American Journal of Botany 94(4): 660–673. 2007.
TRANSFER
OF GLYPHOSATE RESISTANCE: EVIDENCE OF
HYBRIDIZATION IN CONYZA (ASTERACEAE)1
IAN A. ZELAYA,2,4 MICHEAL D. K. OWEN,2
AND
MARK J. VANGESSEL3
2Iowa State University, Department of Agronomy, 2104 Agronomy Hall, Ames, Iowa 50011-1011 USA; and 3University of
Delaware, Department of Plant and Soil Sciences, Research and Education Center, 16684 County Seat Highway, Georgetown,
Delaware 19947-9575 USA
Transfer of herbicide resistance genes between crops and weeds is relatively well documented; however, far less information
exists for weed-to-weed interactions. The hybridization between the weedy diploids Conyza canadensis (2n ¼ 18) and C.
ramosissima (2n ¼ 18) was investigated by monitoring transmission of the allele conferring resistance to N-phosphonomethyl
glycine (glyphosate). In a multivariate quantitative trait analysis, we described the phylogenic relationship of the plants, whereas
we tested seed viability to assess potential postzygotic reproductive barriers (PZRB) thus affecting the potential establishment of
hybrid populations in the wild. When inflorescences were allowed to interact freely, approximately 3% of C. ramosissima or C.
canadensis ova were fertilized by pollen of the opposing species and produced viable seeds; .95% of the ova were fertilized
under no-pollen competition conditions (emasculation). The interspecific Conyza hybrid (FH
1 ) demonstrated an intermediate
phenotype between the parents but superior resistance to glyphosate compared to the resistant C. canadensis parent. Inheritance of
H
H
glyphosate resistance in the selfed FH
1 (F2 ) followed the partially dominant nuclear, single-gene model; F1 backcrosses confirmed
successful introgression of the resistance allele to either parent. Negligible PZRB were observed in the hybrid progenies,
confirming fertility of the C. canadensis 3 C. ramosissima nothotaxa. The implications of introgressive hybridization for
herbicide resistance management and taxonomy of Conyza are discussed.
Key words: allogamy; Conyza canadensis; Conyza ramosissima; gene flow; herbicide resistance; interspecific hybridization;
shikimic acid; transgressive segregation.
diverse ‘‘weedy’’ traits. Current estimates of weed-to-weed
herbicide resistance transfer based on four genera vary from
0.15% to 85% as predicted by the frequency of resistant
individuals in the interspecific hybrid progeny (Table 1). To
address this disparity in knowledge, we investigated hybridization between the weedy diploids Conyza canadensis (L.)
Cronq. (2n ¼ 18) and C. ramosissima Cronq. (2n ¼ 18) by
monitoring transmission of the allele conferring resistance to
the herbicide N-phosphonomethyl glycine (glyphosate).
Asteraceae is one of the largest and most diverse family
within dicotyledonous plants, members of which are present in
every global environment except for aquatic habitats (Cronquist, 1980). The genus Conyza Less., represented by
approximately 60 species, is composed of annual herbaceous
plants that prosper chiefly in the tropical and subtropical
regions of the globe (Nesom, 1990). Species endemic to the
United States include the winter or summer annual forbs C.
ramosissima and C. canadensis (Cronquist, 1980).
Conzya ramosissima was originally described in Illinois as
Erigeron divaricatus (Michaux, 1803); the species was later
annexed to the genus Conyza based on the eligulate character
of its multiseriate pistillate florets (Cronquist, 1980). Both C.
ramosissima and C. canadensis are ubiquitous in disturbed
habitats and play important roles in primary ecological
successions; however, only C. canadensis is reported to reduce
yields in row crops, serve as an alternate host for diverse pests,
and limit grazing by reducing the palatability of pasture forages
(Steyermark, 1963; Cronquist, 1980). Importantly, C. canadensis has evolved resistance to amide, bipyridilium, glycine,
imidazolinone, sulfonylurea, and triazine herbicides in more
than 10 countries worldwide and thus is considered one of the
10 most important herbicide-resistant weeds (Heap, 2006).
Conzya ramosissima is found from North Dakota to Pennsylvania and south from New Mexico to Alabama in the United
Interspecific hybridization refers to the cross-fertilization
between two species that produces a fertile or infertile progeny
with phenotypic traits of both parents; this process of
interspecific gene transfer promotes genetic diversity and
genome evolution (Abbott, 1992; Barton, 2001). This natural
process has been utilized in breeding efforts to improve crop
traits (Anamthawat–Jónsson, 2001). This process, however,
may also facilitate the rapid evolution and adaptation of
introduced plant pathogens and contribute to the genetic
diversity of crop pests, and thus it may increase production
difficulties in current agroecosystems (Teal and Oostendorp,
1995; Schardl and Craven, 2003). Furthermore, interspecific
hybridization may adversely affect crop production and weed
management, as interspecific transfer of herbicide resistance
and genetically engineered genes has been documented (Owen
and Zelaya, 2005).
Much information exists regarding the transgene flow and
transfer of herbicide resistance genes between crops and their
wild relatives (Kwon and Kim, 2001; Ellstrand, 2003; Légère,
2005; Guadagnuolo et al., 2006; Reichman et al., 2006).
However, far less attention has been focused on gene flow
between weed species and the impact on dissemination of
herbicide resistance alleles or the evolution of novel taxa with
1
Manuscript received 14 May 2006; revision accepted 11 February 2007.
The authors thank J. Wendel for his valuable discussions on
evolutionary biology; M. Graham, P. Tranel, J. Gressel, and R. Hartzler
for their critical review of earlier versions of this manuscript; D. Lewis at
the Ada Hayden Herbarium (ISC) for mounting the specimens submitted
to ISC, the Botanical Research Institute of Texas (BRIT), and the New
York Botanical Garden (NY); and P. Knosby, J. Ruhland, and R. van der
Laat for assistance with greenhouse endeavors.
4 Author for correspondence (e-mail: iazelaya@iastate.edu); current
address: Syngenta Ltd., Weed Control Research, Jealott’s Hill
International Research Centre, Bracknell, Berkshire, RG42 6EY, UK
660
April 2007]
TABLE 1.
Z ELAYA
ET AL .—H YBRIDIZATION IN
C ONYZA
661
Cases of confirmed transfer of herbicide resistance gene(s) through introgressive hybridization in weeds.
Pollen donor species
Resistant to
Pollen receptor species
Transfer frequency (%)a
Reference
Avena sterilis
Amaranthus palmeri
Amaranthus hybridus
Ambrosia trifida
Helianthus annuus
fenoxaprop–P–ethyl
imazethapyr
imazethapyr
cloransulam and primisulfuron
imazethapyr and chlorimuron
Avena fatua
Amaranthus rudis
Amaranthus tuberculatus
Ambrosia artemisiifolia
Helianthus tuberosus
70
0.15–1
33–85b
40–60
30–50
Cavan et al., 1998
Wetzel et al., 1999; Franssen et al., 2001
Tranel et al., 2002; Trucco et al., 2005
Vincent and Cappadocia, 1987; Zelaya and Owen, 2006
Natali et al., 1998; Zelaya and Owen, 2006
a
b
Proportion of F1 interspecific hybrids with a confirmed herbicide-resistant phenotype.
Range includes estimates under control environments and field conditions.
States, in southern Canada, and northern Mexico, while the
more cosmopolitan C. canadensis is distributed throughout the
Americas, West Indies, Europe, and Africa (Steyermark, 1963).
Interestingly, the Conyza taxon represents the most successful
case of intercontinental colonization of the Americas to the Old
World, to the extent that C. canadensis and C. floribunda
H.B.K. are probably the most widely distributed species
throughout the world (Thébaud and Abbott, 1995; Pruski and
Sancho, 2006).
Glyphosate is one of the world’s most important herbicides;
glyphosate-resistant crops provide farmers with a simple,
economical, and effective tool to manage a diverse weed flora,
hence favoring the rapid adoption of this technology in many
United States crop production systems (Owen and Zelaya,
2005). The mechanism of glyphosate action is the competitive
inhibition with respect to the phosphate moiety of phosphoenolpyruvate in the reaction mediated by 3-phosphoshikimate
1-carboxyvinyltransferase (EPSPS; EC 2.5.1.19) (Steinrücken
and Amrhein, 1980; Holländer-Czytko and Amrhein, 1983).
The unique mode of action and limited metabolism in plants
are purported reasons for the low frequency of evolved
glyphosate resistance compared to other herbicide chemistries
(Jasieniuk, 1985; Bradshaw et al., 1997). Since the commercial
introduction of glyphosate-resistant crops in 1996 and the
accompanying ubiquitous glyphosate use in these production
systems, 12 weed species resistant to glyphosate have been
identified worldwide, including confirmation in 15 independent
C. canadensis populations within the United States and
populations in Brazil and China (Heap, 2006).
