Content uploaded by Curtis Daehler
Author content
All content in this area was uploaded by Curtis Daehler
Content may be subject to copyright.
INTERFERTILITY BETWEEN HAWAIIAN ECOTYPES OF SIDA FALLAX
(MALVACEAE) AND EVIDENCE OF A HYBRID DISADVANTAGE
Mitsuko Yorkston and Curtis C. Daehler
1
Department of Botany, University of Hawaii, Honolulu, Hawaii 96822, U.S.A.
The flora of the Hawaiian Islands is widely recognized for its spectacular adaptive radiations, yet factors
maintaining or reinforcing the evolutionary divergence associated with radiation have been little studied. In the
Hawaiian Islands, Sida fallax occurs as two genetically distinct forms, a beach ecotype, which grows as
a prostrate shrub with small, pubescent leaves, and a mountain ecotype, which grows as an erect shrub with
larger, nearly glabrous leaves. Plants with intermediate morphology have only occasionally been reported. We
made experimental crosses between the beach and mountain forms to test for fertility barriers and to assess the
morphology and fitness of hybrids. Among beach-mountain crosses, average fruit set (83%) and germination
(80%) were not statistically different from intra-ecotype crosses. Furthermore, pollen stainability among
beach-mountain hybrids (96%) was also not significantly lower than in intra-ecotype crosses, indicating a lack
of fertility barriers. Among eight traits that differed statistically between beach and mountain ecotypes, four
(50%) were intermediate in the hybrids, three (37.5%) showed affinity toward one parental type, and one
(12.5%) was extreme. The survival and growth rate of beach-mountain hybrids did not differ from intra-
ecotype crosses under well-watered or drought greenhouse conditions; however, the beach ecotype had a higher
flowering frequency under drought conditions in comparison with the hybrids and the mountain ecotype. In
the field, low water availability in the beach environment probably selects against hybrids, reinforcing
differentiation between beach and mountain ecotypes. Furthermore, human land use has reduced intermediate
habitats, decreasing opportunities for natural hybridization.
Keywords: ecotype, fertility, germination, hybridization, morphology, pollination, Oahu.
Introduction
Plant biologists have long recognized the important role of
hybridization in plant evolution (Anderson 1949; Stebbins
1950; Baker 1951; Anderson and Stebbins 1954; Heiser
1973; Stace 1975; Grant 1981; Rieseberg 1995; Arnold
1997). Frequent hybridization between taxa can eliminate ge-
netic differentiation, but fertility barriers, even if they are in-
complete, may promote the coexistence of genetically distinct
taxa (Ferdy and Austerlitz 2002). For highly interfertile taxa,
a hybrid fitness disadvantage may maintain differentiation
(Arnold 1997). Habitat partitioning between the interfertile
taxa may also maintain differentiation (Wolf et al. 2001), al-
though when small populations are involved, as is common
on islands, demographic or genetic swamping may eliminate
differentiation via extinction of one of the taxa even if there
is strong habitat partitioning (Levin et al. 1996).
Ecotypes are genetically differentiated infraspecific taxa
that are adapted to particular environments or resources.
They have not attained formal taxonomic recognition, often
due to lack of study by taxonomists. Ecotypes can be highly
dissimilar, suggesting that they could develop fertility bar-
riers, and interecotype crosses can suffer from outbreeding
depression due to disruption of coadapted gene complexes
(Hufford and Mazer 2003). Examples of ecotypic differentia-
tion include adaptation to serpentine or calcareous soils
(Lombini et al. 2003), wet or dry environments (Fenster 1997;
Cavers et al. 2003), and high- or low-elevation environments
(Baruch 1979; Mariko and Koizumi 1993; Weih and Karlsson
1999). Ecotypic differentiation is often envisioned as an early
stage in the process of speciation by adaptive radiation
(Schluter 1996).
The flora of the Hawaiian Islands showcases the process of
adaptive radiation with a number of well-studied examples
that illustrate the evolution of widely divergent growth forms
and morphologies from a single colonizing ancestor (Craddock
2000). For many of these radiations (e.g., in Bidens, Scaevola,
and Lipochaeta), artificial or natural hybridizations indicate
that the species are usually fully interfertile (Gillett 1966;
Rabakonandrianina 1980; Ganders and Nagata 1984), whereas
in the Hawaiian silversword alliance, chromosomal rearrange-
ments have reduced cross-fertility between some members of
the group (Carr and Kyhos 1981), and polyploidy prevents
crossing of some Lipochaeta species (Rabakonandrianina
1980). Among the fully interfertile products of Hawaiian
adaptive radiations, some species are single-island endemics,
and for these, maintenance of differentiation can be simply
explained by allopatry. Among co-occurring interfertile spe-
cies or forms, maintenance of differentiation is usually ex-
plained by adaptations to specific habitats (Carlquist 1974)
and a presumed hybrid disadvantage, although this has rarely
been tested.
1
Author for correspondence; telephone 808-956-3929; fax 808-
956-3923; e-mail daehler@hawaii.edu.
Manuscript received June 2005; revised manuscript received November 2005.
221
Int. J. Plant Sci. 167(2):221–230. 2006.
Ó 2006 by The University of Chicago. All rights reserved.
1058-5893/2006/16702-0005$15.00
Our study focuses on experimental hybridization between
two morphologically distinct forms of Sida fallax Walp.
