(American Journal of Botany. 1999;86:614-633.)
© 1999 Botanical Society of
America, Inc.
Isozyme variability among cryptic species of Botrychium subgenus Botrychium (Ophioglossaceae)1
Warren D. Hauk2 and
Christopher H. Haufler
Department of Botany, University of Kansas, Lawrence, Kansas 66045
Received for publication December 18, 1997.
Accepted for publication October 13, 1998.
 |
ABSTRACT
|
|---|
The systematics of Botrychium subgenus Botrychium has been controversial, primarily because reduction in frond size and complexity has limited the number of characters available for discrimination of species. The recognition of many polyploid species has magnified the difficulty of classification because allopolyploids are often morphologically intermediate between their progenitor diploids. In order to evaluate species limits and sectional boundaries, we surveyed and compared 16 of the 24 currently recognized species for isozymic variation. Little or no intrapopulational variation was detected, but the variation present was consistent with the hypothesis that Botrychium species are primarily inbreeding. Interspecific comparisons distinguished six diploid species and provided evidence of molecular differentiation between the cryptic sister species B. lunaria and B. crenulatum. Evidence of possible progenitor/descendant relationships was found for certain diploid/polyploid relationships. Using enzyme bands shared between species, realignment of the sectional assignment of several species is proposed. Anomalous banding patterns in certain individuals suggested that gene silencing or homoeologous chromosome pairing might be operating in B. minganense, B. hesperium, and B. matricariifolium. Isozyme data also showed that some populations of species presumed to be uniformly diploid possessed isozyme patterns typical of polyploids. All available molecular data indicate that members of Botrychium subgenus Botrychium are actively evolving at diploid and polyploid levels.
Key Words: Botrychium • breeding systems • isozymes • Ophioglossaceae • polyploids
|
INTRODUCTION
|
|---|
Species of Botrychium Swartz subgenus Botrychium, commonly referred to as the moonworts, comprise an inconspicuous and often overlooked component of the North American fern flora. Owing in part to their small size (1–15 cm) and production of a single, epigeal leaf per year, moonworts exhibit a remarkable degree of morphological simplicity and possess a limited number of characters for delimiting species and constructing classifications. The systematics of this subgenus has been particularly problematic and controversial (cf. Clausen, 1938
; Tryon and Tryon, 1982
; Wagner and Wagner, 1993
). Using morphological features, Clausen (1938)
recognized only six species in subgenus Botrychium, whereas W. H. and F. S. Wagner and others have raised that number to 
24 species based on a combination of morphological and cytological studies (F. S. Wagner, 1993
; Wagner and Wagner, 1981
, 1983a
, b
, 1986
, 1990a
, b
, 1994
; Wagner et al., 1984
; Farrar and Johnson-Groh, 1991
; see Table 1, Figs. 1–2
).
View this table:
[in this window]
[in a new window]
|
Table 1. Species of Botrychium subgenus Botrychium. Preliminary sectional assignments based on morphological and cytological studies of Wagner and Wagner (personal communication). The number after each species name indicates the documented ploidy level of the species (2x = diploid, 4x = tetraploid, 6x = hexaploid) with x = 45. Asterisks indicate species sampled in the present study.
|
|
View larger version (25K):
[in this window]
[in a new window]
|
Fig. 1. Hypothetical parents for polyploid species of subgenus Botrychium (after F. S. Wagner, 1993
). Ploidy level indicated by ellipse = 2x, rectangle = 4x, ellipse over rectangle = 6x, where x = 45 chromosomes. ASC = B. ascendens, CAM = B. campestre, CRE = B. crenulatum, ECH = B. echo, GAL = B. gallicomontanum, HES = B. hesperium, LAN = B. lanceolatum, LUN = B. lunaria, MAT = B. matricariifolium, MIN = B. minganense, MON = B. montanum, PAL = B. pallidum, PAR = B. paradoxum, PED = B. pedunculosum, PIN = B. pinnatum, PSE = B. pseudopinnatum, SIM = B. simplex, SPA = B. spathulatum, and WAT = B. x watertonense.
|
|
View larger version (31K):
[in this window]
[in a new window]
|
Fig. 2. The single most-parsimonious Fitch tree produced from analysis of 26 rbcL sequences representing 20 species of subgenus Botrychium (after Hauk, 1995
). All characters were equally weighted and unordered. Sequences of Botrychium multifidum (S. G. Gmelin) Ruprecht and B. lunarioides (Michaux) Swatrz of subgenus Sceptridium were used as outgroups. Segments lengths are indicated above branches. Using only informative characters, the tree had a length of 49 steps, a CI of 0.918 and a RI of 0.974.
|
|
Members of Botrychium subgenus Botrychium have been called "cryptic" species because species delimitations are based on subtle morphological features (Paris, Wagner, and Wagner, 1989
). Cryptic species were defined by Stebbins (1950
, p. 193) as "...population systems which were believed to belong to the same species until genetic evidence showed the existence of isolating mechanisms separating them." Cryptic species are characterized by the following criteria (Paris, Wagner, and Wagner, 1989
): (1) poor morphological differentiation; (2) distinct evolutionary lineages that are reproductively isolated; and (3) historic misinterpretation as members of a single species.
Many species of subgenus Botrychium satisfy these criteria. First, they are notoriously difficult to differentiate without considerable field experience. Second, detection of ploidy level differences and sterile hybrids (Wagner, 1980
, 1991
; Wagner et al., 1984
; Wagner, Wagner, and Beitel, 1985
; Wagner and Wagner, 1988
) has provided evidence of reproductive isolation and hybridization. Third, the majority of recently recognized Botrychium species were once considered to be minor morphological variants within single species. For example, B. minganense, B. spathulatum, and B. ascendens were all considered insignificant variants of B. lunaria.
As emphasized by Stebbins (1950)
and Paris, Wagner, and Wagner (1989)
, cryptic species must be genetically distinct or they cannot be considered natural evolutionary units. Molecular features such as isozymes can provide markers to help substantiate the genetic distinctness of putative cryptic species (Good, 1989
). In ferns, isozyme studies have led to the recognition of new "cryptic" species in the Adiantum pedatum complex (Paris and Windham, 1988
; Paris, 1991
), Gymnocarpium (Pryer and Haufler, 1993
), and Polypodium (Haufler and Windham, 1991
). In each of these cases, isozymic characterization supported hypotheses that diploid progenitors of tetraploid species were distinct species, rather than infraspecific variants (i.e., Adiantum aleuticum (Ruprecht) Paris
, Gymnocarpium appalachianum Pryer and Haufler
, and Polypodium appalachianum Haufler and Windham
).