Transmission of glyphosate resistance via gene flow has been
documented between several crop and weed systems (Watrud et
al., 2004; Légère, 2005; Guadagnuolo et al., 2006; Reichman et
al., 2006). Presently, however, limited information exists
regarding the level of within-species glyphosate resistance
transfer in self-incompatible, wind-pollinated weed species as
Plantago lanceolata, Amaranthus palmeri, and A. tuberculatus
or the between-species transfer in the interfertile Ambrosia
artemisiifolia and A. trifida or Lolium rigidum and L. perenne
(Vincent and Cappadocia, 1987; Tonsor, 1990; Balfourier et al.,
1998; Wetzel et al., 1999). We previously reported that the
within-species glyphosate resistance transfer, based on the
proportion of resistant individuals in the progeny of C.
canadensis plants, ranged from 0% to 14% and 92% to 100%
under pollen competition and no-pollen competition studies,
respectively (Zelaya et al., 2004). Introgressive hybridization
(introgression) in Conyza is well documented in the European
species; however, the existence of Conyza hybrid zones in the
Americas is unknown (Knobloch, 1972; Stace, 1975; McClintock and Marshall, 1988; Thébaud and Abbott, 1995). Considering the importance of glyphosate as a global herbicide and the
pervasive nature of Conyza worldwide, an investigation was
undertaken to assess the potential for glyphosate resistance
transfer from C. canadensis to C. ramosissima through
hybridization. Postzygotic reproductive barriers and phenotypic
characterization of the interspecific Conyza hybrid (FH
1 ) were
also determined because these factors may affect the fitness and
potential of hybrids to establish as important agricultural weeds.
A naturally occurring Conyza hybrid with introgressed resistance
to glyphosate would likely have immediate and considerable
economic implications to United States agriculture. The potential
implications of hybridization on management of glyphosate
resistance and taxonomy of Conyza are discussed.
MATERIALS AND METHODS
Plant materials—The glyphosate-susceptible C. ramosissima population
(PS0 ) was collected from the Ontario Cemetery, in Ames, Iowa, in June 2003.
According to city records, no glyphosate was used on the cemetery premises.
The original C. canadensis glyphosate-resistant population (PR0 ) was that
collected by Mark VanGessel in Delaware (VanGessel, 2001). We previously
reported that glyphosate resistance in this population was governed by an
incompletely dominant, single nuclear gene (R allele) (Zelaya et al., 2004). A
stable, near-homozygous glyphosate-resistant C. canadensis population (RS2)
was isolated in the greenhouse through two cycles of recurrent selection on PR0
at the rate of 2.0 kg acid equivalent (AE) glyphosate/ha (Zelaya et al., 2004).
Complete specimen sets of PS0 , RS2, and the interspecific hybrid progenies
(explained later) were placed in the Ada Hayden Herbarium (ISC; accession
nos. 435636–435644) (Appendices S3–S9, see Supplemental Data accompanying online version of this article). Duplicate specimen sets were also
deposited at the Botanical Research Institute of Texas (BRIT) and the New
York Botanical Garden (NY).
Growth conditions—Twenty PS0 plants were transplanted from the field
into 12-cm diameter pots with a peat : perlite : loam (1 : 2 : 1) soil mix media. In
contrast, 20 RS2 individuals originated from seeds that were germinated in flats
containing the soil mix media and later transplanted to 12 cm diameter pots. All
crosses were performed in growth cabinets set at a 16-h photoperiod, 358: 258C
day : night, 70–90% relative humidity (RH), and 600 lmolm2s1 photosynthetic photon flux density (PPFD) conditions. The RS2, PS0 , and hybrid
progenies (explained next) were grown in a greenhouse set to 258–358C and
50–80% RH diurnal conditions and 208–258C and 50% RH nocturnal
conditions; natural light was supplemented with 600–1000 lmolm2s1
PPFD artificial illumination and photoperiod set to 16 h. Pots were irrigated as
needed and fertilized (Miracle Gro Excel, Scott-Sierra Horticultural Products,
Marysville, Ohio, USA) 1 mo after establishment. Achene production per
capitula and plant, in addition to the germination rate of seeds, were estimated
after crosses and in subsequent generations (explained later).
Conyza interspecific hybridization studies—The gynomonoecious C.
canadensis and C. ramosissima possess white pistillate ray florets in the
capitulum periphery and yellow perfect disk florets in the capitulum core.
Pollen release may occur prior to capitula opening; therefore, both pollen
competition and no-pollen competition studies were conducted pre-anthesis.
Response of parents to glyphosate—Prior to conducting crosses, the
phenotype of parents was assessed to verify that RS2 and PS0 were resistant and
662
A MERICAN J OURNAL
susceptible to glyphosate, respectively. The RS2 parents were treated with 2.0
kg AE of glyphosate/ha (2.3 times the recommended rate), and survival was
assessed 20 d after treatment (AT); resistant parents had no phytotoxicity
symptoms. While the phenotype of susceptible parents was not confirmed prior
to conducting crosses, Ames city records indicated that the PS0 population had
never been exposed to glyphosate. Furthermore, treatment of the progeny of
selfed PS0 plants with 0.85 kg AE of glyphosate/ha resulted in uniform plant
death (data not shown).
Pollen competition studies—Reciprocal crosses were performed between C.
canadensis and C. ramosissima plants (PS0 3 RS2, RS2 3 PS0 ) by placing the
intact inflorescences of both species inside a PQ218 DelNet bag (DelStar
Technologies, Middletown, Delaware, USA) and fastening bags at the
inflorescence base with a wire. The mesh size in DelNet bags allowed for
airflow while preventing the entry of foreign pollen. Under these conditions,
the receptive stigmas of each species could only be fertilized by pollen of either
species, thus the term ‘‘pollen competition’’. Fertilized florets were permitted to
mature on the mother plant, capitula were harvested, seeds germinated on flats,
and the emerged seedlings transplanted to 12-cm pots. Pollen competition tests
assessed the level of allogamy between the evaluated Conyza species under
controlled conditions. Under field conditions, however, the allogamy levels
estimated by this test may increase or decrease depending on factors such as
plant vicinity, insect pollinators, or wind speed and direction.
No-pollen competition studies—Disk florets of PS0 parent plants were
manually removed with forceps (emasculation). Upon stigma protrusion of the
remaining PS0 pistillate florets, fertilization was accomplished by gently rubbing
the intact capitula of the pollen donor RS2 parent plants. Crosses were
performed in one direction with PS0 and RS2 serving as the pollen receptor and
pollen donor parent, respectively. Non-emasculated capitula in PS0 plants were
removed to prevent autogamy. Approximately 50 capitula per PS0 parent plant were
emasculated and fertilized daily for 1 wk. When the pappus became visible,
capitula were harvested from plants, seeds germinated in flats, and the resultant
seedlings were transplanted to 12-cm pots. This test assessed the level of
compatibility between the studied Conyza species in the absence of pollen
competition.
Assessment of emasculation efficiency—Ten capitula per PS0 parent were
emasculated and permitted to develop on inflorescences covered with DelNet
bags, thus preventing fertilization from external pollen. Once the emasculated
capitula reached maturity, capitula were harvested, seeds were planted on flats,
and the germination was monitored. If emasculation was completely effective
in preventing autogamy in Conyza, no viable seeds would be produced.
Crossing scheme—Ten RS2 and PS0 plant pairs (families) were crossed in
isolation for the reciprocal pollen competition studies, and an identical
arrangement was used for the unidirectional no-pollen competition studies.
Estimates of hybridization were based on characterization of the first filial
generation (F1) and quantification of interspecific hybrid (FH
1 ) frequencies. One
FH
1 plant per family was then permitted to self-pollinate in isolation, and
segregation of the R allele was monitored in the resultant generation (FH
2 ); 297
individual FH
2 plants were evaluated in the segregation analysis to develop a
genetic model. Finally, introgression of the R allele was confirmed by
H
H
S
backcrossing one FH
1 plant per family to the original RS2 (BCR ) and P0 (BCS )
parents. The parents used in backcrosses originated from florets of the original
PS0 or RS2 parents that were allowed to self-pollinate and produce seeds. The
phenotype of RS2 parents used in backcrosses was confirmed as indicated
previously. The PS0 parents were confirmed susceptible by treating plants with
the sublethal glyphosate rate of 0.4 kg AE/ha; this caused 60% visual injuries
but did not kill PS0 plants, thus allowing for nondestructive verification of the
susceptible phenotype.