(Malvaceae) in the Hawaiian Islands. The beach ecotype,
which occurs across Oceania, is a prostrate, sprawling shrub
that is typically < 50 cm tall with small (commonly 1–4 cm
long), densely pubescent leaves. In contrast, the mountain
ecotype, which is apparently restricted to the Hawaiian Is-
lands, is an upright shrub that typically reaches 1–1.5 m in
height with large (commonly 4–8 cm long), glabrous or
nearly glabrous leaves. The two ecotypes naturally occur in
dry beach/coastal sites and upland, mesic forest sites, respec-
tively. Major differences between the two ecotypes have a ge-
netic basis, as demonstrated by the maintenance of distinct
forms and ecophysiological traits when grown from seed in
a common garden environment (Stephens 2000).
The first objective of our study was to test the hypothesis
that fertility barriers help maintain the divergent beach and
mountain ecotypes. This hypothesis was addressed by mak-
ing experimental crosses between ecotypes. The only published
evidence that these ecotypes are interfertile is circumstantial,
based on occasional records of intermediate specimens (Bates
1999). The second objective was to determine the morphol-
ogy and relative fitness of any hybrids from interecotype
crosses to examine whether hybrids might have a fitness dis-
advantage. Fitness was assessed by measuring growth and
flowering under well-watered and drought greenhouse condi-
tions. Detailed studies of trait variability and fitness among
the hybrids can help us understand the ecological and evolu-
tionary implications of hybridization (Arnold and Hodges
1995) and the maintenance of ecotypic differentiation.
Material and Methods
Study Species
Sida fallax Walp. is a prostrate to erect perennial shrub
that is indigenous to Oceania, including the Hawaiian Islands
(Bates 1999). It is found on all the main Hawaiian Islands as
well as Nihoa and Midway atoll, occurring in coastal and
lowland dry communities (near sea level to 300 m elevation)
as well as in mesic forests and montane communities up to
2000 m elevation (Bates 1999). Examination of herbarium
specimens at the University of Hawaii at Manoa (HAW) indi-
cates that the beach ecotype, with small (1–4 cm diameter),
densely pubescent leaves, is common across the Pacific,
whereas the mountain ecotype, with large (4–8 cm diameter),
nearly glabrous leaves and an upright growth, appears to be
unique to the Hawaiian Islands. This conclusion was also
reached by Stephens (2000) after examining dozens of her-
barium specimens at the Bishop Museum (BISH). The beach
ecotype can be found along beaches and in dry, coastal shrub
lands. The mountain ecotype is generally found in more me-
sic and forested habitats, further from the coast. These two
ecotypes were given different names by the ancient Hawai-
ians, reflecting the clear morphological differences between
them (Neal 1965). Intermediate morphological forms have
been occasionally reported (Bates 1999); however, they ap-
pear to be rare (M. Yorkston, personal observation). Bates
(1999) did not recognize formal varieties within S. fallax, but
he considered S. fallax to be the most widespread and vari-
able taxon of the Malvaceae in the Hawaiian Islands.
Study Site and Plant Material
Experimental crosses between beach and mountain eco-
types were made in a greenhouse at the University of Hawaii
at Manoa. The plants used for crosses were a subset of the
plants studied by Stephens (2000). These plants were grown
from seeds collected from three beach/coastal sites and three
mountain/upland sites on the island of Oahu (fig. 1; table 1).
Plants from the mountain populations (elevation higher than
266 m) had an upright growth form with nearly glabrous
leaves, while plants from the coastal populations (near sea
level) were prostrate with densely pubescent leaves (Stephens
2000). Ten greenhouse-grown plants of the beach (B) ecotype
and 10 plants of the mountain (M) ecotype were used for the
crossing experiments. Plants from all six localities were in-
cluded to establish a diverse sample of the beach and moun-
tain ecotypes. Possible differences among localities were not
examined because the purpose of this work was to broadly
assess the outcome of interecotype crosses.
Fig. 1 Sida fallax seed collection sites on Oahu, Hawaiian Islands.
Table 1
Collection Sites for Seeds of the Sida fallax Plants Used
for Experimental Crosses, Oahu, Hawaii
Collection site Elevation (m)
Annual
rainfall (mm)
Mountain S. fallax (M):
Wa‘ahila Ridge 366 2000
Hawai‘iloa Ridge trail 366 1250
Kuaokala Forest Reserve 266 900
Beach S. fallax (B):
Makapu‘u beach park Near sea level 800
Ka‘ena Point Natural
Area Reserve Near sea level 800
Sandy Beach Park Near sea level 800
Note. Rainfall data are from Giambelluca et al. (1986).
222
INTERNATIONAL JOURNAL OF PLANT SCIENCES
Experimental Crosses
The flowers of S. fallax open irregularly over a period of
weeks, but each flower lasts only 1 d. Attempts to emasculate
plants prior to crossing resulted in damage to stigmas and
the style due to the large number of anthers in close proxim-
ity to the stigma column; however, initial trials had indicated
that S. fallax was self-incompatible (C. C. Daehler, unpub-
lished data), making emasculation unnecessary in the green-
house. Nevertheless, to test for autogamy in unemasculated
flowers, 160 unmanipultated flowers were monitored for
fruit development in the greenhouse as experimental crosses
were being made during summer 2001. Two types of experi-
mental crosses were made: between ecotype (M
3 B and
B
3 M) and within ecotype (M 3 M and B 3 B) (the first let-
ter in each cross denotes the maternal plant). Interecotype
crosses were made using both ecotypes as the maternal plant,
and the products of these crosses were examined separately,
since some studies have found asymmetry in crossability and
hybrid viability (Duvall and Biesboer 1988). Most crosses
were made in the morning using a sharp-tipped forceps to
collect mature anthers from a pollen donor and transfer pol-
len to the stigmas of a maternal parent. Crosses were made
haphazardly between available individuals. For each cross
type (M
3 B, B 3 M, M 3 M, and B 3 B) between 35–50
unique maternal plant–paternal plant combinations were made
involving a total of 77–137 flowers within each cross type.
