AMERICAN JOURNAL OF BOTANY 103 (4): 1 – 14 , 2016; http://www.amjbot.org/ © 2016 Botanical Society of America • 1
AMERICAN JOURNAL OF BOTANY
Cryptic species groups, which may contain undescribed species
within one historic taxonomic species, pose a problem for taxono-
mists, systematists, ecologists, and conservationists. Paris et al. (1989 ,
p. 47) de ned cryptic species using criteria provided by Stebbins (1950) ,
Grant (1981) , and Wiley (1981) as follows:
“1. Are poorly di erentiated morphologically.
2. Represent distinct evolutionary lineages because they are re-
3. Have historically been misinterpreted as members of a single
Because cryptic speciation obscures morphological delineation
of taxa, researchers are likely to lump similar taxa that have distinct
lines of evolutionary history (in contrast to the splitting of a geneti-
cally cohesive species, which is likely when intraspeci c morpho-
logical variability equals or exceeds interspeci c variability). Paris
et al. (1989) used Botrychium subgenus Botrychium (= Botrychium
s.s.) as an example of cryptic species complexes. Within the genus,
B. matricariifolium provides an excellent case study in which cryp-
tic species and intraspeci c variability have long presented confusion
and controversy with regards to taxon delineation. Taxonomists
now have genetic tools to tease apart such cryptic species groups
and gain insight into their origins.
A clade comprising Botrychium , other members of Ophioglos-
saceae, and Psilotaceae is sister to Marattiales plus leptosporangiate
ferns ( Rai and Graham, 2010 ; Wickett et al., 2014 ; Rothfels et al.,
2015 ). Botrychium s.s. consists of about 35 distinct taxa (recognized
as species, subspecies, or varieties) worldwide, with a center of diver-
sity in northern North America ( Farrar, 2011 ). Closely related gen-
era include Botrypus Michx. (rattlesnake fern), Japanobotrychium
1 Manuscript received 12 June 2015; revision accepted 19 February 2016.
2 Chicago Botanic Garden, Department of Plant Biology and Conservation, 1000 Lake Cook
Road, Glencoe, Illinois 60022 USA;
3 Iowa State University, Department of Ecology, Evolution, and Organismal Biology, 253
Bessey Hall, Ames, Iowa 50011 USA; and
4 Tamarack Studios, P. O. Box 453, Manistique, Michigan 49854-0453 USA
5 Author for correspondence (e-mail: email@example.com)
Cryptic speciation in allotetraploids: Lessons from the
Botrychium matricariifolium complex 1
Evelyn W. Williams 2,5 , Donald R. Farrar 3 , and Don Henson 4
PREMISE OF THE STUDY: Cryptic species are a challenge for botanists and taxonomists. To improve species delineation in the genus Botrychium (Ophioglos-
saceae), which includes multiple instances of allotetraploid speciation, we examined a cryptic species complex using genetics and morphology.
METHODS: We sampled species in the B. matricariifolium complex, concentrating on the Upper Peninsula of Michigan and including multiple proposed
morphospecies. We analyzed over 1500 samples using 10 enzyme systems, measured 42 quantitative and qualitative morphological characters for over
650 individuals, and analyzed 145 samples using AFLPs. We tested for diagnostic enzymes in the morphospecies and calculated the correlation between
morphological and genetic distances to determine whether putatively distinct morphotypes warrant taxonomic recognition.
KEY RESULTS: Allozyme allelic variation corresponded loosely to some morphotypes of B. matricariifolium , but with lower genetic distinction among them
than found between B. matricariifolium and B. michiganense . Botrychium michiganense contains unique alleles, indicating a di erent hybrid origin from
that of B. matricariifolium and supporting its status as a genetically distinct species.
CONCLUSIONS: We showed that B. acuminatum morphology and genetics are accommodated taxonomically within B. matricariifolium ; B. matricariifolium and
B. michiganense likely represent hybridization events between related species; and morphotypes within B. matricariifolium likely represent repeated hybridiza-
tion events between the same two parental species. These hybridizations have resulted in the array of morphotypes observed by eld botanists. By helping
to identify diagnostic morphological characters, genetic analyses also help us understand and resolve morphological variation observed in the eld.
KEY WORDS allozymes; Botrychium matricariifolium ; cryptic species; ferns; Ophioglossaceae; polyploidy; speciation
http://www.amjbot.org/cgi/doi/10.3732/ajb.1500281The latest version is at
AJB Advance Article published on April 7, 2016, as 10.3732/ajb.1500281.
Copyright 2016 by the Botanical Society of America
2 • AMERICAN JOURNAL OF BOTANY
Masam., and Sceptridium Lyon (grapeferns: Kato, 1987 ; Wagner
and Wagner, 1993 ; Hauk et al., 2003 ), all of which were previously
considered to be subgenera in the genus Botrychium s.l. ( Clausen,
1938 ). Of these genera, Botrychium s.s. has the smallest plants and
the most reduced morphology. Species are rarely taller than 20 cm,
with a leaf divided more or less equally into two segments, a spore-
producing sporophore and a photosynthetic nonfertile tropho-
phore ( Fig. 1 ). Stems are underground, upright, and usually
unbranched with a single apical bud (also underground) that pro-
duces a single aboveground leaf during the annual growing season.
Roots are eshy and hairless. Nutrition is derived in part through
an underground association with mycorrhizal fungi of the genus
Glomus ( Winther and Friedman, 2007 ).
Sexual reproduction is via bisexual gametophytes (also under-
ground and supported by mycorrhizae) derived from spores re-
leased from aboveground leaves. Presumably due to the di culty of
sperm movement underground, new sporophyte production is al-
most exclusively by intragametophytic self-fertilization ( Soltis and
Soltis, 1990 ; Farrar, 1998 ; Barker and Hauk, 2003 ). Consequently,
because egg and sperm are derived by mitosis from the same hap-
loid spore, the resulting sporophytes are completely homozygous.
Occasional heterozygous individuals and hybrids attest to rare oc-
currences of cross fertilization between gametophytes, but predom-
inant intragametophytic sel ng and its consequences are a key part
of understanding the population biology of Botrychium .
Diploid species of Botrychium s.s. can be divided into two
groups based on morphology: section Lunariae that is once-pin-
nate with fan-shaped pinnae and section Lanceolatae that is
twice-pinnate with elongate pinnae ( Clausen, 1938 ). is division
is supported by genetic analyses of plastid regions ( Hauk et al.,
2003 , 2012 ; Williams and Waller, 2012 ; Dauphin et al., 2014 ).
Molecular work also suggests the possible subdivision of section
Lunariae , with “Lunaria”, “Simplex”, “Pallidum”, and “Camp-
estre” subgroups variously recognized ( Hauk, 1995 , 2012 ; Wil-
liams and Waller, 2012 ).
More than half of Botrychium taxa are polyploid and unequivo-
cally derived through the process of allopolyploidy—hybridization
between genetically distinct diploid taxa accompanied by genome
doubling ( Hauk, 1995 , 2012 ; Farrar, 1998 , 2011 ; Hauk and Hau er,
1999 ; Williams and Waller, 2012 ). e morphological distinction
among many of these polyploid taxa is subtle and presents serious
identi cation and management problems because some taxa are on
governmentally protected lists. Here we explore genetic and evolu-
tionary origins of problematic morphological variation in the B.
matricariifolium A.Braun complex of allopolyploid taxa in the
western Great Lakes (WGL) region, seeking explanations for ob-
served intra- and interspeci c variation and a consistent basis for
taxonomic recognition. e complex exempli es the problems en-
countered with cryptic species groups; it is widespread and com-
mon in northeastern North America and exists in a frustrating
array of morphological forms.
Of the 17 formally named Botrychium species occurring in the
WGL region (nine diploids and eight polyploids), three formally
named species represent the B. matricariifolium complex: B. matri-
cariifolium , Botrychium michiganense W.H.Wagner ex A.V.Gilman,
Farrar and Zika, and B. acuminatum W.H.Wagner ( Wagner and
Wagner, 1990 , 1993 ; Farrar, 2011 ; Gilman et al., 2015 ). A fourth
member of the complex, B. hesperium (Maxon and Clausen) Wag-
ner and Lellinger is no longer recognized as occurring in eastern
North America ( Gilman et al., 2015 ), but has bearing on the origins
of the eastern taxa.