Postzygotic reproductive barriers—Seed viability was tested according to
the Association of Official Seed Analysts (AOSA) standard germination
procedure recommended for Asteraceae (AOSA, 2003). One hundred seeds per
H
H
H
each of the 10 parents (RS2 and PS0 ) or families (FH
1 , F2 , BCR , and BCS ) were
germinated (n ¼ 1000) on deionized-water-moistened blue blotter circles
(Anchor Paper, St. Paul, Minnesota, USA) inside plastic petri dishes in the
aforementioned growth cabinet conditions. Dishes were monitored daily for 2
wk, and germinated seeds were counted and removed; the remnant seeds were
then classified as dormant or nonviable based on tetrazolium test results
OF
B OTANY
[Vol. 94
(Moore, 1985). Seed viability experiments were repeated in time 1 wk after the
initial assessment (n ¼ 2000). For postzygotic reproductive barriers, hybrid
inviability was tested by assessing viability of FH
1 seeds, hybrid sterility tested
H
H
on the FH
2 seed, and hybrid breakdown on the viability of BCR and BCS seeds.
Characterization of interspecific hybrids—Ten randomly selected RS2
and PS0 plants or one randomly selected FH
1 plant per each of the 10 families
generated through no-pollen competition crosses were sampled to determine
quantitative traits. Five measurements (samples) were determined per plant, and
measurements were repeated in time (n ¼ 100). Rosette measurements were
taken 5–6 wk after emergence. These measurements included rosette diameter,
rosette leaf number, leaf length, leaf width, leaf shape (leaf length 4 leaf
width), leaf dentation, adaxial leaf trichomes, leaf area, leaf mass, and the
specific leaf area (SLA; leaf area 4 leaf mass). Two weeks after stem
differentiation but prior to anthesis, cauline measurements included leaf length,
leaf width, leaf shape, leaf dentation, adaxial leaf trichomes, leaf area, leaf
mass, and SLA, branch number per plant, and stem diameter at 2 cm from the
soil or at the axis of the rosette. Quantitative traits were also recorded postanthesis but prior to seed physiological maturity and included capitula length,
capitula width, capitula shape (capitula length 4 capitula width), the number of
pistillate and perfect florets, and total florets per capitulum. Only the achene
length parameter was measured at maturity. All dimensional measurements
were recorded with a digital caliper; leaf area was estimated by outlining leaves
on millimetric paper (2-mm divisions) and calculating the area therein.
Trichome numbers were determined using a stereoscope by counting the
pubescence of leaf disks (30 mm2) excised with a hole punch.
Response of interspecific Conyza hybrids to glyphosate—The recommended glyphosate rate for Conyza at the 10-cm diameter rosette stage is 0.85
kg AE/ha (Roundup UltraMAX, Monsanto, St. Louis, Missouri, USA)
(Anonymous, 2004). The resistant (R) and intermediate-resistant (IR)
phenotypes comprised those rosettes that developed 30% and 31–69% visual
injury 20 d AT, respectively, when treated with 2.0 kg AE of glyphosate/ha
(Zelaya et al., 2004). After treatment with glyphosate, both R and IR
phenotypes reached reproductive stage; however, only the R phenotype had
visual growth rates equivalent to those observed in untreated Conyza plants.
The susceptible (S) phenotype had 70% visual injury and was killed at the 2.0
kg AE/ha glyphosate rate.
Whole-plant rate response—Plants were treated with deionized water
(dH2O; control) or 0.5, 1, 2, 4, 8, or 16 times the recommended glyphosate rate
of 0.85 kg AE/ha. Treatments were applied 30 cm above the plant canopy
through an 80015-E nozzle (TeeJet Spraying Systems, Wheaton, Illinois, USA)
in a CO2-powered spray chamber (SB5–66, DeVries Manufacturing,
Hollandale, Minnesota, USA) delivering 187 L/ha at a pressure of 2.8 kg/
cm2. Treatments had four replications per rate and were repeated once in time
(n ¼ 8). Glyphosate phytotoxicity symptoms included plant stunting, leaf
chlorosis, and necrosis that developed from the meristems and leaf tips to the
rest of the plant. At 20 d AT, glyphosate efficacy was visually estimated by
comparing glyphosate-treated Conyza plants with the dH2O-treated control
plants (0% ¼ non-injured; 100% ¼ completely necrotic). Plants were then cut at
the soil surface, placed in paper bags, and dried at 808C for 48 h. Biomass was
estimated by weighing the individual Conyza sample per 12 cm diameter pot
and used to determine the glyphosate rate that inhibited plant growth by 50%
(GR50); meristems subsamples were also taken for shikimic acid determination
based on a method previously reported for Conyza (Zelaya et al., 2004).
Statistical analysis—Statistical Analysis Software (SAS, 2000) was
utilized to conduct data analyses. Seed viability tests, arranged in a complete
randomized design (CRD), were subjected to analysis of variance (ANOVA;
PROC GLM) as well as mean separation by Fisher’s least significant difference
(LSDa¼0.05) when ANOVA identified significant taxon effects. Glyphosate rate
response tests were analyzed as a randomized complete block (RCB). GR50, I50
(glyphosate rate resulting in 50% accumulation of the maximum estimable
shikimic acid), LD50 (glyphosate rate inflicting 50% mortality within the
population), jk50j (the absolute difference between two estimated GR50 values),
and v2 goodness-of-fit estimates were done as previously reported (Zelaya et
al., 2004).
Quantitative trait data were tested for normality based on the univariate
Shapiro-Wilk test and accepted if the P value for W100 was 0.05; otherwise,
the variance (r2) of the data were normalized by natural-log transformation
(Shapiro and Wilk, 1965). ANOVA was done on individual quantitative traits
April 2007]
Z ELAYA
ET AL .—H YBRIDIZATION IN
Fig. 1. Main plot: N-phosphonomethyl glycine (glyphosate) wholeplant rate response of Conyza canadensis (RS2), C. ramosissima (PS0 ), the
H
interspecific hybrid (FH
1 ), and the hybrid progeny (F2 ) 20 d after herbicide
treatment. Insert: Endogenous shikimic acid accumulation in the tissue of
treated plants. For either graph, data points represent the mean of four
replications and two experiments (n ¼ 8). Three randomly selected plants
H
per each of the 10 families (n ¼ 30) were used in FH
1 and F2 determinations. Extensions on symbols designate the standard error associated
with individual means (rM).
considering taxon and family nested within taxon as fixed and random effects,
respectively. Fisher’s LSDa¼0.05 tested for differences between taxa means.
When the normality assumption was not met, Kruskal-Wallis analyses (PROC
NPAR1WAY) were conducted, and post hoc non-parametric mean separation
was performed by Dunn’s test (Bonferroni’s method) (Conover, 1999). In
addition, PROC VARCOMP was used to partition the total phenotypic r2 into
the different ANOVA components for traits that converged to Shapiro-Wilk’s
normality assumption; these r2 components were then used to calculate the
intraclass correlation coefficient (t):
t¼
r2b
r2b
þ r2w
ð1Þ
where r2b and r2w corresponded to the r2 between and within taxon, respectively.
S
Phenotypic intermediacy of the FH
1 , as it related to the RS2 and P0 parents,
was tested by a character count procedure using trait means based on an
intermediate vs. non-intermediate one-tailed sign test (Wilson, 1992).
Multivariate normality was tested by Mardia’s kurtosis (c2) analysis (Mardia,
1970). PROC PRINCOMP was invoked to test whether the data covariance
matrix was singular. If rejected, squared Mahalanobis distances (d2t ) from the
centroid were computed for chi-square (v2) quantile–quantile (Q–Q) comparisons of multivariate normal data; 75% confidence intervals were constructed
from standard deviation (r) estimates of g(z), as indicated by Chambers et al.
(1983). Quantitative traits that converged to Mardia’s assumption were then
subjected to canonical discriminant function (CDF) analysis utilizing PROC
CANDISC (Thompson, 1984). Taxon clustering of group averages used
Mahalanobis’ distance matrix based on the unweighted pair-group method with
arithmetic mean (UPGMA) (PROC CLUSTER); concomitantly, PROC TREE
was used to display the phylogenic relationship between taxon.
RESULTS
The Conyza parental populations differed in their
response to glyphosate—Nonsignificant lack-of-fit (LOF) tests
and coefficients of determination estimates for biomass (F ¼
0.33; P ¼ 0.98; R2ðpseudoÞ ¼ 0.82) and shikimic acid (F ¼ 1.36; P ¼
0.15; R2ðpseudoÞ ¼ 0.92) confirmed suitability of the log-logistic
C ONYZA
663
Fig. 2. Main plot: Mortality observed 20 d after application of Nphosphonomethyl glycine (glyphosate) on Conyza canadensis (RS2), C.
ramosissima (PS0 ), the interspecific hybrid (FH
1 ), and the hybrid progeny
(FH
2 ). Each symbol represents the mean of four replications and two
H
experiments (n ¼ 8); the FH
1 and F2 mortality was estimated from three
randomly selected plants per each of the 10 generated families (n ¼ 30).
