- Social Sciences
- African Studies
- American Studies
- Asian Studies
- Communication Sciences
- Ethnic Studies
- European Studies
- Gender Studies
- Physical Sciences
- Life Sciences
- Animal Communications
- Cell Biology
- Evolutionary Biology
- Food Science and Technology
- Human Anatomy
Historical Biogeography of Asplenium adiantum-nigrum (Aspleniaceae) in North America and Implications for Speciation Theory in Homosporous Pteridophytes
by Tom A. Ranker, Sandra K. Floyd, Michael D. Windham, P. Genie Trapp
Historical Biogeography of Asplenium adiantum-nigrum (Aspleniaceae) in North America and Implications for Speciation Theory in Homosporous Pteridophytes
Tom A. Ranker, Sandra K. Floyd, Michael D. Windham, P. Genie Trapp
American Journal of Botany
Updated: December 11th, 2012
THEORY IN HOMOSPOROUS PTERIDOPHYTES '
TOMA. RANKER,^,^,^ SANDRA MICHAELD. WIND HAM,^ AND P. GENIE
K. FLOYD,~ TRAPP~
*University Museum and 3Department of Environmental, Population, and Organismic Biology,
University of Colorado, Boulder, Colorado 80309;
4Utah Museum of Natural History, University of Utah, Salt Lake City, Utah 841 12; and
5Boulder Valley Schools, Boulder, Colorado 80303
Theories of plant speciation have generally recognized the importance of allopatry as a potential precursor to the genetic divergence of populations. The relative importance of long-distance dispersal vs. vicariance events in leading to allopatry, however, has been debated. We examined isozymic variability in highly disjunct populations of allotetraploid Asplenium adiantum-nigrum to test alternative hypotheses on their mode of origin. In addition, we assessed the genetic distinctness of the population from Boulder County, Colorado, which had been proposed as a separate species, A. andrewsii. Our results revealed that samples from all continental U.S. populations were isozymically identical and, with the exception of two samples from Boulder, displayed no intrapopulational variability. Continental U.S. populations were most similar to those from Hawaii, whereas both of these sets of populations were considerably more divergent from samples from Mexico and the Caucasus. The distribution of alleles and genotypes support the hypothesis that populations from different geographical regions had unique origins, resulting from at least several independent hybridization and polyploidization events followed by long-distance dispersal. These results have implications for speciation theory of pteridophytes in documenting the effectiveness of long-distance dispersal in the establishment of disjunct populations which may set the stage for allopatric speciation. In addition, the data suggest that the Boulder population is not sufficiently distinct to be considered a separate
Deciphering the historical factors that have produced present-day patterns of plant distribution can provide important insight into the dynamics of biogeographical processes and their interplay with mechanisms of speci- ation responsible for the origin and evolution of biological diversity. Although numerous detailed mechanisms of speciation have been identified, the general importance ofallopatry is widely accepted (e.g., Coyne and Orr, 1989). The relative importance of long-distance dispersal vs. vicariance events in leading to geographical isolation and as a precursor to speciation, however, has been debated (e.g., Lynch, 1989). A difficult problem to overcome in this regard is that we are attempting to reconstruct his- torical events after the actual advent of speciation, where subsequent changes in geographic distribution could lead to erroneous conclusions on the original causes of geo- graphical isolation.
Long-distance dispersal has been proposed as an im- portant mechanism contributing to the widespread geo- graphic distribution of many species of pteridophytes, the concomitant low rates of endemism generally found in
Manuscript received 25 August 1993;revision accepted 3 January 1994.
The authors thank the City of Boulder Open Space Department, Mar- tha Weiser of Boulder, Zion National Park Research Department, and Hawaii Volcanoes National Park for permission to collect leaf samples; Tamara Naumann of the City of Boulder, Vic Vieira and Pat Buccello of Zion National Park, Linda Cuddihy of Hawaii Volcanoes National Park, and Martha Weiser for facilitating fieldwork; Dr. Robert Bye for providing specimens from Mexico; and Chris Haufler, Bill Weber, and Charlie Werth for providing helpful comments on the manuscript. Fund- ing was provided by NSF grant DEB-9096282to TAR, and the Uni- versity of Colorado.
