| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Biology Department, Santa Clara University, Santa Clara, California 95053 USA
Received for publication December 14, 1999. Accepted for publication June 13, 2000.
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONLITERATURE CITED |
|---|
400% more frequent in taxonomic intermediates. One to four supernumerary chromosomes were found in 13% of the populations. Artificial hybridizations were made in 58 of the 65 possible diploid-level combinations involving the ten varieties of E. lanatum and E. confertiflorum var. confertiflorum. Aberrations, mostly failure to pair normally, were observed in diakinesis or M1 cells of progeny in 23 of 99 crosses. Studies of pollen stainability in cotton blue-lactophenol and other fertility indicators in 886 F1's from 191 crosses involving 81 populations showed that strong (22–40% pollen stainability) to weak (60–76% pollen stainability) barriers to interbreeding existed among diploids of the E. lanatum varieties and among them and E. confertiflorum var. confertiflorum. Pollen stainability was much higher in progenies of tetraploid, hexaploid, and octoploid intra- and interspecific crosses involving E. confertiflorum var. tanacetiflorum, E. jepsonii, and E. latilobum than in diploid ones, supporting the hypothesis that polyploidy has mainly served to stabilize the products of intervarietal and interspecific hybridizations.
Key Words: Asteraceae • barriers to interbreeding • Eriophyllum • Helenieae • polyploidy
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONLITERATURE CITED |
|---|
, and unpublished data).
Eriophyllum lanatum comprises 20 intergrading regional races that occur in many plant communities, from ocean bluffs to treeline. Edaphic and insolation differences strongly influence leaf morphology and the sizes of plants and plant parts.
Polytypism and polymorphism invite diverse taxonomic treatments, especially by those with limited field experience. Seventy-five binomials and trinomials had been used in E. lanatum before Constance (1937)
, without cytological study, reduced the complex to ten varieties.
Eriophyllum lanatum is centered in California, where all but vars. lanatum (Oregon, Washington, Idaho, Montana) and leucophyllum (Oregon, Washington, British Columbia) occur. Varieties integrifolium (Oregon, Washington, Nevada, Idaho, Wyoming, Montana, northwestern Utah) and achillaeoides (Nevada, Oregon) also occur in California. Varieties grandiflorum and lanceolatum occur in California and southern Oregon. The other varieties (arachnoideum, croceum, obovatum, and hallii) are California endemics. All the varieties except the last two are sympatric with at least one other variety. Variety obovatum is uncommon and var. hallii is rare (Skinner and Pavlik, 1994
).
Carlquist (1956)
, while briefly considering Eriophyllum as a doctoral problem, reported chromosome counts of n = 8 and 16 in three E. lanatum varieties. Chromosome counts from 368 plants representing 239 populations and all varieties but hallii revealed nine diploid taxa supporting a superstructure of tetraploid, hexaploid, and octoploid populations (Mooring, 1975
). Variety hallii, with one of its two populations sampled, proved to be diploid (Mooring, 1986
). I concluded that the major role of polyploidy has been to stabilize the products of intra- and intervarietal hybridizations (Mooring, 1975
). Artificial intervarietal crosses made primarily to investigate chromosome pairing showed reduced pollen grain stainability in diploid progeny. Reduced stainability led me to investigate interfertility in the E. lanatum complex and the fertility relationships between members of that complex and E. confertiflorum, E. latilobum, and E. jepsonii.
In this paper I report the results of artificial hybridizations and supplement knowledge of the cytogeography of the complex.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONLITERATURE CITED |
|---|
Treatments
Fruits were germinated in vermiculite or in vermiculite-soil mixtures, and seedlings were potted in "UC Mix" soil in an unheated Santa Clara University greenhouse. Individuals were generally transplanted to the garden after 1 or 2 yr because they soon grew to unmanageable size. Almost all garden plants were allowed to grow until they died of natural causes. Individuals of var. leucophyllum lived for 1–8 yr but seldom flowered.
Pollen viability estimates
Fresh pollen grains were stained overnight in cotton blue-lactophenol. Over 95% of my estimates rest on 300+ grains per sample, with each plant being sampled twice, on different days. The other estimates are based on no less than 120 pollen grains.
Meiotic analyses
Most of the cytological analyses have been of diakinesis or M1 stages of microsporocytes squashed in acetocarmine and examined with a phase contrast microscope. Young heads of wild, garden, or greenhouse plants were fixed in 1:3 acetic ethanol or, rarely, in 1:3:6 acetic-chloroform-ethanol. Quickly placing collections in an ice-filled cooler usually improved fixation (Anderson, 1966
). Beek's (1955)
technique provided clearer preparations. The relatively few root-tip counts came from seedling material placed in 1:1 concentrated HCL-95% ethanol for 7–10 min at room temperature, followed by squashing and warming in acetocarmine. Voucher specimens have been deposited in the Santa Clara University herbarium (SACL), and duplicates of most of them will be distributed elsewhere.
Artificial hybridizations
Capitula of isolated greenhouse plants were rubbed together over 3–8 d or, occasionally, pollen was transferred between bagged heads of garden plants of E. lanatum, E. confertiflorum var. confertiflorum, E. confertiflorum var. tanacetiflorum, E. jepsonii, and E. latilobum. Appendix 2 shows the number of and ploidy level of the hybridizations.