Solid and broken lines represent the mortality estimated by PROBIT and
that expected considering partially dominant monofactorial inheritance in
S
the FH
2 (Tabashnik, 1991), respectively. Insert: Biomass of untreated P0
(white bar), FH
1 (gray bar), and RS2 (black bar) rosettes at the 10–12-cm
diameter stage. Bars represent the mean of 10 randomly selected rosettes
and determination repeated in time (n ¼ 20). Letters above bars represent
minimum statistical differences according to Fisher’s LSD for comparisons between Conyza taxa means (LSDa¼0.05 ¼ 0.24 g); extensions on bars
or symbols represent the standard error associated with individual means
(rM).
model for describing plant response as a function of increasing
glyphosate rates. PS0 was killed at 0.85 kg AE of glyphosate/ha
(Figs. 1 and 2). Compared to PS0 , RS2 had a three fold and seven
fold increase in estimated GR50 and LD50 values, respectively
(Table 2). In addition, when comparing GR50 estimates for PS0
and RS2, a statistically significant jk50j value of 1.60 (Fobs ¼
1.77; P , 0.01) was observed, reaffirming that the parental
populations differed in their response to glyphosate. Shikimic
acid determinations, which indirectly estimate the level of
EPSPS inhibition by glyphosate (Harring et al., 1998),
suggested that twice the rate of glyphosate was required to
inhibit 50% of EPSPS in RS2 plants compared to PS0 (Table 2).
Untreated Conyza plants contained 7–14 lmol of shikimic
acid/g dry mass, which increased sigmoidally to a maximum of
150 lmol at the two highest glyphosate rates (Fig. 1).
Concurrently, endogenous shikimic acid levels correlated
negatively with plant biomass and positively with visual injury,
thus further supporting the differential response to glyphosate
observed between the parental populations (Table 2).
Levels of interspecific hybridization—Pistillate florets in
emasculated PS0 capitula contained ,1% viable seeds in the
absence of pollen (Table 3). Previously, we proposed that
fertilization of pistillate florets under these conditions probably
originated from pollen of perfect florets released prior to
anthesis or from pollen of perfect florets incompletely excised
during emasculation (Zelaya et al., 2004). Under greenhouse
664
A MERICAN J OURNAL
OF
B OTANY
[Vol. 94
H
TABLE 2. Response of Conyza canadensis (RS2), C. ramosissima (PS0 ), the interspecific hybrid (FH
1 ), and the hybrid progeny (F2 ) to glyphosate. Numbers
in parentheses designate the 95% lower and upper confidence intervals (GR50; I50) or fiducial limits (LD50) for the preceding estimated parameter. The
extent of association between shikimic acid levels and biomass or visual injury is reported according to Spearman’s correlation (r2) analysis.
GR50a
Taxon
PS0
RS2
e
FH
1
e
FH
2
0.69
2.29
3.58
3.93
LD50b
(0.38–0.99)
(1.51–3.07)
(2.42–4.74)
(2.27–5.59)
0.68
4.79
6.23
3.16
I50c
(0.35–1.04)
(3.29–7.02)
(4.21–9.20)
(1.86–5.68)
1.84
3.80
4.08
3.27
(1.59–2.10)
(3.44–4.16)
(3.63–4.53)
(2.93–3.61)
r2 (biomass)d
r2 (injury)d
0.75
0.81
0.80
0.61
0.83
0.88
0.87
0.65
a
Glyphosate rate in kg acid equivalents (AE)/ha that reduced biomass accumulation by 50%.
Glyphosate rate in kg AE/ha that inflicted 50% mortality in the population.
c Glyphosate rate in kg AE/ha that resulted in 50% accumulation of the maximum estimable shikimic acid.
d Spearman’s statistics test the H that r ¼ 0. Since the probability of a value greater than jrj was 0.001 for all estimates, the H was rejected.
0
0
e Six randomly selected rosettes per each of the 10 generated families were used to assess the rate response of the FH and FH populations to glyphosate.
1
2
b
environments, approximately 60% of the total achenes on PS0
plants produced viable seeds (Fig. 3). Therefore, we estimated
that the emasculation method was 98% effective at preventing
self-fertilization in PS0 .
Interspecific hybridization under pollen competition conditions ranged from 0% to 7% in the RS2 to PS0 crosses and from
0% to 9% in the reciprocal PS0 to RS2 crosses, although five in
10 PS0 and two in 10 RS2 families contained no identifiable
Conyza hybrids (data not shown). In contrast, high hybridization levels (98.2%) were observed under no-pollen competition
TABLE 3. Frequencies of hybridization in the Conyza canadensis 3 C.
ramosissima cross. Emasculation efficiency was tested by evaluating
the germination of seeds that developed in the absence of pollen in 10
emasculated capitula per each of 10 C. ramosissima (PS0 ) parents
evaluated. Crosses between C. canadensis (RS2) and PS0 were
conducted under pollen and no-pollen competition conditions.
Hybridization was estimated from the frequency of Conyza plants
with a susceptible (S), resistant (R), or hybrid (H) phenotype within
the first filial generation (FH
1 ); data represent the sum of the 10
generated families (see Materials and Methods).
conditions when RS2 served as pollen donor to PS0 (Table 3).
The observed levels of glyphosate-susceptible individuals
(,2%) in the no-pollen competition studies mirrored those of
emasculation efficiency estimates (2%), suggesting that some
level of self-fertilization occurred prior to the observed
anthesis. Collectively, these data suggested that while the
levels of hybridization were relatively low (2.6% to 4.1%)
under pollen competition conditions, C. canadensis and C.
ramosissima are genetically compatible (.95%), and thus the
potential for interspecific hybridization in the field exists.
Anecdotal herbaria records of ‘‘off-type’’ Conyza sp. suggest
that hybridization does occur, albeit infrequently, in natural
plant communities.
Phenotypic variance of parents and hybrid progenies—
Differences among families within taxa were significant for
Assessment of emasculation efficiency in C. ramosissimaa
Cross type
Total no. of
planted achenes
Not crossed
1030
Emerged
C. ramosissima
plants
4
Efficiency of
emasculation (%)
99.6
Estimate of hybridization in Conyza under no-pollen competitionb
Progeny with
an S phenotype
Cross type
RS2 donor
PS0 receptor
5
Progeny with
an H phenotypec
Estimate of
hybridization (%)
267
98.2
Estimate of hybridization in Conyza under pollen competitiond
Progeny with
an S phenotype
Cross type
RS2 donor
PS0 receptor
257
Progeny with
an R phenotype
PS0 donor
RS2 receptor
280
Progeny with
an H phenotype
7
Progeny with
an H phenotype
12
Estimate of
hybridization (%)
2.6
Estimate of
hybridization (%)
4.1
a Capitula were emasculated pre-anthesis and covered with a DelNet
bag; ova therefore matured in the absence of pollen.
b PS capitula were emasculated pre-anthesis and fertilized with RS
2
0
pollen.
c Plants with a hybrid (H) phenotype demonstrated an intermediate
phenotype between both the C. canadensis and C. ramosissima parents;
the H phenotype was also resistant to glyphosate.
d RS and PS inflorescences were covered with a DelNet bag at anthesis
2
0
and permitted to cross freely (see Materials and Methods).
Fig. 3. Proportion of germinated (gray bars), dormant (dark gray bars),
and nonviable (white bars) seeds among the Conyza canadensis (RS2), C.
H
ramosissima (PS0 ), interspecific hybrid (FH
1 ), hybrid progeny (F2 ), and the
H
H
S
FH
backcross
to
RS
(BC
)
or
P
(BC
).
Bars
represent
the
mean
of two
2
1
0
R
S
experiments comprised of 100 seeds per each of the 10 parents (RS2 and
H
H
H
PS0 ) or 10 families (FH
1 , F2 , BCR , and BCS ) generated in the no-pollen
competition studies (n ¼ 2000); extensions above bars designate the
standard error associated with individual means (rM). Letters above bars
represent minimum statistical differences according to Fisher’s LSD for
comparisons within germinated (LSDa¼0.05 ¼ 12%), dormant (LSDa¼0.05
¼ 7%), and nonviable (LSDa¼0.05 ¼ 14%) seeds.
April 2007]
Z ELAYA
ET AL .—H YBRIDIZATION IN
C ONYZA
665
TABLE 4. Partitioning of the total phenotypic variance (r2) into taxon and family effects for the evaluated quantitative traits. Numbers represent F values
and probability (in parentheses) estimates for the analysis of variance (ANOVA). The total r2 between and within the Conyza taxon was utilized to
calculate the intraclass correlation coefficients (t) for each trait; underlined coefficients correspond to traits with a significant family effect (P , 0.05)
in the ANOVA for the individual taxa. Nomenclature: Conyza canadensis (RS2), C. ramosissima (PS0 ), and the interspecific hybrid (FH
1 ).