Author for correspondence.
this group of organisms, and the apparent low rates of speciation compared to angiosperms (Tryon, 1970, 1972, 1986; Smith, 1972). Conversely, long-distance dispersal across particularly great distances could also allow for genetic divergence of newly formed populations in the absence of continued gene flow from source populations.
One avenue of investigation that may provide insight (albeit indirectly) into the relative importance of long- distance dispersal vs. vicarious isolation in speciation phenomena is the study of the origin of geographically disjunct, conspecific populations. Although we cannot know if particular disjunct, extant populations will ulti- mately diverge into separate species, numerous disjunct populations in the past undoubtedly have given rise to divergent evolutionary lineages. Thus, by discerning the mode of origin of disjunct populations, we may gain new knowledge into necessary preconditions for speciation events and of the relative impact oflong-distance dispersal vs. vicariance in setting the stage for such precursors to speciation.
In the present study, we employed genetic data to ex- plore the origins of highly disjunct populations of As
plenium adiantum-nigrum L. ("black spleenwort"). This is one of the rarest ferns in North America where pop- ulations are known only from Elden Mountain, Arizona, Boulder County, Colorado, and Zion National Park, Utah, and several locations in the Mexican states of Chihuahua and Tamaulipas. Outside of North America, A. adiantumnigrum is known from the Hawaiian Islands, Europe, Africa, and Asia. Manton (1950) demonstrated that A. adiantum-nigrum is a tetraploid with n = 72 bivalents at meiosis I. Shivas (1969) provided cytological evidence from experimental hybridizations suggesting thatthis tet- raploid is of hybrid origin, resulting from the hybridiza- tion of two European diploids, A. onopteris L. and A. cuneifolium Viv. The original diploid hybrid is presumed to have been sterile, but through chromosomal duplica- tion (i.e., such as by nondisjunction at meiosis) ultimately produced a fertile allotetraploid. Existing evidence sug- gests that this is the most common mode of production of allopolyploid ferns, referred to as Type I1 polyploid- ization by Harlan and deWet (1975; see also Manton,
1950; Walker, 1979).
Ranker, Floyd, and Trapp (in press) suggested from genetic data that populations of A. adiantum-nigrum in Hawaii have resulted from at least several long-distance dispersal and colonization events from non-Hawaiian source populations. In the present study we address the question of the origin of the disjunct North American populations of this species. These populations could have arisen via long-distance spore dispersal from Old World or Pacific sources (Ewan, 1945) or they may be relictual from a once more continuously ranging species.
In addition to exploring alternative hypotheses on the mode of origin of the North American populations of A. adiantum-nigrum, we addressed the question of the ge- netic distinctness of the Boulder population which has occasionally been treated as a separate species A. andrewsii (Nelson, 1904).