Crossability, Percent Germination,
and Germination Velocity
All pollinated flowers were monitored for 3 wk, and the
success or failure in seed production was recorded. When a
shizocarp was empty or all mericarps appeared to be aborted,
the pollination was considered a failure. When the shizocarp
contained any number of healthy-looking mericarps, the polli-
nation was considered successful. Nonaborted seeds were
tested for germination in August 2001. Before planting, all
seeds were stripped from the mericarps and nicked to enhance
germination. Nicking was done with a razor under a dissecting
microscope. Seeds were then pressed into solidified 2% water
agar that had been poured into covered, clear-plastic petri
dishes. The dishes were randomly positioned on lab benches
at room temperature (23°C) under natural light. Germination
was observed for 60 d, and the germinated seeds were counted
at 2-d intervals. The agar medium was replaced as needed to
prevent desiccation. The germination rate was determined
using the modified Timson’s index of germination velocity
(G)/t, where G is the percentage of seeds that germinated
at any given 2-d interval and t is the total germination period
(Timson 1965; Khan and Rizvi 1994).
Growth Patterns, Survival, and Flowering
Within 1 wk after germination (2–7 d), seedlings were
transplanted to 13
3 13 3 13-cm pots. The growth medium
was a blended soil mix (4 potting soil : 4 peat moss : 1 com-
post : 1 perlite) recommended for native Hawaiian plants
(Culliney and Koebele 1999). Pots were randomly located on
benches in the greenhouse (University of Hawaii at Manoa).
All pots were equally well watered for 3 wk to reduce trans-
plant shock. To compare growth rates (as a measure of fit-
ness) under two environmental conditions, seedlings derived
from each of the four cross types were randomly assigned to
one of two treatments: well-watered control (watered to keep
the soil consistently moist) or drought (watered only when
plants showed visible signs of wilting). Usually, the control
Table 2
Fruit set in Sida fallax beach (B) and mountain (M) ecotypes for unmanipulated, open flowers and
flowers subjected to experimental crossing, Oahu, Hawaii
Cross type
Number of
maternal plants
Number of
distinct crosses
a
Total number
of flowers
% Fruit set
(61 SD)
Normal seeds per
fruit (61 SD)
Aborted seeds
per fruit (61 SD)
B (open) 9 ... 90 6.6 6 9.8
A
1.2 6 1.2
A
4.7 6 2.4
A
M (open) 7 ... 70 2.8 6 4.2
A
3.1 6 1.8
A
4.9 6 0.4
A
M 3 M10 35 77916 15
B
5.2 6 0.8
B
0.9 6 1.3
B
B 3 B 10 50 125 83 6 20
B
5.6 6 0.9
B
0.6 6 0.6
B
B 3 M 9 41 137 71 6 28
B
4.6 6 1.8
B
1.4 6 1.5
B
M 3 B 10 40 128 95 6 13
B
5.1 6 0.8
B
1.2 6 0.8
B
Note. Standard deviations represent variation among individual maternal plants. Values followed by differing superscript letters are statisti-
cally different (P < 0:05).
a
Crosses involving different maternal-paternal plant combinations.
Fig. 2 Germination of seeds from Sida fallax crosses over time.
223
YORKSTON & DAEHLER—INTERFERTILITY AMONG SIDA FALLAX ECOTYPES
pots were watered every day and the drought pots were
watered every 2–3 d. All pots were randomly positioned and
rerandomized monthly. Survival, final number of leaves, and
flowering rates were compared among cross types over 7 mo.
Pollen Stainability
Among the plants that flowered in the well-watered treat-
ment, the percentage of stainable pollen among the four cross
types was calculated to estimate the pollen viability. Pollen
collected from freshly opened flowers was stained for at least
24 h using a solution of aniline blue in lactophenol (Hauser
and Morrison 1964; Kearns and Inouye 1993). Observations
of stainability were made at
332 magnification under a dis-
secting microscope. Only uniformly well-stained grains of
normal size were scored as stained. Estimates of percent
stainable pollen were made from counting a minimum of
300 grains/plant.
Morphological Analyses
After the plants had flowered in 2001 and 2002, the mor-
phology of the four cross types was assessed using 17 charac-
ters. Vouchers of representative maternal plants and crosses
were deposited at HAW (Yorkston 1–6, 11–15, 17, 19–21,
and 26–28). At least one flower and two leaves were sampled
per individual plant. Leaves were collected from the third
and fourth node proximal to the tip of the main (longest)
stem. Leaf width was measured at widest part of the leaf,
and leaf length was measured from the apex to the leaf base.
The leaf width : length ratio was used as an index of leaf
shape, e.g., 1 being roughly circular, <1 elongate. Depth of
the basal leaf lobe was measured from the point of petiole at-
tachment to the edge of the basal lobe. Leaf pubescence on
top and bottom surfaces was assessed by counting the num-
ber of trichomes/mm of the cross section of the leaf (avoiding
the midrib area) under the dissecting microscope at
324 mag-
nification. The leaf margin was categorized as sharp serrate
(1) or round-toothed (0) based on whether the margin teeth
formed acute apices or rounded apices. Flower diameter was
measured at the widest part of the fully opened flower. Petals
were separated from the base of the flower and petal width
was measured at the widest part. Each petal was generally
two-lobed, so two measures of petal length were recorded,
one of each lobe. The asymmetry in petal lobe size was as-
sessed by subtracting the short petal length from the long
petal length. Style length was measured to the nearest
0.01 mm from the tip of the stigma to the end of the style where
it united to the staminal column by using a digital caliper.