In an early genetic study of this complex in Michigan, Farrar
and Wendel (1997) introduced two putatively distinct taxa that
were originally identi ed by Henson and informally referred to
by Farrar and Wendel (1997) as B. matricariifolium “blocky”
morphotype and B. matricariifolium “acuminate” morphotype,
the later displaying intermediacy between B. acuminatum and
typical B. matricariifolium. In this study, we treat Wagner’s B.
acuminatum within the acuminate morphotype. us, in the
WGL, four entities ( B. matricariifolium , B. matricariifolium
“blocky”, B. acuminatum , and
) with similar and
to some degree overlapping morphologies are found that present
a bewildering array of intra- and interspeci c variability ( Fig. 2 ).
Collectively, plants of this complex are among the most common
ferns of the region.
Understanding similarities among species of the B. matricariifo-
lium complex requires consideration of their origins from diploid
Botrychium taxa. e pinnate-pinnati d morphology of all taxa of
the B. matricariifolium complex is intermediate between that of
twice-pinnate members of section Lanceolatae and once-pinnate
members of section Lunariae , supporting a hybrid origin combin-
ing genomes of diploids from these two sections ( Fig. 3 ). Genetic
evidence from allozymes and plastid DNA strongly supports east-
ern B. angustisegmentum as the maternal plastid-donating parent
of eastern B. matricariifolium and B. acuminatum, whereas western
B. lanceolatum is indicated as the maternal parent of western
B. hesperium ( Wagner, 1993 ; Hauk and Haufler, 1999 ; Hauk
et al., 2003 ; Farrar, 2011 ; Williams and Waller, 2012 ). Allozyme
FIGURE 1 Morphology of a Botrychium fern, B. matricariifolium, with tro-
phophore, sporophore, pinnae, common stalk, and morphological mea-
surements made here, following Swartz and Brunsfeld (2002) ; see online
Appendix S3 for more information. Length measurements: (1) sporophore,
(2) sporophore branching point, (3) pinnae length, (4) pinnae width, (5)
rst pinnae pair, (6) trophophore, (7) second pinnae pair, (8) trophophore
width. Angle measurements: blue = pinnae stem angle, brown = pinnae
whole angle, maroon = pinnae partial angle, orange = 2nd pinnae pair
stem angle, green = 2nd pinnae pair partial angle. The circularity of basal
pinnae and the number of pinnae pairs were also measured.
APRIL 2016 , VOLUME 103 • WILLIAMS ET AL. —CRYPTIC SPECIATION IN BOTRYCHIUM • 3
evidence indicates that B. michiganense
in both eastern and western North
America combines alleles that are now
unique to either B. angustisegmentum
or B. lanceolatum ( Gilman et al., 2015 ),
suggesting an origin predating the dif-
ferentiation of these two Lanceolatae
Considering the most probable dip-
loid parents of the B. matricariifolium
complex within section Lunariae , and
drawing on evidence from codominant
allelic expression of metabolic enzymes
and morphology, Gilman et al. (2015)
concluded that the closest match for all
members of the complex is B. pallidum .
e morphological similarities of their
diploid ancestors ( B. pallidum × one of
the Lanceolatae species) likely contrib-
ute in large measure to morphological
similarities among extant members of
the B. matricariifolium complex. Species
bound aries are further confused by
the possibility that each allopoly-
ploid taxon was derived from multiple
hybridization events between the same
two diploid taxa, and from individuals
expressing intraspeci c morphological
variation. e possibility of additional
cryptic species within B. matricariifo-
lium prompted our investigation. Here
we analyze complementary data de-
rived from allozymes, ampli ed frag-
ment length polymorphisms (AFLPs;
Vos et al., 1995 ), and morphometrics to
investigate morphotypes within the
B. matricariifolium complex in the
WGL region. Allozymes and AFLPs
allow intensive analyses of multiple
plants and populations based on nu-
clear genetic markers that re ect the
contributions from all parental taxa of
allopolyploid species. We expected that
observed genetic differences would
correlate with morphology within
groups and would help identify distinct
taxon units in the B. matricariifolium
complex. Our study includes analysis
of the possible genetic bases for the
“acuminate” and “blocky” morphotypes identi ed by Henson and de-
scribed by Farrar and Wendel (1997) .
MATERIALS AND METHODS
Collections — Collections examined in this study were made in 1995
and 1996 for a Botrychium diversity study sponsored by the Ottawa
National Forest and conducted by Henson and Farrar ( Farrar and
Wendel, 1997 ; Fig. 4 , open circles; Appendix S1, see Supplemental
Data with online version of this article). Additional relevant collections
FIGURE 2 Examples of the range of trophophore morphologies found within Botrychium matricariifolium,
including the “acuminate” and “blocky” morphotypes. These examples do not encompass all of the varia-
tion seen in B. matricariifolium , whose morphotypes can include immature individuals that resemble spe-
cies in section Lunariae . “EW” samples determined using AFLPs, and “DF” samples are determined by
documenting range-wide allozyme pro les of all Botrychium taxa
were made by Farrar and colleagues from 1997 through 2011. Williams
and Henson made collections in 2007–2009 ( Fig. 4 , closed circles;
online Appendix S2). Because of the di erent time periods of the
initial collections, the same individual plants could not be analyzed
for both allozyme and AFLP data ( Table 1 ). us, not all species
analyzed with AFLPs were analyzed with allozymes and vice versa.
Voucher specimens for allozyme-studied plants are maintained in
the Iowa State Herbarium. Voucher specimens for AFLP-studied
plants are deposited in the University of Michigan Herbarium. Scans
of all specimens are available from the authors.
4 • AMERICAN JOURNAL OF BOTANY
Allozymes — Samples for enzyme electrophoresis were cut from the
base of fresh leaf stalks, leaving the remainder of the leaf for press-
ing as a herbarium voucher. Electrophoresis procedures followed
Zika and Farrar (2009) . We analyzed 22 gene loci from 10 enzyme
systems using three bu er systems of Soltis et al. (1983) : bu er sys-
tem 7 (0.038 M LiOH, 0.188 M boric acid) for resolving enzyme
systems aspartate aminotransferase (AAT) and triose-phosphate
isomerase (TPI); bu er system 9 (0.065 M -histidine, 0.015–0.016
M citric acid, anhydrous) for resolving enzyme systems malate de-
hydrogenase (MDH), phosphoglucomutase (PGM), 6-phospho-
gluconate dehydrogenase (6-PGD), and phosphoglucoisomerase
(PGI); and bu er system 11 (0.4 M citric acid, trisodium salt) for
resolving enzyme systems aconitase (ACN), diaphorase (DIA),
isocitrate degydrogenase (IDH), and shikimate dehydrogenase
(SKDH). Numbered alleles for each locus were assigned relative to
their distance of migration from the origin, considering all species
of Botrychium ; lower numbers indicating faster migration. To fa-
cilitate discussion of genotypes of TPI-1 and PGI-1, we designated
FIGURE 3 Hypothesized relationships among the diploid and allotetraploid taxa discussed in this study.
Ploidy level is indicated by “2 n ” for diploid species (ellipses) and “4 n ” for derived allopolyploids (rectan-
gles). Trophophores of taxa in the left column exhibit twice-pinnate dissection whereas those of
Botrychium pallidum on the far right are once-pinnate. Trophophores of the allopolyploids, including
morphotypes of Botrychium matricariifolium , exhibit an intermediate condition referred to as pinnate-
pinnati d. Botrychium lanceolatum (“ancestral”) is a proposed extinct taxon that di erentiated into east-
ern B. angustisegmentum and western B. lanceolatum in North America.
FIGURE 4 Map of collection sites in northern Michigan (MI), Wisconsin,
and Minnesota (MN). Open circles are sites used for allozyme analysis
(collections from 1995 and 1996), solid circles are sites used for morpho-
logical and AFLP analyses (collected 2007–2009).
TABLE 1. Numbers of individuals per Botrychium taxon included in the study.
Sites of collection in the western Great Lakes region are shown in Fig. 4 .
B. angustisegmentum (2 n )36 67 43
B. lanceolatum (2 n ) — — 128
B. neolunaria (2 n ) 17 36 524
B. lunaria (2 n ) 5 3 465
B. matricariifolium (4 n ) 59 433 208
B. matricariifolium “blocky”
3 3 100
B. matricariifolium “acuminate”
B. michiganense (4 n ) 5 42 115
B. minganense (4 n ) 5 23 287
B. pallidum (2n) 11 28 121
Total 145 658 2045
a Voucher information in online Appendix S2.
b Alloz yme numbers include representative plants from populations throughout the species
ranges, sampled in ongoing studies of the genus. Voucher information in online Appendix
S1. Speci c collection data are available in a database maintained by Farrar in the Iowa State
University herbarium (http://www.public.iastate.edu/~herbarium/botrychium.html). Additional
species information, including species’ ranges, are available in the following publications:
Farrar (1998) ; Stensvold et al. (2002) ; Stensvold (2008) ; Zika and Farrar (2009) ; Gilman et al.
the alleles for them as fast (F), medium
(M), and slow (S). Genetic identity
(GI) measures between populations
and species and was calculated using
the program POPGENE ( Yeh and
Boyle, 1997 ).