MATERIALS AND METHODS
Because of the rarity of A. adiantum-nigrum in North America, minimal samples were taken to lessen the impact of collecting on the few existing populations. Leaves were collected from 2 1 individuals from each ofthe populations at Zion National Park, Utah (vicinity of Canyon Overlook Trail and Emerald Pools) and Elden Mountain, Arizona (bordering the city of Flagstaff). Twenty individuals were sampled from the population in Boulder County, Colo- rado (White Rocks area). Leaves were kept on ice in plastic bags until processing for electrophoresis. Specimens from the sampled populations are housed in the herbaria of Colorado State University (CS), New York Botanical Gar- den (NY),Northern Arizona University (ASC), U.S. Na- tional Herbarium (US), University of Arizona (ARIZ), University of Colorado (COLO), University of Wyoming (RM), and University of Utah (UT); new voucher spec- imens were not collected for the present study. Fragments of fertile leaves obtained from a population in the State of Chihuahua, Mexico (Mpio. Bocoyna, SE of Creel; Bye #18331) were kindly provided by Dr. Robert A. Bye, Jr. Spores were sown in petri dishes on an agar growth me- dium following Windham, Wolf, and Ranker (1986) to obtain gametophytes for isozymic analyses. Viable spores were also obtained from herbarium specimens collected in Mexico in 1977 (Bye 7384 from Mpio. Guachochi, Chihuahua; COLO accession no. 352634) and in the Sochi District of the Caucasus in 198 1 (Vas6k s.n.; COLO ac- cession no. 441033) and sown on agar growth medium. The specimen from the Caucasus was chosen as a source of spores because its recent date of collection suggested that its spores would still be viable (i.e., see Windham, Wolf, and Ranker, 1986). Although the small samples available frm Mexico and the Caucasus did not allow for estimation of intrapopulational variability, they were sufficient to provide estimates of genetic similarity among populations (see Nei, 1978).
Samples were also collected from nine localities on the island of Hawaii, State of Hawaii. Detailed results from isozymic analyses of these populations are presented else- where (Ranker, Floyd, and Trapp, in press) and only gen- eral comparisons with Hawaiian populations will be made in the present context.
Horizontal starch-gel electrophoresis was conducted on extracts of collected leaf material and cultured gameto- phytes following Ranker et al. (1989). Results were ob- tained from 1 1 enzyme systems including aldolase (ALD), fructose-bisphosphatase (FBP), glutamate oxaloacetate transaminase (GOT), hexokinase (HK), isocitrate dehy- drogenase (IDH), leucine aminopeptidase (LAP), malate dehydrogenase (MDH), 6-phosphogluconate dehydroge- nase (BPGDH), phosphoglucose isomerase (PGI), shiki- mate dehydrogenase (SkDH), and triosephosphate isom- erase (TPI).
Nei's unbiased genetic identity coefficient (I;Nei, 1978) and Rogers' genetic similarity coefficient (S; Rogers, 1972) were calculated for all pairs of populations with the pro- gram BIOSYS- 1 (Swofford and Selander, 1989).
Seventeen putative duplicate locus pairs were scored across the 11 enzyme systems. Each locus pair was given a numeric abbreviation with the most anodal region of activity (when there was more than one such region on a gel) designated "1" (e.g., Pgi-I). The members of a pair were abbreviated with letters, "a" assigned to the more anodal region of activity and "b" assigned to the more cathodal region when two such regions could be distin- guished (e.g., Pgi-la and Pgi- lb; following Werth, 1989). Each locus of a pair was presumed to have been contrib- uted by a single diploid progenitor of this allotetraploid species and, thus, the members of a duplicate locus pair were presumed to be homoeologous to each other (Kle- kowski, 1976; Werth and Windham, 1991). Because of the tetraploid nature of A. adiantum-nigrum, a region of activity that was expressed as a monomorphic, homo- zygous "locus" was interpreted as two monomorphic, homoeologous loci that were homozygous for the same allele. Allele frequencies for all populations are presented in Table 1.
The samples from Elden and Zion exhibited no within- or between-population variability and were fixed for a single allele at each of eight locus-pairs (Fbp-Ialb, Hk- alb, Zdh-alb, Mdh-Zalb, 6Pgdh- lalb, 6Pgdh-2alb, Pgi- Ialb, and Skdh-alb). At the remaining nine locus-pairs, all samples from Elden and Zion expressed multiple- banded patterns that were interpreted as fixed interlocus heterozygosity for functional alleles. Presumably these latter locus-pairs represent cases where the diploid pa- rental taxa of A. adiantum-nigrum possessed different alleles that were combined in the allotetraploids (e.g., Ald-a contributed by one parent and Ald-b by the other parent). When assayed electrophoretically, individual ga- metophytes also expressed fixed heterozygosity at the same nine loci as the sampled populations (unpublished data). Values of Nei's Z and Rogers' Swere both 1.000 between the Elden and Zion populations (Table 2).
