| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
2 Department of Botany, University of Georgia, Athens, Georgia 30602 USA; and Savannah River Ecology Laboratory, Drawer E, Aiken, South Carolina 29802 USA
Received for publication February 9, 1999. Accepted for publication September 30, 1999.
ABSTRACT
We investigated genetic structure in two closely related perennial plants that occur in isolated wetlands: Sagittaria isoetiformis, restricted to the southeastern Coastal Plain of North America, and S. teres, endemic to the northeastern Coastal Plain. Using horizontal starch-gel electrophoresis, we screened 527 individuals from 11 populations of S. isoetiformis and 367 individuals from seven populations of S. teres. A high proportion of the 16 loci were polymorphic (%PS = 93.8% in S. isoetiformis and %PS = 75.0% in S. teres), with higher mean numbers of alleles per polymorphic locus and effective alleles per locus in S. isoetiformis (AP = 3.27, AE = 1.90) than in S. teres (AP = 2.58, AE = 1.30). Species- and population-level expected heterozygosities were higher in S. isoetiformis (HES = 0.399, HEP = 0.218) than in S. teres (HES = 0.177, HEP = 0.101). Jackknife estimates of F statistics indicated moderate levels of inbreeding in S. teres
(
IS =
23.1%). Strong differentiation characterized these geographically
isolated populations (GST = 39.9% in
S. isoetiformis, and GST = 26.1%
in S. teres). Genetic identities varied substantially within
(
= 75%, range = 0.558–0.963 in
S. isoetiformis; = 89%, range
= 0.776–0.963 in S. teres) and among species
( = 81%, range = 0.506–0.882),
leading to the discrimination of four regional population clusters using
nonmetric multidimensional scaling (NMDS). It appears that S.
isoetiformis and S. teres are a progenitor-derivative species
pair.
Key Words: Alismataceae • allozymes • habitat isolation • progenitor-derivative • Sagittaria isoetiformis • Sagittaria teres.
Most plants exist as collections
of fairly distinct populations in habitat patches that are discontinuous in
space and unstable over time (Rice and Jain, 1985
). The degree to which populations are
isolated depends on the magnitude of dispersal relative to life-history
traits, since these traits determine the strength of demographic and
environmental stochasticity and the magnitude of inbreeding depression
(Karieva, 1990
)
and genetic drift (Wright, 1940
; Nei, 1973
). Plants with narrow habitat specificity
and limited dispersal potential are at particular risk for global
extinction as landscapes become mosaics of "habitat islands"
due to anthropogenic activities. Naturally occurring habitat islands
associated with, for example, particular substrates or disturbance regimes
provide the opportunity to explore the effects of long-term isolation on
the evolution and viability of populations.
Isolated wetlands are
islands in a terrestrial landscape to which some species are restricted and
uniquely adapted. The patchy nature of isolated wetlands may influence gene
flow in wetland or aquatic plants, especially if their dispersal mechanisms
are relatively restricted (Barrett, Eckert, and Husband, 1993
). Population genetic
surveys of plant species in isolated habitats can reveal the levels and
distribution of genetic variation within and between populations as well as
the extent to which populations are connected by dispersal (Glover and
Barrett, 1987
;
Affre, Thompson, and Debussche, 1997
). At the species level, aquatic and
wetland angiosperms are generally more polymorphic than terrestrial plants,
despite the fact that most aquatic plants exhibit some form of vegetative
reproduction (Barrett, Eckert, and Husband, 1993
), which may reduce population-level
heterozygosity. At the population level, many aquatic taxa have fewer
polymorphic loci, fewer alleles per locus, and lower levels of variation
than nonaquatics (Barrett, Eckert, and Husband, 1993
; Laushman, 1993
). Thus, high
species-level diversity in aquatic species could be due to high levels of
differentiation among isolated populations. Compared to widespread and
abundant species, however, rare plant species generally have lower levels
of genetic diversity (Hamrick and Godt, 1989
). Consequently, rare taxa that are also
restricted to isolated wetlands might be expected to have high levels of
differentiation among populations but relatively low overall levels of
diversity, unless diversity within populations has been
maintained.
Plants often become rare due to the selective alteration
or destruction of particular habitats. This has been the case for many
plants associated with depression wetlands in the United States, including
Carolina bays, limesinks, and glacial ponds. These wetlands harbor a large
proportion of the endangered and threatened plants in their respective
landscapes (Knox and Sharitz, 1990
; Sutter and Kral, 1994
; Kirkman, 1995
; L. Duthie, Norcross
Wildlife Sanctuary, Massachusetts, personal communication). Carolina bays
are freshwater depression wetlands found in the southeastern Atlantic
Coastal Plain, most abundantly in North and South Carolina (Lide,
1997
; Sharitz and
Gresham, 1998
).
In South Carolina, >90% of the Carolina bays over 0.8 ha in size
have been altered or completely destroyed (Bennett and Nelson,
1991
) primarily
due to agricultural and silvicultural practices; comparable losses are
likely in Alabama, Georgia, and North Carolina. Limesinks are distributed
in the Atlantic and Gulf Coastal Plain from Alabama and Georgia to southern
Florida and have been similarly impacted (Wharton, 1978
; Kirkman, 1995
). In the Northeast,
glacial ponds near coastal areas have been variously affected by
recreation, aquaculture, construction, and storm run-off waters from
developed areas (Massachusetts Endangered and Threatened Species Program
Element Occurrence records for 1998). The alteration and loss of depression
wetlands have contributed to the rarity of species by reducing population
numbers and geographical ranges, and further isolating the remaining
populations.
By exploring the levels and distribution of genetic
variation in closely related rare species that share similar habitat
restrictions, we can investigate whether species of similar phylogenies and
habitats have comparable patterns of genetic variation and, consequently,
provide insight into their management needs (Hamrick and Godt,
1996
; McDonald
and Hamrick, 1996
). Surveys of genetic diversity among
closely related species also can reveal phylogenetic relationships (Ranker
and Schnabel, 1986
; Edwards and Wyatt, 1994
; McClintock and
Waterway, 1994
; Purdy, Bayer, and MacDonald,
1994
). If the
close relatives represent a recent progenitor-derivative species pair, they
are expected to have high genetic identities; the progenitor species should
have higher levels of genetic diversity than the derivative, and the
derivative should contain a subset of the alleles present in the
progenitor, with few, if any, unique alleles (Gottlieb, 1973
; Crawford,
1983
).