Morphology measurements — We made
a number of both quantitative and
qualitative measurements of mor-
phology on the samples from 2007
to 2009 (Appendix S3). We scanned
samples and measured quantitative
traits that include angle and length
measurements of Swartz and Brunsfeld
(2002) using the program ImageJ
( Rasband, 2008 ). We measured ve
angles on the bottom four pinnae and
calculated 13 length ratios ( Davis,
1983 ) ( Fig. 1 ; Appendix S3). Length
ratios allowed us to quantify the over-
all dimensions of the fern; for exam-
ple, we quantified how broad the
trophophore was in comparison to its
height. We measured rough shape
parameters using circularity (where 1 = perfect circle) of a bottom
pinnae and counted the number of pinnae pairs, for a total of 22
Qualitative traits included characters that mirror traditional
species descriptions and keys of Botrychium s.s. (Appendix S3).
To determine morphological distance, we described each qualita-
tive character using binary coding. A Mantel test using the package
vegan ( Oksanen et al., 2015 ) in the program R ( R Core Team,
2014 ) showed that the quantitative and qualitative data sets were
correlated, although at a low level ( r = 0.26, P < 0.001). Given this
correlation, we combined the data sets for subsequent analyses. We
visualized variation in the data using principal component analysis
APRIL 2016 , VOLUME 103 • WILLIAMS ET AL. —CRYPTIC SPECIATION IN BOTRYCHIUM • 5
(PCA) using the program JMP (v. 9, SAS Institute, Cary, North
Carolina, USA) following Zelditch et al. (2004) and Fresnedo-
Ramírez et al. (2011) . To test the sensitivity of the data to analysis
method, we also analyzed morphological data using nonmetric
multidimensional scaling (NMDS) ordination but found no quali-
tative di erences between them and so do not present the latter
Ampli ed fragment length polymorphisms — We extracted DNA
from dried 2007–2009 samples using Qiagen DNeasy Plant Mini
kits (Qiagen, Valencia, California, USA). We used a representa-
tive sampling of the individuals used in the morphometric study,
with a focus on B. matricariifolium individuals in addition to its
putative parent species B. angustisegmentum and B. pallidum .
Data from AFLPs were generated from a library of genomic DNA
digested with the restriction enzymes EcoR1 and MseI following
the method of Williams and Waller (2012) . To analyze the DNA
fragments, we used the program GeneMarker v. 1.4 (So Genetics,
State College, Pennsylvania, USA), creating panels and scoring
fragments by hand, using the conservative criteria of Holland et
al. (2008) : intensity levels over 100, band sizes greater than 100,
symmetric peaks, and small bin sizes. To standardize data across
runs, we scored replicates within each run and examined a set of
alleles that ampli ed across all runs.
We translated the data into appropriate le structures in R
using the source code AFLPdat ( Ehrich, 2006 ). We used NMDS
(following Emshwiller et al., 2009 ) as implemented in vegan
( Oksanen et al., 2015 ) to analyze the data. We examined B. an-
gustisegmentum , B. pallidum , and B. matricariifolium using the
program v. 184.108.40.206 ( Pritchard et al., 2000 ) to determine
population structure. We followed the procedure of Evanno et
al. (2005) for coding AFLPs and ran simulations of 1,000,000
generations, with 10,000 generations of burn-in and inferred the
distribution of alleles ( λ ). We ran each set with the admixture
model with two runs per k (1–10). e change in likelihood as
the number of populations ( k ) increases can be more informa-
tive than the likelihood estimate, so we also calculated Δ k
following Evanno et al. (2005) .
Comparing morphology and genetics — We performed Mantel
tests with R packages vegan ( Oksanen et al., 2015 ) and ape ( Paradis
et al., 2004 ) to test a possible correlation between morphological
and genetic distances using the Gower distance method ( Gower,
1971 ) for morphological traits and Nei–Li genetic distance
( Nei and Li, 1979 ). We also ran partial Mantel tests controlling
for geographic distances between collection sites using vegan
( Oksanen et al., 2015 ).
Allozymes — Of 22 gene loci examined by enzyme electrophoresis,
20 varied within or among members of the B. matricariifolium
complex, their putative diploid parents ( Table 2 ), and related dip-
loid species. At three gene loci in B. matricariifolium and one in B.
michiganense , we found alleles that were not detected in range-wide
sampling of the putative parental species.
We calculated genetic variability parameters of 10% polymor-
phic loci and 1.09 alleles per locus for B. angustisegmentum ,
TABLE 2. Alleles expressed at enzyme-coding loci in B. michiganense and related species. Allele numbers re ect relative migrating positions within genus
Botrychium, with alleles of the presumed diploid parents shaded. Shaded rows indicate loci at which B. angustisegmentum and B. lanceolatum consistently
display di erent alleles. In allotetraploid combinations, the presumed Lanceolatae contribution is listed rst, using the format of the presumed contributing
diploid, or black where the contributor is ambivalent. Allele numbers in parentheses are orphans that were not detected in any of the three diploid species,
presumed to have been contributed by the non- Lanceolatae parent. The species are B. angustisegmentum (ang), B. lanceolatum (lan), B. matricariifolium (mat), B .
michiganense (mich), B. hesperium (hesp), and B. pallidum (pal). Modi ed from Gilman et al. (2015) and reproduced with permission.
Locus ang lan mat mich hesp pal
Aat-2 3 3 3+( 2 ) 3+( 2 ) or 3+ 3 3+ 3 3
Aat-3 0.2 0.5 0.2 + 2 0.2 + 2 or 0.5 + 2 0.5 + 2 2
Aat-4 0.2 0.5 0.2 + 3 0.2 + 3 or 0.5 + 3 0.5 + 3 3
Aco-2 3 3 3+ 3 3+ 3 3+ 3 2 or 3
Dia-1 1 1 1+ 2 1+ 2 1+ 2 2
Dia-2 1 0.5 1 + 1 1 + 1 or 0.5 + 1 0.5 + 1 1
Dia-3 2.5 or 3 3 3+ n 3+ n 3+ n a N
Dia-4 5 6 5 + 8 5 + 8 n + 8 8
Idh-1 1 1 1+ 1 1+ 1 1+ 1 1
Mdh-1 2 2 2+ 1 2+ 1 2+ 1 1
Mdh-2 3 7 3 + 3 7 + 3 7 + 3 3
Mdh-3 2.5 or 3 2.5 or 3 3+ 2 3+ 2 or 2.5+ 2 3+ 2 2
Mdh-4 2 2 2+ 2 2+ 2 2+ 2 2
6Pgd 4 4 4+ 1 4+ 1 4+ 1 1
Pgi-2 4 4 4+( 1 ) or 4+ 2 or 4 +n or n+ 2 4+ 2 4+ 2 or 4+ n 2
Pgm-1 3 3 3+ 1 or 3+ n 3+ 1 or 3+ n 3+ 1 or n+ 1 1
Pgm-2 1.5 1.5 1.5+ 2 1.5+ 2 1.5+ 2 2
Skdh 2 1 2 + 1 or 2 + n 2 + 1 or 1 + 1 1 + 1 1
Tpi-1 3 3 3+( 1 ) or 3+ 3 3+ 3 3+ 3 3
Tpi-2 3 3 3+ 3 3+ 3 3+ 3 3
a It is assumed that tetraploid plants received two homoeologous copies of each gene, but because these genes are redundant in activity, one of the homoeologous copies may have become
silenced (n). We have not attempted to detect this possibility at loci receiving the same allele from both parental diploids here. Where parental diploids di er in potential contribution but
only one allele is expressed, we suggest that silencing has occurred.
b Diploid Botrychium species regularly express banding patterns re ecting four independently migrating loci for Diaphorase, but B. pallidum expresses only three bands for this enzyme family.
It is possible that Dia-3 has become silenced or that it comigrates with one of the other loci.
6 • AMERICAN JOURNAL OF BOTANY
TABLE 3. Distribution of B. matricariifolium morphotypes among potential TPI/PGI genotypes in the western Great Lakes region. Leaf silhouettes are representative specimens of the morphotoypes
with the speci ed genotype. Genotypes without silhouettes had no detected pairs of that genotype/morphotype combination. Genotype formulas at the bottom of boxes are the alleles visualized or
expected followed by the number and percentage of the morphotype with that genotype.
