Journal of Heredity 2014:105(2):265–275
doi:10.1093/jhered/est074
Advance Access publication October 25, 2013
© The American Genetic Association 2013. All rights reserved.
For permissions, please e-mail: [email protected]
Population Genetics of Braun’s
Rockcress (Boechera perstellata,
Brassicaceae), an Endangered Plant
with a Disjunct Distribution
Carol J. Baskauf, Nacole C. Jinks, Jennifer R. Mandel, and David E. McCauley
From Department of Biology, Austin Peay State University, PO Box 4718, Clarksville, TN 37044 (Baskauf and Jinks);
Department of Plant Biology, University of Georgia, Athens, GA (Mandel); and Department of Biological Sciences,
Vanderbilt University, Nashville, TN (McCauley). Nacole C. Jinks is now at Environmental Sciences, Mount Juliet, TN.
Address correspondence to Carol J. Baskauf at the address above, or e-mail: [email protected].
Data deposited at Dryad: http://dx.doi.org/10.5061/dryad.31rq4
Abstract
Boechera perstellata is an endangered plant found only in middle Tennessee and north central Kentucky. After sampling 4
Tennessee and 3 Kentucky populations, genetic variability and population structure were examined for this species using
isozymes, chloroplast DNA, and microsatellites (averaging 35, 29, and 27 individuals per population per class of marker,
respectively). The only genetic variability detected for 23 isozymes was a fixed difference between Tennessee and Kentucky
populations at 1 locus. Fixed differences between populations of the 2 states were also observed for 3 chloroplast markers.
Polymorphism at 19 nuclear microsatellites was 74% at the species level and averaged 21% at the population level. However,
observed heterozygosity was extremely low in all populations, ranging from 0.000 to 0.005. High FIS values (0.93) suggest
that Boechera perstellata is a primarily selfing species. Tennessee populations have more genetic diversity than Kentucky populations of B. perstellata. Microsatellite markers revealed substantial genetic divergence between the states and genetic differences
among populations within each state. Analysis of molecular variance indicates that most variability in this species occurs
between the 2 states (49%) and among populations within each state (42%), with relatively little variation found within populations (9%). These data indicate that there is very little gene flow among populations of B. perstellata and that it is important
to protect as many populations as possible in order to conserve the genetic diversity of this rare species.
Key words: allozymes, Arabis perstellata, chloroplast DNA, conservation, endemic, SSRs
Genetic diversity and population genetic structure have
long-term and short-term evolutionary consequences for
any species, and conservation biologists are particularly
concerned about the level and distribution of genetic variability in rare species. In the long term, potential for adaptation and evolutionary change depends on the magnitude
of genetic variation present within a species. In the short
term, loss of genetic variation can cause inbreeding depression and can have harmful effects on development, growth
rates, disease resistance, survival, and fitness (Frankel 1970,
1974; Allendorf and Leary 1986; Lande and Barrowclough
1987; Barrett and Kohn 1991; Huenneke 1991; Reed and
Frankham 2003).
Rare species tend to have limited genetic variability
(Hamrick and Godt 1990; Gitzendanner and Soltis 2000;
Cole 2003). Numerically small populations are prone to
the loss of alleles through genetic drift, and to inbreeding
(Barrett and Kohn 1991; Ellstrand and Elam 1993; Keller
and Waller 2002). Wind pollination may be less effective for
plants in sparse populations, and insect pollinators may be
less likely to visit such populations (Kunin 1997). The ability
to cross with relatives or even to self-fertilize can be beneficial when populations are small because it may sometimes be
the only way a species can persist, but it does result in the loss
of heterozygosity.
Even if there is little variation within a population of a
species, variation among isolated populations may still exist
due to mutation, random genetic drift, and/or divergent natural selection in different habitats resulting in locally adapted
populations. The effect of the extirpation of one population on species-wide genetic variation therefore depends on
the level of differentiation among populations. Information
265
Journal of Heredity
regarding the level and distribution of genetic variation
within a species can provide valuable guidance and increase
the effectiveness of conservation efforts on behalf of a rare
species (Lande and Barrowclough 1987; Huenneke 1991;
Ellstrand and Elam 1993; Frankham 2005).
Boechera perstellata (E. L. Braun) Al-Shehbaz (Braun’s rockcress) is a rare plant from the mustard family (Brassicaceae).
This small herbaceous perennial grows in full or partial
shade on fairly steep wooded slopes without much ground
cover—often around limestone rock outcrops and on animal trails (USFWS 1995). It is federally endangered due
to its relatively small geographic range, limited number of
populations, and typically small population sizes (although a
few populations have more than a thousand plants; USFWS
2004).
B. perstellata is only known from middle Tennessee and
north central Kentucky, with populations from the 2 states
separated by about 250 km (Figure 1). Although the plants
in the 2 different states were once considered to be separate varieties (then known as Arabis perstellata var. ampla in
Tennessee, var. perstellata in Kentucky; Rollins 1960) and
were listed as such when the species was given the endangered status, they are no longer treated as separate varieties (Rollins 1993; USFWS 2004). The taxonomic change
occurred because the leaf size and pubescence differences
originally observed were not as striking for later Tennessee
collections and so were thought to be the result of environmental factors such as different shade and moisture levels
(Rollins 1993).
At the time of this study, the species was known in
Tennessee from 12 sites (5 “populations”) in 2 counties
(Rutherford and Wilson) (USFWS 2010) in the Central Basin
Physiographic Region of the Interior Low Plateau (USFWS
1995) although recently 2 new sites have been discovered
in Smith County (Lincicome D, personal communication).
Boechera perstellata populations in Tennessee (TN) tend to
occur roughly ¾ of the way up large hills (e.g., at elevations
of 275–320 m, on hills 320–360 m in elevation) and so are
potentially isolated from each other. Kentucky (KY) has
more populations, which tend to be in closer proximity and
may be less isolated from each other than those in Tennessee.
In 2010, 42 populations of the species were known from 3
Kentucky counties (Franklin, Henry, and Owen) (USFWS
2010) in the Blue Grass Physiographic Region of the Interior
Low Plateau (USFWS 1995). In both states, most sites occur
on private property, and loss of habitat via development,
grazing, and logging is a major threat to the species. Because
B. perstellata appears to be a poor competitor, several invasive
species (e.g., European garlic mustard, Alliaria petiolata) also
threaten its populations (USFWS 2004).