We focused on two closely related
plant species that are restricted to isolated habitats and have narrow
geographical ranges. Sagittaria teres S. Watson was first
described in 1890 (Watson and Coulter, 1890), and S. isoetiformis
was described by J. G. Smith in a note (1895b) that accompanied his
revision of the North American Sagittaria (1895a). In subsequent
treatments of the genus, it was thought either that: (1) both taxa were
merely varieties of a third species, S. graminea A. Michaux
(Bogin, 1955
), or
(2) S. isoetiformis simply represented a southern disjunction in
the geographical range of S. teres (Beal, 1960
). Godfrey and Adams
(1964)
argued that all three are morphologically
distinguishable species. Wooten (1973b)
conducted cross-compatibility experiments
among seven taxa in the grass-leaved section of Sagittaria and
found that, although S. isoetiformis was cross-compatible with
S. graminea var. graminea and S. graminea var.
weatherbiana Bogin, viable seeds were not produced from any
interspecific crosses involving S. teres. Only six reciprocal
crosses were performed on a single individual of each species, however, so
these results may have been due to the small sample size. Nevertheless,
Wooten (1973b)
suggested that S. teres is a post-glacial derivative species of
S. graminea var. graminea, with which its geographical
range overlaps. The ranges of S. graminea var. graminea
and S. isoetiformis also overlap, and Smith (1895b)
, based on a few specimens
from Florida, felt that S. isoetiformis resembled S.
graminea var. graminea more than it resembled any other
species. Recently, Haynes and Hellquist (1999)
supported distinguishing
S. isoetiformis and S. teres as species, taxonomically
distinct from the three subspecies (previously varieties) of S.
graminea.
We conducted a population genetic survey of S. isoetiformis and S. teres to address the following questions: (1) What are the levels and distribution of genetic variation in these two rare plants? (2) How does population isolation influence gene flow in these taxa? (3) What is the relationship between S. isoetiformis and S. teres? To examine these questions, we used horizontal starch-gel electrophoresis to quantify the levels and partitioning of genetic variation and relatedness among taxa. We conclude by suggesting guidelines for the preservation of these rare taxa.
MATERIALS AND METHODS
Species
biology
Sagittaria isoetiformis and S. teres are
herbaceous perennials, 5–50 cm tall, that reproduce sexually and
asexually. The plants are monoecious, producing simple racemes arranged in
1–5 whorls of 1–3 unisexual flowers per whorl. Female flowers
typically occur only in the lowermost whorl, whereas male flowers can occur
in all whorls. Floral development is basipetal, with the lower flowers
opening first. Both taxa are self-compatible (Wooten, 1973b
; Edwards, 1999
). Autogamy
(self-fertilization within an inflorescence) is generally avoided because
male and female flowers are rarely open on the same day, but geitonogamy
among clones (ramets) of the same genetic individual (genet) may be common.
In field populations, each female flower of S. isoetiformis
produces up to 200 fruits (achenes); fruits appear to be photosynthetic and
mature ~20–21 d after pollination (Edwards, unpublished data). We
were unable to observe fruit set in the field for S. teres. Both
species produce new ramets from thin stolons 1–30 cm long; stolon
connections between clones do not persist (A. L. Edwards, personal
observation). As with many aquatic and wetland plants, these plants develop
storage tissues (corms). When present, corms of S. isoetiformis
are small (up to 29 mm long and 8 mm wide) and can persist for 1–3 wk
during growing season droughts in dry-to-moist soil (A. L. Edwards,
unpublished data).
Both taxa have narrow geographical ranges
(Fig. 1) and are restricted to
acidic depression wetlands with sandy substrates. Sagittaria
isoetiformis occurs in a subset of herbaceous Carolina bays and
limesinks in the Coastal Plain from southeastern Alabama through the
Carolinas, and south to central Florida. The suite of bays and limesinks
that support S. isoetiformis are high-light environments, with
soils that are low in organic matter and limited in nutrients, particularly
phosphorus and nitrogen (Schalles and Shure, 1989
; Brenner, Binford, and
Deevey, 1990
;
Miller, 1998
).
Water levels in Carolina bays depend primarily on precipitation and
evapotranspiration and generally have little connection to groundwater
(Lide et al., 1995
). The hydrology of limesinks also depends
on precipitation and evapotranspiration, but the karst topography in which
they occur, often overlain with sandy sediments, permits groundwater
influence (Wharton, 1978
; Brenner, Binford, and Deevey,
1990
; Sutter
and Kral, 1994
).
Sagittaria teres is restricted to glacial ponds in Massachusetts,
Rhode Island, New Jersey, and New York that have in common the properties
of Carolina bays and limesinks described above. They differ in that they
are much deeper and receive more groundwater
inflow.
|
). Leaves were ground in chilled buffer
with sterile sand, and the extract was adsorbed through miracloth onto 3MM
chromatography paper wicks. Wicks were stored in microtest plates at
-70°C; wicks were transferred directly from the freezer to
11-12% starch gels for electrophoresis.
|
with some modifications by J. L. Hamrick (personal communication) for the
majority of the enzyme stains. We also used protocols by Cheliak and Pitel
(1984)
for DIA, by Broyles and Wyatt
(1990)
for GDH, and by Wendel and Weeden
(1989)
for G3PD[NADP], with slight
modifications in substrate proportions to optimize
resolution.
|
. Fixation indices
(FIS) at each locus within each population (Li and
Horvitz, 1953
) and
overall heterogeneity in allele frequencies at each locus (Workman and
Niswander, 1970
) were tested using chi-square analyses.
These statistics were averaged across all polymorphic loci to obtain
species-level estimates as described by Edwards and Wyatt
(1994)
.
In addition, Wright's
(1922)
F statistics
(FIS, FST, FIT)
were calculated following Weir and Cockerham (f, 
, and
F, respectively; 1984; Weir, 1996
) to measure deviations from
Hardy-Weinberg equilibrium at each polymorphic locus. These fixation
indices measure levels of inbreeding within individuals in subpopulations
(FIS; subpopulations will be referred to as populations
in the biological sense), inbreeding due to population subdivision
(FST; an indicator of the degree of differentiation
among biological populations), and overall levels of inbreeding
(FIT; equivalent to regional- or species-level
inbreeding). Positive F statistics indicate heterozygote
deficiencies. We used GENEPOP (Version 3.1b; Raymond and Rousset,
1995
) to
calculate the F statistics. We used FSTAT (Goudet et al.,
1996
; Goudet,
1998
) to
compute unbiased jackknifed estimates, standard errors, and P
values for each locus based on 500 permutations of the data over alleles
within loci (IS), and over genotypes within
loci (ST,
IT). Mean
IS,
ST, and
IT values were
considered significant departures from Hardy-Weinberg equilibrium if
95% bootstrap confidence intervals did not overlap zero. Pairwise
estimates of FST and Nm between populations
also were calculated using FSTAT.