TPI Genotype Morphotype FF– FFSS MM– MMSS SSSS or–SS
No. and %
FSSS acuminate FS/F 0 FS/FS 0
FS/M 21 (38.)
FS/MS 5 (9 .3)
FS/SS 0 26 (48.1)
FS/F 13 (13)
FS/FS 28 (28)
FS/MS 10 (10)
FS/SS 0 51 (51)
typical FS/F 0 FS/FS 0 FS/M 0 FS/MS 0 FS/SS 0 0
SSSS or–SS acuminate S/F 0 S/FS 0 S/M 0
S/MS 27* (50)
S/S 1 (1.9)
S/F 9 (9)
S/FS 35 (35)
S/M 2 (2)
S/MS 3 (3)
S/S 0 49 (49)
S/F 13 (9.4)
S/FS 117 (84.8)
S/MS 5 (3.6)
S/S 3 (2.2)
Total no. and %
acuminate 0 0 21 (38.9) 32 (59.3) 1 (1.9) 54
blocky 22 (22) 63 (63) 2 (2) 13 (13) 0 100
typical 13 (9.4) 117 (84.8) 0 5 (3.6) 3 (2.2) 138
Total 35 (12) 180 (61.6) 23 (7.9) 50 (17.1) 4 (1.4) 292
APRIL 2016 , VOLUME 103 • WILLIAMS ET AL. —CRYPTIC SPECIATION IN BOTRYCHIUM • 7
5% and 1.05 for B. lanceolatum , and 5% and 1.05 for B . pallidum ,
respectively. For the allotetraploid taxa, the respective values were
70% and 1.9 for B. matricariifolium (including morphotypes
“acuminate” and “blocky”), 75% and 1.9 for B. michiganense , and
65% and 1.75 for B. hesperium . We observed allelic variability that
was not attributable to xed heterozygosity or gene silencing as-
sociated with allopolyploidy in two loci for B. matricariifolium
and six loci for B. michiganense . Considering only these loci, val-
ues for the percentage of polymorphic loci and alleles per locus
were then 10% polymorphic and 1.1 alleles per locus for B. matri-
cariifolium , 30% and 1.05 for B. michiganense , and 5% and 1.05
for B. hesperium .
The degree of genetic similarity among allopolyploid taxa
expressed as genetic identity (GI: Nei, 1972 , 1978 ) was highest
among the morphotypes of B. matricariifolium , with values of
0.995 between “typical” and “acuminate” morphotypes, and 0.954
between “typical” and “blocky.” GI values between B. michiganense
and the morphotypes of B. matricariifolium ranged from 0.807 to
In addition to allelic variability contributed by parental di er-
ences (homoelogous variability) in morphotypes of Botrychi um
matricariifolium , they also displayed allelic variability in contri-
butions from a single parent (homologous variability) at TPI-1
(with two alleles) and PGI-2 (with three alleles). In both cases,
the variable alleles are presumed to be contributed by the B. pal-
lidum parent as the alleles in question have not been detected in
any species of section Lanceolatae . e distribution of morphot-
ypes “typical”, “blocky”, and “acuminate” among potential TPI/
PGI genotypes is shown in Table 3 , with TPI alleles represented
as F (fast) and S (slow) alleles, and PGI alleles represented as F
(fast), M (medium), and S (slow alleles). One half of the TPI/PGI
combinations were undetected in any of the 292 plants sampled
in the WGL region, and no plants of “typical” morphology dis-
played the TPI F allele. e TPI F allele was expressed in ap-
proximately half of the plants of both “blocky” and “acuminate”
e PGI F allele was expressed in 94.2% of plants with “typical”
morphology, 85% of plants with “blocky” morphology, and in no
plants with “acuminate” morphology. e PGI M allele was ex-
pressed in 3.6% of “typical” plants, 15% of “blocky” plants, but
98.2% of “acuminate” plants. e PGI S allele was expressed in 5.8%
of “typical” plants, 13% of “blocky” plants, and 61.2% of “acumi-
nate” plants. Potential origins accounting for the unbalanced distri-
butions of morphotypes are presented in Fig. 5 .
Morphology — e ordination of ve species using 42 morphological
characters showed a pattern in which diploid species were separated,
but with considerable overlap of individuals between species ( Fig. 6 ).
Individuals with missing morphological data were not included. e
rst three axes in this combined PCA accounted for 25.86% of the
variation ( Table 4 ). e analysis that used only quantitative traits ac-
counted for a higher percentage of the variation in the rst three axes,
but had a similar pattern to the broader analysis. Characters making
up the rst three axes included the spacing of the rst and second
pinnae, broadness of the trophophore, shape of the pinnae, and an-
gles of the second pinnae pair. In general, B. angustisegmentum and
species in section Lunariae species were separated on the perimeter
of the ordination, with allopolyploids B. matricariifolium (including
B. acuminatum and B. michiganense ined.) and B. minganense
(putatively derived from hybridization between B. pallidum and B.
lunaria ) in two broad and overlapping central groups.
AFLPs — e NMDS ordination of AFLP data showed discrete
clustering of B. angustisegmentum , B. pallidum , B. neolunaria , and
FIGURE 5 The minimum number of original hybridizations necessary
to produce the genotypes observed in three morphotypes of
Botrychium matricariifolium . Origins 1 and 2 can account for all geno-
types of the “typical” morphotype. Origin 3 alone can account for all
of the acuminate morphotypes. Origin 4 provides the appropriate
genotypes for 96.4% of “typical” individuals. Genotype formulae refer
to TPI/PGI genotypes possessing fast-, medium-, and slow-migrating
FIGURE 6 Principal component analysis of morphology characters in
Botrychium ferns, including 22 quantitative (e.g., angle of rst pinnae,
trophophore height, and trophophore broadness) and 20 qualitative
characters (i.e., color, surface texture, and trophophore shape). Axes
scores are low, but a pattern with allotetraploid species in between dip-
loid parental species is seen.
8 • AMERICAN JOURNAL OF BOTANY
B. lunaria ( Fig. 7 ). Allotetraploid individuals formed a di use group
between B. angustisegmentum and B. pallidum . For the allotetra-
ploids of interest ( Fig. 8 ), B. michiganense (numbered 1–5) was dis-
tinct from B. matricariifolium , while B. acuminatum and plants with
“acuminate” and “blocky” morphotypes were in the central cluster
with “typical” B. matricariifolium . One outstanding individual
(number 6) may represent allohexaploid B . pseudopinnatum , which
has a genomic contribution from B. neolunaria ined. Notably, this
individual was displaced toward B. neolunaria in Fig. 7 . We were
unable to relocate this plant or similar individuals in the eld for
Within pooled species, there is a clear distinction between B. pal-
lidum , B. angustisegmentum , and the allotetrapolyploid group ( k = 3,
Fig. 9A ). Within the allopolyploids, consistently identi-
ed 10 individuals ( k = 4, Fig. 9B ) that contained more than 0.1 mem-
bership in a di erent genetic group. Individuals of B. michiganense
(numbers 1–5) were separated from B. matricariifolium, as was the
putative species B. pseudopinnatum (number 6). At values of k higher
than three, these individuals were subdivided into two genetic
groups, but no other individual was identi ed so strongly as being
di erent from the main body of B . matricariifolium plants.
Comparing morphology and genetics — Mantel tests found a sig-
ni cant correlation between genetic and morphological distance
across all species and within a data set of B. angustisegmentum and
B. pallidum ( Table 5 ). Correlations between genetic and geographic
distance within B. angustisegmentum , B. pallidum , and B. matricari-
ifolium separately were also signi cant. However, there was no cor-
relation between genetic distance and morphology among B.
matricariifolium individuals or between allopolyploids (including
B. michiganense and B. acuminatum ). Partial Mantel tests that con-
trolled for geographic distance while testing the correlation be-
tween genetic and morphological distance were also not signi cant
in the latter cases ( P > 0.07).
Our investigation of cryptic species in the allotetraploid
B. matricariifolium complex found evidence for repeated hy-
bridizations between similar diploid parents that resulted in
morpholo gically similar species and confusing morphotypes within
TABLE 4. PCA components from morphological analyses of Botrychium species. The qualitative data set uses common key and description characters. The
quantitative data set uses angle measurements and length ratios following Swartz and Brunsfeld (2002) , with details in Appendix S3. The combined data set
corresponds to Fig. 6 .