The genus Boechera is known to include species with various
modes of reproduction and ploidy levels (Dobeš et al. 2006).
Figure 1. Locations of Boechera perstellata sampling sites in Tennessee and Kentucky, United States.
266
Baskauf et al. • Population Genetics of Boechera perstellata
Many species are sexually reproducing, with inbreeding being
common among those species, whereas some species can
reproduce by apomictic reproduction (but not vegetatively).
Many species are diploid, but some species are polyploid (usually triploid), perhaps as the result of cross-species hybridization, and a few are aneuploid. Boechera perstellata is a diploid
species (2n = 14, Rollins 1966). Mating system has not been
studied for this species, and although it is assumed to be insect
pollinated, the pollinators are not known (USFWS 1997).
The small seeds are assumed to be gravity or wind dispersed
because they lack special structures indicating animal dispersal.
The Boechera genus has been the focus of much evolutionary and ecological research in recent years, partly
because of its great diversity and partly because of its
close phylogenetic relationship to another member of the
Brassicaceae family, the model system species Arabidopsis
thaliana. In fact, Boechera is now considered by many to be an
important model system in its own right (Rushworth et al.
2011). The taxonomy of this genus is still in flux, however.
Al-Shehbaz (2003) moved most North American species
from the genus Arabis into the genus Boechera (including
B. perstellata), but Alexander et al. (2013) have recently advocated breaking up the large Boechera genus, a change that
would include moving B. perstellata and its closest relatives
into the genus Borodinia.
In this study, we examined the question of whether the
endangered B. perstellata has low levels of genetic diversity, as is commonly predicted for rare species. In addition,
because of the species’ disjunct geographic distribution in 2
different states, we postulated that the species would show
state-specific genetic differentiation, at the very least. We
expected that there would be within-state population-level
differentiation as well, considering that some populations
appear to be relatively isolated and so may have restricted
gene flow. We examined the genetic variability and population genetic structure of B. perstellata utilizing 3 different
types of genetic markers: the rapidly evolving microsatellites (also known as simple sequence repeats, or SSRs) from
nuclear DNA, the more slowly evolving isozyme genes, and
some noncoding regions of chloroplast DNA (cpDNA).
The comparison of 3 different types of markers not only
allows us to evaluate the consistency of genetic patterns
revealed, but the different markers also provide different
degrees of resolution and different types of information
(e.g., gene flow for maternally inherited cpDNA occurs by
seeds alone vs. gene flow for the other markers also occurs
through pollen, McCauley 1995).
Materials and Methods
Collection of Plant Material
Populations were sampled throughout the range of this species (Figure 1), concentrating on those populations that are
larger and/or geographically more distant from each other
in order to maximize the chances of detecting genetic variability in the species. Populations sampled included 4 populations from Tennessee and 3 populations from Kentucky. The
Table 1 note provides population sizes, and Supplementary
Table S1 provides other summary information. The KY populations represent the eastern, southern, and northern part
of the species range in that state, and although the KY populations are all somewhat small, KY-R is one of the largest
KY populations known (White D, personal communication)
and is similar in size to 2 of the TN populations sampled
(TN-V, TN-G).
Boechera perstellata leaf tissue was collected in 2008 by sampling leaves from individual plants scattered throughout a
population. Isozyme assays were first carried out using fresh
leaf material for 139–274 plants per locus, resulting in an
average sample size of 35 plants per isozyme per population
(Jinks 2009). DNA was extracted from a total of 205 plants
(an average of 29 plants per population). Almost all DNA
samples were used for cpDNA assays (200–205, depending
on primer pair). Table 1 and Supplementary Table S2 give
population and locus-specific microsatellite sample sizes.
Table 1 Genetic variability averaged across 19 microsatellite loci (14 polymorphic and 5 monomorphic) for all sampled Boechera
perstellata populations (standard errors in parentheses)
State-Population
N
A
Ap
P
P95
Ho
He
TN-I
TN-V
TN-G
TN-CK
Mean TN
KY-R
KY-C
KY-H
Mean KY
Mean for species
(all populations)
31.0 (1.4)
29.1 (1.2)
30.1 (1.2)
28.7 (1.5)
29.7
28.5 (1.0)
25.3 (0.7)
13.7 (0.5)
22.5
26.6 (0.6)
1.6 (0.3)
1.4 (0.1)
1.1 (0.1)
1.4 (0.1)
1.4
1.2 (0.1)
1.3 (0.2)
1.1 (0.1)
1.2
1.3 (0.1)
5
6
9
5
6.3
3
5
3
3.7
5.1(0.8)
36.8
31.6
5.3
31.6
26.3
10.5
15.8
10.5
12.3
21.1(4.7)
21.1
21.1
5.3
15.8
15.8
10.5
10.5
10.5
10.5
13.5
0.002 (0.002)
0.003(0.002)
0.000 (0.000)
0.005 (0.004)
0.003
0.005 (0.005)
0.002 (0.002)
0.000 (0.000)
0.002
0.002 (0.001)
0.111 (0.052)
0.079 (0.039)
0.035 (0.035)
0.068 (0.033)
0.073
0.036 (0.029)
0.063 (0.042)
0.021 (0.015)
0.040
0.059 (0.014)
Mean sample size (N), mean number of alleles per locus (A), number of private alleles (Ap), percentage of polymorphic loci (P), percentage of polymorphic loci at 95% criterion (P95), mean direct count heterozygosity (Ho), unbiased estimate (Nei, 1978) mean expected heterozygosity (He). Population sizes
(number of individuals) estimated at time of sampling are as follows: TN-I (2000), TN-V (200), TN-G (200), TN-CK (1300), KY-R (200), KY-C (100),
KY-H (100).
267
Journal of Heredity
Isozyme Extraction, Electrophoresis, and Staining
Starch gel electrophoresis and cellulose acetate gel electrophoresis were utilized to assay for isozyme markers. Methods
generally followed Werth (1985) and Wendel and Weeden
(1989) for staining methods and starch gels and Hebert and
Beaton (1989) for cellulose acetate gels. Leaves were homogenized on ice in a cold room using Werth’s (1985) simple
extraction buffer. Homogenate was either adsorbed onto filter paper wicks to load onto 12% starch gels or stamped onto
cellulose acetate gels using a “Super Z” (Helena Laboratories,
Beaumont, TX).