In species for which dispersal may
be geographically restricted, populations separated by greater distances
are expected to be more distinct genetically. We tested for this
"isolation by distance" (Wright, 1943, 1946
) in both species using
two different methods that employ pairwise estimates of
FST (Slatkin, 1993
; Rousset, 1997
). We compared matrices of pairwise log
geographical distances to Slatkin's
log(), and to Rousset's
ST/(1 -
ST) using 1000
permutations of the data and Mantel's t statistic (NTSYS-PC;
Rohlf, 1994
).
Nei's
(1973)
genetic identities and distances were calculated according to Hillis
(1984)
on pairwise comparisons of all populations by LYNSPROG. We conducted a
nonmetric multidimensional scaling (NMDS) analysis of genetic distances
among populations (PROC MDS; SAS [1997
], Version 6.2) to detect regional
patterns of geographical variation. This ordination technique was chosen
over the more commonly used clustering (e.g., UPGMA) or linear ordination
techniques (e.g., principal components analysis) because it can uncover,
but does not assume, hierarchical relationships among populations, and it
does not require the assumption of linearity among variables (Lessa,
1990
).
RESULTS
Levels
and distribution of genetic variation
At the species level, S.
isoetiformis and S. teres were highly polymorphic. All loci
except Acp-2 were polymorphic in S. isoetiformis,
although Aco was polymorphic for only one population (FLSL), and
6Pgd was fixed in four of the 11 populations (FLCL, FLDP, SCPW,
and SCSB). In S. teres, four loci (Aco, Acp-2, 6Pgd, and
Tpi-1) were monomorphic; Mnr was monomorphic in all but one
population (MALL), Pgi-2 was monomorphic in all but two
populations (MAFM and MALL), and Tpi-2 were fixed in all but two
populations (MALS and RILS). The mean number of alleles per polymorphic
locus and the effective number of alleles per locus were higher in S.
isoetiformis (APS = 3.27 and
AES = 1.90) than in S. teres
(APS = 2.58 and AES =
1.30; Table 3). The lower effective
numbers of alleles indicate the presence of rare alleles in both
taxa.
|
|
Within populations, average observed heterozygosities were lower than expected in S. isoetiformis (HOP = 0.209, HEP = 0.218) and in S. teres (HOP = 0.079, HEP = 0.101; Table 3). Estimates of observed and expected heterozygosities overlapped in Alabama, Florida, Georgia, and South Carolina populations, but were significantly lower in the NCJM population. Although all populations sampled were previously confirmed to be either S. isoetiformis or S. teres, the population at the northern geographical limit of S. isoetiformis (NCJM) was fixed for alleles at Aat and Pgm that were not found in any other population of either species. We therefore repeated analyses, with NCJM excluded from the populations representing S. isoetiformis. Overall, species-level estimates of heterozygosity did not change (HTS = 0.421 and HES = 0.394), but population-level estimates increased (HSP = 0.218 and HEP = 0.227). The percentage of polymorphic loci in S. isoetiformis populations averaged 67.6% (68.2% without NCJM) and ranged from 62.5 to 75.0%, with the exception of the NCJM population (37.5%). The average for the species was 93.8%, with or without NCJM. In S. teres populations, 31.0-62.5% of the loci were polymorphic, averaging 48.2% at the population level and 75.0% at the species level. Population-level estimates for the mean number of alleles per polymorphic locus and effective number of alleles per locus were higher in S. isoetiformis (APP = 2.33 and AEP = 1.40; APP = 2.33 and AEP = 1.41 without NCJM) than in S. teres (APP = 2.19 and AEP = 1.17; Table 3).
Significant allele frequency heterogeneity was detected at most loci using chi-square goodness-of-fit tests. All 16 polymorphic loci in S. isoetiformis had significantly heterogeneous allele frequencies across populations. Sixteen loci were polymorphic in the Alabama and Florida populations, but only 14 loci were polymorphic in the Georgia and South Carolina populations. Of those 14 polymorphic loci, all but 6Pgd exhibited allele frequency heterogeneity, and 6Pgd was either monomorphic or nearly so (second allele at frequencies <2%). In S. teres, ten of the 12 polymorphic loci exhibited significant allele frequency heterogeneity, and the two loci that did not were monomorphic in most populations or nearly so (Mnr and Tpi-1; second allele at frequencies <1%).
None of the 249 fixation
indices for the polymorphic loci within populations was significantly
negative. Roughly one-fourth of all fixation indices for polymorphic loci
were significantly greater than zero in S. isoetiformis
(22.4%) and in S. teres (23.8%), indicating an
excess of homozygotes relative to expected Hardy-Weinberg equilibrium (data
not shown). Jackknife
IS values were not
significantly different than zero in S. isoetiformis (4.8 ±
4.6%, with or without NCJM; Table
4). Conversely, significant levels of inbreeding within
populations were detected in jackknife estimates for S. teres
(23.1 ± 10.5%). Estimates of inbreeding overlapped zero among
peripheral and central populations of S. isoetiformis, with the
Alabama/Florida population cluster averaging near zero inbreeding
(IS = 0.7
± 5.9%) and the Georgia/South Carolina population cluster
averaging an inbreeding coefficient an order of magnitude higher
(IS = 8.7
± 4.8%). No regional jackknife estimates could be obtained
for the northern periphery of the geographic distribution, since there was
only a single northern population sampled
(IS = 2.5
± 24.0% for NCJM). Averaged over all polymorphic loci,
estimates of inbreeding calculated arithmetically (FIS)
were all within the 95% bootstrap confidence intervals (data not
shown).
|
ST; Table 4) were equivalent in the two species
(ST = 0.473
± 0.067 and ST
= 0.473 ± 0.092 for S. isoetiformis and S.
teres, respectively). The estimate for S. isoetiformis was
lower when NCJM was excluded
(ST = 0.446
± 0.067). The values for the Alabama/Florida cluster were somewhat
lower than the average for S. isoetiformis
(ST = 0.381
± 0.063). Population differentiation within the Georgia/South
Carolina cluster of S. isoetiformis, however, was significantly
lower (ST =
0.191 ± 0.042) than the species average. Overall,
IT estimates mirrored
the ST values (data not
shown).
Jackknife estimates of population differentiation
(ST) were higher than
those calculated arithmetically (GST; Table 4). Averaged across populations within
species, three of the 16 polymorphic loci in S. isoetiformis
(Aco, Gdh, and 6Pgd) and five of the 12 polymorphic loci
in S. teres (Gdh, Idh, Mnr, Pgi2, and Tpi1) had
a single allele with frequencies >95%, and a sixth locus in
S. teres (Pgi1) was dominated by a single allele at
frequencies of >90%. As a consequence of the low variability in
these loci, ST values
were inflated, particularly for S. teres. Calculated as the
proportion of the total genetic variation found among populations, mean
differentiation among populations was much higher in S.
isoetiformis (GST = 38.8%;
GST = 36.7% without NCJM) than in S.
teres (GST = 26.1%). Populations in
the Alabama/Florida cluster (GST = 28.0%)
were more distinct from one another than populations in Georgia/South
Carolina (GST =
13.7%).