Data set PCA axis one Correlation PCA axis two Correlation PCA axis three Correlation
Qualitative 9.3% of variation 8% of variation 6% of variation
Pinnae margins entire or
−0.29 Pinnules round −0.40 Similar 1st and 2nd
symmetrical or not
0.32 Terminal trophophore
−0.31 3 or more terminal pinnae −0.22
Sporophore branching 0.33 Color −0.27 1st and 2nd pinnae pairs
di erent because
Sporangial thumbing 0.33 Pinnae shape 0.27 1st and 2nd pinnae pairs
di erent because of
Pinnae margins not lobed 0.35 Trophophore ovate/
0.36 1st and 2nd pinnae pairs
di erent because of length
Quantitative 31.9% of variation 17.4% of variation 12.6% of variation
1st pinnae pair to trophophore
end: length of trophophore
−0.36 Trophophore broadness 0.35 Length of pinnae: width
1st pinnae pair: trophophore
−0.36 2nd pinnae whole angle 0.34 Trophophore length 0.38
2nd pinnae pair to trophophore
end: length of trophophore
0.36 Number of pinnae pairs 0.33 Pinnae broadness 0.34
2nd pinnae pair: trophophore
0.37 Trophophore length 0.32 1st pinnae partial angle −0.34
1st pinnae pair: sporophore
0.42 2nd pinnae partial angle 0.56 1st pinnae stem angle 0.29
Both 11.4% of variation 7.98% of variation 6.5% of variation
1st pinnae pair to trophophore
end: length of trophophore
−0.35 1st pinnae whole angle 0.28 Trophophore length 0.30
1st pinnae pair: trophophore
−0.35 2nd pinnae stem angle −0.25 Sporophore length 0.28
2nd pinnae pair to trophophore
end: length of trophophore
0.25 Trophophore broadness 0.25 Pinnae margins −0.25
2nd pinnae pair: trophophore
0.25 2nd pinnae partial angle 0.25 Pinnae shape 0.23
1st pinnae pair: width of
0.32 Number of pinnae pairs 0.24 Deeply dissected 0.23
APRIL 2016 , VOLUME 103 • WILLIAMS ET AL. —CRYPTIC SPECIATION IN BOTRYCHIUM • 9
B . matricariifolium. ese taxa and their origins were clari ed us-
ing genetic evidence derived through enzyme electrophoresis
and AFLP markers.
Cryptic species in the B. matricariifolium complex — Morphologi-
cal intermediacy strongly supports the origin of the tetraploid B .
matricariifolium complex by hybridization events between diploid
species in sections Lanceolatae and Lunariae (twice-pinnate and
once-pinnate sections, respectively). Range-wide analysis of all
Botrychium diploids using codominant allozyme alleles (the al-
leles referred to below) points to B. angustisegmentum and/or B.
lanceolatum ( Lanceolatae ) and B. pallidum ( Lunariae ) as the most
likely parents of all members of the B. matricariifolium complex
( Table 2 ). e “best match” for the Lanceolatae parent of tetraploid
B. hesperium from allozyme data is western B. lanceolatum . Lan-
ceolatae parentage of B. michiganense is best explained as having
an origin that predates the eastern–western di erentiation of
these two species because it contains alleles of both B. angustiseg-
mentum and B. lanceolatum ( Gilman et al., 2015 ). e best match
for the Lanceolatae parent of B . matricariifolium , including mor-
photypes “typical”, “blocky”, and “acuminatum” is B. angustiseg-
mentum (see Fig. 2 and Table 2 ).
Within the western Great Lakes (WGL) area where B. michi-
ganense and B. matricariifolium co-occur, we found that each
species had unique alleles. Of 20 loci examined, ve had alleles
unique to B. michiganense , and two had alleles unique to B . mat-
ricariifolium . ese allelic di erences translate into a genetic
identity (GI: Nei, 1972 , 1978 ) between the two allotetraploids of
0.905, which is high relative to GI numbers typically found be-
tween diploid species ( Crawford, 1985 ; Hamrick and Godt,
1990 , 1996 ). ese high GI values are the result of a shared
B. pallidum parentage common to both tetraploid species. ese
allotetraploids also share alleles common to both Lanceolatae
parents, but B. michiganense also contains ve Lanceolatae al-
leles that are now present only in B. lanceolatum and two only in
. The North American distribution of
B. michiganense is consistent with an origin that pre-dated the
di erentiation of western and eastern Lanceolatae species.
Botrychium michiganense occurs sympatrically with B. lanceola-
tum in the Rocky Mountains from Wyoming northward to southern
British Columbia and Alberta. In boreal regions from northern
Alberta eastward, B. matricariifolium occurs sympatrically with
B. angustisegmentum .
Several observations support recognizing B. michiganense and
B. matricariifolium as distinct species. ese include the relatively
lower GI between B. michiganense and B. matricariifolium com-
pared with GI values between B. matricariifolium morphotypes, the
seven unique alleles that di erentiate B. michiganense and B. matri-
cariifolium , and likely di erent diploid parentage of B. michi-
ganense and B. matricariifolium .
Allelic di erentiation between the morphotypes within B. mat-
ricariifolium does not match the level of genetic distinction ob-
served between B. matricariifolium and B. michiganense . Although
genotype combinations of TPI and PGI have a level of genotype
correlation with morphotype, these genotypes are neither con-
stant within morphotypes nor limited in most cases to particular
morphotypes. Genetic identity values between the morphotypes
were 0.989 between “typical” and “blocky,” and 0.995 between
“typical” and “acuminate,” the latter morphotype including for-
mally described B. acuminatum . Genetic similarities and mor-
phological gradations observed between the morphotypes support
retaining them all, including B. acuminatum , within the species B.
Our multivariate analyses of a mixture of quantitative and quali-
tative traits (see Table 4 ) were only moderately useful in separating
entities within the Botrychium genus. Morphological characters
produced clustering of diploid species, but with overlap between
them, and there was little clustering within the B. matricariifolium
complex ( Fig. 6 ). Instead, these allopolyploids plotted in a di use
array, with individuals arranged across a broad morphospace be-
tween and overlapping with putative parent species. is pattern
updates our understanding of the range of variation seen in this
group (and the reasons for its frustrating taxonomic issues). It is
important to recognize that this analysis is based on unbiased selec-
tion of characters and includes many characters that the allopoly-
ploid taxa likely share because of intermediacy between two very
di erent parents. Such undiagnostic characters may compromise
the e ectiveness of the relatively few characters that are diagnostic
for taxa across the genus. Separating diagnostic from nondiagnostic
characters is greatly aided by genetic analysis that rst groups the
same set of plants on the basis of genetic similarity, allowing subse-
quent recognition of character states correlated with speci c groups.
FIGURE 7 Nonmetric multidimensional scaling (NMDS) ordination of
AFLP data for six species of Botrychium in the Great Lakes area, including
two species of the allotetraploid B. matricariifolium complex and their
putative diploid parents. Two diploid species ( B. lunaria and B. neolu-
naria ) and an allotetraploid species ( B. minganense ), all unrelated to the
B. matricariifolium complex are included for scaling purposes. Allotetra-
ploids B. matricariifolium and B. michiganense are positioned between
putative parents B. pallidum and B. angustisegmentum . Allotetraploid B.
minganense is positioned between putative parents B. neolunaria and B.
pallidum ( Farrar, 2011 ). Botrychium lunaria and B. neolunaria occupy dis-
crete areas as proposed by Stensvold (2008) .