|TABLE1.||Allele frequencies in populations surveyed.|
|Locus||Boulder Zion Elden Mexico Caucasus||Hawai*|
|Fbp- 1 b|
|Mdh- 1 b|
|Locus||Boulder||Zion||Population Elden Mexico||Caucasus||HawaiP|
|6Pgdh- la (N) 1|
|6Pgdh-3 (N) 2|
|Pgi- 1 a (N)1 2 3|
|Pgi- 1 b (N) 1 2 3|
|Pgi-2a (N)1 2|
|Pgi-2b (N) 2 3 4|
|(N)1||20 1.000||2 1 1.000||2 1 1.000||2 -||1 1.000||236 0.971|
|(N)1||20 -||2 1 -||2 1 -||2 -||1 1.000*||236 -|
|(N)3||20 -||2 1 -||2 1 -||2 -||1 1.000*||236 -|
|(N)1||20 -||21 -||2 1 -||2 -||1 -||236 0.005*|
|(N)3||20 -||2 1 -||2 1 -||2 -||1 1.000*||236 -|
a Values for Hawaii represent means of nine populations. N = sample size per population per locus.
* Alleles unique to a region.
The Boulder sample was nearly identical to those from Elden and Zion, expressing a difference at only a single locus-pair. Whereas all samples from Elden and Zion and 18 of the individuals from Boulder expressed a "heter- ozygotic" genotype for Lap-a/b (combining Lap-all and La~-b~~,
where superscripts refer to presumed genotypes), two individuals from the latter population were "ho- mozygous" for allele 1 at this locus-pair (i.e., Lap-all: Lap-blL). The only allele unique to continental U.S. pop- ulations, therefore, was allele 1 at Lap-b, found only in the Boulder population (but see discussion below on gene silencing). Rogers' Sbetween Boulder and each of Elden and Zion was 0.997 (Table 2); S was a more sensitive indicator of the difference at Lap-alb among these pop- ulations than Nei's I,which was 1.000 for both compar- isons.
The samples from Hawaii exhibited more allozymic variability than the continental U.S. populations but gen- erally showed high degrees of genetic similarity to them. Samples from Hawaii were invariant at eight locus-pairs . (Table I), four of which showed fixed "homozygosity" and four of which showed fixed "heterozygosity." The remaining loci exhibited varying degrees of variability within and among populations. Two of the heterozygous locus-pairs expressed three alleles and one expressed four alleles across individuals and populations, whereas only two alleles per locus were expressed at heterozygous locus- pairs in the continental U.S. populations. Twelve alleles unique to Hawaiian populations (so-called "private al- leles") were observed across nine locus-pairs (Table l).
Three of those private alleles (i.e., at Hk-alb, Skdh-alb, and Tpi-2alb) were each found at only a single locality among the nine populations sampled on the island of Hawaii and were at frequencies of less than or equal to
0.1 10 across the entire Hawaiian sample. The remaining nine private alleles were discovered at two or more local populations and ranged in frequency from 0.004 to 0.72 1. The two private alleles observed at Idh-alb (alleles 1 and 3) were found in all Hawaiian populations sampled. The mean values of I and S (Table 2) between Hawaiian and continental U.S. populations were 0.930 and 0.903, re- spectively. At three locus-pairs (Hk-alb, Pgi-lalb, and Pgi-Zalb), alleles were shared uniquely between Hawaiian and continental U.S. populations. Allele 3 at both Hk-a and Hk-b was fixed in all continental U.S. samples and was nearly fixed in all Hawaiian populations. At Pgi- lalb all samples from both the continental United States and Hawaii were fixed for the same alleles. At Pgi-2alb all samples from the continental United States and most of those from Hawaii were fixed for alleles 2 and 4, which were shared uniquely between these two regions; a subset of individuals from three Hawaiian populations were "ho- mozygotes" for allele 2.