Relationship among taxa
There were
50 alleles in S. isoetiformis and 35 in S. teres totaled
across all 16 loci (Edwards, 1999
). Of those alleles, 16 were unique to
S. isoetiformis, whereas only one was unique to S. teres.
All but five of the unique S. isoetiformis alleles occurred at
frequencies >5%. The single unique allele in S. teres
occurred in one individual in a single population (MALL). Summed over
populations, 11 alleles occurred at frequencies <5% in each
species. There were no fixed differences between the two taxa. The same
allele was the most common in both species for ten of the 16 loci. For the
remaining six loci, the most common allele in one species was always
present in some populations of the other species, with the exception of
idh-a. This allele occurred in seven of the 11 S.
isoetiformis populations at frequencies of 0.02–1.00 but was not
found in any S. teres populations.
There were several
populations of both species in which the most common allele was rare or
absent in other populations. The most dramatic shifts in allele frequencies
contributing to the differentiation of populations and population clusters
occurred at the Aat, Acp1, Dia, Idh, Pgi-1, Pgi-2, Pgm, and
Tpi-2 loci (Edwards, 1999
). The Alabama/Florida populations were
most strongly differentiated from the Georgia/South Carolina populations by
Dia, Idh, and Tpi-2, and differentiated from NCJM by
Aat, Dia, Idh, and Pgm. The NCJM population was also
distinguished from the Georgia/South Carolina populations by unique alleles
at Aat and Pgm. The S. teres populations were
most divergent from the Alabama/Florida populations at Idh, Pgi-1,
Pgm, and Tpi-2, and from NCJM at Aat, Pgi-1, Pgi-2,
and Pgm. There were no consistent differences at particular loci
between S. teres and the Georgia/South Carolina
populations.
The mean genetic identity () between
taxa was 81% (with or without NCJM), ranging from 0.506 to 0.882.
Genetic identities among populations within species ranged from 0.558 to
0.963 ( = 75%) in S. isoetiformis
and 0.776 to 0.963 ( = 89%) in S.
teres (Edwards, 1999
). Average identities among the population
clusters were lowest between Alabama/Florida vs. NCJM (
= 63%), Georgia/South Carolina ( =
66%), and S. teres ( = 57%).
Average identities were higher between NCJM vs. Georgia/South Carolina
( = 69%), and S. teres
( = 74%). Average identities were highest
between the Georgia/South Carolina vs. S. teres (
= 81%).
Isolation by distance was suggested for some
groups. Using Slatkin's (1993)
method, the log geographic distances
among S. isoetiformis populations were negatively correlated with
the number of migrants per generation (r = -0.85,
P = 0.001), even when NCJM was excluded (r
= -0.86, P = 0.001). This strong correlation
was due to the hierarchical arrangement of populations. The relationship
was less apparent at smaller geographic scales, when the same test was
performed on the Alabama/Florida (r = -0.88,
P = 0.083) and Georgia/Florida (r =
-0.77, P = 0.069) population clusters. Populations of
S. teres had similarly weak correlations (r =
-0.75, P = 0.020). Using Rousset's
(1997)
method, we found a positive correlation
between log geographic distance and population differentiation in S.
isoetiformis (r = 0.22, P = 0.004 with
NCJM; r = 0.22, P = 0.065 without NCJM),
and in the Alabama/Florida (r = 0.96, P =
0.039) and Georgia/Florida (r = 0.89, P =
0.009) populations clusters. Evidence for isolation by distance in S.
teres (r = 0.26, P = 0.087) was not
strongly supported.
DISCUSSION
Levels
and distribution of genetic variation
Compared to the average for
other narrowly distributed plant species (HES =
0.137 ± 0.011; Hamrick and Godt, 1989
), S. isoetiformis has very high
levels of diversity and S. teres has average levels of diversity.
Indeed, the expected heterozygosity for the former (HES
= 0.399) is twice the average for widespread plants
(HES = 0.202 ± 0.015), which generally
have higher levels of diversity than species with more restricted
geographical distributions (Karron, 1987
; Hamrick and Godt, 1989
). Species-level expected
heterozygosity for S. teres (HES =
0.177) is higher than the average reported for endemic species
(HES = 0.096 ± 0.010; Hamrick and Godt,
1989
). These
high estimates are due to the high proportion of polymorphic loci evident
in both species, as well as higher means for effective number of alleles
(%PS = 93.8%,
AES = 1.90 in S. isoetiformis, and
%PS = 75.0%,
AES = 1.30 in S. teres) relative to the
averages reported for widespread (%PS =
58.9% ± 3.1, AES = 1.31 ±
0.03) or endemic species (%PS =
40.0% ± 3.2, AES = 1.15 ±
0.04; Hamrick and Godt, 1989
). A little over half of the total
variation present in both species was distributed within populations
(HS = 0.233 and 0.134 in S.
isoetiformis and S. teres, respectively). Although species
with restricted geographical ranges typically have lower levels of genetic
diversity, present geographical range does not necessarily predict
diversity levels for specific cases, especially if phylogenetic history has
a strong influence (Lewis and Crawford, 1995
; Hamrick and Godt, 1996
).
The low-to-moderate FIS levels in these species probably indicate some selfing and/or biparental inbreeding caused by limited seed and pollen dispersal, combined with vegetative reproduction. There was a trend towards inbreeding within populations and population clusters in S. isoetiformis, albeit not significant. Significant inbreeding was indicated in S. teres. Both species are self-compatible, and insect pollinators tend to visit closely adjacent flowers (A. L. Edwards, unpublished data); thus, geitonogamy among ramets of the same genet probably contributes to inbreeding. Local population substructuring could also be a factor, influenced by differences in flowering phenology along hydrologic gradients. In addition, seeds are gravity dispersed, and generally sink below the water surface to the soil around the parent plant within 30 min (A. L. Edwards, personal observation). Asexual reproduction is common in both species, and although we sampled individuals that were at least 0.5 m apart (stolon lengths were much shorter), large clones within populations could be especially important in S. teres.
Population differentiation
These
sagittarias are characterized by strong divergence among populations,
whether calculated according to Nei (1972)
or using jackknife estimates. Although
not yet commonly used in the population genetic literature, the obvious
advantages of using resampling techniques to estimate F statistics
are that the resulting estimates should be less biased and supply a test
for significant departures without making assumptions about population
distributions. However, such estimates can be inflated when a large
proportion of the polymorphic markers used are not sufficiently variable.