Source of varianceb
Stagea
Rosette
Cauline
Anthesis
Senescence
Quantitative trait
Diameter (mm)
Leaf number
Leaf length (mm)
Leaf width (mm)
Leaf shape
Leaf dentationd
Leaf trichomes (cm2)
Leaf area (mm2)
Leaf mass (mg)
SLAf (mm2/mg)
Stem diameter (mm)
Branch numberd
Leaf length (mm)
Leaf width (mm)
Leaf shape
Leaf dentationd
Leaf trichomes (cm2)
Leaf area (mm2)
Leaf mass (mg)
SLAf (mm2/mg)
Capitula length (mm)
Capitula width (mm)
Capitulum shape
Pistillate florets
Perfect florets
Total florets
Achene length (mm)
Between taxa
85.26
365.34
40.15
142.91
30.64
314.24
56.14
106.13
132.37
302.88
508.08
769.94
197.69
231.86
13.19
367.94
294.88
189.96
202.89
19.85
43.42
210.59
196.55
494.67
13.74
252.9
89.01
(,0.0001)
(,0.0001)
(,0.0001)
(,0.0001)
(,0.0001)
(,0.0001)
(,0.0001)
(,0.0001)
(,0.0001)
(,0.0001)
(,0.0001)
(,0.0001)
(,0.0001)
(,0.0001)
(,0.0003)
(,0.0001)
(,0.0001)
(,0.0001)
(,0.0001)
(,0.0001)
(,0.0001)
(,0.0001)
(,0.0001)
(,0.0001)
(,0.0002)
(,0.0001)
(,0.0001)
Intraclass correlation coefficient (t)c
Family within taxa
2.33
0.4
2.45
2.34
3.8
1.35
3.5
1.56
1.75
1.85
2.01
1.21
3.79
2.09
2.7
0.65
8.13
2.92
1.82
1.55
1.56
3.16
2.17
1.46
0.5
0.84
5.27
(0.0151)
(0.9364)
(0.0109)
(0.015)
(0.0002)
(0.2128)
(0.0004)
(0.1282)
(0.077)
(0.0591)
(0.0386)
(0.291)
(0.0002)
(0.0306)
(0.0051)
(0.7537)
(,0.0001)
(0.0026)
(0.0645)
(0.129)
(0.1279)
(0.0012)
(0.0246)
(0.1612)
(0.8734)
(0.5814)
(,0.0001)
RS2
FH
1
PS0
0.98
0.00
1.00
1.00
1.00
NPe
0.93
1.00
1.00
0.96
1.00
NP
1.00
0.65
1.00
NP
1.00
0.72
0.29
0.00
0.00
1.00
0.96
1.00
0.75
1.00
1.00
1.00
0.00
0.95
1.00
1.00
NP
1.00
0.00
0.00
1.00
1.00
NP
1.00
1.00
1.00
NP
0.92
1.00
1.00
1.00
0.96
1.00
0.99
0.00
1.00
0.00
0.71
0.99
0.00
1.00
1.00
1.00
NP
0.97
1.00
1.00
0.87
0.91
NP
0.94
1.00
0.00
NP
1.00
0.86
0.79
0.19
1.00
1.00
1.00
0.00
1.00
0.95
1.00
a Phenological stages: rosette, determined 5–6 wk after seedling emergence; cauline, 2 wk post-stem elongation but prior to anthesis; anthesis, postanthesis but prior to maturation; senescence, post-physiological maturity of seeds but prior to the anemochorous stage.
b Mean squares were derived from type I sums of squares. F values for the random family effect were calculated by dividing mean squares by the
residual error; per contra, the fixed taxon effect used the family within taxon mean squares.
c r2 components were estimated by the restricted maximum likelihood method (REML) and were then used to estimate the intraclass correlation
coefficient (t) according to Eq. 1.
d Data neglected to converge to Shapiro-Wilk’s test of normality; thus r2 was normalized through natural-log transformation prior to ANOVA.
e Nonparametric (NP); normality H was rejected; thus ANOVA was restricted to elucidating differences among taxa.
0
f Specific leaf area (SLA) calculated the proportion of the total leaf area per unit fresh mass.
approximately half of the quantitative traits evaluated (Table
4). Maternal family effects were significant (P , 0.05) in nine
of 27 and 10 of 27 quantitative traits within the PS0 and FH
1
populations, respectively, and in approximately half of the RS2
traits evaluated. Only the characters rosette leaf length, cauline
leaf trichomes, and achene length had significant maternal
family effects across RS2, PS0 , and FH
1 (Table 4). Intraclass
correlation coefficients (t) were generally (85%) different from
zero and most (72%) were greater than 0.90 (Table 4).
Combined with the ANOVA, data suggested that most of the
observed phenotypic variance (r2) was accounted for differences between the studied Conyza taxa. Furthermore, we
assumed that this r2 was primarily attributable to genetic
differences between the studied taxa rather than to the
interaction with the environment because the Conyza populations developed under controlled greenhouse conditions
(Appendix S1, see Supplemental Data accompanying online
version of this article).
Phenotype of FH
1 —Statistical differences (P , 0.05) between
the studied Conyza taxa were obtained for all 27 quantitative
traits evaluated (Table 5). RS2 produced wider rosettes with
fewer leaves than PS0 . Furthermore, both RS2 leaf dimorphisms
were longer and wider and possessed more dentations than PS0
(Table 5). In contrast, PS0 was profusely branched, had narrower
stems, and developed more pubescent leaves with greater
density (SLA) than RS2. Capitula dimensions and the number
of individual and total florets were greater in RS2; however, PS0
produced longer achenes (Table 5).
To investigate whether the phenotype of FH
1 plants was
intermediate to both parents, a character count one-tailed sign
test was conducted. Rosette leaf number, cauline branch
number, cauline SLA, and capitula length FH
1 measurements
tended to exceed those of either parent; however, only
capitulum shape was significantly different (Table 5). Hence,
five of the 27 evaluated quantitative traits did not meet the
character count assumption that FH
1 was intermediate to both
parents (D ¼ 22). Because conditions for the DeMoivre–
Laplace theorem were met (np and nq 5), normal
approximation was used for probability estimation rather than
a binomial distribution. The calculated one-tailed sign test
value for phenotypic intermediacy was significant (zþ ¼ 3.27; P
666
A MERICAN J OURNAL
OF
B OTANY
[Vol. 94
TABLE 5. Comparison of quantitative traits among Conyza canadensis (RS2), C. ramosissima (PS0 ), and the interspecific hybrid (FH
1 ). Values represent the
mean of five observations per each of 10 sampled plants and assessments repeated in time (n ¼ 100); numbers in parentheses designate the standard
error associated with means (rM). A one-tailed sign test of intermediate vs. non-intermediate was used to test the hypothesis of phenotypic FH
1
intermediacy.
Stagea
Rosette
Cauline
Anthesis
Senescence
Quantitative trait
Diameter (mm)
Leaf number
Leaf length (mm)
Leaf width (mm)
Leaf shape
Leaf dentationc
Leaf trichomes (cm2)
Leaf area (mm2)
Leaf mass (mg)
SLAd (mm2/mg)
Stem diameter (mm)
Branch numberc
Leaf length (mm)
Leaf width (mm)
Leaf shape
Leaf dentationc
Leaf trichomes (cm2)
Leaf area (mm2)
Leaf mass (mg)
SLAd (mm2/mg)
Capitula length (mm)
Capitula width (mm)
Capitulum shape
Pistillate florets
Perfect florets
Total florets
Achene length (mm)
FH
1
RS2
107.09
25.53
35.51
9.85
3.66
3.93
1410
157.11
71.08
2.22
9.01
2.07
74.71
7.81
9.89
3.9
444
265.84
50.58
5.53
4.90
2.01
2.45
36.90
15.68
52.58
0.87
(2.00)
(0.67)
(1.03)
(0.27)
(0.09)
(0.14)
(55.56)
(7.89)
(3.61)
(0.01)
(0.23)
(0.24)
(1.76)
(0.24)
(0.17)
(0.16)
(17.50)
(11.10)
(2.73)
(0.07)
(0.03)
(0.02)
(0.02)
(0.54)
(0.33)
(0.77)
(0.01)
76.79
87.57
29.48
4.96
6.59
1.39
2754
67.17
21.57
3.14
2.66
21.34
42.41
3.30
13.30
0.61
874
83.34
13.53
6.27
4.95
1.60
3.09
17.62
14.44
32.06
1.13
(1.98)
(2.44)
(0.57)
(0.17)
(0.25)
(0.12)
(89.86)
(2.26)
(0.73)
(0.03)
(0.04)
(1.01)
(1.18)
(0.08)
(0.37)
(0.07)
(29.06)
(3.57)
(0.64)
(0.08)
(0.03)
(0.01)
(0.01)
(0.26)
(0.18)
(0.38)
(0.02)
PS0
40.18
83.42
17.82
2.88
6.50
0.35
3491
26.88
7.76
3.52
0.99
18.79
31.63
2.38
14.01
0.51
3237
40.19
6.73
6.12
4.50
1.58
2.87
12.49
12.84
25.33
1.26
(0.72)
(2.51)
(0.35)
(0.07)
(0.18)
(0.06)
(69.82)
(0.83)
(0.27)
(0.03)
(0.02)
(1.00)
(0.65)
(0.05)
(0.43)
(0.06)
(58.80)
(1.28)
(0.25)
(0.07)
(0.04)
(0.02)
(0.03)
(0.27)
(0.26)
(0.41)
(0.01)
MSDb
FH
1 different from parent
Phenotypic intermediacy
10.78
5.39
4.22
0.89
0.89
0.43
418
19.23
8.60
0.11
0.59
2.40
4.74
0.57
1.80
0.38
260
25.82
4.92
0.26
0.11
0.05
0.07
1.72
1.14
2.65
0.06
both
RS2
both
both
RS2
both
both
both
both
both
both
RS2
both
both
RS2
RS2
both
both
both
RS2
PS0
RS2
both
both
both
both
both
þ
þ
þ
þ
þ
þ
þ
þ
þ
þ
þ
þ
þ
þ
þ
þ
þ
þ
þ
þ
þ
þ
27 : 22
a Phenological stages: rosette, determined 5–6 wk after seedling emergence; cauline, 2-wk post-stem elongation but prior to anthesis; anthesis, postanthesis but prior to maturation; senescence, post-physiological maturity of seeds but prior to the anemochorous stage.