The samples from Mexico and the Caucasus showed the greatest amount of genetic divergence compared to each other and relative to all other populations sampled. Nine private alleles were found in the Mexican samples across six locus-pairs, and ten such alleles were discovered in the Caucasian sample at six locus-pairs (Table 1). Al- though the two Mexican samples were collected from different localities in Chihuahua, they were identical to each other at every locus. Mean values of Iand Sbetween the Mexican samples and those from the continental Unit- ed States were 0.647 and 0.646, respectively (Table 2), and those between the Caucasus and continental United States were both 0.618. The mean values of I and S between Hawaiian and Mexican populations were 0.566 and 0.553, respectively, and those between Hawaiian and Caucasian populations were 0.5 77 and 0.573, respective- ly. The values of Iand Sbetween Mexican and Caucasian populations were both 0.529. Seven alleles were shared uniquely among Hawaiian and all North American pop- ulations, six were found only in samples from the Cau- casus, Hawaii, and the continental United States, and four alleles were shared uniquely between the Mexican and Caucasian samples.
The nearly complete genetic identity among the three continental U.S. populations of A. adiantum-nigrum (Boulder, Elden, and Zion) suggests that they share a com- mon and recent evolutionary history. All of the individ- uals sampled from Elden and Zion and most ofthose from Boulder may be descendants of a single hybrid individual that combined the particular parental genotypes expressed in those populations. The production of such an individ- ual in one locality with subsequent spore production and dispersal to the present population sites would account for the observed distribution of this genotype in North America. (Ranker, Floyd, and Trapp, in press, present evidence suggesting that A. adiantum-nigrum is highly
TABLE2. Matrix of Nei's unbiased genetic identity (above diagonal) and Roger's genetic similarity (below diagonal) coefficients between pairs of populations.
Mean Nei's unbiased genetic identity among Hawaiian populations.
inbreeding, if not totally selfing, and thus does not ex- perience sexual recombination. The equivalence of the number of genotypes with the number of hybrid origins, therefore, appears to be a valid assumption in this species.) Alternatively, the existing populations could be remnants of a once more continuous, southern Rocky Mountain metapopulation, most of which has gone extinct. Even under the latter scenario, however, one must invoke dis- persal from a single original point of hybridization with subsequent reduction in population size.
The genetic identities among the continental U.S. pop- ulations may be particularly significant in unraveling the biogeographic history of this species in light of its tetra- ploid nature. Werth and Windham (1991) proposed a model of speciation of allopatric populations of polyploid pteridophytes involving reciprocal silencing of alternate alleles at homoeologous loci, and several empirical studies have documented such reciprocal silencing in disjunct, conspecific populations of allopolyploids (e.g., Werth, Guttman, and Eshbaugh, 1985b; Bryan and Soltis, 1987). No differential gene silencing has occurred between the Elden and Zion populations ofA. adiantum- nigrum, how- ever, supporting the recent common evolutionary history of these populations. The variant Lap-b genotype ob- served in two of the samples from the Boulder population may have arisen through the mutational silencing of allele 4 at this locus in an individual tetraploid plant or via a second hybridization event with both of the hybridizing individuals of the parental species contributing allele 1, and subsequent dispersal (directly or indirectly) to the Boulder locality. The gene silencing hypothesis seems more likely, given the otherwise complete genetic identity of the two variant samples from Boulder with all other Boul- der, Elden, and Zion samples (i.e., across all other locus- pairs), in contrast to the great genetic differentiation ob- served among populations from different regions.