This was particularly true for S. teres, for which only six of the
12 polymorphic loci were very variable, resulting in a twofold difference
in estimates of differentiation
(ST =
47.3%, GST = 26.1%). The estimates
for S. isoetiformis were much closer
(ST =
47.3%, GST = 38.8% with NCJM;
ST =
44.6%, GST = 36.7% without NCJM).
Within geographical regions of S. isoetiformis, the populations in
the periphery of the range were more highly differentiated than populations
in the central portion of the range
(ST =
38.1%, GST = 28.0% in
Alabama/Florida, but ST = 19.1%, GST = 13.7% in Georgia/South Carolina). Nonetheless, the levels observed in both species were between those reported for predominately selfing species (GST = 51.0%) and plants with mixed-mating systems (GST = 21.6%; Hamrick and Godt, 1989
); the GST estimates calculated by geographic region are more indicative of mixed mating.
Geographically isolated populations with limited means of dispersal are typically more highly differentiated (Hamrick and Godt, 1996
). Colonization history also could contribute to regional differences. For example, Glover and Barrett (1987)
found lower levels of genetic variation, but higher population differentiation, in more recently colonized Jamaican populations of Eichhornia paniculata than in continental Brazilian populations. Affre, Thompson, and Debussche (1997)
found that terrestrial island populations of Cyclamen balearicum were more highly differentiated than those on true islands, despite more recent isolation. In both cases, historical factors played a prominent role: the more recently colonized islands probably represent multiple founder events from different population sources (Wade and McCauley, 1988
; Whitlock and McCauley, 1990
).
The high levels of differentiation in S. isoetiformis and S. teres are likely driven by limited gene flow and genetic drift as a consequence of the isolated distribution of specialized habitats and perhaps recent substantial habitat loss. As "island" populations, a relationship between geographical distance and gene flow is suggested both within regions and overall in S. isoetiformis. There is little doubt that S. isoetiformis populations were more numerous and less isolated prior to the widespread destruction of isolated wetlands in the southeastern United States. In contrast, S. teres has probably had a very restricted geographical range at least since the last glacial maximum, facilitating dispersal among populations. Several populations of S. teres, especially in the periphery of its range, are considered extinct or consist of very few individuals (Massachusetts, New Jersey, and New York heritage programs). Indeed, 40–50 populations occur in just two counties in Massachusetts, with <20 populations in the remaining range (Fig. 1). Additional sampling of populations in the extreme southern end of the range of S. teres could reveal stronger evidence for isolation by distance in this species.
Relationship among taxa
The relationship between S. isoetiformis and S. teres appears to fit the criteria for recent progenitor-derivative species pairs (Gottlieb, 1973
). Sagittaria teres has lower overall levels of diversity, possesses a subset of the alleles present in S. isoetiformis, and has one unique, low-frequency allele. Further support for this hypothesis comes from the fact that the ponds in which S. teres occurs were created during the last glacial retreat, in a landscape that was almost entirely glaciated 14 000–10 l000 yr BP (Oldale, 1992
). This hypothesis is also consistent with the few studies that have investigated whether colonists of regions glaciated during the Pleistocene have lower levels of diversity compared to close congeners in nonglaciated regions (Loveless and Hamrick, 1984
; Pleasants and Wendel, 1989
; Lewis and Crawford, 1995
). Given the genetic evidence presented here, the most likely scenario appears to be multiple long-distance founder events in a very specific set of appropriate habitats. The proposed progenitor-derivative relationship is complicated, however, by the high degree of divergence among populations within species. The genetic identity between these two taxa was 81%, which is low; genetic identities between recent progenitor-derivative species typically are >88% (Gottlieb, 1981
; Crawford, 1983
; cf. Loveless and Hamrick, 1984
). Even within species, however, three-fourths of S. isoetiformis (78.2% with NCJM and 73.3% without NCJM) and one-third of S. teres (33.0%) identities were <88%. Consequently, these guidelines may be too conservative for species with extraordinary levels of population differentiation. The fact that S. teres contains strictly a subset of the alleles present in S. isoetiformis is the strongest evidence for a progenitor-derivative relationship, but we cannot rule out Wooten's (1970)
hypothesis that S. graminea ssp. graminea could be a progenitor of either species. There are striking habitat differences that discriminate S. graminea subspecies from S. isoetiformis and S. teres (Wooten, 1970, 1973a
; C. B. Hellquist, Massachusetts College of Liberal Arts–North Adams, and R. R. Haynes, University of Alabama–Tuscaloosa, personal communication). However, we were unable to produce seed from over 30 hand-pollinations between the two taxa (unpublished data), confirming Wooten's (1973b)
findings that S. teres is not cross-compatible. The loss of alleles within widespread species after migration into previously glaciated areas also has been documented (Schwaegerle and Schaal, 1979
; Cwynar and MacDonald, 1987
; Dolan, 1994
; Broyles, 1998
). Could it be that the taxa surveyed here represent a single hyperdispersed species?
We support the assignment of these taxa to species rank and suggest on the basis of the additional evidence presented below that S. isoetiformis is most likely the progenitor species of S. teres. The S. teres population cluster is more closely related to the Georgia/South Carolina cluster of S. isoetiformis than the Georgia/South Carolina cluster is to the other S. isoetiformis clusters. However, as Godfrey and Adams (1964)
pointed out, the two species are morphologically distinguishable in the field and geographically disjunct by >1500 km. Wooten (1973) also noted that "stringent habitat specificity, demonstrated in some members of the S. graminea group, probably exist in all taxa [within this group]." Sagittaria isoetiformis and S. teres are found in acidic, CO2-limited wetlands with sandy substrates, whereas all of the S. graminea subspecies are generally found in environments with heavy organic substrates. In addition, S. isoetiformis and S. teres share the ability to shift from C3 photosynthesis when emergent to CAM photosynthesis when submerged; S. graminea ssp. graminea does not accumulate enough malic acid when submerged to suggest that this photosynthetic shift occurs (J. E. Keeley and A. L. Edwards, unpublished data). In aquatic and wetland plants, the ability to shift between C3 and CAM photosynthesis only has been observed in certain aquatic habitats, characterized by marked CO2 fluctuations (Keeley, 1990
). Presumably, diurnal fluctuations in CO2 levels are much less extreme in the more highly organic wetland soils that S. graminea ssp. graminea inhabits. If there is a genetic basis to substrate specificity, as Wooten (1973a)
suggested and as is likely to be the case for C3/CAM shifting, these ecological and physiological characteristics suggest that S. isoetiformis and S. teres are more closely related to each other than either is to S. graminea ssp. graminea. Complicating the taxonomic relationship between these taxa, however, are the high levels of divergence between population clusters of S. isoetiformis. Clearly, a phylogenetic analysis of all of the grass-leaved Sagittaria taxa (sect. Gramineae) is required to clarify these relationships.