Statistical analyses
These used the statistical package Statistix, Version 4.0, Analytical Software, St. Paul, Minnesota, USA.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONLITERATURE CITED |
|---|
Supernumerary chromosomes
From one to four full-sized, stainable chromosomes in excess of the basic complement occurred in 10% of the 426 individuals and in 13% of the 271 populations sampled (Table 1). They were present in 11% of the 307 diploid individuals and in 9% of the 119 polyploid ones. Their frequency in garden populations was almost double that found in natural ones. Certain populations, regions, and varieties had significantly higher frequencies. Supernumeraries were not found in vars. arachnoideum, croceum, hallii, lanceolatum, and obovatum and were most frequent in individuals of vars. integrifolium (16/100) and lanatum (11/32).
|
Polyploidy
Diploids comprised 72% of the 426 plants sampled, followed by tetraploids (22%), hexaploids (3%), octoploids (3%), and a triploid, which occurred in an otherwise diploid population (Table 1). The frequency of diploid, tetraploid, hexaploid, and octoploid populations in the 271 populations sampled was almost exactly that for individuals (Table 2). Polyploid populations were about four times more frequent (26/31) among those judged as intergrades between varieties than they were among populations assignable to varieties (50/240). Varieties hallii and obovatum appeared to be entirely diploid, and varieties grandiflorum and integriflorum are the only ones known to have four ploidy levels. Diploid populations were not only three times more frequent than polyploids, they occurred throughout the range of the complex, whereas most polyploids occurred in northern California, western Oregon, and near the Columbia River (Fig. 1).
|
|
50–95% dark and stiff fruits. The same procedure in most intraspecific and interspecific pollinations, however, yielded fruits ranging from dark and stiff to light and limp. I classified fruit quality at 1x magnification as "good" (dark and stiff), "fair" (moderately dark and stiff), "poor" (pale and flexible), and "NG" (obviously without an embryo). Varying proportions of "good" and "fair" fruits failed to germinate, whereas as much as 5% of "poor" ones did (Appendix 3). Clearly, germination rate data must be interpreted cautiously. I computed percentage germination for almost all crosses. Variability was so high within groups that the only significant difference found between groups was in var. lanatum, where germination in progenies of intravarietal hybridizations was 15%, compared to 43% in intervarietal ones.
Progeny of diploid artificial hybridizations: meiotic aberrations
Sixty-five possible combinations can be obtained at the diploid level by crossing the ten varieties of E. lanatum to one another and to E. confertiflorum. I obtained progeny from one or more crosses within each of 58 of these combinations and cytologically examined one or more representatives of 99 crosses. The frequency of meiotic aberrations in progenies of intravarietal, intervarietal, and interspecific crosses was, respectively, 3/19, 15/68, and 5/12 (Appendix 3). Aberrations in progenies of intravarietal hybridizations occurred in varieties integrifolium and lanceolatum and involved failures to form bivalents consistently and occasional connections between bivalents (Table 3). Aberrations in interspecific hybrids with E. confertiflorum occurred in varieties achillaeoides, croceum, grandiflorum, and integrifolium and involved failure to form bivalents consistently. Progenies of intervarietal hybridizations, on the other hand, displayed more varied aberrations, including unequal bivalents and, in an integrifolium x obovatum cross, maximum pairing of only two bivalents (Table 3).
|
|
Progenies from intravarietal hybridizations
Representatives of all but vars. hallii and leucophyllum were available for intravarietal crosses. Mean pollen stainability of intravarietal progenies from the other eight varieties ranged from 86% (arachnoideum)to 62% (grandiflorum), 
= 75 ± 8, when the mean for populations was computed (Table 4), and 85% (arachnoideum) to 67% (lanatum), = 75.1 ± 7 when the mean for all offspring was calculated (Table 4). Mean pollen stainability in nonhybrid seed-grown greenhouse and garden plants ranged from 83% (var. grandiflorum) to 91% (var. obovatum), = 86 ± 12 (Table 5). Greatly overlapping standard deviations show that with the exception of var. obovatum the means for progenies from intravarietal hybridizations do not differ significantly from those of cultivated nonhybrids (Tables 4 and 5). Progenies were obtained from interpopulation crosses within all the varieties except hallii and leucophyllum, and from intrapopulation crosses in all but vars. grandiflorum, hallii, leucophyllum, and obovatum. Stainability in progenies from intrapopulation crosses was lower than in interpopulation ones in vars. croceum, integrifolium, and lanatum, the same in var. achillaeoides, and higher in vars. arachnoideum and lanceolatum. (The data from intra- and interpopulation crosses are combined under intravarietal crosses in Table 4.)
|
= 51 vs. 75%) or the means of all offspring in that combination ( = 50 vs. 75%). The difference between the means was highly significant, P = 0.0000. Stainability was significantly lower in 26 of the 36 combinations in which t tests could be made, with P values from 0.0000 to 0.0402 (Table 4).
Differences in stainability by comparing the results of intravarietal with intervarietal crosses could not be made for vars. hallii and leucophyllum, because flowering individuals of these taxa were not available for intravarietal crosses. Mean pollen stainability percentages were very low in progenies of intervarietal crosses involving var. hallii ( = 33 ± 9%), compared to the average for all progenies of intervarietal crosses ( = 44 ± 16%). Mean percentage stainability for progenies of intervarietal crosses involving var. leucophyllum ( = 51 ± 19%), however, was higher (Table 4).
Hybrids between E. lanatum varieties and E. convertiflorum var. confertiflorum
Crosses involving vars. lanceolatum and leucophyllum with E. confertiflorum var. confertiflorum could not be attempted because flowering representatives of vars. lanceolatum and leucophyllum were absent. Pollen stainability in the other interspecific combinations averaged lowest in var. lanatum and highest in var. hallii, 22 vs. 65% for populations and 14 vs. 65% for all hybrids (Table 4). For each variety the difference between the means for populations and that for all hybrids was 0–21 percentage points, but overall the means were almost the same, 43 ± 16% for populations and 41 ± 18% for all hybrids.
Mean pollen grain stainability was significantly lower in these interspecific hybrids than in the progeny of the intravarietal crosses, with P values of 0.0000–0.0163 (Table 4).