In this study, we used isozyme variation to evaluate species and sectional concepts within subgenus Botrychium and compared the results of these analyses to those of morphological and rbcL studies. These isozyme data provided evidence for (1) assessing relationships among diploid species, (2) characterizing the breeding system(s) of diploid species, (3) identifying possible diploid progenitors of polyploid species, and (4) evaluating hypotheses concerning classification of subgenus Botrychium species.
|
MATERIALS AND METHODS
|
|---|
Collections representing 16 currently recognized Botrychium species were obtained in the continental United States and Canada (Table 2). A total of 440 diploid, 770 tetraploid, four hexaploid, and 44 individuals of unknown ploidy were surveyed. With the exception of one population, all identifications were verified by Warren H. Wagner (vouchers of a Pike's Peak, Colorado collection were inadvertently destroyed by the U.S. Postal Service, but subsequent collections from the same location were identified as B. lineare Wagner).
View this table:
[in this window]
[in a new window]
|
Table 2. Population abbreviations, collection site, and voucher information for 51 populations of subgenus Botrychium species. All Hauk vouchers are deposited at NC, all Wagner vouchers are at MICH, and all Lang vouchers are at ORE.
|
|
Whenever possible, live plants were transported to the laboratory on wet ice. When shipped by mail, live plants were enclosed in plastic bags with fresh Fagus sp. or Quercus sp. leaves to help prevent desiccation without promoting fungal growth. Plants deteriorated rapidly after collection, reducing enzymatic activity. Accordingly, when it was not possible to electrophorese the samples immediately on their return to the laboratory, plant material was ground in an appropriate buffer, the homogenate absorbed into filter paper wicks, and the wicks stored at -80°C. This procedure preserved the activity of some enzymes for up to 18 mo. Representative specimens (Table 2) were pressed and deposited in one of the following herbaria: University of North Carolina (NCU), University of Michigan (MICH), or University of Oregon (ORE).
Isozyme banding patterns were obtained using starch gel electrophoresis for the following enzymes: aspartate amino transferase (AAT), diaphorase (DIA), fructose 1,6-bis-phosphatase (F1-6), glyceraldehyde 3-phosphate dehydrogenase (G3PDH), isocitrate dehydrogenase (IDH), leucine amino peptidase (LAP), malate dehydrogenase (MDH) phosphoglucoisomerase (PGI), phosphoglucomutase (PGM), 6-phosphogluconate dehydrogenase (6PGD), shikimate dehydrogenase (SKDH), and triosephosphate isomerase (TPI). Band resolution was best for TPI, PGM, and GOT on system 6 of Soltis et al. (1983)
, for PGI, LAP, IDH, and DIA on system 8 (Haufler, 1985
), for G3PDH, F1-6, 6PGD, MDH, and SKDH on either system 11 (Haufler, 1985
) or a morpholine/citrate system (Werth, 1985
). Of the 12 enzyme systems surveyed, five yielded consistent, interpretable banding patterns for most populations (TPI, PGI, 6PGD, MDH, and DIA). Data from the other seven loci (Aat, F1-6, G3pdh, Idh, Lap, Pgm, and Skdh) were not included in the analyses. Photographs were taken of all gels using Kodak Technical Pan film and a red filter. Genetic interpretations of banding patterns assumed that subunit structure and cellular compartmentalization in Botrychium are similar to those of other plants (Gottlieb, 1982
). Loci and alleles were numbered sequentially from the most anodal to the most cathodal (e.g., Mdh-1, Mdh-2, etc.). Loci are reported using standard numbers (e.g., 2, 3), whereas alleles are denoted by superscripts (e.g., 1,1). All band profiles presented in Table 3 reflect the presumed ploidy of individual sporophytes. Thus, diploids are shown with two alleles (e.g., 2,2) at a single locus, whereas tetraploids have four alleles (e.g., 2,2;3,3), two at each of the two duplicated loci, and hexaploids have six alleles across three loci (e.g., 2,2;3,3;4,4). However, in the single hexaploid species, B. pseudopinnatum Wagner, banding patterns were identical to those seen in tetraploids and the genetic constitution of the two alleles presumably encoded at a third locus could not be inferred and are reported in Table 3 as (?,?). Homozygotes (Fig. 3) have two copies of the same allele (1,1), whereas heterozygotes (Fig. 4) have different alleles (1,3). Allelic heterozygosity (1,3) segregates during meiosis, whereas fixed interlocus heterozygosity (1,1;3,3) does not. When heterozygosity was fixed (i.e., all sporophytes sampled revealed the same multiple-banded pattern; Fig. 4) in a species known to be polyploid, it was assumed that the heterozygous patterns resulted from duplicate gene expression rather than allelic heterozygosity at a single locus. Because Botrychium gametophytes are difficult to culture (Whittier, 1981
), studies of allelic segregation to verify the interpretation of banding patterns were not attempted.
View this table:
[in this window]
[in a new window]
|
Table 3. Catalog of individual, sporophytic genotypes detected within populations of subgenus Botrychium species. The number of individuals carrying each genotype is indicated in parentheses after the abbreviation of the population name. The number after each species name indicates the documented ploidy level of the species (2x = diploid, 4x = tetraploid, 6x = hexaploid) where x = 45. Genotypes of species with unknown ploidy are reported as diploid. See Materials and Methods for explanation of the interpretation of banding patterns. Population numbers and collection sites are referenced in Table 2.
|
|
View larger version (117K):
[in this window]
[in a new window]
|
Figs. 3–4. Representative gel photographs showing Pgi banding patterns typical of diploid and polyploid species. The upper set of bands was designated Pgi-1 and the lower set Pgi-2. 3. The diploid B. lunaria showed homozygous, one-banded patterns at both loci. 4. The tetraploid B. matricariifolium showed homozygous patterns at Pgi-1 and heterozygous, three-banded patterns at Pgi-2.
|
|
Statistics of genetic variation
Once genetic interpretation of banding patterns was obtained, a data matrix was developed according to the specifications of the BIOSYS-L program manual (Swofford and Selander, 1989
). Botrychium lineare was the only species included in this study for which ploidy was not known. The banding patterns of individuals of B. lineare observed on the isozyme gels were typical of diploids (i.e., no fixed heterozygosity at any locus) and B. lineare was assumed to be diploid for the purpose of data analysis. BIOSYS-L was used to calculate five statistics of genetic variation (Tables 4–6): P (the percentage loci polymorphic), A (the mean number of alleles per locus), H (the mean expected heterozygosity), I (Nei's [1978]
genetic identity), and D (Nei's [1978]
genetic distance). UPGMA cluster analyses were performed on isozyme profiles of all diploid populations.
View this table:
[in this window]
[in a new window]
|
Table 4. Mean heterozygosity per locus (H), mean number of alleles per locus (A), and percentage of loci polymorphic (P) for 16 species of subgenus Botrychium.
|
|
View this table:
[in this window]
[in a new window]
|
Table 5. Values of Nei's (1978) unbiased genetic identity and distance for infraspecific comparisons among populations of six diploid species of Botrychium subgenus Botrychium. Values above diagonal are genetic identity and values below diagonal are genetic distance.
|
|
View this table:
[in this window]
[in a new window]
|
Table 6. Values of Nei's (1978) unbiased genetic identity for interspecific comparisons among six diploid species of Botrychium subgenus Botrychium.
|
|
Evaluation of polyploid species
Because meaningful comparisons among polyploid isozyme profiles are often constrained by a lack of relevant information (Werth, Hilu, and Langner, 1994
) polyploid banding patterns were analyzed by reconstructing ancestral diploid genotypes. Ranker et al. (1989)
and Werth, Hilu, and Langner (1994)
compared isozyme patterns of diploids and polyploids by reconstructing two hypothetical, ancestral diploid genotypes from tetraploid isozyme patterns. The hypothetical diploid genotypes were analyzed with diploid isozyme profiles in UPGMA cluster analyses to establish genetic affinities and generate hypotheses of polyploid origins.