Genetic data were collected for 14 enzyme systems: acid
phosphatase (ACPH) (3.1.3.2), adenylate kinase (ADK)
(2.7.4.3), aldolase (ALD) (4.1.2.13), aspartate aminotransferase (AAT) (2.6.1.1), glucose-3-phosphate dehydrogenase
(G3PDH) (1.2.1.12), isocitrate dehydrogenase (IDH) (1.1.1.42),
malate dehydrogenase (MDH) (1.1.1.37), malic enzyme (ME)
(1.1.1.40), menadione reductase (MDR) (1.6.99.2), phosphoglucoisomerase (PGI) (5.3.1.9), phosphoglucomutase (PGM)
(5.4.2.2), 6-phopshogluconate dehydrogenase (6PGDH)
(1.1.1.44), shikimate dehydrogenase (SKDH) (1.1.1.25), and
triose-phosphate isomerase (TPI) (5.3.1.1).
Three buffer systems were used to resolve these enzymes
on starch gels: a modified tris borate ethylenediaminetetraacetic acid (EDTA) pH 9.1 (“dilute Salamander B,”
Baskauf et al. 1994) (for ACPH, G3PDH, and PGM); tris
borate EDTA pH 8 (for IDH and PGI), and morpholine
citrate pH 6.1 (for ADK, ALD, MDH, MDR, ME, and
6PGDH). A tris EDTA borate buffer system from Graham
(1994) was used for cellulose acetate gels (for AAT, SKDH,
and TPI).
DNA Extraction and Polymerase Chain Reaction
Protocols
Total genomic DNA was extracted from 100 mg of fresh leaf
tissue using the DNeasy plant mini kit (Qiagen, Valencia, CA)
following grinding with a Retsch (Newtown, PA) MM301
bead mill and was stored at −80 °C.
cpDNA
In order to obtain large sample sizes, polymerase chain reaction (PCR) followed by restriction enzyme digest techniques
was used for analysis of variation in cpDNA, with random
fragment length polymorphisms (RFLPs) visualized on agarose gels. Some sequencing was done to verify interpretation
of the gels (see below).
DNA was amplified from each of 4 different noncoding
regions of the chloroplast genome. In separate reactions, various primer pairs amplified products from 3 different intergenic spacer regions: primers e and f of Taberlet et al. (1991)
for the region between the trnL and the trnF genes; primers
(called “G” and “S” here) of Hamilton (1999) for the region
between the trnG and trnS genes, and primers (called “H”
and “P” here) of Hamilton (1999) for the region between the
trnH and psbA genes. Primers c and d of Taberlet et al. (1991)
268
were used to amplify the fourth region, an intron within the
trnL gene.
PCR for cpDNA was performed in a total volume of
50 µl, consisting of 5 µl of tricine Taq buffer (300 mM tricine,
500 mM KCl, 20 mM MgCl2), 0.2 µl dNTPs (each at 25 mM),
0.3 µM forward primer, 0.3 µM reverse primer, 2.5 units
of Taq DNA polymerase (GoTaq Flexi DNA Polymerase,
Promega, Madison, WI), and 1 µl template DNA (3–48 ng/
µl). PCR conditions were as follows: 3 min at 95 °C; 10 cycles
of 30 s at 94 °C, 30 s at 65 °C (reduced by 1 °C per cycle), and
45 s at 72 °C; followed by 30 cycles of 30 s at 94 °C, 30 s at
55 °C, 45 s at 72 °C; followed by 20 min at 72 °C.
Ten microliters of each PCR product were then digested
separately with 10 units of MseI restriction enzyme at 37 °C
for at least 1 h in a 20 µl reaction volume that included 2 µl of
NEBuffer 4 and 1 µg bovine serum albumin. New England
Biolabs (Ipswich, MA) reagents were used. Following digestion, a 10 µl aliquot of digested DNA from each sample was
electrophoresed on a 4% MetaPhor (Lonza, Walkersville,
MD) agarose gel and then stained with ethidium bromide to
observe the resulting DNA fragments.
Interpretation of the gels was confirmed by sequencing PCR products of 1 individual from each of 2 TN and
2 KY populations. Sanger sequencing was carried out by the
Vanderbilt University DNA Sequencing Facility (Nashville,
TN) using ABI 3730xl/3730 DNA analyzers (Applied
Biosystems, Life Technologies, Carlsbad, CA). Sequences
(Genbank accession KF638563KF638568) were then aligned
manually and trimmed, and MseI restriction sites (TTAA)
were identified using the BioEdit software package (Hall
1999).
Nuclear Microsatellites
In this study, we transferred microsatellites (most often gene
based) that were previously developed for genetic mapping
and population genetic analyses in other Boechera/Arabis species to our study system. A total of 50 primer pairs were
surveyed, with most from Schranz et al. (2007) (with primer
sequences given in supplementary tables of Schranz et al.
2007), but also some from Clauss et al. (2002) and Dobeš
et al. (2004). Information from these sources regarding the
repeated unit size and locations on chromosomes is summarized in Supplementary Table S3.
The protocol for fluorescent labeling of PCR products
generally followed Schuelke (2000), although the 18 bp M13
universal sequence “tag” used in this study (to attach to the
fluorophore and to the 5′ end of the forward primer) was
5′-CAC GAC GTT GTA AAA CGA-3′. The fluorophores
used were 6FAM, VIC, NED, or PET (Applied Biosystems,
Life Technologies).
PCR for microsatellites was performed in a total volume of 15 µl, consisting of 1.5 ml tricine buffer (300 mM
tricine, 500 mM KCl, 20 mM MgCl2), 0.06 µl dNTPs (each at
25 mM), 0.04 µM M-13-tagged forward primer (0.6 pmoles),
0.2 µM reverse primer (3 pmoles), and 0.2 µM M-13 tag with
fluorophore (3 pmoles). PCR conditions were as described
Baskauf et al. • Population Genetics of Boechera perstellata
for the cpDNA PCR. Fluorescently labeled PCR products
were visualized by GENEWIZ (South Plainfield, NJ) using
an Applied Biosystems (Life Technologies) 3730xl Genetic
Analyzer, with the LIZ-500 size standard run in each lane.
Alleles were called manually using GeneMarker v. 1.97 software (SoftGenetics LLC, State College, PA, 2010), with 2
people verifying the allele calls.