Petal color was assigned to three categories (1 ¼ pale yellow,
2 ¼ medium yellow, 3 ¼ dark yellow). Presence or absence of
a red spot at the base of the flower was recorded in three cat-
egories (0 ¼ absent, 1 ¼ light, 2 ¼ medium, 3 ¼ dark).
Statistical Analyses
All statistical analyses were conducted with SYSTAT 9
(SPSS, Chicago). Differences in crossability, viable seed
Fig. 3 Habit of representative plants derived from intra-ecotype
crosses (M
3 M and B 3 B) and interecotype crosses (M 3 B).
Table 3
Mean values (61 SD) of traits exhibiting significant differences between beach (B) and mountain (M) forms of
Sida fallax and experimental hybrids when grown together from seed in a greenhouse
Cross type
Character B
3 B mean B 3 M mean M 3 B mean M 3 M mean Summary
a
Leaf length (cm) 4.05
A
6 1.06 4.86
B
6 1.13 5.19
B
6 1.29 5.56
B
6 1.03 d
Leaf base lobe (cm) 0.54
A
6 0.17 0.36
B
6 0.14 0.43
B
6 0.27 0.06
C
6 0.10 i
Hair top (trichome/mm) 9.27
A
6 3.35 3.42
B
6 3.60 3.44
B
6 2.94 0.44
C
6 0.51 i
Hair bottom (trichome/mm) 13.30
A
6 4.61 7.21
B
6 4.30 8.71
B
6 3.61 2.44
C
6 0.73 i
Leaf margin
b
0.24
A
6 0.52 0.85
B
6 0.36 0.91
B
6 0.28 1.00
B
6 0.00 d
Petal width (mm) 15.37
A
6 3.52 12.48
A
6 3.60 13.35
A
6 2.34 10.42
B
6 2.82 d
Pedicel length (cm) 3.68
A
6 0.66 3.30
B
6 0.54 2.94
B
6 0.69 1.80
C
6 0.75 i
Dark spot of flower
c
1.29
A
6 0.49 1.77
A
6 0.83 2.25
B
6 0.60 1.56
A
6 0.73 e
Standard deviations reflect variation obtained when crosses were produced from different maternal plants. For each character, means sharing
the same superscript letter do not differ significantly (P < 0:05).
a
d ¼ hybrids showed affinity toward one ecotype, i ¼ hybrids intermediate, e ¼ hybrids had extreme phenotype.
b
0 ¼ soft crenate, 1 ¼ sharp serrate.
c
0 ¼ absent, 1 ¼ light, 2 ¼ medium, 3 ¼ dark.
224
INTERNATIONAL JOURNAL OF PLANT SCIENCES
production, germination rate, growth rate, pollen viability,
and plant morphology among the four cross types (B
3 B,
B
3 M, M 3 B, and M 3 M) were analyzed by ANOVA. Prior
to analyses, all measurements made on the four cross types
were first averaged at the level of the individual plant if mul-
tiple measures had been taken per plant (e.g., leaf width).
Next, plants derived from the same maternal parent and
cross type were averaged to obtain one average value for
each maternal parent across each of the four cross types.
These averages by maternal plant were treated as replicates
in the ANOVA. Significant differences between each of the
four cross types were assessed using post hoc comparisons
employing a Bonferroni correction for multiple comparisons.
A multidimensional scaling (MDS) analysis, based on the
Fig. 4 Representative photos of (A) beach ecotype leaves, (B) mountain ecotype leaves, (C, D, E) F1 hybrids leaves, (F) beach ecotype flower,
(G) hybrid (B
3 M) flower, and (H) mountain ecotype flower. Black scale bars ¼ 1 cm, smaller red lines represent 1 mm spacing.
225
YORKSTON & DAEHLER—INTERFERTILITY AMONG SIDA FALLAX ECOTYPES
morphological variables, was also employed to visualize dif-
ferences between the cross types and relationships among the
traits. MDS was chosen because it does not require that the
data be distributed as multivariate normal, and it is often
more effective at providing visual separation of groups if
there are nonlinear relationships among the underlying varia-
bles (SYSTAT 2002). Up to 50 iterations were allowed, or iter-
ations were terminated if convergence reached 0.005. Stress
was evaluated using a Kruskal monotonic (nonparametric) loss
function. We specified only two dimensions for the MDS anal-
ysis to allow for a simple graphical presentation.
Results
Crossability, Percent Germination,
and Germination Velocity
Fruit set rates for unmanipulated, open-pollinated flowers
in the greenhouse were <6% compared with 71%–95% for the
experimental crosses (table 2). Among the few open-pollinated
flowers that set fruit, most of the seeds were aborted, as indi-
cated by a shriveled seed coat and flattened seed, which is in
contrast to the plump (normal) seeds that were predominant
in the crossing treatments (table 2). Among the four cross types,
there were no significant differences in percent fruit set, nor-
mal seeds per fruit, or aborted seeds per fruit (table 2).