Before reconstruction of ancestral isozyme profiles from polyploid banding patterns, it was necessary to establish isozyme profiles for the diploid species surveyed (Table 8). To simplify computations, we selected the most common isozyme profile(s) observed in each diploid species. However, we excluded diploid populations with patterns of fixed heterozygosity because they may have been of hybrid origin (see below). All populations were standardized to a sample size of ten individuals to minimize the influence of sample size on the cluster analyses.
View this table:
[in this window]
[in a new window]
|
Table 7. Chi-square values calculated to test levels of heterozygosity for departure from Hardy-Weinberg expectations for the isozyme locus Pgi-2.
|
|
View this table:
[in this window]
[in a new window]
|
Table 8. Isozyme profiles of six diploid subgenus Botrychium species used for cluster analyses with hypothetical diploid genotypes reconstructed from polyploid isozyme patterns. Population abbrevations are listed in Table 2.
|
|
Following Ranker et al. (1989)
and Werth, Hilu, and Langner (1994)
, we separated isozyme patterns from each polyploid into a "C" or chloroplast parental genotype and an "N" or nonchloroplast parental genotype (Table 9). Although maternal inheritance of the cpDNA has not been demonstrated in subgenus Botrychium, Gastony and Yatskievych (1992)
showed that in cheilanthoid ferns inheritance of the cpDNA is maternal. Hauk (1995)
found that polyploid subgenus Botrychium species possessed a single rbcL gene sequence and that the sequence is similar or identical to that of a single diploid. Thus, the probable chloroplast parent of many subgenus Botrychium polyploids was identified. In most cases, the isozyme profile of the diploid identified as the chloroplast parent by rbcL data served as the basis for reconstruction of hypothetical, ancestral genotypes for each polyploid (see the Appendix).
View this table:
[in this window]
[in a new window]
|
Table 9. Hypothetical diploid isozyme profiles reconstructed from polyploid isozyme profiles of ten subgenus Botrychium species. "C" and "N" stand for chloroplast and nonchloroplast genotypes, respectively. Population abbreviations are listed in Table 2.
|
|
|
RESULTS
|
|---|
Isozyme loci
The five enzyme systems resolved for all populations yielded a total of six loci for analysis. Table 3 lists isozyme genotypes interpreted from isozyme banding patterns. Five of six loci revealed infraspecific variation, with most variation partitioned among rather than within populations. In the following section, we present our interpretation of each enzyme system with a summary of the distribution of allelic and genotypic variation across the 16 species surveyed.
Pgi
Two banding regions were present (Pgi-1 and Pgi-2; Figs. 3–4), but the more anodal region (Pgi-1) did not yield consistently interpretable bands for all taxa and was not included in the analyses. Five alleles were observed in Pgi-2 among the 16 species surveyed. Pgi-2 was monomorphic within six species [B. lunaria (L.) Swartz, B. crenulatum Wagner, and B. pinnatum St. John contained Pgi-24; B. spathulatum Wagner, B. pumicola Coville, and B. lineare had Pgi-23]. One of two different alleles was fixed in populations of B. ascendens Wagner (Pgi-24 or 1) and B. simplex (Pgi-23 or 4). Pgi-23,3:5,5 and 2,2:5,5 heterozygotes were present in B. hesperium, and B. acuminatum, whereas B. matricariifolium contained Pgi-21,1:5,5 and 2,2:5,5 heterozygotes. Pgi-22,2:5,5 heterozygotes were observed but not fixed in B. lanceolatum, whereas Pgi-22,2:5,5 was fixed in B. pedunculosum and B. echo. Botrychium pseudopinnatum was the only species with a fixed Pgi-22,2:4,4 pattern. Botrychium minganense was the most variable species and the only taxon that possessed all five Pgi-2 alleles. Most populations were fixed for either Pgi-2 3,3;3,3, Pgi-24,4:4,4 or Pgi-25,5:5,5. However, population 2 of B. minganense was uniformly heterozygous (Pgi-22,2;4,4) and populations 1 and 10 contained both homozygotes (Pgi-21,1:1,1 or 4,4:4,4) and heterozygotes (Pgi-21,1:4,4).
Tpi
Other studies have demonstrated that Tpi is compartmentalized in plant cells (Gottlieb, 1982
: Gastony and Darrow, 1983
), and we interpreted the two banding regions present in subgenus Botrychium Tpi patterns as separate loci, the more anodal region designated Tpi-1, and the more cathodal set Tpi-2. In contrast to nearly all other pteridophytes, subgenus Botrychium species had simple banding patterns. There was no evidence of either duplications or post-translational modifications of either Tpi locus (see Gastony, 1988
; Hickey, Guttman, and Eshbaugh, 1989
). However, Watano and Sahashi (1992)
reported multiple Tpi banding patterns in four species of Botrychium subgenus Sceptridium and interpreted the bands as duplicated loci. The absence of multiple Tpi banding patterns in subgenus Botrychium may indicate the reversal of a pleisiomorphic Tpi duplication common to other pteridophytes. Tpi-1 had three alleles and was the only locus that showed no infraspecific variability. Tpi-11 was present for B. simplex, and Tpi-13 was observed in B. lunaria and B. pumicola. All other species were monomorphic for Tpi-12.
Most taxa were monomorphic for Tpi-22 (B. lanceolatum, B. matricariifolium, B. echo, B. hesperium, B. acuminatum, B. pedunculosum, B. pinnatum, B. pseudopinnatum, B. pumicola, B. crenulatum, and B. lineare), but Tpi-21 was present in B. simplex. Six populations of B. lunaria were monomorphic for Tpi-23 and population 5 was heterozygous (Tpi-22,3). Most individuals of B. minganense were Tpi-22,2;3,3 heterozygotes, although a few Tpi-23,3;3,3 and/or 2,2;2,2 homozygotes were observed in populations 3, 4, 6, and 8. Populations 5, 7, 9, and 11 of B. minganense were fixed for Tpi-22,2;2,2.
Dia
There were several regions of Dia activity on the gels, although the most anodal set of bands (Dia-1) was the only one that was resolved consistently. Two alleles were detected. Dia-12 was monomorphic in nine taxa (B. lunaria, B. crenulatum, B. ascendens, B. spathulatum, B. matricariifolium, B. acuminatum, B. pseudopinnatum, B. pumicola, and B. lineare). Ten of the 11 populations of B. minganense were fixed for Dia-12,2;2,2; Dia-13 was observed in population 1 of B. minganense and was heterozygous (Dia-11,1;3,3). Depending on ploidy, Dia-11,2 or Dia-11,1;2,2 heterozygotes were present in B. lanceolatum, B. echo, B. hesperium, B. pedunculosum, and B. pinnatum. Five populations (four of B. lanceolatum and one of B. simplex) were monomorphic for Dia-11,1, but the allele was never monomorphic across all populations of a single species.