Pollen stainability was significantly lower in progenies of interspecific crosses than in intervarietal ones in all but nine combinations (Table 4). Stainability in the seven combinations involving var. hallii was significantly lower than in the var. hallii x E. confertiflorum var. confertiflorum progeny, with P values ranging from 0.0000 (hallii x integrifolium) to 0.04 (hallii x croceum). Stainability was also significantly lower in arachnoideum x integrifolium (P = 0.0175) and arachnoideum x leucophyllum (P = 0.0432) progenies.
F2 diploid progenies
Only 27 diploid F2 progenies were studied (Appendix 2); variety-for-variety comparisons could not be made with F1 progenies. Overall, mean percentage pollen stainability did not differ significantly between the F1 and F2 progenies, being 75 ± 10% vs. 72 ± 20% for intravarietal and 50 ± 8% vs. 52 ± 19% for intervarietal crosses. Figures for interspecific crosses did vary significantly, 40 ± 16% vs. 66 ± 15%. Standard deviations in the F2 progenies generally exceeded those in the F1's. In five of the 19 intervarietal crosses, stainability was 5–31 percentage points higher than that shown for F1's (Table 4); in the other crosses stainability was 1–34 percentage points lower.
Progenies of artificial hybridizations of polyploids: heteroploid crosses
I made relatively few crosses at the polyploid level. Of these, 23 homoploid and 11 heteroploid crosses yielded progeny (Appendices 2 and 3).
Three diploid x tetraploid hybridizations gave triploids; a fourth (303–3, diploid, var. grandiflorum x S28–102, tetraploid, var. lanatum) yielded triploids when the diploid 303–3 was the seed parent, but a tetraploid when 303–3 was the pollen parent. Meiotic configurations in triploids ranged from 8 II + 8 I to 10 II + 4 I. The tetraploid formed 16 II (Appendix 3). Mean germination of diploid x tetraploid progenies (27%) did not differ significantly from that of diploid (22%) or tetraploid (24%) progenies and was much higher (35 vs. 8%) when the diploid was the seed parent. Pollen stainability in triploid progenies averaged 5–36%, compared to 65% in the tetraploid arising from the diploid x tetraploid cross. Perhaps stainability over 50% indicates tetraploidy in this combination (Appendix 3).
The diploid x hexaploid intervarietal cross (S164–1, diploid, var. integrifolium x S211–1, hexaploid, var. achillaeoides-grandiflorum intermediate) gave tetraploids forming 14 II + 4 I to 16 II with pollen stainability averaging 73% (Appendix 3).
Progenies of artificial hybridizations of polyploids: homoploid crosses
Tetraploid x tetraploid intraspecific crosses yielded meiotically normal tetraploids forming 16 II, except for the intervarietal hybrid S149–1 (E. lanatum var. achillaeoides) x S264–13 (E. lanatum var. integrifolium), which had an unequal bivalent. The interspecific cross 230–1 (E. lanatum var. achillaeoides-arachnoideum intermediate) x 4–3 (E. latilobum) gave meiotically normal, 16 II progeny (Appendix 3). The latter species probably originated by hybridization between E. lanatum var. arachnoideum and E. confertiflorum var. confertiflorum (Constance, 1937
).
The progenies of hexaploid x hexaploid or octoploid x octoploid intraspecific or interspecific crosses formed bivalents only or bivalents plus two univalents (Appendix 3). Germination averaged about three times higher in hexaploid (72%) and octoploid (67%) offspring than in diploid (22%) or tetraploid (24%) ones.
Variety-for-variety pollen stainability comparisons between progenies of diploid and polyploid crosses were not possible because only 30 such polyploid combinations were available. Percentage pollen stainability of polyploid offspring (Table 6) averaged significantly higher than that for diploid ones (Table 4), especially for interspecific hybrids. Pollen stainability in progenies from intraspecific crosses of tetraploids was lower (76%) than in interspecific hybrids (88 or 95%). In hexaploid and octoploid intraspecific crosses, progenies averaged 93 and 90% stainable pollen, respectively, compared to 80 and 88% for interspecific hybrids (Table 6).
|
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONLITERATURE CITED |
|---|
). In many, they can be transmitted to progeny or increased in number, e.g., in Clarkia unguiculata (Mooring, 1960
). Transmission has been shown to be genetically controlled in wild maize populations (Rosato et al., 1996
). The origin, role, or means of persistence of supernumerary chromosomes in the E. lanatum complex is mostly unknown. Their 11% frequency in individuals and 13% frequency in populations suggest that they may be advantageous. Their apparent absence in vars. arachnoideum, croceum, hallii, lanceolatum, and obovatum may reflect inadequate sampling because only 15% of the 426 plants sampled came from these taxa. On the other hand, in var. lanatum their presence in 11 of 32 individuals and in five of the 16 populations suggests ability to persist (e.g., meiotic drive) or adaptive value, or both (Table 1). Certain areas had higher frequencies. The highest frequencies in var. lanatum were in four Oregon and Idaho populations adjacent to the Snake River Canyon; four of six plants in 1969 and six of nine plants in 1986 had supernumeraries. Two lightly sampled areas with high frequencies of populations with supernumeraries were Trinity County, California, where three of four var. grandiflorum populations had them, and a Nevada-Idaho transect, where all three var. integrifolium populations had them.
The 9 + 9 anaphase distribution of an individual that formed 8 II + 1 I shows that their number can be multiplied, and crossing results show that they can be transmitted (Mooring, 1975
). More definitive answers to the adaptive value of supernumerary chromosomes in E. lanatum would require extensive studies like those carried out in wild populations of maize (see Rosato et al., 1998
).