After 60 d, percent seed germination averaged significantly
lower for the M
3 M cross (63%) compared with the other
cross types (Kruskal-Wallis Test, P ¼ 0:048), which averaged
between 75% and 82% germination and were not statisti-
cally different from each other. The M
3 M cross type also
had an overall slower rate of germination (Timson’s index of
germination velocity ¼ 22 vs. 28–31 for the other cross
types). The slower pattern of germination for M
3 M can be
seen from the shallower slope in the plot of percent germina-
tion versus time, especially during the first 30 d (fig. 2). The
pattern over time for the interecotype hybrids was more simi-
lar to that of the B
3 B crosses.
Hybrid Morphology
Hybrids generally had a stature that was intermediate be-
tween the erect mountain ecotype and the prostrate beach
ecotype (fig. 3). Eight out of 17 morphological traits were
statistically different between the B
3 B and M 3 M cross
types (table 3). Compared with the beach ecotype, the moun-
tain ecotype had longer leaves with less pronounced basal
lobing and sharp-serrate leaf margins (table 3; fig. 4A,4B).
Leaves of the mountain ecotype also had lower densities of
trichomes on both the upper and lower leaf surfaces. Consid-
ering the floral traits, the mountain ecotype had shorter pedi-
cels and narrower petals. Among the eight traits that differed
between the beach and mountain ecotypes, hybrids showed
intermediacy (additivity) for four of the traits: basal leaf
lobe, trichome density on upper leaves, trichome density on
lower leaves, and pedicel length (table 3; fig. 5). Three traits
showed evidence of dominance, as indicated by a hybrid af-
finity toward one of the two parental types. The longer leaves
and coarser-toothed, sharp-serrate leaf margin of the moun-
tain ecotype was apparent in the hybrids (fig. 4C–4E), while
the hybrids possessed the wider petal length of the beach eco-
type. One trait showed some evidence of extremism (epista-
sis) in the hybrids. The hybrids tended to have a darker red
spot at the center of their flowers (fig. 4F–4G). This trend
was apparent for both M
3 B and B 3 M hybrids but was sta-
tistically significant only for the M
3 B hyrids. The following
traits were not statistically among cross types: leaf width,
width/length ratio, petiole length, flower diameter, petal
length long, petal length short, petal lobe, filament length,
and petal color.
Survival, Growth, Flowering, and Pollen Stainability
Survival in the drought treatment (fig. 6) and well-watered
treatment (data not shown) did not vary significantly among
the cross types. During the 7-mo period when growth was
Fig. 5 Multidimensional scaling analysis separating intra-ecotype
crosses (M 3 M and B 3 B) from interecotype crosses (M 3 B and
B 3 M) based on morphology. Vectors for the most important traits
are shown. The relative importance of each trait is indicated by the
length of its vector. Output was obtained after 10 iterations;
stress ¼ 0:161, R
2
¼ 0:868.
Fig. 6 Survival of Sida fallax crosses over time in the drought
treatment. Differences among cross types were not statistically
significant.
226
INTERNATIONAL JOURNAL OF PLANT SCIENCES
monitored, there were no significant differences in growth
rate between cross types as measured by leaf number in the
drought treatment, although there was a trend toward fewer
leaves in the M
3 M crosses (fig. 7). Leaf numbers in the
well-watered treatment were not statistically different from
the drought treatment (data not shown). Despite the lack of
statistical differences in vegetative growth (number of leaves)
between crosses or between the drought and well-watered
treatments, the drought treatment had a significantly lower
rate of flowering for all crosses (P < 0:05), except for B
3 B,
which did not exhibit a reduced flowering rate in the drought
treatment (fig. 8). Age at first flowering did not differ signifi-
cantly among cross types or between watering treatments,
averaging 243 d. With the exception of a single plant derived
from the M
3 B cross, which had a mean pollen stainability
of 12.5%, the progeny derived from interecotype crosses had
high pollen stainability, ranging from 92%–100%. The rate
of pollen stainability for interecotype progeny was not statis-
tically different from intra-ecotype progeny (table 4).
Discussion
Breeding System of Sida fallax
The fruit-set rate of unmanipulated flowers in the green-
house was an order of magnitude lower than the experimen-
tal crosses, indicating that Sida fallax is not generally
autogamous. This was confirmed by a complete lack of fruit
set among 32 bagged flowers that received no pollen manipu-
lation (Yorkston 2005). On a single occasion a wasp was ob-
served flying in the greenhouse near the flowers, so it is
possible that the low fruit set rate among unmanipultated,
unbagged flowers could have been from stray pollinations by
Hymenoptera. No insects were regularly seen around the
plants in the greenhouse, except for small ants, and these did
not appear to contact the reproductive parts. Deliberate
transfer of self pollen between flowers on the same plants
(geitonogamous selfing) resulted in low fruit set rates (aver-
aging 14.5%), with the majority of seeds in those fruits being
aborted (Yorkston 2005). These results in composite indicate
that S. fallax is not autogamous, and the plants are effec-
tively self incompatible. The low fruit-set rate of unmanipu-
lated flowers indicates that only a small fraction of the
crosses may have been of the wrong cross type. Three speci-
mens on the MDS plot could be examples of this (fig. 5; two
B
3 B near [0, 1] and one M 3 M near [3, 5]). We did not
use genetic markers to confirm the genetic status of progeny;
however, the fact that interecotype crosses differed statisti-
cally in morphology from intra-ecotype crosses demonstrates
that the crossing treatments were effective.
Hybrid Morphology
The morphology of F1 hybrids is often intermediate be-
tween the two parental types (Marhold et al. 2002; Rocas
et al. 2004), but ecologically important traits are sometimes
expressed in a nonadditive pattern in F1 hybrids (Burke et al.
1998). Examples of the latter include the sucrose : hexose ra-
tio in nectar of Costus hybrids (Sytsma and Pippen 1985),
which potentially influences pollinator attractiveness, and the
presence of novel phytochemicals in F1 Helianthus hybrids
(Buschmann and Spring 1995), which could affect herbivory.