Mdh
Mdh had several regions of activity on the gels, but only the most anodal set of bands was consistently interpretable (Mdh-1). Three alleles were distributed among the 16 species surveyed. Mdh-11 was monomorphic in six species (B. simplex, B. pumicola, B. crenulatum, B. lineare, B. spathulatum, and B. echo). The most variable species was B. minganense, which possessed individuals monomorphic for each of the three Mdh-1 alleles. The majority of B. minganense populations had fixed Mdh-11,1;3,3 patterns, and population 10 was fixed for Mdh-11,1;2,2. Botrychium matricariifolium, B. acuminatum, and B. pseudopinnatum had invariant Mdh-11,1;2,2 patterns. Six of seven populations of B. lanceolatum were monomorphic for Mdh-12, and Mdh-11,2 was fixed in population 2. Five of seven populations of B. lunaria were monomorphic for Mdh-13; populations 4 and 5 had fixed Mdh-11,3 patterns. One population each of B. hesperium and B. pinnatum was fixed for Mdh-12,2;2,2, and the second population of each showed fixed Mdh-11,1;2,2 patterns. Of the three B. ascendens populations, one had a Mdh-11,1;3,3 pattern, and the other two were fixed for Mdh-11,1;2,2.
6Pgd
Two 6Pgd banding regions have been reported in many taxa (Gottlieb, 1982
). In Botrychium, however, only the more cathodal of the two areas (6Pgd–2) produced consistently interpretable bands. Four alleles were detected across all species. Four taxa were monomorphic: B. pumicola and B. lineare for 6Pgd–21, B. pedunculosum for 6Pgd–23, and B. crenulatum for 6Pgd–24. Botrychium simplex was homozygous (6Pgd–21,1) in all populations except population 1, which was heterozygous (6Pgd–21,2). Botrychium matricariifolium, B. echo, B. hesperium, and B. acuminatum were heterozygous for 6Pgd–21,1;2,2. Botrychium ascendens, B. spathulatum, and B. pseudopinnatum were fixed for 6Pgd–21,1;4,4. 6Pgd–23,3;4,4 patterns were found in only one species, B. pinnatum. Botrychium minganense was heterozygous (6Pgd–21,1;4,4) in all but population 5 where it was homozygous (6Pgd–21,1;1,1). Five populations of B. lanceolatum showed fixed 6Pgd–22,2 patterns and two populations had 6Pgd–21,2 patterns. Botrychium lunaria was homozygous (6Pgd–24,4) in five populations and heterozygous (6Pgd–21,2) in two.
Statistics of genetic variation
Across all species there was a mean of 3.33 alleles per locus. Within species, however, there was a mean of 1.30 alleles per locus (diploids = 1.08, tetraploids = 1.45, hexaploid = 1.50, unknown ploidy = 1.00). Mean heterozygosity per locus (H) for diploids was 3.5%, for tetraploids 21.6%, for the hexaploid B. pseudopinnatum 28.6%, and for B. lineare (uncertain ploidy) 0% (Table 4). The percentage of loci polymorphic (P) was 7.33 for diploids, 43.01 for tetraploids, 50.0 for the hexaploid, and 0.0 for B. lineare (Table 4).
Infraspecific values of Nei's unbiased genetic identity and distance were calculated among 21 populations of four diploid species: five populations of B. simplex, seven of B. lunaria, seven of B. lanceolatum, and two of B. crenulatum (Table 5). These diploids represented three of the sections proposed by Wagner and Wagner (Table 1). Infraspecific genetic identity values ranged from 0.67 to 1.00 with a mean of 0.912. Interspecific values were computed, based on mean allele frequencies, from the same database as infraspecific comparisons, except for the addition of single populations of B. lineare and B. pumicola (Table 6). The values for interspecific genetic identity ranged from 0.045 to 0.833 with a mean of 0.409.
Chi-square analysis for goodness-of-fit to Hardy-Weinberg equilibrium was calculated for the genotype frequencies (Pgi-2) observed within population 4 of B. simplex and populations 1, 2, and 7 of B. lanceolatum (Table 7). All chi-square values were significant (P < 0.05) when compared to the critical value, and the null hypothesis of random mating was rejected.
UPGMA cluster analysis of interpopulational genetic identities produced a dendrogram of diploid population relationships (Fig. 5). Populations of B. lanceolatum, B. lunaria, B. crenulatum, and B. simplex associated most strongly with other populations of the same species. Populations of B. lanceolatum had low genetic identities with all other populations included in the analysis. Populations of B. crenulatum clustered more closely to populations of B. lunaria than to populations of any other species. The single populations of B. lineare and B. pumicola were more similar to each other than to populations of any other species.
View larger version (14K):
[in this window]
[in a new window]
|
Fig. 5. Cluster analysis of isozyme data from 23 populations of six diploid subgenus Botrychium species based on Nei's genetic identity (1978)
of isozyme data, using the unweighted pair group method. CRE = B. crenulatum, LAN = B. lanceolatum, LIN = B. lineare, LUN = B. lunaria, PUM = B. pumicola, SIM = B. simplex. Population abbreviations are referenced in Table 2
. The scale represents the range of genetic identity values.
|
|
Cluster analyses of hypothetical diploid isozyme genotypes of polyploid species
UPGMA cluster analyses of Nei's (1978)
genetic identities were conducted on the reconstructed hypothetical diploid genotypes plus the diploid isozyme profiles. A single analysis of all 73 diploid and hypothetical diploid genotypes was not conducted because BIOSYS-L permits simultaneous analysis of only 60 populations. To circumvent this limitation, we conducted four separate analyses: (1) diploids plus hypothetical genotypes from B. spathulatum and B. ascendens, (2) diploids plus hypothetical genotypes from B. minganense, (3) diploids plus hypothetical genotypes from B. matricariifolium, B. acuminatum, B. hesperium, and B. echo, (4) diploids plus hypothetical genotypes from B. pedunculosum, B. pinnatum, and B. pseudopinnatum (Tables 8, 9).
Reconstructed genotypes from the tetraploids B. spathulatum and B. ascendens were included in the same cluster analysis (Fig. 6) because they belonged to the same rbcL subclade (Hauk, 1995
). The B. ascendens 1 N genotype was identical to profiles of B. crenulatum populations. The B. ascendens 2 and 3 N genotypes grouped closely to B. crenulatum profiles. All three B. ascendens C genotypes clustered and formed a group independent of any sampled diploid. The B. spathulatum C genotype associated with B. lineare, and the N genotype grouped closest to the B. ascendens N genotypes and the B. crenulatum profiles.