Supernumerary chromosomes similar in appearance and behavior to those in E. lanatum are present in E. confertiflorum (Mooring, 1994
) and some annual species of Eriophyllum (Strother, 1972, 1976
; Keil and Pinkava, 1976
; Johnson, 1978
). Eriophyllum has base chromosome numbers of x = 4, 5, 7, 8, 15, and 19 (Mooring, 1997
). Perhaps supernumerary chromosomes have played a role in the evolution of its base numbers.
Role of polyploidy
Diploid populations of Eriophyllum lanatum outnumber polyploid ones 3:1 (Table 2) and occupy the geographical and environmental extremes of the complex (Fig. 1). The overall geographic distribution of the polyploids is not correlated with environmental factors. Instead, ten diploid taxa support a three-level polyploid superstructure, similar to the "pillar" complexes (Stebbins, 1950
) of Phacelia magellanica (Heckard, 1960
) and Sanicula crassicaulis (Bell, 1954
).
Polyploidy has facilitated intervarietal hybridization in the E. lanatum complex. Polyploid populations are four times more frequent among taxonomically intermediate populations than among populations assignable to varieties (Table 2). Most polyploid populations occur where the ranges of the varieties overlap significantly (Fig. 2). (Identification problems also arise from environmental modifications caused by differences in insolation and edaphic factors.) Northern California, western Oregon, and near the Columbia River between Oregon and Washington had the highest concentrations of polyploid populations(Figs. 1, 2). Varieties arachnoideum, achillaeoides, and grandiflorum occur, respectively, in the western, eastern, and northern sections of the North Coast Ranges of California. In these sections, 38 of the 55 populations sampled were diploid. Taxonomically intermediate populations occurred in a central zone through a transition from redwood and mixed evergreen forest to oak woodland and chaparral. Polyploids dominated in the central region; only one of the 21 populations intermediate between vars. achillaeoides and arachnoideum or achillaeoides and grandiflorum proved to be diploid, and of the 34 plants that could be accommodated in vars. achillaeoides, arachnoideum, or grandiflorum, 14 were polyploids, mostly tetraploids. One large-headed, rayless entity (var. aphanactis J. T. Howell) appears to be entirely tetraploid and of achillaeoides-grandiflorum ancestry. It also has a coherent geographic distribution. The densest concentration of different chromosome numbers was along an 11-km section of a secondary road passing from an oak woodland-chaparral ecotone to chaparral to mixed evergreen forest. Diploid, octoploid, diploid, hexaploid, and diploid populations occurred in that order. The diploids were assignable to var. achillaeoides and the polyploids to var. grandiflorum. Much of the region has been disturbed, especially roadside habitats. Possibly the mixing of different cytotypes here began only a few decades ago, especially as a consequence of road building. "Hybridized habitats" (Anderson, 1948
) facilitate hybridization. Hybridization and polyploidy also seem to have blurred distinctions between vars. achillaeoides and leucophyllum, leucophyllum and lanatum, and lanatum and integrifolium where they meet in the Pacific Northwest. Constance (1937)
submerged var. leucophyllum in var. lanatum, commented on the absence of a distinct taxonomic boundary between vars. lanatum (sensu lato, s.l.), and achillaeoides in southern Oregon, and noted that intermediates between vars. lanatum (s.l.) and integrifolium were especially common east of the Cascade Mountains. Seven of the 21 populations of var. leucophyllum in my sampling were polyploid; all abutted the range of var. lanatum or var. achillaeoides. As to intermediates between vars. lanatum and integrifolium, the Oregon and Washington hexaploid and octoploid populations are clustered where the intermediates occur (Figs. 1, 2).
|
The appearance of a tetraploid individual among progeny in an artificial tetraploid x diploid cross (Appendix 3) suggests that intra- or intervarietal crosses between diploids are not the only way of generating polyploids. An artificial tetraploid–diploid hybridization in E. confertiflorum also produced tetraploids (Mooring, 1994
). Unreduced male gametes are a likely explanation for tetraploid formation, one that Stutz and Sanderson (1983)
used to account for polyploids in Atriplex confertifolia.
Polyploidy has also played two less prominent roles in the Eriophyllum lanatum complex: (1) preventing or decreasing interbreeding between individuals of different varieties in mixed stands and (2) probably producing species by hybridization. Preventing interbreeding is exemplified by the decades-long coexistence of vars. croceum and grandiflorum on Banner Hill near Nevada City, California. Hall and Essig's 1916 collection #10170 noted the lack of intermediates there, and my Banner Hill samples showed that var. croceum was tetraploid and var. grandiflorum was diploid (Appendix 1). The microenvironments and flowering times differ slightly at the site and may also have prevented natural hybridization.
Probable "speciation by fusion" (Mayr, 1970
, p. 248) is exemplified by Eriophyllum latilobum and E. jepsonii. The former may have originated by hybridization between E. lanatum var. arachnoideum and E. confertiflorum var. confertiflorum (Constance, 1937
; Munz, 1959
). Eriophyllum latilobum is tetraploid (Carlquist, 1956
; Mooring, 1973
), and hybrids formed 16 II when it was crossed to tetraploid E. confertiflorum var. confertiflorum (Mooring, 1994
) or to an E. lanatum var. achillaeoides-grandiflorum intermediate (Appendix 3). Pollen grain stainability averaged 89% in the former combination and 95% in the latter. Eriophyllum jepsonii is octoploid (Mooring, 1973
). Constance (1937
, p. 106) referred to it as a possible "evolutionary link" between E. lanatum and E. confertiflorum. Munz (1959)
believed it might have originated by hybridization between the two species. Artificial hybridizations between E. jepsonii and octoploid individuals of E. lanatum var. grandiflorum yielded F1's averaging 83% pollen stainability and forming 32 II + 1 I (Appendix 3). Artificial hybridizations between E. jepsonii and octoploid individuals of E. confertiflorum var. confertiflorum produced progeny whose mean pollen stainability was 87% (Mooring, unpublished data). Eriophyllum confertiflorum var. tanacetiflorum may also have originated by hybridization between E. lanatum and E. confertiflorum var. confertiflorum. It is octoploid (Carlquist, 1956
; Mooring, 1973
) and is sympatric with diploid E. lanatum and tetraploid and hexaploid E. confertiflorum var. confertiflorum (Mooring, 1994
). Crossing E. confertiflorum var. tanicetiflorum to octoploid E. lanatum var. grandiflorum yielded F1's that averaged 92% pollen stainability and formed 31 II + 2 I (Appendix 3). Crossing octoploid individuals of E. confertiflorum var. confertiflorum to E. jepsonii yielded F1's that averaged 95% pollen stainability (Mooring, unpublished data).