If our hypothesis of a progenitor-derivative relationship is correct, there remain two unresolved problems: (1) where the refugial populations of S. isoetiformis persisted during the last glacial epoch, and (2) how and when S. teres came to occupy northern sites. Estimates for the age of Carolina bays range from 250 000 to 10 000 yr BP (Sharitz and Gresham, 1998
), but it is unlikely that S. isoetiformis populations could have persisted in Carolina bays during or since even the last glacial maximum. These habitats have not been consistent hydrologically over the past 10 000 yr due to several-hundred-year cycles of drier and wetter weather (Stahle, Cleavland, and Hehr, 1988
; Brooks, Taylor, and Grant, 1996
; Gaiser, 1997
). Given the narrow habitat specificity of these species, we suggest that there were multiple isolated refugia in sand-based acidic wetlands, such as those found in the Florida "high ridge" during the last glacial maximum, where seeds could easily disperse as sea levels rose and conditions became very wet (Webb, 1990
). Other refugia may have included habitats exposed during the last glacial period that are now submerged under the Atlantic Ocean. Multiple isolated rufugia combined with extremely long-lived seed banks could explain the high levels of genetic variation observed within populations, but additional paleoecological evidence would help to substantiate this speculation. Rare long-distance dispersal of seeds ingested by migrating ducks (DeVlaming and Proctor, 1968
) or other highly vagile animals feeding in muck may be the most likely dispersal mechanism for S. teres from southern S. isoetiformis and for the migration of seeds among populations within species.
We were surprised to discover a population (NCJM) that was monomorphic at two loci for alleles not found in any other populations of either S. isoetiformis or S. teres. There are several possible explanations for this phenomenon: (1) we did not sample enough populations to detect the alleles elsewhere; (2) the NCJM population has been repeatedly misidentified as S. isoetiformis, but it actually represents a known species in the Gramineae section of the genus; (3) the NCJM population represents a cryptic species that has not been morphologically distinguished from S. isoetiformis, or (4) the NCJM population is S. isoetiformis, but contains remnant genetic input from past introgression with another species. It is unlikely that none of the other 17 populations surveyed contained the unique fixed NCJM alleles, however it is not impossible (there was almost a unique difference between the Alabama/Florida and Georgia/South Carolina regions for Tpi-2). The possibility that NCJM has been consistently misidentified is feasible, but several authorities independently examined and identified this population as S. isoetiformis. The possibility that this population represents a cryptic species is appealing, but we consider this with caution. The NCJM population was unusual in that it was growing in a millpond created over 160 yr ago, with water-level fluctuations attenuated by stream inputs, and resulting in some peat accumulations. Unlike all of the other habitats we visited, the NCJM population grew very close to another Sagittaria species, S. englemanniana (J. G. Small); this habitat may differ from typical S. isoetiformis and S. teres habitats in other important ways. One of the consequences of the variable hydrologies of most S. isoetiformis wetlands is that those habitats with shallower water levels and with shorter hydroperiods do not support this species in dry years. For example, on the 28 000-ha Savannah River Site (SRS) near Aiken, South Carolina, in over 400 isolated wetlands, 11 populations have been documented, but fewer than half of those are extant in an average year, and all disappear from the standing vegetation during prolonged droughts. The sand-based, herbaceous bays that support S. isoetiformis in NC tend to be shallower and have shorter hydroperiods than bays on the SRS. Other populations have been documented in the area of NCJM, but they were not extant during our sampling periods. Nonetheless, the possibility that NCJM represents an unrecognized taxon cannot be rejected and warrants additional genetic surveys in the area. Comparative surveys of all the Sagittaria species that occur in the eastern United States may also uncover evidence of past introgression.
Conservation implications
For the most part, these taxa occur in habitats that are fixed in space, isolated, and fluctuate in suitability over time. Fluctuations in population sizes and the frequent extinction and recolonization of habitats would be expected to decrease effective population sizes severely, causing genetic bottlenecks and the subsequent loss of genetic variation (Hartl and Clark, 1989
), particularly in populations experiencing limited gene flow. Even very large populations (>100 000 individuals) of S. isoetiformis can disappear from the extant vegetation due to stochastic regional weather patterns in a single year (A. L. Edwards, unpublished data). Although we were unable to observe S. teres plants under drought conditions, we have observed that the corms of S. isoetiformis do not persist through prolonged drought. We can only assume, since levels of genetic variation are fairly high for the endemic S. teres and extremely high in S. isoetiformis, that populations are buffered to some degree against bottlenecks and the associated loss of diversity by persistent seed banks. The importance of the seed bank in maintaining genetic variation in temporally variable habitats has been demonstrated both theoretically and empirically (Templeton and Levin, 1979
; Brown and Venable, 1986
; Levin, 1990
; Ellner and Hairston, 1994
; McCue and Holtsford, 1998
).
Recurrent extinction–recolonization cycles and genetic drift drive the large allele frequency differences among these Sagittaria populations. The Carolina bays and limesinks of S. isoetiformis are habitat patches that vary cyclically in suitability—they are metapopulations recolonized primarily by persistent seed banks. The glacial ponds of S. teres do not dry out completely, because they are deeper and are fed by groundwater sources. As a consequence, the plants often occur along much steeper hydrologic gradients. Sagittaria teres plants are therefore more likely to be extant in any given year, but marked water-level fluctuations cause mortality at the gradient extremes. In very high-water years S. teres plants do not flower, so the two species share the environmentally driven trait of not producing seed in years of extreme drought (S. isoetiformis) or extreme high water conditions (S. teres). Both could be categorized as rare species with narrow habitat specificity and sometimes large populations (Rabinowitz, 1981
). Nevertheless, whereas S. isoetiformis was probably once much more abundant and widespread, S. teres may have always been restricted to the narrow geographical range where coastal plain glacial ponds were formed.
Relatively high levels of genetic diversity and population differentiation, limited dispersal ability, and persistent seed banks characterize the plants we studied. Neither species is federally listed, but both are regarded as threatened, endangered, or of special concern in portions of their geographical ranges. The regional status of the more restricted S. teres has been unaffected by its confusion with S. isoetiformis. Nonetheless, protection and management of habitats should be undertaken in each geographical region we identified. Furthermore, sites with recently documented but not necessarily extant populations should be considered for protection. This would encompass preserving the more hydrologically ephemeral habitats to include genetic diversity not captured in this survey. Preservation of hydrologically diverse S. isoetiformis and S. teres habitats would have the additional benefit of preserving co-occurring rare species in these local "hot spots" of plant biodiversity and rarity (Dobson et al., 1997
).