View larger version (16K):
[in this window]
[in a new window]
|
Fig. 6. Cluster analysis of isozyme data from 23 populations of six diploid species plus the reconstructed chloroplast (C) and nonchloroplast (N) hypothetical genotypes from the tetraploids B. ascendens and B. spathulatum, based on Nei's (1978)
genetic identity, using the unweighted pair-group method. See Table 8
for diploid isozyme profiles, and Table 9
for reconstructed hypothetical diploid genotypes. The scale represents the range of genetic identity values.
|
|
Because rbcL analysis did not associate B. minganense with any diploid species surveyed isozymically, the hypothetical B. minganense N and C genotypes were analyzed separately from those of all other polyploids (Fig. 7). All populations except 5 and 11 contained hypothetical N genotypes identical to the B. crenulatum isozyme profile. The population 5 N and C genotypes were identical to the isozyme profile of B. lineare, and the population 11 N genotype grouped closest to the B. crenulatum profiles. The hypothetical B. minganense C genotypes clustered to form a group distinct from all diploids sampled. However, the B. minganense C genotypes were not identical; similar levels of genetic distance were observed among B. minganense C genotypes as were detected between the diploids B. lineare and B. crenulatum.
View larger version (17K):
[in this window]
[in a new window]
|
Fig. 7. Cluster analysis of isozyme data from 23 populations of six diploid species plus the reconstructed chloroplast (C) and nonchloroplast (N) hypothetical genotypes from the tetraploid B. minganense, based on Nei's (1978)
genetic identity, using the unweighted pair-group method. See Tables 8
and 9
for diploid isozyme profiles and reconstructed hypothetical diploid genotypes. The scale represents the range of genetic identity values.
|
|
The four tetraploids, B. matricariifolium, B. hesperium, B. echo, and B. acuminatum, had rbcL sequences identical to that of B. lanceolatum (Hauk, 1995
) and were included in the same cluster analysis (Fig. 8). The hypothetical B. matricariifolium, B. echo, and B. hesperium C genotypes were identical to the B. lanceolatum isozyme profiles. The two B. acuminatum C genotypes clustered separately from, but closely related to, the B. lanceolatum profiles. The N genotypes clustered as four distinct groups: (1) the four B. matricariifolium NA genotypes, (2) the four B. matricariifolium NB genotypes plus the B. echo 1 and B. acuminatum 2 N genotypes, (3) the B. acuminatum 1 and B. hesperium 1 N genotypes plus the B. lineare profile, and (4) the B. hesperium 2 N genotype.
View larger version (16K):
[in this window]
[in a new window]
|
Fig. 8. Cluster analysis of isozyme data from 23 populations of six diploid species plus reconstructed chloroplast (C) and nonchloroplast (N) hypothetical genotypes from the tetraploids B. matricariifolium, B. hesperium, B. echo, and B. acuminatum, based on Nei's (1978)
genetic identity, using the unweighted pair group method. See Table 8
for diploid isozyme profiles and Table 9
for reconstructed hypothetical diploid genotypes. The scale represents the range of genetic identity values.
|
|
Reconstructed N and C genotypes of B. pedunculosum, B. pinnatum, and B. pseudopinnatum (Fig. 9) were analyzed together because they possessed rbcL sequences identical to that of B. lanceolatum, but each of these polyploids lacked isozyme alleles present in B. lanceolatum (e.g., both B. pinnatum and B. pseudopinnatum lack Pgi-15, the diagnostic marker for B. lanceolatum). The B. pinnatum C genotypes clustered closest to B. crenulatum profiles, whereas the N genotypes did not cluster close to any diploid profile. Neither hypothetical B. pseudopinnatum genotype grouped with B. lanceolatum profiles. The B. pseudopinnatum C genotype clustered with the B. pinnatum N genotype, and the B. pseudopinnatum N genotype grouped closest to the B. lineare profile. The B. pedunculosum N and C genotypes clustered closer to each other than to any extant diploid.
View larger version (16K):
[in this window]
[in a new window]
|
Fig. 9. Cluster analysis of isozyme data from 23 populations of six diploid species plus the reconstructed chloroplast (C) and non-chloroplast (N) hypothetical genotypes from the polyploids B. pinnatum, B. pseudopinnatum, and B. pedunculosum, based on Nei's (1978)
genetic identity, using the unweighted pair group method. See Table 8
for diploid isozyme profiles, Table 9
for reconstructed hypothetical diploid genotypes. The scale represents the range of genetic identity values.
|
|
Diploid populations with polyploid isozyme patterns
Four diploid populations exhibited fixed patterns of heterozygosity (Table 3): populations 4 and 5 of B. lunaria, and populations 1 and 2 of B. lanceolatum. Patterns of fixed heterozygosity were typical of polyploid species and infrequent in diploid species. Therefore, it is possible that these diploid populations are either of hybrid origin, or that they are polyploids "masquerading" as diploids. In order to determine possible sources of the fixed heterozygosity observed in these four "diploid" populations, we separated the diploid isozyme profiles of these populations into two hypothetical genotypes: one representing the typical profile of the diploid species to which the populations were identified (N = near), and a second profile that contained the alleles not usually found in either B. lunaria or B. lanceolatum (F = far). The hypothetical profiles were analyzed in the same fashion as the hypothetical genotypes generated from polyploid isozyme patterns (see the Appendix). The B. lunaria 4 and 5 hypothetical F genotypes associated with populations of B. crenulatum, B. lineare, and B. pumicola (Fig. 10). The B. lanceolatum 2 F genotype clustered with B. lineare, and the B. lanceolatum 1 F genotype did not cluster close to any diploid profile (Fig. 10).
View larger version (16K):
[in this window]
[in a new window]
|
Fig. 10. Cluster analysis of isozyme data from 19 populations of six diploid subgenus Botrychium species plus hypothetical genotypes from four "diploid" populations, where "N" and "F" refer to the near and far reconstructed genotypes from each of the "diploid" populations. Based on Nei's genetic identity (1978)
, using the unweighted pair-group method. CRE = B. crenulatum, LAN = B. lanceolatum, LIN = B. lineare, LUN = B. lunaria, PUM = B. pumicola, and SIM = B. simplex.
|
|
|
DISCUSSION
|
|---|
Infraspecific comparisons for diploid species
Infraspecific comparisons can be used to develop hypotheses concerning the reproductive biology and evolutionary history of species (Hamrick, 1982
). In pteridophytes, mean genetic identities for conspecific populations range from 0.780 to 0.996 (Soltis and Soltis, 1990
) and the mean for diploid Botrychium species (0.912) fell within this range (Table 5). The populations sampled of the diploid species represented a substantial portion of the geographic distribution of each species, yet genetic identities did not always correlate with geographic distance between populations. For example, in B. simplex, the genetic identity between two populations from Oregon (4 and 5) was 0.833, whereas population 4 was allelically identical to population 3 from Michigan. Gene flow between the Oregon (4) and Michigan (3) populations is not a likely explanation for the high genetic identity detected. Reasons for the lack of correlation between diploid isozyme profiles and geographic distribution are unclear.