GenAlEx ver. 6.41 (Peakall and Smouse 2006) was used
to calculate population genetic statistics for microsatellite data—the percentage of polymorphic loci, number of
alleles per locus, and heterozygosity levels (observed and
expected)—as well as for most other analyses of microsatellite data, unless otherwise noted. Estimates of genetic
similarity were based on Nei’s (1978) unbiased identity and
distance. To investigate the genetic structuring of B. perstellata, analysis of molecular variance (AMOVA) was carried
out with the input option of codominant genotypic distance matrix and using the allelic distance matrix to estimate
F-statistics. Significance in AMOVA was tested for, using
9999 permutations. Hedrick’s (2005) G’ST was also estimated,
using GenAlEx ver. 6.5 (Peakall and Smouse 2006, 2012).
A principal coordinate analysis was performed to transform
genetic distances between all B. perstellata individuals into a
2-dimensional form that explains as much of the observed
variance as possible. Chi-square goodness-of-fit tests for
deviations of genotype frequencies from Hardy–Weinberg
expectations were carried out (with the Levene correction for
small samples) using BIOSYS-1 (Swofford & Selander 1989).
Linkage disequilibrium (LD) coefficients, as measured
by r2, and significance testing (via 1000 permutations) were
calculated for pairwise comparisons between polymorphic
microsatellite markers using TASSEL v. 3 (Bradbury et al.
2007). Given the high degree of population structure among
populations (see Results) and because population structure
can lead to inflated levels of LD among markers and/or
spurious associations between markers (Pritchard et al. 2000;
Buckler and Thornsberry 2002), measures of LD were also
calculated within populations separately for those loci that
were polymorphic within a given population.
Population structure in B. perstellata was investigated
further using a Bayesian, model-based clustering algorithm
as implemented in the software package STRUCTURE
(v. 2.3.3) (Pritchard et al. 2000). For this analysis, individuals were assigned to K population genetic clusters based on
their multilocus genotypes. The proportion of membership
in each cluster was estimated, and this analysis did not rely
on prior population information (i.e., USEPOPINFO was
turned off). For each analysis, K = 1–7 population genetic
clusters were evaluated with 5 runs per K value, and the probability values were averaged across runs for each cluster. For
each run, the initial burn-in period was set to 50 000 with
100 000 MCMC iterations. We employed the DeltaK method
of Evanno et al. (2005) in order to determine the most likely
number of clusters in our dataset. However, it is known that
the DeltaK method often identifies the highest level of structure in the dataset (Coulon et al. 2008); therefore, we also
examined the next most likely value of K from this analysis
(see Results). The online program STRUCTURE Harvester
was used to plot likelihood values and DeltaK (Earl and
­vonHoldt 2012). In fulfillment of data archiving guidelines
(Baker 2013), we have deposited the primary data underlying
these analyses with Dryad.
Results
Isozyme Results
Twenty-three isozymes were resolved from 14 enzyme systems. Despite sampling an average of 35 plants per population, populations within each state showed no variability at
any of the 23 isozyme loci; however, the Tennessee and the
Kentucky populations were fixed for different alleles at the
IDH locus. Thus, only 4% (1/23) of assayed isozyme loci
were polymorphic at the species level, and no heterozygotes
were observed because no loci were polymorphic within any
population. Genetic identities were therefore extremely high:
1.000 for population pairs within the same state and 0.957 for
TN versus KY population pairs.
cpDNA Results
The entire chloroplast genome essentially represents a single locus, from which we sampled 4 regions. Three of the 4
chloroplast PCR/RFLP markers (those using the H/P, e/f,
and c/d primers) were fixed for different banding patterns
in TN and KY populations, whereas the fourth chloroplast
marker (from the G/S primers) was monomorphic across all
sampled populations (Figure 2). Sequencing of representative
PCR products revealed state-specific differences in the number of MseI cut sites for H/P, e/f, and c/d PCR products
(Supplementary Table S4) but no differences in cut sites for
the G/S PCR product (data not shown), consistent with the
fragments seen on gels. Sequencing also revealed some other
variability not apparent from PCR/RFLP gels, both state-specific differences (e.g., a mononucleotide substitution for the
H/P product, a dinucleotide substitution for e/f, and a few
mono or dinucleotide insertion/deletions for c/d) and also
within-state population-specific differences (some mono or
dinucleotide insertion/deletions within both states for c/d).
Nuclear Microsatellite Results
Nineteen microsatellite markers (“loci”) were resolved in a
survey of 50 primer pairs from previously published Boechera/
Arabis studies (Supplementary Table S3). Boechera perstellata
and Boechera stricta have the same number of chromosomes,
and assuming that the chromosome structure for B. perstellata
is also similar to that of B. stricta (Schranz et al. 2007), these
microsatellites represent loci from all chromosomes (x = 7) of
the species. Of these 19 markers, only 5 were monomorphic
across all populations; thus, 74% were polymorphic at the species level. Sample sizes and allele frequencies for the resolved
microsatellite markers are shown in Supplementary Table S2.
TN and KY did not share any alleles at more than half
(8/14) of the polymorphic loci. Furthermore, there were
major genetic differences among the TN populations for
13 of the 14 polymorphic microsatellite loci. In each case,
269
Journal of Heredity
Figure 2. Examples of restriction fragment patterns produced by digestion of PCR product produced by each of 4 primer
pairs (e/f, H/P, c/d, or G/S) with MseI. Samples are plants from TN populations and KY populations.
at least one TN population (often more) did not share
alleles with the other TN populations. Every population of
B. ­perstellata sampled had private alleles at one or more loci
(Table 1, Supplementary Table S5), with the number of private alleles ranging from 3 to 9 for a given population and a
mean of 5.6 private alleles per population (3.4 loci per population having private alleles). For TN populations, the number of private alleles for all loci combined ranged from 5 (for
TN-CK) to 9 (for TN-G), whereas the number of private
alleles in the KY populations ranged from 3 (for KY-R and
KY-H) to 5 (for KY-C).
As indicated by the population-level variability estimates
in Table 1, the mean polymorphism per population (21%)
was considerably lower than the 74% polymorphism estimated at the species level. There was no consistent relationship between population size and microsatellite variability
(see Table 1). Tennessee populations averaged higher genetic
variability than Kentucky populations for all these measures,
including more than twice as many polymorphic loci (despite
low levels for the TN-G population). Many microsatellite
studies exclude species-wide monomorphic loci from diversity estimates, and if only the 14 loci polymorphic at the species level for B. perstellata are included in calculations, then
28% of the loci were polymorphic per population, with a
mean number of alleles per polymorphic locus of 1.4, a mean
observed heterozygosity of 0.003, and a mean expected heterozygosity of 0.080.