The general pattern of trait expression in the F1 S. fallax
hybrids was similar to that reported for Carduus hybrids
(Warwick et al. 1992): the F1 hybrids were intermediate for
some traits but showed affinities toward one parent or the
other for other traits. The red spot at the center of S. fallax
flowers seemed to be accentuated in interecotype hybrids. A
central, prominent dark spot occurs on the flowers of other
Malvaceae, such as Hibiscus calyphylus. Presumably, it func-
tions as a nectar guide, but we do not know whether the de-
gree of accentuation observed in the hybrids has ecological
Fig. 7 Growth of Sida fallax crosses over time in the drought
treatment, as measured by number of leaves. Differences among cross
types were not significant.
Fig. 8 Flowering frequencies in Sida fallax intra-ecotype crosses
(M
3 M and B 3 B) and interecotype crosses (M 3 B and B 3 M) after
7 mo. Asterisks indicate significant difference between control and
drought treatments. Plus sign indicates significant difference from
other cross types.
227
YORKSTON & DAEHLER—INTERFERTILITY AMONG SIDA FALLAX ECOTYPES
significance. Some individuals of the beach ecotype have
flowers that completely lack red markings (C. C. Daehler,
personal observation); factors contributing to this color-
pattern polymorphism merit further investigation.
In our experiment we analyzed the M
3 B and B 3 M cross
types separately because some hybridization studies have
found that the F1 progeny more closely resemble the mater-
nal (or paternal) parent, especially during the early stages
of growth (Hawley and Dehayes 1985). This phenomenon,
which could be due to either differential transmission of
DNA elements (e.g., via mitochondria or chloroplasts) or
nonheritable maternal effects, was not apparent in our study.
Barriers to Hybridization
No evidence was found of genetic barriers that would pre-
vent hybridization or introgression between the beach and
mountain ecotypes of S. fallax for either M
3 BorB3 M
crosses. We did not test for possible distorted chromosome
segregation (Whitkus 1998), but a lack of genetic barriers to
introgression is further supported by high pollen stainability
among second-generation backcrosses (Yorkston 2005). Nev-
ertheless, the two ecotypes showed clear genetic differences
in morphology (see also Stephens 2000), and the two types have
remained distinct in the field, segregated by habitat type.
Differences in pollinator preference have sometimes been
reported as a barrier to hybridization (Dawar et al. 1994;
Schemske and Bradshaw 1999), but this is unlikely to be
the case for S. fallax ecotypes, as the flowers of both eco-
types are primarily visited by honey bees (Apis melifera) and
other generalist Hymenoptera (C. C. Daehler, personal ob-
servation).
If genetic or pollinator barriers do not prevent hybridiza-
tion and introgression among ecotypes, how are the distinct
ecotypes maintained in the field? Two factors are probably
important. The first relates to the pattern of human distur-
bance across the landscape. In many parts of the Hawaiian
Islands, humans have transformed the transition zone be-
tween beach and mountain habitats through urbanization,
suburbanization, agriculture, and industrial developments. Thus,
human activities have created a physical barrier between
beach and mountain populations of S. fallax, reducing the
overall potential for crossing between ecotypes,
The second, perhaps more important factor maintaining
the two ecotypes appears to be selection against hybrids.
Selection against hybrids is generally weaker in controlled
laboratory environments as compared to field conditions
(Schluter 1996), so fitness differences we detected in the
greenhouse would likely be magnified in the field. We found
no growth or survival disadvantage for hybrids in the green-
house; however, the beach ecotype flowered at a higher fre-
quency than interecotype hybrids and the mountain ecotype
in the drought treatment. This indicates that interecotype hy-
brids and the mountain ecotype would be at a disadvantage
in dry habitats. In fact, the beach ecotype is generally found
in dryer leeward habitats (Stephens 2000; see also table 1).
Stephens (2000) compared the ecophysiology of beach and
mountain ecotypes and found that leaves of the beach eco-
type had a lower osmotic potential at the turgor loss point,
suggesting a greater tolerance of drought compared with the-
mountain ecotype. Stephens (2000) also found higher reflec-
tance of solar radiation by leaves of the beach ecotype.
Higher reflectance, probably due to leaf pubescence, could
potentially decrease leaf temperature in hot, dry environ-
ments, helping to conserve water. The pubescence may also
create a boundary layer that reduces leaf evaporation. Al-
though our study did not measure growth under competition,
hybrids tended to have intermediate stature (fig. 4), which is
likely to be a disadvantage in mesic montane environments,
relative to the more erect mountain growth form.
Spatial and temporal variation in water availability may
commonly control the distribution of hybrids and the pat-
terns of introgression (Vanbolkenburgh et al. 1998; Orians
et al. 1999). Intermediate habitats may favor hybrids (Arnold
1997), but the zone of higher hybrid fitness is often quite nar-
row (Briggs 1962), and much intermediate habitat has been
lost, as mentioned above. Future field experiments are needed
to measure selection differentials for each ecotype and for hy-
brids in beach and montane habitats. Jordan (1991) out-
planted experimental hybrids between coastal and inland
ecotypes of Diodes teres, allowing selection of particular mor-
phological traits to be measured in coastal and inland habi-
tats. The selection differential between habitats indicated that
the observed differentiation between ecotypes could be ex-
pected to develop within 25–100 generations (Jordan 1991).
Thus, morphological patterns can segregate rather quickly
across habitats even when there are abundant contact zones.