Breeding system(s) of diploid subgenus Botrychium species
Based on accumulated data from isozyme studies (e.g., Hamrick, Linhart, and Mitton, 1979
; Hamrick and Godt, 1989
), inbreeding species usually have most of their isozymic variation partitioned among populations, whereas outcrossers generally have most isozymic variation contained within individual populations. Following this line of reasoning, the relatively high mean value for genetic identity among diploid subgenus Botrychium populations indicated a tendency toward outcrossing. However, in diploid subgenus Botrychium species, little or no isozyme variation was observed within populations (Table 3); only four of 23 diploid populations possessed more than a single genotype. With such a small amount of infraspecific variation, it is possible to have high genetic identities between populations and still have most of the total genetic variation partitioned among populations (rather than contained within them). Thus, the limited data available on within-population structuring of genetic variation in these diploids could not exclude either inbreeding or outcrossing as the predominant breeding system of subgenus Botrychium.
Although low amounts of allozyme variation were detected within subgenus Botrychium populations, four diploid populations could be used to evaluate reproductive modes. Population 6 of B. simplex and populations 1, 2, and 7 of B. lanceolatum were the only diploid populations that contained more than one allele at any locus without exhibiting fixed heterozygosity (Table 3). For example, in population 6 of B. simplex, 24 of 30 individuals were homozygous for one Pgi-2 allele, four were homozygous for a second, and two individuals were heterozygous. A chi-square analysis of these data (Table 7) yielded a highly significant value (P < 0.001). These data did not support the null hypothesis of random mating but rather indicated that nonrandom mating (in this case a high frequency of inbreeding) is predominant in this population. Populations 1, 2, and 7 of B. lanceolatum exhibited similar nonrandom mating patterns (Table 7).
Data from four populations of two species cannot be extrapolated to characterize the breeding systems of an entire subgenus of species, but the information obtained from the B. simplex and B. lanceolatum populations substantiates other evidence concerning breeding systems in Botrychium. McCauley, Whittier, and Reilly (1985)
and Watano and Sahashi (1992)
reported inbreeding in diploid species of the closely related subgenus Sceptridium, and Soltis and Soltis (1986)
concluded that B. virginianum (subgenus Osmundopteris) is predominantly inbreeding. Although interspecific hybrids between species of subgenus Botrychium are documented (Wagner, 1980
, 1991
; Wagner et al., 1984
; Wagner, Wagner, and Beitel, 1985
; Wagner and Wagner, 1988
), individuals of several species typically grow sympatrically and yet hybridization appears to be rare. All Botrychium species have subterranean gametophytes that may hinder sperm transfer (Tryon and Tryon, 1982
) and reduce the opportunity for outcrossing. Therefore, all available evidence is consistent with inbreeding as the most probable breeding system operating in Botrychium.
Interspecific comparisons for populations of diploid species
Interspecific values of Nei's (1978)
genetic identity for six diploid species ranged from 0.045 to 0.833 with a mean of 0.409 (Table 6). These genetic identities were relatively low when compared to conspecific values (Table 5) and indicated that considerable genetic differentiation has occurred among the diploids (Fig. 5). Gottlieb (1981)
and Crawford (1983)
reported a mean genetic identity of 0.67 for congeneric angiosperm species, whereas congeneric ferns have much lower values, averaging 0.33 (Soltis and Soltis, 1990
). The Botrychium mean (0.409) was consistent with typical values for pteridophytes. Assuming a constant rate of mutation, the low genetic identity values suggested that diploid Botrychium species are older than many angiosperm species and similar in age to most extant fern species.
Among some diploid species, interrelationships suggested by isozymes correspond to those based on rbcL analysis (Hauk, 1995
). In the isozyme analysis (Fig. 5), the species pair B. lunaria and B. crenulatum is more similar to the group composed of B. simplex, B. pumicola, and B. lineare than either group is to B. lanceolatum. Similarly the rbcL tree (Fig. 2) placed the "lunaria" and "simplex-campestre" clades as sister groups and the "lanceolatum" clade as the earliest diverging lineage of subgenus Botrychium. Morphology supports the position of B. lanceolatum as sister to all other species of subgenus Botrychium; the leaves of B. lanceolatum are ternate, a condition that occurs within the closely related subgenera Sceptridium and Osmundopteris, whereas all other subgenus Botrychium species exhibit a presumably more derived pinnate leaf construction. Concordance of isozymic, DNA, and morphological data indicates that the "lanceolatum" clade was the earliest diverging lineage of subgenus Botrychium and that the "lunaria" and "simplex-campestre" clades are more derived species assemblages.
Levels of genetic identity among species, however, did not always correlate with morphological similarity, or with hypotheses of relationships suggested by rbcL sequence analysis. Botrychium lineare and B. pumicola, for example, aligned with different subclades in rbcL analyses (Hauk, 1995
) and are morphologically distinct, but had the highest isozymic interspecific genetic identity value (0.833) among the species sampled, differing only at Tpi-1. In contrast, B. simplex and B. pumicola, which are similar morphologically and possess identical rbcL sequences, had a mean genetic identity of only 0.589. Similarly, the formerly conspecific B. crenulatum and B. lunaria have identical rbcL sequences, yet the two species had a genetic identity value of 0.535. Species with low genetic identity values and identical rbcL sequences may have divergent nuclear genotypes but share a common chloroplast ancestor, possibly as a result of introgression (Rieseberg, Beckstrom-Sternberg, and Doan, 1990
; Rieseberg and Brunsfeld, 1991
) or lineage sorting (Doyle, Doyle, and Brown, 1990
).
Diploid/polyploid relationships
Isozymic patterns among polyploid species were not easily interpretable because overlapping of isozyme profiles among polyploid species was prevalent. Undoubtedly this is a consequence of some diploid species (i.e., B. lanceolatum) contributing to the origin of several polyploids (Fig. 1). Separating polyploid isozyme patterns into hypothetical diploid genotypes and analyzing these hypothetical genotypes with diploid isozyme profiles provide one method of unraveling reticulate relationships among morphologically cryptic species. Furthermore, this method may be the only currently available means of simultaneously analyzing diploid and polyploid isozyme data.
The origins of B. spathulatum and B. ascendens
Morphological characters indicate that B. spathulatum is an allopolyploid derivative of the diploids B. lunaria and B. campestre (Fig. 1), and rbcL data identified B. campestre as the chloroplast parent of B. spathulatum (Fig. 2). Botrychium campestre (not surveyed isozymically) and B. lineare (surveyed) are sister species based on both morphological (Wagner and Wagner, 1994
) and rbcL data (Hauk, 1995
). Although sister species frequently do not possess identical isozyme profiles (e.g., B. lunaria and B. crenulatum), at every isozyme locus examined, B. spathulatum C genotype possessed all alleles found in B. lineare. If B. campestre is the chloroplast parent of B. spathulatum, the lack of isozyme divergence between the B. spathulatum C genotype and B. lineare suggests that B. campestre and B. lineare possess similar or identical isozyme profiles.