Significant heterozygote deficits (chi-square goodnessof-fit tests, P < 0.001) were found at all polymorphic loci
in every population except for the C09 locus in population
TN-V. This lack of heterozygotes is reflected in the very high
FIS value for this species (FIS = 0.933, as estimated from
AMOVA, P < 0.001). Although null alleles at a locus can
270
make it difficult to detect heterozygotes and thus inflate FIS
estimates, it is unlikely that null alleles are responsible for
all 14 polymorphic loci showing this same lack of heterozygotes, and even less so considering that mostly gene-based
microsatellites were assayed, for which null alleles are less
likely to be an issue (Leigh et al. 2003; Liewlaksaneeyanawin
et al. 2004; Rungis et al. 2004).
F-statistics indicate how variability is partitioned at different levels of population structure, and FST indicates the
proportion of the species’ genetic variability that is due to
populations differing from each other genetically. A very
high proportion of B. perstellata microsatellite variability was
manifested as differences among populations, with about
91% of the variability of the species being due to population genetic differentiation (FST = 0.906, as estimated from
AMOVA, P < 0.001; G’ST, Hedrick’s (2005) standardized
analogue for GST and FST, was similar, at a value of 0.931,
P = 0.001). This overall population differentiation can be further broken down to estimate the proportion of the species’
variability that is due to populations from different regions
(states) being genetically different from each other. AMOVA
(P < 0.001 for all estimates) indicated that genetic differences
between the 2 states accounted for 49% of the variability of
the species, whereas differences among populations within
states accounted for 42% of the variability.
Estimates of the pairwise genetic identities varied quite
widely among populations of this species (Table 2). In general, KY populations were relatively similar to each other,
whereas TN populations were much less similar to each
other. Pairwise comparisons between TN and KY populations tended to show the lowest genetic identity values of all
although the TN-G population was fairly different from the
other TN populations.
Baskauf et al. • Population Genetics of Boechera perstellata
The mean pairwise r2 (LD coefficient), considering the
total pooled sample, was 0.26 and ranged from 0 to 1,
with 1 locus pair being in complete LD. However, high
levels of population structure can inflate LD estimates.
Calculation of r2 within populations separately for those
populations that were polymorphic for at least 2 loci
revealed a mean r2 of 0.064 and very few pairs of loci
with r2 values that were significantly different from zero.
Following a Bonferroni correction for multiple testing
(α = 0.05), only 2 comparisons in the TN-I population
remained significant.
For principal coordinates analysis (Figure 3), the first
axis accounted for 59.4% of the variation and clearly separated populations from the 2 states. Individuals from each
TN population tended to cluster together although TN-CK
and TN-V overlapped. There was much more overlap
among individuals from KY populations, however. The
TN-G population was separated from other TN populations along the axis for coordinate 2 (accounting for 23.6%
of the variation).
The STRUCTURE and DeltaK analysis (Figure 4A)
further illustrates this differentiation between the states,
providing support for the presence of 2 genetically distinct
clusters (i.e., K = 2) that entirely separated the TN and KY
individuals. The next most significant number of clusters
was K = 5 (Figure 4B), with each TN population forming its
own cluster, whereas all KY individuals form a single cluster,
emphasizing the relative lack of population genetic structure among populations in KY compared with populations
in TN.
Discussion
Boechera perstellata has very low levels of genetic variability.
In fact, this species has almost no detectable genetic variability at isozyme loci—much less even than many other
rare plant species (Hamrick and Godt 1990, but see Peakall
et al. 2003, and various studies in Cole 2003 for examples
of plant species with no isozyme variability). This is despite
the fact that the species’ unusual disjunct distribution would
seem to provide sufficient isolation to cause much more
genetic divergence between the 2 states’ populations than
the one fixed difference observed, even given the fact
Table 2 Below the diagonal are Nei’s (1978) unbiased genetic identity values, and above the diagonal are Nei’s (1978) unbiased genetic
distance values for each Boechera perstellata population pair
State-Population
TN-I
TN-V
TN-G
TN-CK
KY-R
KY-C
KY-H
TN-I
TN-V
TN-G
TN-CK
KY-R
KY-C
KY-H
—
0.695
0.445
0.666
0.508
0.480
0.508
0.364
—
0.517
0.680
0.384
0.359
0.384
0.811
0.660
—
0.538
0.276
0.315
0.271
0.407
0.386
0.619
—
0.357
0.328
0.358
0.677
0.957
1.286
1.029
—
0.903
0.968
0.733
1.024
1.156
1.115
0.103
—
0.891
0.678
0.956
1.307
1.027
0.033
0.116
—
Figure 3. Principal coordinates analysis using a genetic distance matrix without data standardization for individuals from
Boechera perstellata populations (population names given in the legend) based on microsatellite data. The first coordinate explains
59.36% of the variation, and the second axis explains 23.61% of the variation (cumulative 83%).
271
Journal of Heredity
Figure 4. STRUCTURE results for the 7 Boechera perstellata populations (a) for K = 2 clusters and (b) for K = 5 clusters.
that there is no within-population variability. McKay et al.
(2001) also found low isozyme diversity for Boechera fecunda,
a rare congener found in 2 different regions of Montana
separated by about 100 km (Song and Mitchell-Olds 2007).
Polymorphism was detected for only 1 of 14 allozymes
(7.1%), with very low levels of heterozygosity present at
this 1 locus.
Microsatellite markers proved to be more variable and
to provide much better resolution of population structure
for B. perstellata. Overall, 74% of the sampled microsatellite
loci were polymorphic for this species compared with 4% of
isozyme loci. Even for microsatellites, however, the population averages for polymorphic loci, allelic richness, and
heterozygosity were all low compared with many other rare
plant species that have been studied (although Peakall et al.,
2003, detected no microsatellite variability for 2 species in the
ancient conifer family Araucariaceae). In a review of microsatellite studies (using only polymorphic loci), Nybom (2004)
reported mean expected heterozygosities for endemics (7
272
studies), short-lived perennials (29 studies), and selfing species (15 studies) that were each more than 5 times higher than
the mean expected heterozygosity estimate for B. perstellata.
Furthermore, the mean observed heterozygosity reported for
endemics and short-lived perennials was more than 2 orders
of magnitude greater than that for B. perstellata, and even
the selfers’ mean observed heterozygosity was more than an
order of magnitude greater. Despite this comparatively low
variability within populations for B. perstellata, microsatellites
were able to reveal genetic differences among regions (states)
and even among populations within a state for B. perstellata.
