. This is rather to be expected because
(1) the two have different flower structures suited to different kinds of pollinating agents (insects vs. humming birds, for example), and
(2) they do not flower at the same time although their periods of flowering overlap somewhat.

A. D. MEEUSE (The Netherlands): Did you, apart from recording survival at your different experimental stations, notice significant differences in seed production? The biological flower types of the two species of Mimulus—most probably controlled by several genes—strongly suggest to me different pollinators and under natural conditions this difference would be very important in preventing or encouraging the invasion of suitable habitats by one of the forms and not by the other.

W. M. HIESEY: Yes, there is a wide range of segregation in the degree to which individual F2 segregants are able to set seed without artificial assistance on the part of the experimenter. This is a result of recombination in the flower structure found in the F2 because of the different pollinating agents to which M. cardinalis and M. lewisii flowers are adjusted.

O. H. FRANKEL (Australia): A comparison of the results reported by Gajewski and Hiesey illustrates the point I made after Gajewski's paper. Gajewski's characters are seed characters, and as Linné knew, remarkably invariate: they tend to acquire a blocking-heritance. Hence parental types reappear quickly and in surprising large proportions. The characters measured by the Stanford workers are perhaps more superficial in relation to survival. Hence their inheritance is complex, and there is a great deal of recombination with a lesser representation of parental types. It seems necessary in such comparisons to consider the nature of the characters used by investigations.

W. M. HIESEY: The differences between Gajewski's example and those of the California group are of degree rather than of kind. The Achillea, Potentilla and Mimulus differences are not superficial. Oil the contrary, they appear to be crucial with regard to survival in these climates.