Barriers to interbreeding
Studies of interbreeding in Mimulus guttatus and M. glabratus (e.g., Vickery, 1969
; Tai and Vickery, 1970
) revealed barriers to interbreeding in these geographically widespread species complexes. Artificial hybridization programs in each involved intensive studies of representatives from 12 populations of these annual/facultative herbaceous perennials that are limited to wet sites. Unlike the Mimulus species, however, most of the individuals in the ±20 races of E. lanatum are short-lived perennials that occur in a variety of habitats. Using too few populations would give an exceedingly coarse-grained and superficial picture of crossing relationships. Moreover, my 6 x 4 m greenhouse was inadequate for the intensive artificial hybridization studies such as those carried out in Mimulus. I therefore undertook an experimental study of scores of populations, realizing that the number of progeny examined in each cross would usually be limited to five to ten.
Generally, within the same ploidy level what appeared to be an unfavorable environment and at least 200 m separated stands of convarietal plants, whereas kilometres separated populations of different varieties. Nevertheless, as noted under the section Role of Polyploidy, phenotypic intergrades often occur where the ranges of different varieties approach each other. Some of the intergrades probably represent environmental modifications resulting from substrate and insolation differences, others almost certainly are intervarietal hybrids. Many of the intergrade plants I examined proved to be polyploids, usually tetraploids (Table 2).
Factors working against effective interbreeding within a ploidy level include spatial separation, habitat differences, pollinators, genetic or chromosomal disharmonies that affect pollen tube growth, seed and fruit formation in the seed parent, germination, seedling growth, flower, seed, and fruit formation in the F1, and inviability, weakness, and reduced fertility in subsequent generations. Ornduff (1969)
epitomized interbreeding as a succession of three related events: "relative crossability," (pollination, fertilization, and maturation of the seed), "progeny establishment," (germination, growth, and maturation of the offspring), and "fertility of the progeny." Using his terms, my data on relative crossability, progeny establishment, and fertility of the progeny come from greenhouse and garden populations and cannot be uncritically applied to interbreeding in natural populations. The relative crossability and the progeny establishment phases in cultivated eriophyllums probably were much more successful than they would have been in nature. Fertility of the progeny, on the other hand, if estimated by pollen stainability, presumably would be similar to that in nature. My assessments of interbreeding in Eriophyllum, therefore, rest lightly on fruit quality and germination, and heavily on pollen stainability.
Reduced synapsis and chromosomal aberrations in progenies from artificial hybridizations
Almost all tetraploid, hexaploid, and octoploid F1's formed only bivalents at diakinesis or MI, whereas 21% of diploid F1's showed reductions in pairing or meiotic aberrations (Appendix 3). The frequency of reduced pairing and probable chromosome repatterning among diploid F1's increased from intravarietal (3/19), to intervarietal (15/68), to interspecific (5/12) crosses (Table 3). Synapsis can be under genetic control (e.g., Jackson, 1982
), but the presence of unequal bivalents, connections between bivalents, and a bridge + fragment suggest translocations and an inversion (Table 3). Minor, rather than major, chromosome restructuring seems to have predominated. As Jackson (1984)
has observed, however, analysis of pachytene pairing sometimes reveals structural differences not observable in later stages. What I have termed "loose" bivalents (i.e., univalents close to each other and oriented as if to pair, but not observed to pair) were present in some diploid intervarietal hybrids. A possible example of major chromosomal differentiation is an integrifolium (S158–1) x obovatum (231–1) combination. The cross yielded 91 fruits judged as fair to good, but only two germinated. The two F1's produced configurations from 16 I to 2 II + 12 I, plus unequal bivalents (Table 3). Pollen stainability was only 14% (Appendix 3).
Reduced synapsis in diploid F1's was much higher when certain varieties were used as parents. Variety leucophyllum led with four appearances (Table 3) out of a total of seven intervarietal hybridizations (Appendix 3), followed by vars. hallii (three of nine), achillaeoides (four of 15), integrifolium (five of 19), and obovatum (four of 17). Variety lanatum, on the other hand, was used in 13 intervarietal crosses (Appendix 3), but all F1's formed 8 II. Varieties achillaeoides, croceum, grandiflorum, and integrifolium had reduced pairing in progeny of crosses with E. confertiflorum, but only vars. achillaeoides and integrifolium were overrepresented in intervarietal crosses (Table 3). Some reductions in pairing and aberrations probably represent individual or population chromosomal or genetic differentiation, rather than characterizing entire varieties.