All 3 classes of genetic markers and several forms of data
analysis revealed clear genetic differences between TN and
KY plants for this species. In addition to the fixed difference
between the states for 1 isozyme locus, TN and KY shared
no alleles at 57% of the polymorphic microsatellite loci, and
almost half of the species’ microsatellite diversity was due to
genetic differences between the 2 states. Further evidence of
this regional distinctiveness was provided by the chloroplast
Baskauf et al. • Population Genetics of Boechera perstellata
genome, with fixed differences at the 3 polymorphic regions.
The genetic data, therefore, have potential taxonomic implications in that they support the original separation of the
species into 2 varieties, despite the fact that morphological
differences have been discounted as inconsistent and probably environmentally influenced (Rollins 1993; USFWS 2004).
For populations within a given state, neither isozymes nor
chloroplast PCR/RFLP markers detected any genetic variability. At microsatellites, however, substantial differences in
allele frequencies were evident among TN populations for
93% (13/14) of the polymorphic loci, and each of the TN
populations had 5 or more “private” alleles. KY populations,
on the other hand, were much less differentiated, with only
21% (3/14) of the loci showing much among-population
variability and with fewer private alleles. Genetic identity values and STRUCTURE analysis (Figure 4) confirmed that KY
populations were fairly similar to each other and much less
similar to any TN populations. TN populations were clearly
more heterogeneous than the KY populations. The TN-G
population had the lowest genetic identity in comparison
with other TN populations and is also the most geographically distant from the other TN populations (32 km northeast
of the closest population, TN-I).
AMOVA estimated that for B. perstellata, only 9% of all
the genetic variability was due to differences of individuals within populations. This means that an extremely high
proportion of the species’ genetic variability (91%) was due
to genetic differences among populations, revealing marked
population structure in B. perstellata and indicating that there is
very little gene flow among populations. The presence of private alleles within each population (Table 1, Supplementary
Table S5)—some at very high frequencies, or even fixed—
also indicates a lack of gene flow among populations. FST for
B. perstellata was several times higher than mean FST values
reported by Nybom (2004) for endemics, short-lived perennials, and selfers (3.5, 2.9, and 2.2 times higher, respectively).
There is dispute (Whitlock 2011) as to whether or not G’ST
is a better measure of population differentiation than FST
for genetic markers with high mutation rates such as many
microsatellites, but in the case of B. perstellata, FST and G’ST
estimates were similar. This may be due in part to the fact
that microsatellites used in this study are mostly gene based
and thus may have lower mutation rates.
Two congeners of B. perstelllata that also have been assayed
using microsatellites are B. fecunda (the previously mentioned
rare species; Song and Mitchell-Olds 2007) and B. stricta (a
widespread species; Song et al. 2006). Estimates of genetic
variability were higher for both these congeners than for
B. perstellata, and FIS and FST estimates were lower (although
still substantial); nonetheless, the overall features of relatively low genetic diversity and high FIS and FST values found
in this study for B. perstellata appear to be shared by these
2 congeners. Comparing estimates based only on the loci
polymorphic at the species level, population-level means for
B. perstelllata, B. fecunda (Song and Mitchell-Olds 2007), and
B. stricta (Song et al. 2006), respectively, were 27.6%, 41.5%,
and 63.5% for polymorphic loci; 1.4, 1.9, and 2.2 for number
of alleles per polymorphic locus; 0.003, 0.051, and 0.029 for
observed heterozygosity; 0.08, 0.20, and 0.26 for expected
heterozygosity; 0.93, 0.82, and 0.89 for FIS, and 0.91, 0.57,
and 0.56 for FST. Sexually reproducing Boechera are often selfcompatible (Dobeš et al. 2006), and the very high FIS estimates for all 3 Boechera congeners would be consistent with
all 3 being predominantly selfing species. In fact, substantial
self-fertilization has been shown to occur in B. fecunda populations (Hamilton and Mitchell-Olds 1994); however, B. perstellata has not been studied in terms of mating system. The
ability to self could be beneficial for species in small or sparse
populations or for pollinator-limited populations. Clearly,
there is also a high degree of population genetic structure at
microsatellite loci for each of these 3 species, with large FST
values indicating extreme population differentiation.
Conclusion
All 3 types of markers indicated genetic differences between
populations of the 2 states—data that lend support to the
original varietal status for TN and KY plants and suggest
that future phylogenetic work involving this species should
include samples from both states. Microsatellites, however,
have proved to be particularly useful genetic markers for
B. perstellata, as only microsatellites were able to detect variability for populations within each state and thus to show
that populations in TN had more genetic variability at these
markers than did populations in KY. Microsatellite data also
provided evidence that very little outcrossing was occurring,
suggesting that B. perstellata may be primarily a selfing species, although this merits further investigation as the mating
system has not been studied directly for this species. The
fact that we were able to successfully use a number of previously developed gene-based microsatellites from related
Boechera/Arabis species to study the B. perstellata illustrates the
value of such transferable SSRs when studying rare relatives
of better studied species (Ellis and Burke 2007).
It is important to protect as many B. perstellata populations as possible, and to ensure that populations in both TN
and KY are protected, because most of the genetic variability
of this species exists as differences among populations and
between states. This is a major challenge, as most populations
occur on privately owned lands. Protecting as many populations as possible in both states would have the additional benefit of lowering the risk of extinction from potential threats
such as diseases, pests, fire, localized drought, etc., that could
decimate a population without affecting more distant populations. Although neutral markers might not reflect patterns
of variability at ecologically important genes, the marked
population structure revealed by microsatellites for B. perstellata would nevertheless caution against mixing plants or seeds
from different states (or even different populations from the
same state), if management strategies include augmenting
established populations. Likewise, if new populations are
established from a mixture of naturally occurring populations, then it may be advisable to check several generations
for outbreeding depression (Edmands 2007). In the future, it
would be useful to carry out within-state and between-state
273
Journal of Heredity
experimental crosses among greenhouse-grown B. perstellata plants and estimate the fitness of offspring from these
various crosses in order to evaluate the risk of outbreeding
depression.
Cole CT, 2003. Genetic variation in rare and common plants. Ann Rev Ecol
Syst. 34:213–237.
Supplementary Material
Dobeš C, Koch M, Sharbel TF. 2006. Karyology, and modes of reproduction in the North American genus Boechera (Brassicaceae): a compilation of
seven decades of research. Ann Mo Bot Gar. 93:517–534.