b The minimum significant difference (MSD) was estimated when ANOVA identified significant taxon effects. Mean separation was conducted by
Fisher’s least significant difference (LSDa¼0.05) when data converged to Shapiro-Wilk’s assumption of normality. Otherwise, significance was estimated
by Kruskal-Wallis’ analysis, and post hoc mean separation was conducted by Dunn’s test (Bonferroni’s method).
c The r2 of nonparametric traits was normalized by natural-log transformation prior to ANOVA.
d Specific leaf area (SLA) calculated the proportion of the total leaf area per unit fresh mass.
¼ 0.001) and thus prompted rejection of the null hypothesis
(H0) for equal sign proportions of the evaluated quantitative
traits. The collective phenotypic data therefore suggested that
FH
1 represented an intermediate between both parents.
Intermediacy of FH
1 —Multivariate normality analysis determined that only 17 of the 27 evaluated morphometric traits
converged to Mardia’s statistics (c2 ¼ 328.2; P ¼ 0.08) and thus
were combined for multivariate discriminant analysis; the
pistillate florets trait was eliminated from the analysis as
inclusion instigated a singular covariance matrix. The H0 that
data arose from populations with a common distribution was
tested by Q–Q plots; linearity of data points along the expected
mean vectors and allocation within the estimated 75%
confidence intervals suggested nondeparture from normality
(Fig. 4). Most (81%) of intraclass correlation coefficients (t) in
the multivariate normally distributed morphometric traits
possessed values greater than 0.70, hence suggesting negligible
interference of correlations among families that could potentially distort interpretation of the canonical discriminant
function (CDF) analysis (Table 4).
Univariate statistics for differences among taxa means were
significant (P , 0.0001) for all traits, as were Wilks’ exact
multivariate statistics for comparison among taxa (k ¼ 0.004; F
¼ 229.9; P , 0.0001). The CDF analysis estimated that the first
(Can1) and second (Can2) canonical variates accounted for
97% and 82% of the total quantitative trait r2, respectively.
Taxa clustered in discrete sections within the canonical graph;
PS0 (6.8) and the FH
1 (1.3) were positive along the Can1 axis,
while RS2 was diametrically negative (8.2). Fewer taxa
divergence was obtained along the Can2 axis (Fig. 4; Appendix
S2, see Supplemental Data accompanying online version of
this article).
Taxa separation was also discernable by group averages
based on the UPGMA. Divergence in Mahalanobis distances
was greatest among the RS2 and PS0 parents (d2t ¼ 226.5; F ¼
630.2; P , 0.0001), and approximately half and one-fourth
that square distance was estimated for comparisons between
2
H
the FH
1 and RS2 (dt ¼ 107.5; F ¼ 299.1; P , 0.0001) and F1
2
S
and P0 (dt ¼ 53.7; F ¼ 149.4; P , 0.0001), respectively.
Hierarchical clustering formed a dendrogram with a common
root node for the taxon (Fig. 4). The PS0 and FH
1 leaves clustered
within a single branch, whereas RS2 clustered to the common
parent node. While the character count procedure lent credence
to the thesis of an intermediate hybrid FH
1 phenotype, CDF
analysis and UPGMA strongly emphasized that the FH
1 was
more similar to PS0 than to RS2.
April 2007]
Z ELAYA
ET AL .—H YBRIDIZATION IN
C ONYZA
667
Fig. 4. (A) Multivariate discriminant function analysis of canonical (CDF) variates 1 and 2 for 17 morphometric traits that converged to Mardia’s
multivariate normality assumption. Data points represent the score of individuals (n ¼ 100) within the Conyza canadensis (RS2), C. ramosissima (PS0 ), and
2
interspecific hybrid (FH
1 ) populations. Inserts: Chi-square (v ) quantile–quantile (Q–Q) plots of taxon observations. The continuous and dashed lines
represent expected mean vectors and 75% confidence intervals for the data, respectively. (B) Clustering analysis of Mahalanobis’ distance matrix based on
taxon averages by the unweighted pair-group method with arithmetic mean (UPGMA).
668
A MERICAN J OURNAL
Postzygotic reproductive barriers—Partial correlation estimates (r2 ¼ 0.76; P ¼ 0.001) associated with the sums of
squares and crossproducts (SSCP) matrix suggested a strong
relationship between the seed viability experiments repeated in
time. Univariate (F ¼ 0.14; P ¼ 0.71) and multivariate (Wilks’
k ¼ 0.99; P ¼ 0.71) tests for the between-time effects were not
significant, therefore no difference in experiments repeated in
time was inferred and data for the seed viability experiments
were combined in further analyzes. Previous research reported
near 100% C. canadensis germination under light and constant
288C conditions (Shontz and Oosting, 1970). Viability of RS2
seeds under our conditions ranged from 17% to 71% within
families with a mean of 46%, which was not statistically
different (LSDa¼0.05 ¼ 12%) from the 57% mean viability
estimated for PS0 seeds (Fig. 3).
Evidence for hybrid inviability was apparent in the 23%
increase in nonviable FH
1 seeds (Fig. 3). No-pollen competition
tests estimated hybridization levels of 98% in the viable hybrid
zygotes within families (Table 3). Viability tests nonetheless
confirmed that approximately half of the available ova in PS0
were not fertilized by RS2 pollen or ova were fertilized but
aborted prematurely, an argument that some level of genetic
incompatibility existed among the Conyza parents (Fig. 3).
While physical damage of PS0 capitula during emasculation could
have contributed to the inviability of hybrid zygotes, epistatic,
homeotic transformations, or interactions of complementary
genes may have also resulted in embryo abortion of FH
1 zygotes.
The presence of a hybrid sterility reproductive barrier was
disregarded because the FH
2 had embryos with approximately
twice the level of viability of their progenitor (FH
1 ) (Fig.
viability
were
significantly
less
than for
3). Estimates of FH
2
either RS2 or PS0 , an indication that some level of self-infertility
remained within hybrids and evidence that the inviable
FH
1 zygotes probably arose from genetic incompatibilities rather
than the physical stress of emasculation (Fig. 3). Backcrosses of
S
FH
1 to RS2 or P0 produced a viable progeny, 21% and 25%,
respectively, confirming successful introgression of the R
allele to either of the evaluated Conyza taxa through the
intermediate hybrid (FH
1 ). Establishment of hybrid interspecific populations in the environment is often limited by
postzygotic reproductive barriers (Barton, 2001). While the
S
germination of FH
1 seeds was less than that of the RS2 or P0
parents, the collective viability data established that the
Conyza hybrid was fertile and probably capable of establishment in natural environments.