Fruit quality and germination in progenies from artificial hybridizations
My method of estimating fruit quality as "good," or "fair," is fallible, as described above. Examination of scores of individual crosses, however, shows repeated pollinations producing relatively few "good" or "fair" fruits, and low percentage germination. Moreover, the seedlings sometimes died early or matured but did not flower. In contrast, 50–95% of "good" fruits of nonhybrids germinated, and most of the offspring flowered. In some artificial hybridizations, no germination occurred in a reciprocal cross; in others, germination was very low. For example, in the three instances when representatives of populations 330 (var. arachnoideum) and 43 (var. integrifolium) were crossed and 330 served as the seed parent, germination was 0–8%, and the seedlings died or matured without flowering. When 43 furnished the seed parent, however, germination was 20 or 37%, and at least 50% of the plants survived to flower (Appendix 3).
Pollen stainability in progenies from artificial hybridizations
In my study pollen stainability, although widely used as an indication of pollen viability (fertility), probably overestimates fertility for three reasons. (1) Some stainable pollen may not germinate (Stace, 1980
, p. 144). (2) More than two categories of sizes and colors are obvious in progenies of some of my hybridizations. Pollen from E. lanatum nonhybrids generally is easy to classify. The grains are approximately the same size, dark blue is viable, and colorless is inviable. Pollen from progeny of my artificial hybridizations was much more difficult to score, especially in some intervarietal and interspecific combinations. Some grains were up to 300% larger, and color ranged from dark blue to pale blue to colorless. Taylor (1967)
noted similar variation in artificial, interspecific hybrids of Aquilegia. My data (Tables 4–8; Appendix 3) probably overestimate fertility because I rated some less than dark blue shades as "viable." (3) Some progeny of my artificial hybridizations may be selfs rather than hybrids. Stace (1980)
described the difficulty of detecting hybrids in some combinations. Despite the high degree of self-incompatibility in E. lanatum in bagging tests (Mooring, 1975
), and in observation of hundreds of plants isolated from other plants (Mooring, unpublished data) some self-pollinations could have occurred. Selfs would be difficult to detect in progeny of intravarietal crosses and in some intervarietal crosses, and my scoring them as hybrids could result in overestimating fertility.
Pollen stainability in progenies from artificial, intraspecific hybridizations of diploids
Percentage pollen stainability in intrapopulation crosses was approximately equal to, or averaged lower than, that in interpopulation crosses in four of the six varieties in which both intra- and interpopulation crosses were made. As with meiotic aberrations, geographic distance between populations does not seem to be correlated with pollen stainability.
Pollen stainability was significantly lower in progeny from intervarietal hybridizations than in intravarietal ones in 26 of the 36 combinations (Table 4). Average pollen grain stainability in the progeny of intervarietal crosses ranged from 22 ± 8% (hallii x integrifolium) to 76 ± 19% (arachnoideum x lanceolatum)(Table 4, lower half of each cell). Dividing that 54-point range of means in three equal ranges gives 22–40% stainability as low, 42–58 as medium, and 60–76% as high stainability (Table 7). Some varieties are overrepresented, vars. achillaeoides and lanceolatum in the relatively high-fertility portion, and vars. hallii, integrifolium, and obovatum in the low-fertility zone. Strong barriers to gene exchange apparently exist between varieties occurring in the low-fertility zone, especially in var. hallii. Only one of its intervarietal combinations occurred out of the low-fertility zone (croceum x hallii, 47% stainable pollen).
|
= 34 ± 9). The differences were significantly higher; P values ranged from 0.0000 to 0.0409. Percentage pollen stainability was also significantly higher for var. arachnoideum x E. confertiflorum var. confertiflorum progeny (54 ± 16) than for arachnoideum x hallii (31 ± 11), arachnoideum x integrifolium (40 ± 17), or arachnoideum x leucophyllum (31 ± 20), with P values of, respectively, 0.0000, 0.0175, and 0.0432. Variety achillaeoides x E. confertiflorum var. confertiflorum hybrids also averaged significantly higher pollen stainability than achillaeoides x hallii progeny, 58 ± 10 vs. 30 ± 13%, P = 0.0000. A caveat exists: fertility conclusions about var. leucophyllum are inconclusive. Intravarietal and interspecific crosses were not possible, and only 34 plants from four populations were used in the seven crosses. Moreover, plant S239–2 was used in three of the crosses, and plant 167–12 in two others. When plant 167–12 was used, the mean percentage of stainable pollen in progenies of intervarietal crosses ranged from 29 (obovatum) to 68 (integrifolium), and the mean for these hybrids ( = 51 ± 18%) exceeded the mean for all intervarietal crosses (44 ± 16%). The difference, however, was not significant (Table 4).
Parent–progeny pollen stainability comparisons
Pollen stainability was known for both parents in 61 diploid-level artificial hybridizations, making parent–progeny stainability comparisons possible for these combinations (Table 8). Average stainability in the progeny increased over the average of the parents in five of the 61 crosses by 4–16%. Three of these were intravarietal crosses. Average stainability decreased in the progeny of the other 56 crosses, by 1–68% ( = 44) in progeny of E. lanatum x E. confertiflorum var. confertiflorum combinations, and by 9–73% ( = 47) in progeny of intervarietal crosses within E. lanatum (Table 8). These data support conclusions drawn from Tables 4 and 7 about vars. integrifolium and obovatum being associated with low fertility. Eight intervarietal crosses generated progenies with mean pollen stainability at least 50% less than that of the average of the parents. In this low-fertility category, variety obovatum occurred four times and vars. hallii and integrifolium three each, together comprising ten of the 16 possible parents. Varieties arachnoideum, lanatum, and grandiflorum were each represented twice, and vars. achillaeoides, croceum, lanceolatum, and leucophyllum not at all, in the low-fertility range. Of the unrepresented varieties, achillaeoides and lanceolatum were in the higher fertility zone discussed above (Table 7).