FOOTNOTES
1 The authors thank Philip Dixon, Lisa Donovan, Jim Hamrick, Barbara Taylor, Robert Wyatt, and three anonymous reviewers for helpful suggestions on this manuscript; Jim Hamrick, Mary Jo Godt, and J Vaun McArthur for generously permitting us to use their electrophoresis facilities; and the ever-helpful folks working for state agencies and The Nature Conservancy who monitor rare species. This research was funded in part by a Financial Assistance Award (number DE-FCO9-96SR18546) between the US Department of Energy and the University of Georgia, the Massachusetts Natural Heritage and Endangered Species Program, a Savannah River Ecology Laboratory Graduate Fellowship, The Society of Sigma Xi, The Society of Wetland Scientists, and the University of Georgia Botany Department Small Grants Program. 
3 Current address: Southeastern Environmental Research Center, OE 148, Florida International University, Miami, Florida 33199 USA.
4 Author for correspondence: Savannah River Ecology Laboratory, Drawer E, Aiken, South Carolina 29802 USA.
LITERATURE CITED
Affre, L., J. D. Thompson, and M. Debussche. 1997 Genetic structure of continental and island populations of the Mediterranean endemic Cyclamen balearicum (Primulaceae). American Journal of Botany 84: 437–451.[Abstract]
Barrett, S. C. H., C. G. Eckert, and B. C. Husband. 1993 Evolutionary processes in aquatic plant populations. Aquatic Botany 44: 105–145.[CrossRef]
Beal, E. O. 1960 The Alismataceae of the Carolinas. Journal of the Elisha Mitchell Scientific Society 76: 68–79.
Bennett, S. H., and J. B. Nelson. 1991 Distribution and status of Carolina bays in South Carolina. Nongame and Heritage Trust Publication Number 1. South Carolina Wildlife and Marine Resources Department.
Bogin, C. 1955 Revision of the genus Sagittaria (Alismataceae). Memoirs of the New York Botanical Garden 9: 179–233.
Brenner, M., M. W. Binford, and E. S. Deevey. 1990 Lakes. In R. L. Myers and J. J. Ewel [eds.]. Ecosystems of Florida, 364–390 University of Central Florida Press, Orlando, Florida.
Brooks, M. J., B. E. Taylor, and J. A. Grant. 1996 Carolina bay geoarcheology and Holocene landscape evolution on the Upper Coastal Plain of South Carolina. Geoarcheology 11: 481–504.[CrossRef]
Brown, J. S., and D. L. Venable. 1986 Evolutionary ecology of seedbank annuals in temporally varying environments. American Naturalist 127: 31–47.[CrossRef][ISI]
Broyles, S. B. 1998 Post-glacial migration and the loss of allozyme variation in northern populations of Asclepias exaltata (Asclepiadaceae). American Journal of Botany 85: 1091–1097.[Abstract]
———, and R. Wyatt. 1990 Paternity analysis in a natural population of Asclepias exaltata: multiple paternity, functional gender, and the "pollen donation hypothesis." Evolution 44: 1454–1468.[CrossRef][ISI]
Cheliak, W. M., and J. A. Pitel. 1984 Techniques for starch gel electrophoresis of enzymes from forest tree species. Petawawa National Forestry Institute, Information Report P1-X-42. Canadian Forestry Service, Agriculture, Chalk River, Ontario, Canada.
Crawford, D. J. 1983 Phylogenetic and systematic inferences from electrophoretic studies. In S. O. Tanksley and T. J. Orton [eds.]. Isozymes in plant genetics and breeding, 257–287. Elsevier, Amsterdam.
Cwynar, L. C., and G. M. MacDonald. 1987 Geographical variation of lodgepole pine in relation to population history. American Naturalist 129: 463–469.[CrossRef][ISI]
DeVlaming, V., and V. W. Proctor. 1968 Dispersal of aquatic organisms: viability of seeds recovered from the droppings of captive killdeer and mallard ducks. American Journal of Botany 55: 20–26.[CrossRef][ISI]
Dobson, A. P., J. P. Rodriguez, W. M. Roberts, and D. S. Wilcove. 1997 Geographic distribution of endangered species in the United States. Science 275: 550–553.
Dolan, R. W. 1994 Patterns of isozyme variation in relation to population size, isolation, and phytogeographic history in royal catchfly (Silene regia: Caryophyllaceae). American Journal of Botany 81: 965–972.[CrossRef][ISI]
Edwards, A. L. 1999 Population genetics of the rare Sagittaria isoetiformis and S. teres (Alismataceae), and the pollination ecology of the progenitor species, S. isoetiformis. Ph.D. dissertation, Department of Botany, University of Georgia, Athens, Georgia, USA.
———, and R. Wyatt. 1994 Population genetics of the rare Asclepias texana and its widespread sister species A. perennis. Systematic Botany 19: 291–307.
Ellner, S., and N. G. Hairston, Jr. 1994 Role of overlapping generations in maintaining genetic variation in a fluctuating environment. American Naturalist 143: 403–417.[CrossRef][ISI]
Gaiser, E. 1997 Paleolimnological reconstruction of Holocene environments in wetland ponds of the Upper Atlantic Coastal Plain using siliceous microfossils. Ph.D. dissertation, University of Georgia, Athens, Georgia, USA.
Glover, D. E., and S. C. H. Barrett. 1987 Genetic variation in continental and island populations of Eichhornia paniculata (Pontederiaceae). Heredity 59: 7–17.[ISI]
Godfrey, R. K., and P. Adams. 1964 The identity of Sagittaria isoetiformis (Alismataceae). Sida 1: 269–273.
Gottlieb, L. D. 1973 Genetic differentiation, sympatric speciation, and the origin of a diploid species of Stephanomeria. American Journal of Botany 60: 545–553.[CrossRef]
———. 1981 Electrophoretic evidence and plant populations. Progress in Phytochemistry 7: 1–45.
Goudet, J. 1998 FSTAT, (Version 1.2): a program for IBM PC compatibles to calculate Weir and Cockerham's (1984) estimators of F-statistics.
———, M. Raymond, T. de Meeüs, and F. Rousset. 1996 Testing differentiation in diploid populations. Genetics 144: 1933–1940.[Abstract]
Hamrick, J. L., and M. J. W. Godt. 1989 Allozyme diversity in plant species. In A. H. D. Brown, M. T. Clegg, A. L. Kahler, and B. S. Weir [eds.]. Plant population genetics, breeding and genetic resources, 43–63. Sinauer, Sunderland, Massachusetts, USA.
———, and ———. 1996 Conservation genetics of endemic plant species. In J. C. Avise and J. L. Hamrick [eds.], Conservation genetics: case histories from nature, 281–304. Chapman and Hall, New York, New York, USA.
Hartl, D. L., and A. G. Clark. 1989 Principles of population genetics, second edition. Sinauer, Sunderland, Massachusetts, USA.