Transgressive segregation in FH
2 —Most hybrid progenies
demonstrate marked traits of transgression (Rieseberg et al.,
1999). The majority of FH
2 plants appeared similar to either the
canadensis or ramosissima epithets. However, specific characteristics attributable to transgression were observed, ranging
from glabrous to pubescent leaves and stems, oblanceolate to
subulate leaves with dentated to strait margins, green to
purplish midribs, irregular arrangements of lateral leaf
nervures, and profusely branched single-stem plants. Generally, flower morphology was a preserved trait among FH
2
individuals. Transgressive segregation was also discernable in
whole-plant rate responses to glyphosate. Compared to the
biomass accumulated by RS2 (r2 ¼ 0.67), PS0 (r2 ¼ 0.42), or
H
2
the FH
1 (r ¼ 0.87), F2 rosettes demonstrated greater variance
(r2 ¼ 1.17) in response to glyphosate. Similarly, greater
variations in the level of endogenous shikimic acid accumulation and visual injury were observed in the FH
2 (data not
OF
B OTANY
[Vol. 94
shown). The FH
2 also was characterized by the unexpected
death of plants (,5%) at the early rosette stages. We theorize
that these low lethal frequencies reflect possible deleterious
homeotic transformations or epistatic interactions in FH
2 plants.
Phenotype of backcrosses—Visual observations confirmed
H
H
that the progeny of FH
1 backcrosses to parents (BCR and BCS )
were phenotypically similar to either C. canadensis or C.
ramosissima. For example, BCH
R resembled RS2 at the rosette
and cauline stages, except that the plant developed lateral
branches below the center-main stem. In addition, BCH
R
produced serrated and nonserrated leaves, while RS2 produced
only the serrated dimorphism. BCH
S possessed a well-defined
axis that was absent in PS0 and generated slightly larger
leaves and thicker stems than PS0 . BCH
S was also precocious
compared to BCH
R , completing the reproductive cycle in 2–3 mo.
Inheritance of glyphosate resistance—Application of 0.40
kg AE of glyphosate/ha to PS0 resulted in visual injury levels of
30–60% and effectively delayed plant development compared
to the dH2O-treated Conyza plants, although treated plants
recovered from glyphosate injuries within 2–4 wk and reached
reproductive stage. Rates of 0.85 kg AE of glyphosate/ha
resulted in uniform kill of PS0 or the selfed progeny of PS0 ,
suggesting that the population was near-homozygous susceptible to glyphosate. We previously confirmed that RS2
represented a near-homozygous glyphosate-resistant population (Zelaya et al., 2004).
H
Overdominance of FH
1 —The F1 produced larger rosettes
S
compared to P0 and more and denser (SLA) leaves than RS2
(Table 5). The difference in these two leaf parameters resulted
in an apparent heterotic response of FH
1 recorded at the 10–12cm diameter rosettes stage (Fig. 2). Compared to PS0 and RS2,
greater glyphosate rates were required to reduce biomass
accumulation or cause mortality in FH
1 (Table 2; Figs. 1 and 2).
This divergence in response to glyphosate was confirmed by
jk50j values for comparisons of FH
1 and RS2 (Fobs ¼ 1.74; P ,
S
and
P
(F
¼
1.76;
P , 0.01). Not only did FH
0.01) or FH
obs
1
0
1
rosettes demonstrate a vigorous growth that probably required
greater glyphosate rates to inhibit, but leaves produced more
trichomes than RS2, which could have hindered glyphosate
H
absorption into FH
1 plants (Table 5). The response of F1 to
glyphosate therefore failed to obey the additive, dominant, or
hybrid susceptibility models that describe resistance in hybrid
populations; rather, the response was best explained by the
hybrid resistance hypothesis, which predicts greater resistance
in hybrid populations compared to their progenitors (Fritz et
al., 1994).
Segregation of the R allele—Glyphosate resistance in RS2 is
conferred by the incompletely dominant, nuclear R allele
(Zelaya et al., 2004). This model for glyphosate resistance in
the Conyza hybrid was tested by monitoring the segregation
H
H
ratios of FH
2 , BCR , and BCS to glyphosate. Efficacy tests at 20
d AT with 2.0 kg AE of glyphosate/ha, a rate that differentiated
R and S phenotypes in the parental populations, identified three
distinct segregates in FH
2 families—R, IR, and S phenotypes.
Exact goodness-of-fit (GOF) tests based on the H0 that the
observed segregation ratios followed a 1 : 2 : 1 Mendelian
distribution provided nonsignificant v2 values for all FH
2
families, substantiating appropriateness of the partially dominant monogenic model (Table 6). Concomitantly, homogeneity
April 2007]
Z ELAYA
ET AL .—H YBRIDIZATION IN
C ONYZA
669
H
TABLE 6. Segregation for glyphosate resistance of the R allele in the progeny (FH
2 ) of one interspecific hybrid plant (F1 ) per family allowed to selfH
H
pollinate in isolation. The 10 F2 families evaluated arose from a single F1 plant isolated from the Conzya canadensis (RS2) to C. ramosissima (PS0 ) noH
H
S
pollen competition cross. For back-crosses, FH
1 served as pollen donor to RS2 (BCR ) or P0 (BCS ).
Observed phenotypea
Cross type
Selfed
FH
1
Selfed FH
1
BCH
R
BCH
S
Expectedb
Family no.
R
IR
S
Total
R : IR : S
v2
P . v2
1
2
3
4
5
6
7
8
9
10
Combinedc
Combinedc
Combinedc
4
12
10
11
6
3
8
5
7
10
76
88
—
16
15
17
15
18
13
23
14
17
16
164
70
83
7
7
5
5
2
5
4
8
9
5
57
—
110
27
34
32
31
26
21
35
27
33
31
297
158
193
6.75 : 13.5 : 6.75
8.5 : 17 : 8.5
8 : 16 : 8
7.75 : 15.5 : 7.75
6.5 : 13 : 6.5
5.25 : 10.5 : 5.25
8.75 : 17.5 : 8.75
6.75 : 13.5 : 6.75
8.25 : 16.5 : 8.25
7.75 : 15.5 : 7.75
74.25 : 148.5 : 74.25
79 : 79 : 0
0 : 96.5 : 96.5
1.59
1.94
1.69
2.35
5.08
1.57
4.37
0.70
0.27
1.64
5.67
2.05
3.78
0.45
0.38
0.43
0.31
0.08
0.46
0.11
0.70
0.87
0.44
0.06
0.15
0.05
a Susceptible (S) comprised individuals that developed 70% visual injury 20 d after treatment of 2.0 kg acid equivalents of glyphosate/ha; in contrast,
resistant (R) and intermediate-resistant (IR) developed 30% and 31–69% visual injury at this same rate, respectively. Both R and IR phenotypes reached
reproductive stage after treatment with glyphosate.
b Expected Mendelian ratios (1 : 2 : 1) for the single allele, incompletely dominant model.
c Homogeneity v2 test was nonsignificant, thus data among families were combined for the v2 goodness-of-fit test. Combined FH families, v2 ¼ 5.72, P
2
H
2
2
¼ 0.77; combined BCH
R families, v ¼ 4.40, P ¼ 0.88; combined BCS families, v ¼ 5.81, P ¼ 0.76.
analysis (v2 ¼ 5.72; P ¼ 0.77) confirmed that the combined FH
2
data converged to the 1 : 2 : 1 genetic model.
H
H
Backcrossing of F1 to RS2 (BCR ) and subsequent progeny
treatment with 2.0 kg AE of glyphosate/ha identified only R
and IR phenotypes (Table 6). The observed segregation ratios
2
among BCH
R families were homogenous (v ¼ 4.40; P ¼ 0.88),
and the collective data were consistent with the expected 1 : 1
H
S
(R : IR) ratio. Similarly, analysis of FH
1 backcrosses to P0 (BCS )
2
resulted in homogenous IR and S segregation ratios (v ¼ 5.81;
P ¼ 0.76) that obeyed the partially dominant model (Table 6).
Because the backcross data followed the monofactorial model
of inheritance, results corroborated our previous assertion that
RS2 and PS0 represented near-homozygous lineages with regard
to their response to glyphosate.
Further substantiation of the incompletely dominant, monogenic model was obtained graphically from the pattern of
observed FH
2 mortality by comparing to that mortality expected
as suggested by Tabashnik (1991): Yv ¼ WR (0.25) þ WIR (0.50)
þ WS (0.25). Three distinct response phases were predicted
based on partially dominant, monofactorial inheritance (Fig. 2).
No mortality was recorded from 0.0 to 0.42 kg AE/ha,
suggesting that both homozygous (RR and rr) and the
heterozygous (Rr) genotypes were present at these glyphosate
rates. A second segment was discernable at glyphosate rates of
0.85–3.38 kg AE/ha, which corresponded to approximately
one-fourth (23–37%) mortality of the putative homozygous
susceptible genotype (Fig. 2). The heterozygous (71%) and
homozygous (100%) resistant genotypes were controlled at
glyphosate rates of 6.77 and 13.54 kg AE/ha, respectively.