The flora of the Hawaiian Islands is well known for its
adaptive radiations accompanied by speciation (Carlquist
1974). The mountain ecotype of S. fallax, which appears to
be endemic to the Hawaiian Islands, may still be in the pro-
cess of differentiating from the more widespread beach eco-
type. A shift in dispersal syndrome could be one mechanism
that accelerates the differentiation process. Awns were found
to be either absent or < 1 mm on mericarps of the beach eco-
type, while the mountain ecotype commonly had awns 2–3 mm
long (Yorkston 2005), suggesting an adaptive shift in the
mountain ecotype toward incidental seed dispersal by ani-
mals. Similar shifts in dispersal syndrome have also been as-
sociated with the adaptive radiation and speciation of Bidens
in the Hawaiian Islands (Ganders and Nagata 1984). Sida
fallax is increasingly being used for horticulture and land-
scaping in Hawaii, and it is likely that growers are experi-
menting with hybrids between beach and mountain ecotypes.
Future gene flow from abundantly cultivated hybrids could
influence the evolutionary trajectories and long-term coexis-
tence of the beach and mountain ecotypes, but the predic-
tions are complicated by hybridization’s dual influence as
a homogenizer and an introducer of evolutionary novelty.
Table 4
Mean pollen stainability for progeny derived from
experimental Sida fallax crosses
Cross type
No. individuals
sampled Mean 6 SD (%) Range (%)
B
3 B 14 98.3 6 1.8 94.1–100
M 3 M 10 87.4 6 14.7 77–100
B 3 M 10 98.0 6 2.5 92–100
M 3 B 30 93.8 6 16.5 12.5–100
228
INTERNATIONAL JOURNAL OF PLANT SCIENCES
Acknowledgments
We thank Michelle Stephens for providing us with the Sida
fallax plants used to make crosses. Shahin Ansari, Dan Bardo,
Stephanie Joe, Huang-Chi Kuo, Lauren Weisenberger, and
Malcolm Yorkston provided greenhouse assistance. Gerry
Carr and Cliff Morden and two anonymous reviewers pro-
vided helpful comments on an earlier draft of the manuscript.
Literature Cited
AndersonE 1949 Introgressive hybridization. Wiley, New York.109 pp.
Anderson E, GL Stebbins Jr 1954 Hybridization as an evolutionary
stimulus. Evolution 8:378–388.
Arnold ML 1997 Natural hybridization and evolution. Oxford
University Press, New York.
Arnold ML, SA Hodges 1995 Are natural hybrids fit or unfit relative
to their parents? Trends Ecol Evol 10:67–71.
Baker HG 1951 Hybridization and natural gene-flow between higher
plants. Biol Rev 26:302–337.
Baruch Z 1979 El evati onal differ entiation in Espeletia schultzii
Compositae, a giant rosette plant of the Venezuelan Paramos.
Ecology 60:85–98.
Bates DM 1999 Malvaceae Mallow family. Pages 868–903 in WL
Wagner, DR Herbst, SH Sohmer, eds. Manual of the flowering
plants of Hawaii. University of Hawaii Press, Honolulu.
Briggs BG 1962 Interspecific hybridization in the Ranunculus lappa-
ceus group. Evolution 16:372–390.
Burke JM, SE Carney, ML Arnold 1998 Hybrid fitness in the
Louisiana irises: analysis of parental and F1 performance. Evolution
52:37–43.
Buschmann H, O Spring 1995 Sesquiterpene lactones as a result of
interspecific hybridization in Helianthus species. Phytochemistry
39:367–371.
Carlquist S 1974 Island biology. Columbia University Press, New
York. 660 pp.
Carr GD, DW Kyhos 1981 Adaptive radiation in the Hawaiian
silversword alliance Compositae Madiinae. 1. Cytogenetics of
spontaneous hybrids. Evolution 35:543–556.
Cavers S, C Navarro, AJ Lowe 2003 A combination of molecular
markers identifies evolutionarily significant units in Cedrela odorata
L. (Meliaceae) in Costa Rica. Conserv Genet 4:571–580.
Craddock EM 2000 Speciation processes in the adaptive radiation of
Hawaiian plants and animals. Pages 1–43 in MK Hecht, RJ
Macintyre, MT Clegg, eds. Evolutionary biology. Vol 31. Kluwer
Academic/Plenum, New York.
Culliney JL, BP Koebele 1999 A native Hawaiian garden: how to
grow and care for island plants. University of Hawaii Press,
Honolulu. 164 pp.
Dawar R, T Ali, M Qaiser 1994 Hybridization in the Sida ovata
complex. II. Evidence from breeding studies. Pak J Bot 26:83–97.
Duvall MR, DD Biesboer 1988 Nonreciprocal hybridization failure
in crosses between annual wild-rice species Zizania palustris
3
Zizania aquatica Poaceae. Syst Bot 13:229–234.
Fenster CB 1997 Ecotypic differentiation for flood-tolerance and its
morphological correlates in Chamaecrista fasciculata. Aquat Bot
56:215–231.
Ferdy J, F Austerlitz 2002 Extinction and introgression in a commu-
nity of partially cross-fertile plant species. Am Nat 160:74–86.
Ganders FR, KM Nagata 1984 The role of hybridization in the
evolution of Bidens on the Hawaiian Islands. Pages 179–194 in WF
Grant, ed. Plant biosystematics. Academic Press, New York.
Giambelluca TW, MA Nullet, TA Schroeder 1986 Rainfall atlas of
Hawaii. State of Hawaii Department of Land and Natural
Resources Report No. R76, Honolulu, 267 pp.
Gillett GW 1966 Hybridization and its taxonomic implications in
the Scaevola gaudichaudiana complex of the Hawaiian Islands.