10 • AMERICAN JOURNAL OF BOTANY
Our AFLP data corroborate allozyme results relative to genetic
diversity in the B. matricariifolium complex. Diploid species of
Botrychium clustered together into more discrete groups in genetic
space than in morphological space ( Fig. 7 ). Botrychium matricari-
ifolium, including the “blocky” and “acuminate” morphotypes, and
B. michiganense together form a distinct genetic cluster between
their putative parent species but do not show strong separation
within the group ( Fig. 8 ). Individuals of presumed B . michiganense
are somewhat removed from the central cluster, as is one individual
presumed to be B. pseudopinnatum . Within the tetraploid group,
identi ed the ve plants of B. michiganense , but with
only partial distinction from B. matricariifolium , which might be
expected as they both likely share a B. pallidum genome.
e AFLP data also support hypotheses of putative parental spe-
cies for other allotetraploids. In addition to the B. matricariifolium
complex being plotted between Lanceolatae and B. pallidum parents,
the allotetraploid B. minganense places between B. lunaria and B.
pallidum , which are its two putative parents ( Wagner, 1993 ; Hauk
and Hau er, 1999 ). e outlier of the B. matricariifolium complex,
which we presume to be B . pseudopinnatum , places toward B. neolu-
naria . e genomic constitution of this allohexaploid contains alleles
FIGURE 8 Nonmetric multidimensional scaling (NMDS) ordination of AFLP data for the B. matricariifo-
lium complex. Plants of B. michiganense (numbered 1–5) occupy space peripheral to plants of B. matri-
cariifolium , including morphotypes “blocky” and “acuminate”. The outlying plant numbered (6) displays
morphology consistent with that of allohexaploid B. pseudopinnatum , which contains a genome
putatively derived from B. neolunaria ( Farrar, 2011 ). Geographic location (indicated by symbol color)
does not correlate with taxon or morphology. Abbreviations: Co., County; Mtn, Mountain; Nat’l For.,
that are otherwise unique to B. neolu-
naria , and we presume was derived by
hybridization between a diploid mem-
ber of section Lanceolatae and allotet-
raploid B. minganense , the latter of
which contains a genome from B. neo-
lunaria ( Wagner and Wagner, 1990 ;
Farrar, 2011 ). e AFLP separation of
European and North American acces-
sions of B. “ lunaria ” supports Stens-
vold’s (2008) proposed recognition of a
distinct North American taxon, B. neo-
lunaria Stensvold ined.
Inferred polyploid formation in
Despite the high genetic
and morphological similarity between
B. matricariifolium morphotypes, the
nonrandom distribution of morphot-
ypes among possible TPI/PGI geno-
types ( Table 3 ) requires explanation.
ree alleles of PGI are present through-
out the range of B. matricariifolium ,
but in the WGL study area, the M allele
is universally present in the “acumi-
nate” morphotype. e TPI F allele
was not detected outside the WGL
area, nor was it detected in any plants
with “typical” morphology within
the WGL area. A high frequency within
the area (it is found in approximately
half of the “blocky” and “typical” mor-
photypes) suggests a local origin of
this allele. It seems unlikely that de
novo mutation within B. matricariifo-
lium producing the TPI F allele would
also contribute the correlated morpho-
logical changes. Conversely, it is more
likely that the new allele was contrib-
uted by a somewhat di erentiated diploid parent. Because this allele
has not been detected in section Lanceolatae and is present in other
species of section Lunariae , B. pallidum is the most likely contributor
of TPI F in B. matricariifolium .
e generation of the genotypes observed in B. matricariifolium
required a minimum of four independent hybridizations between pa-
rental diploids ( Fig. 5 ). ree of these origins required alleles not cur-
rently present in putative parents B. angustisegmentum or B. pallidum .
ese alleles, TPI F and PGI F, have not been detected in species of
section Lanceolatae in our range-wide sampling, but are present in
other species of section Lunariae . However, these species contain
other alleles and morphologies that are not compatible with them
contributing to the parentage of B. matricariifolium . e most parsi-
monious assumption is that these alleles were therefore at some time
present in B. pallidum , PGI F as a widespread allele and TPI F as a
more local and possibly more recent allele in the WGL area. Based on
these assumptions, we hypothesize the following steps in evolution of
the observed genetic diversity within B. matricariifolium ( Fig. 5 ):
(1) A Botrychium pallidum parent with the TPI/PGI genotype SS/
FF combined in hybridization with a B. angustisegmentum
parent with the genotype SS/SS to produce allotetraploid B.
APRIL 2016 , VOLUME 103 • WILLIAMS ET AL. —CRYPTIC SPECIATION IN BOTRYCHIUM • 11
matricariifolium genotypes SSSS/FFFF and SSSS/FFSS that
are displayed by the most common and widespread “typical”
plants of the species, and rare plants with a SSSS/SSSS
(2) e PGI F allele was lost or mutated to current M, and
B. pallidum in the western Great Lakes area that had geno-
types FS/MS combined with B. angustisegmentum SS/SS in
multiple independent hybridizations to produce B. matri-
cariifolium “blocky” and “acuminate” morphotypes with
their multiple allozyme genotypes, and the uncommon
SSSS/MMSS genotype of “typical” B. matricariifolium .
FIGURE 9 Results from STRUCTURE analysis comparing Botrychium species including the Botrychium matri-
cariifolium complex. (A) All individuals ordered by genetic similarity ( k = 3). Numbers correspond to Fig. 8 .
The horizontal line indicates B. michiganense individuals, numbered 1–5. Individual no. 6 is likely the
hexaploid B. pseudopinnatum . (B) Allopolyploids dispersed among geographic populations. Numbers
indicate the same individuals as in (A). Botrychium michiganense individuals are distinct from other indi-
viduals, with high membership in other groups.
TABLE 5. Results of Mantel tests between genetic and morphological distances of Botrychium species. Genetic distance is the Nei–Li genetic distance based on
AFLP data. Morphological distance is the Gower distance based on 42 quantitative and qualitative traits.
Species Distances compared No. individuals No. permutations Correlation P
All Morphological: quantitative vs.
121 10000 0.262 <0.0001
All Genetic vs. morphology 121 10000 0.350 <0.0001
Botrychium pallidum (2 n ),
B. angustisegmentum (2 n )
Genetic vs. morphology 36 10000 0.648 <0.0001
Botrychium allotetraploids a (4 n ) Genetic vs. morphology 55 10000 0.104 0.108
B. matricariifolium b (4 n ) Genetic vs. morphology 50 10000 −0.074 0.783
Note: Bold text indicates signi cance at P = 0.05 level.
a Includes Botrychium michiganense .
b Includes “blocky” and “acuminate” morphologies.
(3) A er a reduction to its current
SS/MM genotype following the
loss of the TPI F allele, B. palli-
dum may have hybridized with
current B. angustisegmentum
SS/SS to produce the formally
described B. acuminatum , which
is the most localized genotype
of the “acuminate” morpho-
type, described by W. H. Wagner
from a single locality in northern
In considering the possibility of
multiple origins as an explanation for
the morphological variation in allo-
tetraploid B. matricariifolium , we
recognize that the observed morpho-
logical variation within taxa likely
includes genetically based variation
that is not detected in our allozyme
analysis. Significant morphological
variation is often present among
plants of the same size in populations
occupying apparently uniform envi-
ronmental conditions ( Williams and
Waller, 2015 ). Allotetraploids of dif-
ferent morphology can arise through
multiple hybridizations between dif-
ferent morphologies of the same two
parental diploids. Such variation is
displayed in Fig. 10 , with morpho-
logically variable but allozymically
identical plants of B. angustisegmen-
tum and B. pallidum adjacent to B. matricariifolium plants of “typ-
ical”, “blocky”, and “acuminate” morphologies to illustrate how
allotetraploids of di erent morphology might arise through mul-
tiple hybridizations between di erent morphologies of the same
two parental diploids.
Genetic data aid in identi cation of diagnostic morphological
While molecular genetic studies of plant diversity
over the last half-century have largely supported previous con-
cepts based on morphology, some have profoundly altered previ-
ous conclusions. e latter include groups where morphology was
12 • AMERICAN JOURNAL OF BOTANY
FIGURE 10 Confusing morphological variation within allotetraploid Botrychium matricariifolium (middle
row) is explained by multiple origins from its diploid parents B. angustisegmentum (top row) and B. palli-
dum (bottom row).
either too di erent from the remainder of ferns ( Rothfels et al.,
2015 ) (e.g., relationships among Equisetales, Psilotales, and
Ophioglossales and between these clades), or too similar (as in
cryptic species) to permit clear morphological distinction of spe-
cies and clades (e.g., cheilanthoid clades recently identi ed
through molecular data, Grusz et al., 2014 ). Revelation of natural
groups using genetic data allows us to identify morphological
characters diagnostic of a particular species or clade that were
previously obscured in a matrix of nondiagnostic variation. is
is the case in the Botrychium matricariifolium complex, where our
genetic distinction results aide in the recognition of the most di-
agnostic morphological characters ( Fig. 11 ) allowing recognition
of B. michiganense ( Gilman et al., 2015 ). By calling our attention
to unusual genotypes and their correlation with certain morpho-
logical characters in B. matricariifolium , genetic analysis also al-
lowed recognition of genetically based morphological variation
and led to hypotheses of multiple origins of the species by sequen-
tial hybridization events involving the same diploid parents. us,
genetic analyses can aid eld botanists by facilitating identi ca-
tion of diagnostic morphological characters and by generating
explanatory hypotheses regarding the evolution of the diversity
that we see.