The fact that the most common genotypes in individ- uals from Hawaii were shared with those sampled from the continental United States suggests a common geo- graphical and evolutionary origin for the populations in these two regions. This hypothesis is strongly supported by the presence of high frequency alleles (i.e., at Hk-alb and Pgi-2alb) that are shared between these two regions but not with the Caucasian population. Since the Ha- waiian Islands arose through suboceanic volcanism in the mid-Pacific, we know that A. adiantum-nigrum is capable of successful long-distance migration to and establishment on previously uninhabited sites (Ranker, Floyd, and Trapp, in press). Whether the continental U.S. populations served as a source of migrants for the colonization of Hawaii (or vice versa) or both regions received migrants from a third common source cannot be completely resolved from these data. Given that continental U.S. ~o~ulations
contain only a subset of the genetic ~ariabilitf~resent
in Hawaiian populations, however, it is more likely that migrants from Hawaii were the source of colonists to North America rather than the other way around, if dispersal has occurred from one region to the other.
The extreme genetic differences among populations from the continental United States and Hawaii vs. those from Mexico and the Caucasus suggest that each set of popu- lations originated independently of the others. The pres- ence of unique alleles and genotypes at single locus-pairs and across locus-pairs in each region (combining Hawaii and continental United States) provides evidence that each set of populations arose from separate hybridization events between individuals of the parental species of dif- ferent genetic constitution. The present European distri- bution of the putative parental species (A. onopteris and
A. cuneifolium) suggests that hybrid individuals originally arose in Europe with subsequent dispersal events to Asia, Hawaii, and North America (either directly or via a "step- ping-stone" sequence). Within North America, the unique alleles and genotypes distinguishing the populations from the United States and those from Mexico are best ac- counted for by hypothesizing separate hybridization events combining different parental genotypes with subsequent independent dispersals to the two regions. This hypothesis is most strongly supported by the distribution ofgenotypes between the two regions at those locus pairs that expressed completely different genotypes, i.e., Hk-alb, Lap-alb, Mdh-1alb, Mdh-2alb, Pgi- 1alb, Pgi-2alb, and Skdh-a/
b. Alternatively, the parental diploids could have occurred in earlier times in the continental regions where the tet- raploids presently occur, with hybrids being produced in situ and subsequent extinction of the diploids from all regions but Europe. The nearly complete lack of genetic variability in continental U.S. populations combined with the presence of variability among populations from dif- ferent geographic regions (thus indicating the polymor- phic nature of the ancestral diploids) argues against the alternative hypothesis of multiple in situ hybridizations at the present North American localities. Although the alternative hypothesis of relictualism from ancient pop- ulations cannot be completely rejected, under that hy- pothesis one might expect more genetic variability within populations of A. adiantum-nigrum.
The recognition of the Boulder population as a species distinct from other populations of A. adiantum-nigrum (as A. andrewsii; Nelson, 1904) appears to be unwarranted in light of the present genetic data. It is worth noting, however, that the only genetic variability discovered among continental U.S. populations was found in the Boulder population. The unique genetic attributes of some members of this population are significant in terms of the conservation management of this rare species in the Unit- ed States. Similarly, the individuals sampled from Mexico are genetically unique.
The results of this case study have implications for speciation theory in pteridophytes in favoring the hy- pothesis that the process of long-distance dispersal has led to the origin of the disjunct North American popu- lations of A. adiantum-nigrum. The geographic isolation of these populations from each other and from Old World and Pacific conspecific populations may allow for their ultimate genetic divergence leading to the origin of distinct phylogenetic lineages. The maximum levels of genetic divergence observed among populations in the present study are similar to those observed between the diploid species A. montanum and A. rhizophyllum of the Appa- lachian Asplenium complex (Werth, Guttman, and Esh- baugh, 1985a). Each of the genetically and geographically distinct entities within A. adiantum-nigrum in North America and elsewhere may experience different evolu- tionary fates, and because of the potential for the devel- opment of reproductive isolation (i.e., through differential gene silencing, other mutational events, and natural se- lection), these may represent cases ofincipient speciation.
BRYAN, F. A., AND D. E. SOLTIS. 1987. Electrophoretic evidence for allopolyploidy in the fern Polypodium virginianum. Systematic Bot- any 12: 553-561.