Supplementary material can be found at http://www.jhered.
oxfordjournals.org/.
Funding
This work was supported by two Tennessee Department of
Environment and Conservation contracts to CJB, ED-0825730-00 and EG-11-34928-00, which were funded using a
portion of US Fish and Wildlife Service Traditional Section
6 grants (TN-E-4-22 and TN-E-4-25 (F11AP00790).
Acknowledgments
We thank Tennessee and Kentucky State Natural Heritage Program staff who
kindly took C.J.B. and N.C.J. to Boechera perstellata sites and sometimes helped
in collecting leaves: A. Bishop, R. McCoy, and S. Mathes from Tennessee
Department of Environment and Conservation, and D. White and
T. Littlefield from the Kentucky State Nature Preserves Commission (who
also collected all the leaves from the Henry County, KY population). Thanks
also to S. Baskauf, who helped create the map figure. A Faculty Professional
Development Leave awarded to C.J.B. by Austin Peay State University provided time for the microsatellite work. We appreciate the suggestions of six
anonymous reviewers whose suggestions improved the manuscript.
Coulon A, Fitzpatrick W, Bowman R, Stith BM, Makarewich CA, Stenzler
LM, Lovette IJ. 2008. Congruent population structure inferred from dispersal behavior and intensive genetic surveys of the threatened Florida scrubjay (Aphelocoma coerulescens). Mol Ecol. 17:1685–1701.
Dobeš C, Mitchell-Olds T, Koch MA. 2004. Intraspecific diversification in
North American Boechera stricta (= Arabis drummondii), Boechera x divaricarpa,
and Boechera holboellii (Brassicaceae) inferred from nuclear and chloroplast
molecular markers – an integrative approach. Am J Bot. 91:2087–2101.
Earl DA, vonHoldt BM. 2012. STRUCTURE HARVESTER: a website
and program for visualizing STRUCTURE output and implementing the
Evanno method. Conserv Genet Resour. 4:359–366.
Edmands S. 2007. Between a rock and a hard place: evaluating the relative
risks of inbreeding and outbreeding for conservation and management. Mol
Ecol. 16:463–475.
Ellis JR, Burke JM. 2007. EST-SSRs as a resource for population genetic
analyses. Heredity. 99:125–132.
Ellstrand NC, Elam DR. 1993. Population genetic consequences of small
population size: implications for plant conservation. Ann Rev Ecol Syst.
24:217–242.
Evanno G, Regnaut S, Goudet J. 2005. Detecting the number of clusters of
individuals using the software STRUCTURE: a simulation study. Mol Ecol.
14:2611–2620.
Frankel OH. 1970. Variation, the essence of life. Sir William Macleay
Memorial Lecture. Proc Linn Soc. 95:158–169.
Frankel OH. 1974. Genetic conservation: our evolutionary responsibility.
Genetics. 78:53–65.
Frankham R. 2005. Genetics and extinction. Biol Conserv. 126:131–140.
References
Gitzendanner MA, Soltis PS. 2000. Patterns of genetic variation in rare and
widespread plant congeners. Am J Bot. 87:783–792.
Alexander PJ, Windham MD, Beck JB, Al-Shehbaz IA, Allphin L, Bailey
CD. 2013. Molecular phylogenetics and taxonomy of the genus Boechera and
related genera (Brassicaceae: Boechereae). Syst Bot. 38:192–209.
Graham M. 1994. Genetic population structure of a glade endemic, the
Missouri bladder-pod (Lesquerella filliformis); Brassicaceae [master’s thesis].
[Kirksville (MO)]: Northeast Missouri State University. p. 61.
Allendorf FW, Leary RF. 1986. Heterozygosity and fitness in natural populations of animals. In: Soulé ME, editor. Conservation biology. Sunderland
(MA): Sinauer. p. 57–76.
Hall TA. 1999. BioEdit: a user-friendly biological sequence alignment editor and analysis program for Windows 95/98/NT. Nucleic Acids Symp Ser.
41:95–98.
Al-Shehbaz IA. 2003. Transfer of most North American species of Arabis
to Boechera (Brassicaceae). Novon. 13:381–391.
Hamilton MB. 1999. Four primer pairs for the amplification of chloroplast
intergenic regions with intraspecific variation. Mol Ecol. 8:521–523.
Baker CS. 2013. Journal of Heredity adopts joint data archiving policy.
J Hered 104:1. doi: 10.1093/jhered/ess137
Hamilton MB, Mitchell-Olds T. 1994. The mating system and relative performance of selfed and outcrossed progeny in Arabis fecunda (Brassicaceae).
Am J Bot. 81:1252–1256.
Barrett SCH, Kohn JR. 1991. Genetic and evolutionary consequences of
small population size in plants: implication for conservation. In: Falk DA,
Holsinger KE, editors. Genetics and conservation of rare plants. Oxford:
Oxford University Press. p. 3–30.
Hamrick JL, Godt MJW. 1990. Allozyme diversity in plant species. In: Brown
AHD, Clegg MT, Kahler AL, Weir BS, editors. Plant population genetics,
breeding, and genetic resources. Sunderland (MA): Sinauer. p. 43–63.
Baskauf CJ, McCauley DE, Eickmeier WG. 1994. Genetic analysis of a rare
and a widespread species of Echinacea (Asteraceae). Evolution. 48:180–188.
Hebert PDN, Beaton MJ. 1989. Methodologies for allozyme analysis using
cellulose acetate electrophoresis. Beaumont (TX): Helena Laboratories.
Bradbury PJ, Zhang Z, Kroon DE, Casstevens TM, Ramdoss Y, Buckler
ES. 2007. TASSEL: software for association mapping of complex traits in
diverse samples. Bioinformatics. 23:2633–2635.
Hedrick PW. 2005. A standardized genetic differentiation measure.
Evolution. 59:1633–1638.
Buckler ES, Thornsberry J. 2002. Plant molecular diversity and applications
to genomics. Curr Opin Plant Biol. 5:107–111.
Huenneke LF. 1991. Ecological implications of genetic variation in plant
populations. In: Falk DA, Holsinger KE, editors. Genetics and conservation
of rare plants. New York: Oxford University Press. p. 31–44.
Clauss MJ, Cobban H, Mitchell-Olds T. 2002. Cross-species microsatellite
markers for elucidating population genetic structure in Arabidopsis and Arabis
(Brassicaceae). Mol Ecol. 11:591–601.