|
Barriers to interbreeding may exist within vars. grandiflorum and obovatum. Variety grandiflorum populations have the lowest mean percentage pollen stainability (62 ± 17%) in intravarietal crosses (Table 4). In progenies of five crosses involving populations 303 and 313 and 303 and 320, germination was 4–23%, and in two progenies the seedlings died; in the survivors pollen stainability was 40–57%. Population 320 is morphologically close to var. achillaeoides; 303 and 313 are typical var. grandiflorum. Populations of variety obovatum, unlike those of var. grandiflorum, occur in two regions 220 airline km apart. In the interregional crosses, percentage pollen stainability was high (66 ± 12%) in one progeny (322–20 x 289–27), but germination failed in its reciprocal. Another cross (231B–1 x 289–44) and its reciprocal yielded apparently good fruits, but germination was 0. In a third cross (289–28 x 322–20), only 2% of the poor-to-fair fruits germinated, and the two seedlings died (Appendix 3).
Interbreeding at diploid level
Formidable barriers to gene exchange exist among diploid populations of the E. lanatum complex at the intervarietal and interspecific levels. Referring to Ornduff's (1969)
"fertility" component, mean pollen stainability percentage decreased from nonhybrids (86 ± 12%), to progenies of intravarietal (75 ± 10%), intervarietal (50 ± 9%), and interspecific (40 ± 16%) crosses (Tables 4 and 5). The difference between the means for nonhybrids and progenies of intravarietal crosses, and between the means for intervarietal and interspecific progenies, was not significant. The difference between the means for progenies of intravarietal and intervarietal crosses, and those of intravarietal and interspecific crosses, however, was highly significant (P = 0.0000). With respect to Ornduff's (1969)
"relative crossability" and "progeny establishment" components, scores of artificial intervarietal hybridizations in E. lanatum show: (1) Repeated pollinations yield surprisingly small numbers of viable-looking fruits. (2) Percentage germination is usually low compared to that of nonhybrids (Table 5). (3) Percentage survival to flowering is occasionally low. (4) Average pollen stainability in F1's is relatively low and often 50–75% less than the parental average. (5) Some F1's show meiotic aberrations indicating either chromosomal repatterning or genetic control of pairing, or both. Sometimes germination and survival of seedlings to flowering reinforced low pollen viability. For example, the lowest pollen stainability percentages in intervarietal crosses were hallii x integrifolium (S306A–2 x 114P), hallii x lanatum (S306A–1 x 324), and integrifolium x obovatum (289P x 114P), with 22 ± 8, 26 ± 4, and 21 ± 12%, respectively (Appendix 3). Reductions in pollen stainability below that of the parental average, were, respectively, 75, 73, and 71%. Germinations were, respectively, 19, 50, and 45%. The combined effects shown in cultivated populations suggest that in natural populations intervarietal gene exchange may be low at the diploid level. Moreover, observations in areas where varieties approach or overlap geographically suggest that environmental factors probably decrease chances of interbreeding. Two California examples involve vars. achillaeoides and grandiflorum. About 2 km separate them near Weed, and near Quincy plants of both varieties mingled (1968) or were within 1 km (1999) along a roadside. I could not find intermediates in either locality. At the Quincy site var. grandiflorum was in full flower and var. achillaeoides was in bud.
I conclude that gene or chromosomal disharmonies, or both, restrict interbreeding by lowering fruit quality in the maternal parent, and by reducing germination, growth, maturation, and, especially, pollen viability in the F1 generation. Vickery (1969
, p. 327) noted similar phenomena in Mimulus: "(1) Barriers that limit the ability of interpopulation combinations to set seed. (2) Barriers that lower the capacity of the F1 hybrid seeds formed to germinate. (3) Barriers that restrict the growth and development of the hybrid seedlings. (4) Barriers that reduce the vigor, flowering, or fertility of the F1 hybrid plants."
The efficacy of barriers to interbreeding among the Eriophyllum lanatum varieties resembles that found in Hulsea, another western American genus of the Helenieae. Hulsea consists of seven species, apparently all diploid (n = 19). All but H. vestita show little intraspecific morphological differentiation. The mean and range of pollen stainability in progenies from E. lanatum intervarietal hybridizations (50% and 22–76%)(Table 4, Fig. 3) are close to those in Hulsea interspecific hybrids (49% and 22–88%)(Fig. 20 in Wilken, 1975
). Wilken (1975)
treated the taxa as species despite their degree of interfertility.
|
As noted under the section Role of Polyploidy, interspecific hybridization and polyploidy may also have brought about the fusion of E. lanatum and E. confertiflorum into E. jepsonii (Constance, 1937
; Munz, 1959
), into E. latilobum (Constance, 1937
), into a local race of populations in E. confertiflorum var. confertiflorum that closely resemble E. latilobum, and, possibly, into E. confertiflorum var. tanacetiflorum (Mooring, 1994
).
Systematic relationships
At the diploid level, the Eriophyllum lanatum varieties are, to varying degrees, incompletely reproductively isolated from one another and from E. confertiflorum var. confertiflorum (Table 7, Fig. 3). Moreover, the sampling of chromosome numbers suggests that at the diploid level the varieties are allopatric or marginally sympatric with one another and with E. confertiflorum var. confertiflorum. To use Mayr's (1970)
terminology, at the diploid level Eriophyllum lanatum is a superspecies, and its varieties are semispecies. Polyploidy, however, by facilitating numerous hybridizations, has created a pillar complex that is taxonomically frustrating and biologically fascinating. Clearly, those defining species in plants must appreciate the array of interbreeding relationships. Such definitions will not be simple ones.