Haynes, R. R., and C. B. Hellquist. 1999 Alismataceae É. P. Ventenat–Arrowhead family. In N. R. Morin [convening ed.], Flora North America, vol. 22, Butomaceae-Haemadoraceae. Oxford University Press, Oxford, UK.
Hillis, D. M. 1984 Misuse and modification of Nei's genetic distance. Systematic Zoology 33: 238–240.
Kareiva, P. 1990 Population dynamics in spatially complex environments: theory and data. Philosophical Transcripts of the Research Society of America 330: 175–190.
Karron, J. D. 1987 A comparison of levels of genetic polymorphism and self-compatibility in geographically restricted and widespread plant congeners. Evolutionary Ecology 1: 47–58.
Keeley, J. 1990 Photosynthetic pathways in freshwater aquatic plants. Trends in Ecology and Evolution 5: 330–333.[CrossRef]
Kirkman, L. K. 1995 Impacts of fire and hydrological regimes on vegetation in depression wetlands of Southeastern USA. In S. I. Cerulean and R. T. Engstrom [eds.], Fire in wetlands: a management perspective. Proceedings of the Tall Timbers Fire Ecology Conference, Number 19, 10–20. Tall Timbers Research Station, Tallahassee, Florida, USA.
Knox, J. N., and R. R. Sharitz. 1990 Endangered, threatened, and rare vascular flora of the Savannah River Site. Department of Energy's Savannah River Site National Environmental Research Park Publication Number 20. Savannah River Ecology Laboratory, Aiken, South Carolina, USA.
Laushman, R. H. 1993 Population genetics of hydrophilous angiosperms. Aquatic Botany 44: 147–158.[CrossRef]
Lessa, E. P. 1990 Multidimensional analysis of geographic genetic structure. Systematic Zoology 39: 242–252.[CrossRef][ISI]
Levin, D. A. 1990 The seed bank as a source of genetic novelty in plants. American Naturalist 135: 563–572.[CrossRef][ISI]
Lewis, P. O., and D. J. Crawford. 1995 Pleistocene refugium endemics exhibit greater allozymic diversity than widespread congeners in the genus Polygonella (Polygonaceae). American Journal of Botany 82: 141–149.
Li, C. C., and D. G. Horovitz. 1953 Some methods of estimating the inbreeding coefficient. American Journal of Human Genetics 5: 107–117.[ISI][Medline]
Lide, R. F. 1997 When is a depression wetland a Carolina bay? Southeastern Geographer 37: 90–98.
———, V. G. Meentemeyer, J. E. Pinder, and L. M. Beatty. 1995 Hydrology of a Carolina bay located on the upper coastal plain of western South Carolina. Wetlands 15: 47–57.[ISI]
Loveless, M. D., and J. L. Hamrick. 1984 Genetic organization and evolutionary history in two North American species of Cirsium. Evolution 42: 254–265.
McClintock, K. A., and M. J. Waterway. 1994 Genetic differentiation between Carex lasiocarpa and C. pellita (Cyperaceae) in North America. American Journal of Botany 81: 224–231.[CrossRef][ISI]
McCue, K. A., and T. P. Holtsford. 1998 Seed bank influences on genetic diversity in the rare annual Clarkia springvillensis (Onagraceae). American Journal of Botany 85: 30–36.[Abstract]
McDonald, D. B., and J. L. Hamrick. 1996 Genetic variation in some plants of Florida scrub. American Journal of Botany 83: 21–27.[CrossRef][ISI]
Miller, S. P. 1998 The dynamics of the grass-mycorrhizal fungi association in nutrient-poor wetlands. Ph.D. dissertation, Institute of Ecology, University of Georgia, Athens, Georgia, USA.
Nei, M. 1972 Genetic distance between populations. American Naturalist 106: 283–292.[CrossRef][ISI]
———. 1973 Analysis of gene diversity in subdivided populations. Proceedings of the National Academy of Sciences, USA 70: 3321–3323.
———. 1977 F-statistics and analysis of gene diversity in subdivided populations. Annals of Human Genetics 41: 225–233.[ISI][Medline]
Oldale, R. 1992 Cape Cod and the islands: the geologic story. Parnassus Imprints, East Orleans, Massachusetts, USA.
Pleasants, J. M., and J. F. Wendel. 1989 Genetic diversity in a clonal narrow endemic, Erythronium propullans, and in its widespread progenitor, Erythronium albidum. American Journal of Botany 76: 1136–1151.[CrossRef][ISI]
Purdy, B. G., R. J. Bayer, and S. E. MacDonald. 1994 Genetic variation, breeding system evolution, and conservation of the narrow sand dune endemic Stellaria arenicola and the widespread S. longipes (Caryophyllaceae). American Journal of Botany 81: 904–911.[CrossRef][ISI]
Rabinowitz, D. 1981 Seven forms of rarity. In H. Synge [ed.], The biological aspects of rare plant conservation, 205–217. Wiley, Chichester, UK.
Ranker, T. A., and A. F. Schnabel. 1986 Allozymic and morphological evidence for a progenitor-derivative species pair in Camassia (Liliaceae). Systematic Botany 11: 433–445.[CrossRef][ISI]
Raymond, M., and F. Rousset. 1995 GENEPOP (Version 1.2): population genetics software for exact tests and ecumenicism. Journal of Heredity 86: 248–249.
Rice, K., and S. Jain. 1985 Plant population genetics and evolution in disturbed environments. In S. T. A. Pickett and P. S. White [eds.]. The ecology of natural disturbance and patch dynamics, 265–286. Academic Press, New York, New York, New York, USA.
Rohlf, F. J. 1994 Numerical taxonomy and multivariate analysis system. Version 1.80. Exeter Software, Setauket, New York, USA.
Rousset, F. 1997 Genetic differentiation and estimation of gene flow from F-statistics under isolation by distance. Genetics 145: 1219–1228.[Abstract]
SAS. 1997 SAS/STAT software: changes and enhancements through release 6.11. SAS Institute, Cary, North Carolina, USA.
Schalles, J. F., and D. J. Shure. 1989 Hydrology, community structure, and productivity patterns of a dystrophic Carolina bay wetland. Ecological Monographs 59: 365–385.[CrossRef]
Schwaegerle, K. E., and B. A. Schaal. 1979 Genetic variability and founder effect in the pitcher plant Sarracenia purpurea L. Evolution 33: 1210–1218.[CrossRef][ISI]
Sharitz, R. R., and C. A. Gresham. 1998 Pocosins and Carolina bays. In M. G. Messina and W. H. Conner [eds.]. Southern forested wetlands: ecology and management, 343–377. Lewis Publishers, Boca Raton, Florida, USA.
Slatkin, M. 1993 Isolation by distance in equilibrium and non-equilibrium populations. Evolution 47: 264–279.