Jinks N. 2009. Genetic variability within and among populations of the rare
plant species, Arabis perstellata E. L. Braun (Brassicaceae) [master’s thesis].
[Clarksville (TN)]: Austin Peay State University.
274
Baskauf et al. • Population Genetics of Boechera perstellata
Keller LF, Waller DM. 2002. Inbreeding effects in wild populations. Trends
Ecol Evol. 17:230–241.
Kunin WE. 1997. Population biology and rarity: on the complexity of
density-dependence in insect-plant interactions. In: Kunin WE, Gaston KJ,
editors. The biology of rarity: causes and consequences of rare-common
differences. London (UK): Chapman and Hall. p. 150–173.
Lande R, Barrowclough GF. 1987. Effective population size, genetic variation, and their use in population management. In: Soulé ME, editor. Viable
populations for conservation. Cambridge (UK): Cambridge University
Press. p. 87–124.
Leigh F, Lea V, Law J, Wolters P, Powell W, Donini P. 2003. Assessment
of EST- and genomic microsatellite markers for variety discrimination and
genetic diversity studies in wheat. Euphytica. 133:359–366.
Liewlaksaneeyanawin C, Ritland CE, El-Kassaby YA, Ritland K. 2004.
Single-copy, species-transferable microsatellite markers developed from
loblolly pine ESTs. Theor Appl Genet. 109:361–369.
McCauley DE. 1995. The use of chloroplast DNA polymorphism in studies
of gene flow in plants. Trend Ecol Evol. 10:198–202.
McKay JK, Bishop JG, Lin JZ, Richards JH, Sala A, Mitchell-Olds T.
2001. Local adaptation across a climatic gradient despite small effective
population size in the rare sapphire rockcress. Proc R Soc Lond [Biol].
268:1715–1721.
Nei M. 1978. Estimation of average heterozygosity and genetic distance
from a small number of individuals. Genetics. 89:583–590.
Nybom H. 2004. Comparison of different nuclear DNA markers for estimating genetic diversity in plants. Mol Ecol. 13:1143–1155.
Peakall R, Ebert E, Scott LJ, Meagher PF, Offord CA. 2003. Comparative
genetic study confirms exceptionally low genetic variation in the ancient
and endangered relictual conifer, Wollemia nobilis (Araucariaceae). Mol Ecol.
12:2331–2343.
Peakall R, Smouse PE. 2006. GENALEX 6: genetic analysis in Excel.
Population genetic software for teaching and research. Mol Ecol Notes.
6:288–295.
Peakall R, Smouse PE. 2012. GenAlEx 6.5: genetic analysis in Excel.
Population genetic software for teaching and research—an update.
Bioinformatics. 28:2537–2539.
Rungis D, Berube Y, Zhang J, Ralph S, Ritland CE, Ellis BE, Douglas
C, Bohlmann J, Ritland K. 2004. Robust simple sequence repeat markers for spruce (Picea spp.) from expressed sequence tags. Theor Appl
Genet.109:1283–94.
Rushworth CA, Song B-H, Lee C-R, Mitchell-Olds T. 2011. Boechera, a model
system for ecological genomics. Mol Ecol. 20:4843–4857.
Schranz ME, Windsor AJ, Song B, Lawton-Rauh A, Mitchell-Olds T. 2007.
Comparative genetic mapping in Boechera stricta, a close relative of Arabidopsis.
Plant Phys. 144:286–298.
Schuelke M. 2000. An economic method for the fluorescent labeling of PCR
fragments. Nature Biotechnol. 18:233–234.
Song B-H, Clauss MJ, Pepper A, Mitchell-Olds T. 2006. Geographic patterns
of microsatellite variation in Boechera stricta, a close relative of Arabidopsis.
Mol Ecol. 15:357–369.
Song B-H, Mitchell-Olds T. 2007. High genetic diversity and population
differentiation in Boechera fecunda, a rare relative of Arabidopsis. Mol Ecol.
16:4079–4088.
Swofford DL, Selander RK. 1989. BIOSYS-1, a computer program for the
analysis of allelic variation in genetics. Release 1.7. User’s manual. Urbana
(IL): University of Illinois.
Taberlet P, Gielly L, Pautou G, Bouvet J. 1991. Universal primers for amplification of three non-coding regions of chloroplast DNA. Plant Mol Biol.
17:1105–1109.
US Fish and Wildlife Service (USFWS). 1995. Endangered and threatened
wildlife and plants; determination of endangered status for Arabis perstellata.
Fed Reg. 60:56–61.
US Fish and Wildlife Service (USFWS). 1997. Recovery plan for Arabis
­perstellata Braun (Braun’s Rockcress). Atlanta (GA): USFWS. p. 21.
US Fish and Wildlife Service (USFWS). 2004. Endangered and threatened
wildlife and plants; designation of critical habitat for Arabis perstellata. Fed
Reg. 69:31459–31496.
US Fish and Wildlife Service (USFWS). 2010. Braun’s Rockcress (Arabis
­perstellata) Five year review: summary and evaluation. Frankfort (KY): US
Fish and Wildlife Service Southeast Region, Kentucky Ecological Services
Field Office. May 28, 2010.
Pritchard JK, Stephens M, Donnelly P. 2000. Inference of population structure using multilocus genotype data. Genetics. 155:945–959.
Wendel JF, Weeden NF. 1989. Visualization and interpretation of plant
isozymes. In: Soltis DE, Soltis PS, editors. Plant biology. Portland (OR):
Dioscorides Press. p. 6–45.
Reed DH, Frankham R. 2003. Correlation between fitness and genetic diversity. Conserv Biol. 17:230–237.
Werth CR. 1985. Implementing isozyme laboratory at a field station. VA J
Sci. 36:53–73.
Rollins RC. 1960. Arabis perstellata in Tennessee. Rhodora. 62:242–244.
Whitlock MC. 2011. G’ST and D do not replace FST. Mol Ecol. 20:
1083–1091.
Rollins RC. 1966. Chromosome numbers of Cruciferae. Contributions from
the Gray Herbarium of Harvard University. 197:43–65.
Rollins RC. 1993. The Cruciferae of continental North America: Systematics
of the mustard family from the Arctic to Panama. Stanford (CA): Stanford
University Press.
Received December 14, 2012; First decision February 22, 2013;
Accepted September 16, 2013
Corresponding Editor: David B. Wagner
275