Evolution 20:506–516.
Grant V 1981 Plant speciation. 2nd edition. Columbia University
Press, New York. 563 pp.
Hauser EJP, JH Morrison 1964 The cytochemical reduction of nitro
blue tetrazolium as an index of pollen viability. Am J Bot 51:
748–752.
Hawley GJ, DH Dehayes 1985 Hybridization among several North
American firs. 2. Hybrid verification. Can J For Res 15:50–55.
Heiser CB 1973 Introgression re-examined. Bot Rev 39:347–366.
Hufford KM, SJ Mazer 2003 Plant ecotypes: genetic differentiation
in the age of ecological restoration. Trends Ecol Evol 18:147–155.
Jordan N 1991 Multivariate analysis of selection in experimental
populations derived from hybridization of two ecotypes of the
annual plant Diodia teres W. Rubiaceae. Evolution 45:1760–1772.
Kearns CA, DW Inouye 1993 Techniques for pollination biologists.
University Press of Colorado, Niwot. 583 pp.
Khan MA, Y Rizvi 1994 Effect of salinity, temperature, and growth
regulators on the germination and early seedling growth of Atriplex
griffithii var. stocksii. Can J Bot 72:475–479.
Levin DA, J Francisco-Ortega, RK Jansen 1996 Hybridization and
the extinction of rare plant species. Conserv Biol 10:10–16.
Lombini A, M Llugany, C Poschenrieder, E Dinelli, J Barcelo
2003 Influence of the Ca/Mg ratio on Cu resistance in three Silene
armeria ecotypes adapted to calcareous soil or to different, Ni- or
Cu-enriched, serpentine sites. J Plant Physiol 160:1451–1456.
Marhold K, J Lihova, M Perny, R Grupe, B Neuffer 2002 Natural
hybridization in Cardamine (Brassicaceae) in the Pyrenees: evidence
from morphological and molecular data. Bot J Linn Soc 139:
275–294.
Mariko S, H Koizumi 1993 Respiration for maintenance and growth
in Reynoutria japonica ecotypes from different altitudes on Mt Fuji.
Ecol Res 8:241–246.
Neal MC 1965 In garden s of Hawaii. Bishop Muse um Press,
Honolulu, HI. 924 pp.
Orians CM, DI Bolnick, BM Roche, RS Fritz, T Floyd 1999 Water
availability alters the relative performance of Salix sericea, Salix
eriocephala, and their F1 hybrids. Can J Bot 77:514–522.
Rabakonandrianina E 1980 Infrageneric relationships and the origin
of the Hawaiian endemic genus Lipochaeta (Compositae). Pac Sci
34:29–39.
Rieseberg LH 1995 The role of hybridization in evolution: old wine
in new skins. Am J Bot 82:944–953.
Rocas G, DE Klein, EA deMattos 2004 Artificial hybridization
between Pitcairnia flammea and Pitcairnia corcovadensis (Brome-
liaceae): analysis of the performance of parents and hybrids. Plant
Species Biol 19:47–53.
Schemske DW, HD Bradshaw Jr 1999 Pollinator preference and the
evolution of floral traits in monkeyflowers (Mimulus). Proc Natl
Acad Sci USA 96:11910–11915.
Schluter D 1996 Ecological causes of adaptive radiation. Am Nat
148(suppl):S40–S64.
Stace CA, ed 1975 Hybridization and the flora of the British Isles.
Academic Press, London.
Stebbins GLJ 1950 Variation and evolution in plants. Columbia
University Press, New York. 643 pp.
Stephens ML 2000 The comparative ecophysiology of mountain and
coastal populations of Sida fallax Walp. (Malvaceae) in Hawaii. MS
thesis. University of Hawaii, Honolulu.
229
YORKSTON & DAEHLER—INTERFERTILITY AMONG SIDA FALLAX ECOTYPES
SYSTAT 2002 SYSTAT 10.2 statistics I. SYSTAT, Richmond, CA.
665 pp.
Sytsma KJ, RW Pippen 1985 Morphology and pollination biology of
an intersectional hybrid of Costus (Costaceae). Syst Bot 10:353–362.
Timson J 1965 New method of recording gemination data. Nature
207:2016–2017.
Van Volkenburgh E, R Stahlberg, L Bultynck 1998 Physiological
mechanisms controlling the rate of leaf growth. Pages 41–56 in H
Lambers, H Poorter, MMI Van Vuuren, eds. Inherent variation in
plant growth. Backhuys, Leiden.
Warwick SI, BK Thompson, LD Black 1992 Hybridization of
Carduus nutans and Carduus acanthoides (Compositae): morpho-
logical variation in F-1 hybrids and backcrosses. Can J Bot 70:
2303–2309.
Weih M, PS Karlsson 1999 Growth response of altitudinal ecotypes
of mountain birch to temperature and fertilisation. Oecologia 119:
16–23.
Whitkus R 1998 Genetics of adaptive radiation in Hawaiian and
Cook Islands species of Tetramolopium (Asteraceae). II. Genetic
linkage map and its implications for interspecific breeding barriers.
Genetics 150:1209–1216.
Wolf DE, N Takebayashi, LH Rieseberg 2001 Predicting the
risk of extinction through hybridization. Conserv Biol 15:
1039–1053.
Yorkston M 2005 Experimental hybridization between mountain
and coastal forms of Sida fallax Walp., and between S. fallax and S.
rhombifolia L. (Malvaceae). MS thesis. University of Hawaii,
Honolulu.
230
INTERNATIONAL JOURNAL OF PLANT SCIENCES