Dealing with allopolyploids and cryptic species — e Botrychium
matricariifolium complex joins the ranks of multiple other taxa
in which independently derived polyploid lineages have been
documented ( Soltis, et al., 1993 , 2014 ; Haufler et al., 1995 ; Beck
et al., 2012 ; Sigel et al., 2014 ). The role of polyploid formation is
increasingly recognized as common and an important process of
diversification and speciation ( Soltis et al., 2014 ). The increase
in allelic variability of allopolyploids
after combination of genomes from
two parental diploids may confer
adaptive exibility ( Soltis et al., 2004 ).
In B. matricariifolium , observed di-
versity measures are much increased
over those recorded for either pa-
rental diploid (see Table 2 ). e large
range and frequency of occurrence
of B. matricariifolium also argues for
ecological success of allotetraploid
Systematic studies in taxa that
have multiple independent lineages
of morphologically small simple
plants can be particularly challeng-
ing ( Perrie and Brownsey, 2005 ; Per-
rie et al., 2010b ; Grusz et al., 2014 ).
e combined evidence presented
here supports recognition of a single
and morphologically variable spe-
cies, B. matricariifolium , that in-
cludes B. acuminatum , which was
derived from multiple hybridization
events between the same two ances-
tral diploids. It also is consistent
with recognition of a distinct B.
michiganense species that was de-
rived from genetically di erent an-
cestral diploids that were morphologically similar to those of
B. matricariifolium . While a case could be made for recognizing
“acuminate” and “blocky” morphotypes as varieties of B. matri-
cariifolium , this would introduce signi cant taxonomic prob-
lems. Indeed, for the eld botanist, the frequent presence of a
continuum of morphologies between such varieties would be
seriously problematic ( Fig. 2 ). Continued studies of allopoly-
ploid origins in Botrychium should concentrate on the patterns
and processes of multiple hybrid origins and how these may cor-
relate with morphology ( Perrie et al., 2010a ; Sigel et al., 2014 ).
Studies on the genetic expression of diagnostic morphological
characters across the genus would also be extremely useful. We
therefore recommend continued recognition of B. matricariifo-
lium , with all its variability, as a single uni ed species that pres-
ents a fascinating window into the origins of allotetraploid
variability in ferns.
e authors thank P. Diggle and two anonymous reviewers for
invaluable comments; D. Waller, D. Baum, C. Ané, and E. Emshwiller
for review of previous dra s; A. Bochte for image analysis assistance;
J. Wendel, J. Nason, and C. Skelton for electrophoresis expertise and
facilities; and L. Loope, B. Leutscher, and S. Trull for help with
permits. is research was supported by the Clarence R. & Florence
N. Hanes Foundation, the Huron Mountain Wildlife Foundation,
the Department of Botany John J. Davis Grant, the Michigan Rare
Plant Fund, the UW-Madison Vilas Travel Award, and the USDA
APRIL 2016 , VOLUME 103 • WILLIAMS ET AL. —CRYPTIC SPECIATION IN BOTRYCHIUM • 13
Barker , M. S. , and W. D. Hauk . 2003 . An evaluation of Sceptridium dissectum
(Ophioglossaceae) with ISSR markers: Implications for Sceptridium system-
atics. American Fern Journal 93 : 1 – 19 .
Beck , J. B. , J. R. Allison , K. M. Pryer , and M. D. Windham . 2012 . Identifying
multiple origins of polyploid taxa: A multilocus study of the hybrid cloak fern
( Astrolepis intege rrima ; Pteridaceae). American Journal of Botany 99 : 1857 – 1865 .
Clausen , R. T. 1938 . A monograph of the Ophioglossaceae. Memoirs of e
Torrey Botanical Club 19 : 1 – 177 .
Crawford , D. J. 1985 . Electrophoretic data and plant speciation. Systematic
Botany 10 : 405 – 416 .
Dauphin , B. , J. Vieu , and J. R. Grant . 2014 . Molecular phylogenetics supports
widespread cryptic species in moonworts ( Botrychium s.s., Ophioglossaceae).
American Journal of Botany 101 : 128 – 140 .
Davis , J. I. 1983 . Phenotypic plasticity and the selection of taxonomic charac-
ters in Puccinellia (Poaceae). Systematic Botany 8 : 341 – 353 .
Ehrich , D. 2006 . AFLPDAT: A collection of R functions for convenient han-
dling of AFLP data. Molecular Ecology Notes 6 : 603 – 604 .
Emshwiller , E. , T. eim , A. Grau , V. Nina , and F. Terrazas . 2009 . Origins
of domestication and polyploidy in oca ( Oxalis tuberosa ; Oxalidaceae). 3.
AFLP data of oca and four wild, tuber-bearing taxa. American Journal of
Botany 96 : 1839 – 1848 .
Evanno , G. , S. Regnaut , and J. Goudet . 2005 . Detecting the number of clusters
of individuals using the so ware : A simulation study. Molecular
Ecology 14 : 2611 – 2620 .
Farrar , D. R. 1998 . Population genetics of moonwort Botrychium . In N. Berlin,
P. Miller, J. Borovansky, U. S. Seal, and O. Byers [eds.], Population and habi-
tat viability assessment (PHVA) for the goblin fern ( Botrychium mormo ),
109–113. Conservation Biology Specialist Group, Apple Valley, Minnesota,
Farrar , D. R. 2011 . Moonwort ( Botrychium ) systematics. Website www.pub-
lic.iastate.edu/~herbarium/botrychium.html [accessed August 2015]. Ada
Hayden Herbarium, Iowa State University, Ames, Iowa, USA.
Farrar , D. R. , and J. F. Wendel . 1997 . Ottawa National Forest Botrychium mat-
ricariifolium study . Final Report to the Ottawa National Forest, Ironwood,
Fresnedo-Ramírez , J. , S. Segura , and A. Muratalla-Lúa . 2011 . Morphovariability
of capulín ( Prunus serotina Ehrh.) in the central-western region of Mexico
from a plant genetic resources perspective. Genetic Resources and Crop
Evolution 58 : 481 – 495 .
Gilman , A. V. , D. R. Farrar , and P. F. Zika . 2015 . Botrychium michiganense sp.
nov. (Ophioglossaceae), a new North American moonwort. Journal of the
Botanical Research Institute of Texas 9 : 295 – 309 .
Gower , J. C. 1971 . A general coe cient of similarity and some of its properties.
Biometrics 27 : 857 – 871 .
Grant , V. 1981 . Plant speciation, 2nd ed . Columbia University Press, New
York, New York, USA.
Grusz , A. L. , M. D. Windham , G. Yatskievych , L. Huiet , G. J. Gastony , and K.
M. Pryer . 2014 . Patterns of diversi cation in the xeric-adapted fern genus
Myriopteris (Pteridaceae). Systematic Botany 39 : 698 – 714 .
Hamrick , J. L. , and M. J. W. Godt . 1990 . Allozyme diversity in plant species. In
A. H. Brown, M. T. Clegg, A. L. Kahler, and B. S. Weir [eds.], Plant popula-
tion genetics, breeding, and genetic resources, 43–63. Sinauer, Sunderland,
Hamrick , J. L. , and M. J. W. Godt . 1996 . E ects of life history traits on genetic
diversity in plant species. Philosophical Transactions of the Royal Society of
London, B, Biological Sciences 351 : 1291 – 1298 .
Hau er , C. H. , M. D. Windham , and E. W. Rabe . 1995 . Reticulate evolution in
the Polypodium vulgare complex. Systematic Botany 20 : 89 – 109 .
Hauk , W. D. 1995 . A molecular assessment of relationships among cryptic spe-
cies of Botrychium subgenus Botrychium (Ophioglossaceae). American Fern
Journal 85 : 375 – 394 .
Hauk , W. D. , and C. H. Hau er . 1999 . Isozyme variability among cryptic
species of Botrychium subgenus Botrychium (Ophioglossaceae). American
Journal of Botany 86 : 614 – 633 .
Hauk , W. D. , L. Kennedy , and H. M. Hawke . 2012 . A phylogenetic investiga-
tion of Botrychium s.s. (Ophioglossaceae): Evidence from three plastid DNA
sequence data sets. Systematic Biology 37 : 320 – 330 .
Hauk , W. D. , C. R. Parks , and M. W. Chase . 2003 . Phylogenetic studies of
Ophioglossaceae: evidence from rbcL and trnL-F plastid DNA sequences
and morphology. Molecular Phylogenetics and Evolution 28 : 131 – 151 .
Holland , B. R. , A. C. Clarke , and H. M. Meudt . 2008 . Optimizing automated
AFLP scoring parameters to improve phylogenetic resolution. Systematic
Biology 57 : 347 – 366 .