The evidence from experimental studies suggests a taxonomic treatment that largely follows Constance's (1937)
, but eliminates his var. cuneatum, merges vars. lanceolatum and obovatum, and replaces his varieties with subspecies. (Constance stated that he retained "variety" to avoid making further new combinations.) Variety cuneatum appears to consist of populations intermediate between vars. grandiflorum and integrifolium and is centered near Donner Pass in the Sierra Nevada. The two populations I sampled there were tetraploid (Appendix 1). Some Plumas County, California, specimens have been referred to var. cuneatum. The populations that I have examined cytologically are diploid and referable to either var. achillaeoides or grandiflorum. These var. achillaeoides populations resemble those of northeastern California and southwestern Oregon and differ markedly from those of the California Coast Ranges. I suspect that the Plumas County var. grandiflorum populations represent var. grandiflorum introgressed by var. integrifolium. Varieties lanceolatum and obovatum, separated by 800 airline km, are nevertheless scarcely separable morphologically, and the mean pollen stainability (73%) in the progeny of the lanceolatum x obovatum cross was among the highest recorded (Table 7). One could also make a case for merging the morphologically very similar vars. arachnoideum and croceum, separated by 150 airline km, and whose mean pollen stainability in the progeny of the arachnoideum x croceum artificial hybrids was 74% (Table 7). Any treatment of the E. lanatum complex will have to realize that intergrades exist. The long list of synonyms (see Constance, 1937
) reflects attempts to recognize variants. Of these, perhaps only E. ternatum Greene deserves recognition. It connects the northern variants of var. achillaeoides with the southern populations of var. leucophyllum, which otherwise merge in and near Douglas County, Oregon.
A hypothetical phylogeny for the Eriophyllum lanatum complex
Using the geographic distributions of taxa, base chromosome numbers, habitat considerations, and artificial hybridizations among species, Mooring (1997)
hypothesized a phylogeny for Eriophyllum that has E. lanatum being derived from an E. confertiflorum var. confertiflorum plexus or having both species come from a common ancestor. To summarize the relevant part of this hypothesis, E. lanatum and E. confertiflorum var. confertiflorum bridge the chromosome number, distribution, and habitat gaps between the maritime, relatively long-lived perennial species E. nevinii (n = 19) and E. staechadifolium (n = 15) on the one hand, and the annual species (n = 7, 5, or 4) of mostly interior plant communities on the other. Both species have a nonrandom distribution of diploid and tetraploid populations. Eighteen of the 23 E. confertiflorum var. confertiflorum populations sampled from Los Angeles County southward are diploid (Mooring, 1994
) as are 25 of the 26 southernmost California E. lanatum populations. Assuming a diploid to polyploid progression, the distribution of cytotypes suggests a southern origin (Mooring, 1997
). Eriophyllum confertiflorum var. confertiflorum extends into Baja California, whereas E. lanatum, once found in Mexico, is now probably extinct there (Moran, 1996
). Below, I present an hypothesis for the origins of the E. lanatum varieties that extends that of Constance (1937)
.
Constance (1937
, p. 72) observed that var. achillaeoides "seems most closely related to the other perennial species and its occurrence fits in with the suggested center of generic dispersal." Today, var. achillaeoides also occurs farther south than all but vars. hallii and obovatum, from the central California Coast Ranges to southern Oregon. Its diploid populations occur in communities as diverse as chaparral, oak woodland, and coniferous forest. In 1875, however, Palmer collected a specimen on Guadalupe Island that is referable to var. grandiflorum, but whose fruits were more like those of var. achillaeoides (Constance, 1937
). The Mexican entity was last seen in 1893 and is presumed extinct owing to overgrazing by feral goats (Moran, 1996
). Constance (1937
, p. 72) conjectured that "it is not improbable that a population ancestrally common to both was the true primitive stock."
My proposed phylogeny for E. lanatum relies heavily on the mean pollen stainability of intervarietal hybrids, geographic distribution, and morphological, especially floral, features. It assumes that relatively high fertility (estimated by pollen stainability) reflects a closer relationship than relatively low fertility.
I hypothesize that E. lanatum originated in southwestern North America, with populations ancestral to var. achillaeoides migrating northward along the Peninsular and Transverse Ranges, then along the Coast Ranges and the west side of the Sierra Nevada–Cascade axis to southern British Columbia, giving rise to vars. hallii, obovatum, arachnoideum, grandiflorum, croceum, lanceolatum, and leucophyllum (Fig. 4). Except for var. hallii, pollen stainability supports the ancestral nature of var. achillaeoides. Stainability percentages for the progeny of intervarietal crosses between it and the varieties listed just above are, respectively, 34, 63, 74, 57, 63, 66, and 66% (Table 7). Five of these figures fall in the high-stainability column, two in the medium, and only one (var. hallii) in the low-stainability column (Table 7). The high stainability for the progeny of the grandiflorum x lanceolatum cross (68%) and the near sympatry of the varieties allows the possibility of var. lanceolatum being derived from var. grandiflorum instead of from var. achillaeoides. The other two varieties not cited above are largely extra-Californian. I posit the origin of var. integrifolium from var. achillaeoides or an achillaeoides-like ancestor east of the Sierra Nevada and the origin of var. lanatum from either var. achillaeoides or var. integrifolium. The relatively high pollen stainability percentages for the achillaeoides x integrifolium and the integrifolium x lanatum F1's (58 and 60%, respectively) support this assumption. Alternatively, var. achillaeoides may have given rise to var. lanatum. They are now widely allopatric, but achillaeoides x lanatum artificial hybrids average 66% stainable pollen.
|
30 yr after Constance (1937)
published his alpha taxonomic study. Now, another 30 yr later, molecular systematics dominates the taxonomic stage. Perhaps newer techniques can better elucidate relationships in Eriophyllum, thus illustrating Constance's (1964)
observation that systematic botany is an unending synthesis.
|
|
|
|
|
|
|