Isozyme data did not lend support to the hypothesis that B. lunaria is the nonchloroplast parent of B. spathulatum. Confusion between polyploid derivatives of B. lunaria and B. crenulatum is possible because the two species have similar fan-shaped pinnae and were once considered conspecific. However, B. lunaria is generally larger, more robust, and a darker green than the smaller, more delicate, yellow-green B. crenulatum. Some populations of B. spathulatum strongly resemble B. lunaria morphologically, whereas other populations more closely approximate B. campestre in form (W. D. Hauk, personal observation); population 1 of B. spathulatum is smaller and more delicate than populations of B. spathulatum collected near Marathon, Ontario. As currently circumscribed, B. spathulatum may represent a taxon composed of allopolyploid derivatives of B. campestre x B. lunaria and B. campestre x B. crenulatum. This hypothesis is consistent with the morphological variation present in B. spathulatum and should be tested. Additional sampling of B. spathulatum populations may reveal populations that possess the diagnostic alleles of B. lunaria.
Botrychium ascendens is hypothesized to have originated as an allotetraploid involving B. montanum and B. crenulatum (Fig. 1), although it did not possess the rbcL sequence of either putative diploid parent (Fig. 2). Similarly, the three B. ascendens C genotypes did not associate with any diploid surveyed isozymically (Fig. 6). Thus, both rbcL and isozyme evidence indicate that an unsampled and perhaps unknown diploid contributed its chloroplast genome to B. ascendens. Clustering of the three B. ascendens N genotypes with B. crenulatum is consistent with the hypothesis that B. crenulatum is the source of the B. ascendens N genotype.
The origin of B. minganense
The most isozymically variable taxon sampled was B. minganense, a tetraploid species distributed across northern and western regions of North America (Wagner and Wagner, 1993
). Traditional morphologically based hypotheses implicate the diploids B. lunaria and B. pallidum as progenitors of B. minganense (Fig. 1). However, only one of these putative ancestors, B. lunaria, was available for isozyme analysis. Despite the apparent close relationship between B. minganense and B. lunaria, no population of B. minganense possessed the marker allele Tpi-13 of B. lunaria (Fig. 11). However, all populations of B. minganense (except IP) possessed a complement of isozyme bands consistent with B. crenulatum ancestry (Fig. 11), and B. crenulatum appears to be the source of the B. minganense N genotype.
View larger version (106K):
[in this window]
[in a new window]
|
Fig. 11. Gel photographs showing Tpi banding patterns of the diploid species B. crenulatum (CRE) and B. lunaria (LUN), and the putative allotetraploid B. minganense (MIN). The upper set of bands was designated Tpi-1 and the lower set Tpi-2. Botrychium crenulatum and B. minganense share Tpi-12, whereas B. lunaria contains Tpi-13.
|
|
The origin of the B. minganense C genotype is more uncertain. Analysis of rbcL sequences failed to identify an extant diploid species that could be the chloroplast progenitor of B. minganense (Fig. 2). Unlike the nearly identical B. minganense N genotypes, the B. minganense C genotypes contained more genetic variation than that observed in any diploid sampled (Table 10). Clustering of all B. minganense C genotypes (Fig. 7) indicated that a single, variable species could explain the chloroplast parentage of B. minganense. If equal mutation rates occur in both the N and C genomes of a polyploid, nearly equal amounts of genetic variation should accumulate over time in each genome. The relatively high genetic variability among the B. minganense C genotypes and the low variability among the N genotypes (and in B. crenulatum) may indicate that the genetic variability observed in B. minganense was inherited from its diploid chloroplast progenitor(s), rather than accumulating after speciation. If this scenario is correct, then B. minganense arose through multiple hybridization events that captured and preserved subsets of the genetic variability contained in the unidentified chloroplast parent (see Werth, 1989
; Werth, Gutman, and Eshbaugh, 1985
).
View this table:
[in this window]
[in a new window]
|
Table 10. Mean values of Nei's (1978) unbiased genetic identity for infraspecific comparisons among populations of diploid species and hypothetical diploid genotypes derived from polyploid isozyme patterns. "C" and "N" stand for chloroplast and nonchloroplast genotypes, respectively.
|
|
Although rbcL evidence demonstrated that B. minganense and B. lunaria are not members of the same rbcL clade, the diploid most closely associated with the B. minganense C genotypes was B. lunaria (Fig. 7). This apparent conflict between isozyme and rbcL data may be a consequence of two factors: (1) the most likely source of the B. minganense C genotype is B. pallidum, a diploid that was not surveyed isozymically, and (2) the small number of isozyme loci analyzed may limit the resolution of the analysis. Isozyme analysis of B. pallidum may reveal that it more closely resembles the B. minganense C genotype than does B. lunaria. Alternatively, resolution of more isozyme loci may differentiate more clearly the B. minganense C genotypes from B. lunaria.
A single population (5) of B. minganense from Independence Pass, Colorado, did not exhibit the interlocus heterozygosity observed in all other B. minganense populations and possessed simple isozyme patterns more typical of diploid species. Individuals of population 5 were indistinguishable morphologically from other collections of B. minganense, yet the isozyme profile of population 5 clustered with B. lineare (Fig. 7). Population 5 was the most isozymically divergent of all the B. minganense populations sampled. The distinctive isozymic profile of population 5 supported, in part, the hypothesis that there may be hidden geographic variants within B. minganense (W. H. Wagner, Jr., personal communication), although the absence of rbcL sequence variation between eastern and western populations of B. minganense did not support the hypothesis of undetected cryptic species (Hauk, 1995
).
The origins of B. acuminatum, B. hesperium, B. matricariifolium, and B. echo
The diploid B. lanceolatum is thought to be one diploid progenitor of the tetraploids B. matricariifolium, B. acuminatum, B. hesperium, and B. echo (Fig. 1), and rbcL data indicate that B. lanceolatum is the chloroplast parent of each polyploid (Hauk, 1995
). Isozyme data provided further support for this relationship because B. matricariifolium, B. acuminatum, B. hesperium, and B. echo shared Pgi-25 with B. lanceolatum (Fig. 12), and the C genotypes of each polyploid clustered closely with B. lanceolatum population profiles (Fig. 8). However, clustering of the N genotypes from these four tetraploids (Fig. 8) did not correlate well with current hypotheses of species boundaries. The lack of association between N genotypes from different populations of B. hesperium (or B. acuminatum) suggested that this species arose from more than one hybridization event. The clustering of N genotypes from B. matricariifolium, B. echo, and B. acuminatum (Fig. 8) did not lend support to the hypothesis that each of these three tetraploids arose from a different nonchloroplast parent species (Fig. 1). Furthermore, the complex relationships among these N genotypes did not resolve the nonchloroplast parentage of B. matricariifolium, B. acuminatum, B. hesperium, or B. echo.