Kato , M. 1987 . A phylogenetic classi cation of Ophioglossaceae. Garden’s
Bulletin Singapore 40 : 1 – 14 .
Nei , M. 1972 . Genetic distance between populations. American Naturalist 106 :
283 – 292 .
Nei , M. 1978 . Estimation of average heterozygosity and genetic distance from
a small number of individuals. Genetics 89 : 583 – 590 .
FIGURE 11 Morphological comparison of plants determined genetically
as Botrychium michiganense and B. matricariifolium . Note that relative to
B. matricariifolium, the second pinnae pair in B. michiganense is less dis-
sected than the rst pinnae pair. Also, the trophophore in B. michiganense
is sessile, while B. matricariifolium trophophores are stalked. Exaggera-
tion of the basal pinnae is usual in B. michiganense but also can occur in
Botrychium matricariifolium . “EW” samples determined using AFLPs, and
“DF” samples are determined by allozymes.
14 • AMERICAN JOURNAL OF BOTANY
Nei , M. , and W. H. Li . 1979 . Mathematical model for studying genetic varia-
tion in terms of restriction endonucleases. Proceedings of the National
Academy of Sciences, USA 76 : 5269 – 5273 .
Oksanen , J. , F. G. Blanchet , R. Kindt , P. Legendre , R. B. O’Hara , G. L. Simpson ,
P. Solymos , et al. 2015 . vegan: Community ecology package, version 2.3-2.
Website https://CRAN.R-project.org/package=vegan .
Paradis , E. , J. Claude , and K. Strimmer . 2004 . APE: Analyses of phylogenetics
and evolution in R language. Bioinformatics 20 : 289 – 290 .
Paris , C. A. , F. S. Wagner , and W. H. Wagner . 1989 . Cryptic species, species
delimitation, and taxonomic practice in the homosporous ferns. American
Fern Journal 79 : 46 – 54 .
Perrie , L. R. , and P. J. Brownsey . 2005 . Genetic variation is not concordant
with morphological variation in the fern Asplenium hookerianum sensu lato
(Aspleniaceae). American Journal of Botany 92 : 1559 – 1564 .
Perrie , L. R. , D. J. Ohlsen , L. D. Shepherd , M. Garrett , P. J. Brownsey , and M. J.
Bayly . 2010b . Tasmanian and Victorian populations of the fern Asplenium
hookerianum result from independent dispersals from New Zealand.
Australian Systematic Botany 23 : 387 – 392 .
Perrie , L. R. , L. D. Shepherd , P. J. De Lange , and P. J. Brownsey . 2010a . Parallel
polyploid speciation: Distinct sympatric gene-pools of recurrently derived
allo-octoploid Asplenium ferns. Molecular Ecology 19 : 2916 – 2932 .
Pritchard , J. K. , M. Stephens , and P. Donnelly . 2000 . Inference of population
structure using multilocus genotype data. Genetics 155 : 945 – 959 .
R Core Team . 2014 . R: A language and environment for statistical computing,
version 2.8.1. Website: http://www.r-project.org/ .
Rai , H. S. , and S. W. Graham . 2010 . Utility of a large, multigene plastid data
set in inferring higher-order relationships in ferns and relatives (monilo-
phytes). American Journal of Botany 97 : 1444 – 1456 .
Rasband , W. 2008 . ImageJ, version 1.41o. Website http://rsb.info.nih.gov/ij .
Rothfels , C. J. , F.-W. Li , E. M. Sigel , L. Huiet , A. Larsson , D. O. Burge , M.
Ruhsam , et al. 2015 . e evolutionary history of ferns inferred from 25 low-
copy nuclear genes. American Journal of Botany .
Sigel , E. M. , M. D. Windham , and K. M. Pryer . 2014 . Evidence for reciprocal
origins in Polypodium hesperium (Polypodiaceae): A fern model system for
investigating how multiple origins shape allopolyploid genomes. American
Journal of Botany 101 : 1476 – 1485 .
Soltis , D. E. , C. H. Hau er , D. C. Darrow , and G. J. Gastony . 1983 . Starch gel
electrophoresis of ferns: A compilation of grinding bu ers, gel and electrode
bu ers, and staining schedules. American Fern Journal 73 : 9 – 27 .
Soltis , D. E. , P. S. Soltis , and L. H. Rieseberg . 1993 . Molecular data and
the dynamic nature of polyploidy. Critical Reviews in Plant Sciences 1 2 :
243 – 273 .
Soltis , D. E. , P. S. Soltis , and J. A. Tate . 2004 . Advances in the study of poly-
ploidy since Plant speciation. New Phytologist 161 : 173 – 191 .
Soltis , D. E. , C. J. Visger , and P. S. Soltis . 2014 . The polyploidy revolu-
tion then and now: Stebbins revisited. American Journal of Botany 101 :
1057 – 1078 .
Soltis , P. S. , and D. E. Soltis . 1990 . Genetic variation within and among popula-
tions of ferns. American Fern Journal 80 : 161 – 172 .
Stebbins , G. L. 1950 . Variation and evolution in plants. Columbia University
Press, New York, New York, USA.
Stensvold , M. C. 2008 . A taxonomic and phylogeographic study of the
Botrychium lunaria complex . Ph.D. dissertation, Iowa State University,
Ames, Iowa, USA.
Stensvold , M. C. , D. R. Farrar , and C. Johnson-Groh . 2002 . Two new species
of moonworts ( Botrychium subg. Botrychium ) from Alaska. American Fern
Journal 92 : 150 – 160 .
Swartz , L. M. , and S. J. Brunsfeld . 2002 . e morphological and genetic dis-
tinctness of Botrychium minganense and B. crenulatum as assessed by
morphometric analysis and RAPD markers. American Fern Journal 9 2 :
249 – 269 .
Vos , P. , R. Hogers , M. Bleeker , M. Reijans , T. Vandelee , M. Hornes , A. Frijters ,
et al. 1995 . AFLP: A new technique for DNA ngerprinting. Nucleic Acids
Research 23 : 4407 – 4414 .
Wagner , F. S. 1993 . Chromosomes of North American grapeferns and moon-
worts (Ophioglossaceae: Botrychium ). Contributions from the University of
Michigan Herbarium 19 : 83 – 92 .
Wagner , W. H. , and F. S. Wagner . 1990 . Moonworts ( Botrychium subg.
Botrychium ) of the Upper Great Lakes region, U.S.A. and Canada, with de-
scriptions of two new species. Contributions from the University of Michigan
Herbarium 17 : 313 – 325 .
Wagner , W. H. , and F. S. Wagner . 1993 . Botrychium . In Flora of North America
Editorial Committee [ed.], Flora of North America north of Mexico, vol. 2,
Oxford University Press, New York, New York, USA.
Wickett , N. J. , S. Mirarab , N. Nguyen , T. Warnow , E. Carpenter , N. Matasci , S.
Ayyampalayam , et al. 2014 . Phylotranscriptomic analysis of the origin and
early diversi cation of land plants. Proceedings of the National Academy of
Sciences, USA 111 : E4859 – E4868 .
Wiley , E. O. 1981 . Phylogenetics: e theory and practice of phylogenetic sys-
tematics. John Wiley, New York, New York, USA.
Williams , E. W. , and D. M. Waller . 2012 . Phylogenetic placement of species
within the genus Botrychium (Ophioglossaceae) based on chloroplast se-
quences, AFLPs, and ow cytometry. International Journal of Plant Sciences
173 : 516 – 531 .
Williams , E. W. , and D. M. Waller . 2015 . Tracking morphological change and
demographic dynamics in ephemeral Botrychium s.s. (Ophioglossaceae)
populations. Journal of the Torrey Botanical Society 142 : 152 – 165 .
Winther , J. L. , and W. E. Friedman . 2007 . Arbuscular mycorrhizal symbi-
onts in Botrychium (Ophioglossaceae). American Journal of Botany 9 4 :
1248 – 1255 .
Yeh , F. C. , and T. J. B. Boyle . 1997 . Population genetic analysis of co-dominant and
dominant markers and quantitative traits. Belgian Journal of Botany 126 : 157 .
Zelditch , M. L. , D. L. Swiderski , H. D. Sheets , and W. L. Fink . 2004 . Geometric
morphometrics for biologists: A primer. Elsevier Academic Press, San
Diego, California, USA.
Zika , P. F. , and D. R. Farrar . 2009 . Botrychium ascendens W. H. Wagner
(Ophioglossaceae) in Newfoundland and notes on its origin. American Fern
Journal 99 : 249 – 259 .