COYNE,J. A., AND H. A. ORR. 1989. Two rules of speciation. In D. Otte and J. A. Endler [eds.], Speciation and its consequences, 180-
207. Sinauer, Sunderland, MA. EWAN, J. 1945. Sources of the fern flora of Colorado. American Fern Journal 35: 114-128. HARLAN, J. R., AND J. M. J. DEWET. 1975. On 0.Winge and a prayer: the origins of polyploidy. The Botanical Review 41: 361-390.
KLEKOWSKI,E. J., JR. 1976. Homoeologous chromosome pairing in ferns. In K. Jones and P. E. Brandham [eds.], Current chromosome research, 175-1 84. Elsevier/North Holland Biomedical, Amster- dam.
LYNCH, J. D. 1989. The gauge of speciation: on the frequencies of modes of speciation. In D. Otte and J. A. Endler [eds.], Speciation and its consequences, 527-553. Sinauer, Sunderland, MA.
MANTON, I. 1950. Problems of cytology and evolution in the pteri- dophyta. Cambridge University Press, Cambridge. NEI, M. 1978. Estimation of average heterozygosity and genetic dis- tance from a small number of individuals. Genetics 89: 583-590.
NELSON, A. 1904. Plantae Andrewseae. Proceedings of the Biological Society of Washington 17: 173-180.
RANKER, T. A., S. K. FLOYD,AND P. G. TRAPP. In press. Multiple colonizations of Asplenium adiantum-nigrum onto the Hawaiian Archipelago. Evolution. ,C. H. HAUFLER, P. S. SOLTIS, AND D. E. SOLTIS. 1989. Genetic evidence for allopolyploidy in the neotropical fern Hemionitis pin- nat~jida(Adiantaceae) and the reconstruction of an ancestral ge- nome. Systematic Botany 14: 439447.
ROGERS,J.S. 1972. Measures ofgenetic similarity and genetic distance. Studies in Genetics, University of Texas Publication 72 13: 145-1 53.
SHNAS, M. G. 1969. A cytotaxonomic study of the Asplenium adian-
tum-nigrum complex. British Fern Gazette 10: 68-80. SMITH, A. R. 1972. Comparison of fern flowering plant distributions with some evolutionary interpretations for ferns. Biotropica 4: 4-9.
SWOFFORD,AND SELANDER. 1989. BIOSYS-1. A computer program for the analysis of allelic variation in population genetics and bio- chemical systematics. Release 1.7. Illinois Natural History Survey, Champaign, IL.
TRYON,R. 1970. Development and evolution of fern floras of oceanic islands. Biotropica 2: 76-84. . 1972. Endemic areas and geographic speciation in tropical American ferns. Biotropica 4: 12 1-1 3 1. . 1986. The biogeography of species, with special reference to ferns. The Botanical Review 52: 1 17-1 56.
WALKER, T. G. 1979. The cytogenetics of ferns. In A. F. Dyer [ed.], The experimental biology of ferns, 87-132. Academic Press, Lon- don.
WERTH, C. R. 1989. The use of isozyme data for infemng ancestry of polyploid pteridophytes. Biochemical Systematics and Ecology 17: 1 17-1 30. ,S. I. GUTTMAN, AND W. H. ESHBAUGH. 1985a. Electrophoretic evidence of reticulate evolution in the appalachian Asplenium complex. Systematic Botany 10: 184-192. ,AND-. 1985b. Recurring origins of allopolyploid species in Asplenium. Science 228: 73 1-733. ,AND M. D. WINDHAM. 199 1. A model for divergent, allopatric speciation of polyploid pteridophytes resulting from silencing of duplicate-gene expression. The American Naturalist 137: 5 15-526.
WINDHAM,M. D., P. G. WOLF, AND T. A. RANKER. 1986. Factors affecting prolonged spore viability in herbarium collections of three species of Pellaea. American Fern Journal 76: 141-148.