View larger version (41K):
[in this window]
[in a new window]
|
Fig. 12. Gel photograph showing Pgi banding patterns of eight subgenus Botrychium species. The upper set of bands represents Pgi-1 and the lower set Pgi-2. The Pgi-25 allele of B. lanceolatum (LAN) is present in the polyploids B. hesperium (HES), B. echo (ECH), B. pedunculosum (PED), B. matricariifolium (MAT), and B. acuminatum (ACU), but is absent from B. pinnatum (PIN) and B. pseudopinnatum (PSE).
|
|
The origins of B. pedunculosum, B. pinnatum, and B. pseudopinnatum
Botrychium lanceolatum is the chloroplast ancestor of three other polyploid species: B. pedunculosum, B. pinnatum, and B. pseudopinnatum (Fig. 2; Hauk, 1995
). However, these three polyploids did not possess isozyme profiles consistent with B. lanceolatum ancestry. For example, neither B. pinnatum nor B. pseudopinnatum possessed Pgi-25 (Fig. 12), an allele found only in B. lanceolatum and known polyploid derivatives of B. lanceolatum (B. matricariifolium, B. hesperium, B. echo, and B. acuminatum). Likewise, B. pedunculosum was homozygous for 6Pgd–23, an allele not detected in B. lanceolatum (or any other diploid surveyed). Consequently, the C genotypes of B. pedunculosum, B. pinnatum, and B. pseudopinnatum did not cluster closely with populations of B. lanceolatum (Fig. 9). The isozyme contribution of B. lanceolatum to these polyploids may have been obscured by amino acid substitutions in critical isozyme alleles. If this hypothesis is correct, at least three separate substitutions in B. lanceolatum isozyme alleles must have occurred: (1) the modification of Pgi-25 to Pgi-24 in B. pinnatum (presumably B. pseudopinnatum inherited this modification from B. pinnatum), (2) the conversion of 6Pgd–22 to 6Pgd–24 in B. pinnatum, and (3) the alteration of 6Pgd–22 to 6Pgd–23 in B. pedunculosum. Other equally parsimonious scenarios may explain the origin of these isozyme alleles, but simple inheritance of isozyme alleles from diploid progenitor directly to polyploid derivative cannot explain the isozyme composition of B. pinnatum, B. pseudopinnatum, and B. pedunculosum. Unfortunately, Dia-1, Mdh-1, Tpi-1, and Tpi-2 were not variable enough to provide further evidence concerning the uncertain origins of these three polyploid species.
The ages of B. pedunculosum, B. pinnatum, and B. pseudopinnatum are not known, although the absence of rbcL sequence divergence suggested that these polyploids were of relatively recent origin, certainly more recent than divergence among the primary diploids, B. lanceolatum, B. lunaria, B. simplex, and B. campestre (Hauk, 1995
). The absence of critical B. lanceolatum marker alleles in B. pinnatum, B. pseudopinnatum, and B. pedunculosum may indicate that these three polyploid species are of more ancient origin than the polyploids that retained the B. lanceolatum isozyme alleles (B. matricariifolium, B. hesperium, B. acuminatum, and B. echo). Clear patterns of relationship in rbcL data and the lack of clear patterns in isozyme data provide circumstantial evidence that isozymes are evolving more rapidly than rbcL sequences in these subgenus Botrychium polyploids.
Infraspecific variation for populations of polyploid species
More isozymic variation was detected within polyploid species than within diploid species, and infraspecific isozyme variation in polyploids ranged from almost none in B. pedunculosum to the relatively high level seen in B. minganense (Table 3). Presuming that the C and N hypothetical diploid genotypes are a true representation of the variation present in each polyploid, it is possible to estimate levels of infraspecific variation in polyploids by assessing independently the variation present in the hypothetical diploid genotypes (Werth, Hilu, and Langner, 1994
).
Infraspecific isozyme variation in hypothetical diploid genotypes was greater than that observed among diploid populations (Table 10). Several hypotheses can be developed to explain higher infraspecific isozymic variation among populations of polyploid species than among populations of diploid species: (1) polyploid species may be the result of several independent hybridization events that captured and preserved genetic diversity no longer present in the putative progenitor diploids, (2) extinct diploid progenitors of extant polyploids may have contained more isozymic variability than extant diploids, (3) polyploid subgenus Botrychium species may have diverged isozymically more rapidly than diploids, (4) founder effects may be exaggerated in polyploids relative to diploids due to inbreeding, (5) polyploids, by nature of their fixed heterozygosity, preserve alleles that would have been selected against if homozygous in a diploid individual. In the last case, extant diploids are products of a higher degree of natural selection than a diploid genome "hidden" (by heterozygosity) from such selective forces in a polyploid. Although these data neither refute nor support any of the hypotheses, the observed patterns of infraspecific variation between diploids and polyploids suggest that some mechanism is preserving (or promoting) isozyme variation in the polyploids.
Polyploid banding patterns in diploid species
Diploid subgenus Botrychium species typically showed monomorphic isozymic banding patterns, whereas polyploid taxa frequently showed fixed patterns of heterozygosity. The allelic variation present in diploids should segregate during meiosis and the fixed heterozygosity of polyploids should not. However, apparent fixed heterozygosity was present at some loci in certain populations of the diploids B. lanceolatum and B. lunaria. In other diploid species heterozygotes were rare and never fixed (Table 3). Analysis of hypothetical, ancestral genotypes from populations 1 and 2 of B. lanceolatum and populations 4 and 5 of B. lunaria (Fig. 10) showed that the anomalous alleles may have originated from hybridization with the diploids B. crenulatum, B. lineare, or B. pumicola. Several hypotheses may explain the fixed heterozygosity detected in these populations of B. lanceolatum and B. lunaria: (1) the "diploid" populations with fixed heterozygosity may be allopolyploid, (2) hybridization or introgression may have introduced new alleles into these diploid populations, (3) these populations may contain individuals with duplicated gene loci, or (4) the small sample sizes of these populations may misrepresent the true distribution of alleles. Further investigations will be necessary to evaluate these hypotheses.
Anomalous gene banding patterns in polyploids
In addition to detecting duplicate banding patterns in diploid species, some banding patterns typical of diploids were observed in certain polyploids. Botrychium pedunculosum is a tetraploid species that had homozygous banding patterns at five of six loci; Pgi-2 was the only locus that showed fixed heterozygosity. Because the putative parents of B. pedunculosum (B. lanceolatum and B. montanum; Fig. 1) belong to different rbcL clades (Fig. 2) and members of the "lanceolatum" and "simplex" clades have different isozyme profiles, it is reasonable to assume that B. pedunculosum should possess heterozygous banding patterns additive of the two presumed ancestors. However, the absence of heterozygosity in B. pedunculosum did not support this line of reasoning.
Equally puzzling are the occasional homozygous banding patterns observed in a few individuals of polyploid populations otherwise fixed for two alleles. For example, in population 2 of tetraploid B. matricariifolium, 41 individuals showed fixed heterozygosity between Pgi-21 and 5, and three between Pgi-22 and 5. However, three individuals from this population had homozygous banding patterns. Additional examples of anomalous homozygous banding patterns in polyploid populations are found in B. minganense where single Tpi-1 alleles were found in otherwise fixed heterozygous populations (3, 4, 6, and 8). Similar observations were made for Pgi-2 in population 10 of B. minganense and population 2 of B. hesperium.
Explanations for these enigmatic banding patterns include reciprocal gene silencing (Werth and Windham, 1991
TOP
ABSTRACT
INTRODUCTION