DISCUSSION
Hybridization of Conyza in nature—The phylogenic
boundaries in Conyza are not clearly understood and
considerable phenotypic variation has been reported, particularly in response to adverse environmental stimuli (Nesom,
1990). We initiated a project to assess potential hybridization
of Conyza species and better understand the phylogenic
relationship between C. canadensis and C. ramosissima, two
weedy species in United States agroecosystems despite nothing
described in the literature. Our work herein suggests that the
studied taxa are genetically compatible, capable of transferring
the R allele, and producing interspecific hybrid progenies that
are vigorous and fertile. Given that C. canadensis has become a
major economic weed problem in the Midwestern United
States and the apparent vigor of the hybrid identified in this
research, the ramifications of an interspecific Conyza hybrid
that has resistance to glyphosate are potentially significant.
Glyphosate-resistant crop systems are suggested to be simple
and without great environmental consequences. However, we
have demonstrated that there are major ecological and
economic consequences from these presumed simple systems.
New weeds typically evolve over a long period of time, and
existing weeds tend to adapt to new agroecosystems slowly.
We propose that if hybridization of new taxa with glyphosate
resistance as a semi-dominant trait can occur with relative ease,
the current agroecosystem is at considerable jeopardy. While
the hybrid demonstrated in this research has not been
specifically identified in the field, we have illustrated the
potential for the occurrence.
We speculate that the relatively high genetic compatibility
among the studied taxa is associated with the common diploidy
structure (2n ¼ 18) that allows for successful chromosome
pairing during meiosis. Both C. canadensis and C. ramosissima represent sibling species as ascertained by nrDNA internal
transcribed spacers (ITS) analysis, which clustered both taxa in
a single branch within group VI of the Erigeron and allied
Asteraceae cladogram (Noyes, 2000). This phylogenic analysis
estimated a recent speciation event between the ramosissima
and canadensis epithets and provides further support to our
thesis of genetic compatibility between the species.
The likelihood of native hybridization in Conyza is probably
low given the enclosed involucre arrangement in Conyza and
the autogamous nature of the genus. Entomophily was reported
in C. canadensis, although the relative importance to
670
A MERICAN J OURNAL
interspecific gene transfer remains unknown (Weaver, 2001).
To our knowledge, there is no prior documentation of
hybridization between C. canadensis and C. ramosissima.
However, since early descriptions, plants with common traits
between both species have been cited (Michaux, 1803).
Voss (1996) reported depauperate C. canadensis biotypes in
the upper peninsula of Michigan with glabrous to pubescent
patterns and ramification near or below the main stem,
characteristics that we observed in the FH
2 . In Canada,
Darbyshire (1990) described profusely branched C. canadensis
biotypes that resembled our FH
1 , but attributed those abnormalities to the loss of apical dominance. Because environmental
factors can impact phenotype, some argue that C. canadensis
and C. ramosissima are conspecific and that C. ramosissima
represents a depauperate extreme of the highly plastic
canadensis epithet (S. Darbyshire, Agriculture and Agri-Food
Canada, personal communication). However, because C.
canadensis and C. ramosissima plants grown in the greenhouse
preserved their distinctive phenotypes, we suggest that these
extreme phenotypes possibly represent transgressive hybrids.
Interspecific gene transfer and the interactions of genetic and
environmental stimuli affecting the phenotypic expression of
Conyza deserve further investigation (Voss, 1996).
Documentation of hybridization in Conyzinae is at present
restricted to Europe. For example, in the British Isles
hybridization between C. canadensis and Erigeron acer L.
was reported to produce weak and apparently sterile plants with
few capitula; the nothotaxa was classified as Conyzigeron 3
huelsenii (Vatke) Rauschert (Stace, 1975). Furthermore,
hybrids of unknown fecundity, namely Conyza 3 flahaultiana
(Thell.) Sennen and Conyza 3 daveauiana Sennen in Spain and
France, originated from crosses between C. canadensis and C.
bonariensis (L.) Cronq. and between C. bonariensis and C.
sumatrensis (Retz.) E. Walker, respectively (McClintock and
Marshall, 1988). In the Iberian Peninsula, Conyza 3 rouyana
Sennen arose from the hybridization between C. albida Willd.
ex Spreng. and C. canadensis (J. L. Carretero, Universidad
Politécnica de Valencia, personal communication). Additionally, Thébaud and Abbott (1995) stated that in France, C.
sumatrensis and C. blakei Cabr. cross under natural environments and produce hybrid progenies with moderate (30%)
fertility. Importantly, significant phenotypic variability exists
in C. blakei (Laı́nz, 2001). In Belgium, Verloove and Boullet
(2001) reported that some xenophyte Conyza populations
identified as C. floribunda were in fact C. canadensis 3 C.
sumatrensis hybrids. Furthermore, Conyza 3 mixta Fouc. &
Neyr. reportedly arose from crosses between C. canadensis and
C. bonariensis in Belgium, France, Great Britain, and Portugal
(F. Verloove, National Botanic Garden of Belgium, personal
communication). More recently, Šı́da (2003) reported a
putative hybrid between C. bonariensis and C. triloba Decne.
in the Czech Republic.
Loss of vigor is apparently a common trait to European
Conyza hybrids. We speculate that ploidy differences may be a
significant barrier determining successful hybridization in
Conyzinae. For example, more compatible and vigorous
hybrids would be expected from crosses between the
allopolyploids (2n ¼ 54) C. sumatrensis, C. floribunda, and
C. bonariensis compared to crosses with the diploid (2n ¼ 18)
C. canadensis. Just as geography delimits the major groups in
Asteraceae, spatial isolation is probably the principal species
barrier within Asteraceae (Nesom, 1990). Therefore, disturbances of these barriers will likely stimulate interactions
OF
B OTANY
[Vol. 94
between phylogenetically related taxa separated by geography
(allopatry) and allow for the exchange of genetic material
between unstructured random-mating (panmictic) Asteraceae
populations.
Ploidy of the Conyza hybrid—Hybridization represents an
important mechanism for plant speciation whereby fertile and
stable taxa can arise by either chromosome number doubling
(allopolyploidy) or recombinational speciation without polyploidization (homoploidy) (Rieseberg et al., 2000). In the
absence of karyological studies, the ploidy of the C. canadensis
3 C. ramosissima hybrid remains unknown. Nonetheless, we
propose that the segregation and compatibility data allude to
the formation of an interspecific hybrid without an increase in
ploidy.
Man-made allopolyploids often display homeotic transformations and aberrant chromosomal rearrangements that result
in gene silencing, hybrid instability, and lethality (Comai,
2000). These phenomena were absent from the Conyza hybrid,
as the interspecific hybrids demonstrated a vigorous growth
and marginal (,5%) FH
2 mortality (Figs. 2 and 3). Additionally,
homeologous recombination in neo-allopolyploids typically
destabilizes genomes by chromosomal deletions or expansions,
and homeologous pairing often hinders meiosis through the
formation of univalent or trivalent chromosomes (Comai,
2000). The genome of FH
1 was apparently stable as hybrids
demonstrated moderate to no postzygotic reproductive barriers
and were genetically compatible, with capacity for introgression with either the C. canadensis or C. ramosissima parent
(Table 6; Fig. 3). Finally, allopolyploids often preserve
homeologous integrity by hindering intergenomic recombination, whereas in homoploids, numerous chromosome combinations form through chiasmata, thus allowing recombination
of the parental genomes (Comai, 2000). Given that RS2 and PS0
represented near-homozygous lineages and that FH
2 families
segregated for glyphosate resistance, the C. canadensis and C.
ramosissima genomes may have coalesced in the FH
1 . This
characteristic of segregation is typically associated with
homeologous recombination in homoploids.
Establishment of the Conyza hybrid in the environment—
The low hybridization levels (0% to 12%) in the pollen
competition studies and the estimated high genetic compatibility between C. canadensis and C. ramosissima (98%)
suggest that the resultant hybrid will probably evolve in nature
as a contiguous population in a common geographic range
(parapatry) (Table 3). Under natural environments, hybridization may occur at different frequencies than those observed in
the greenhouse, depending on insect- and wind-pollination
levels and other environmental factors such as competition and
resource availability, which can affect flowering time in
Conyza (Thébaud et al., 1996). Regardless, the fact that FH
1
was fertile implied that fitness and habitat divergence may be
two essential factors determining the adaptation of Conyza
hybrids to the environment. Certainly, the overdominance for
resistance to glyphosate provides the proposed Conyza hybrid
considerable ecological fitness, given the widespread use of
glyphosate in current agroecosystems.
Models for homoploid speciation suggest that superior
hybrid competitiveness results in the rapid displacement of
the parental taxa, while under a fitness disadvantage, either the
hybrid taxa becomes extinct or it coexists with the parents,
provided adequate niche differentiation in the environment
April 2007]
Z ELAYA
ET AL .—H YBRIDIZATION IN
(Rieseberg, 1997). Hybrids are rarely better fit than their welladapted congeners. Low initial frequencies, reduced fertility
and viability, and competitive disadvantage with respect to
parents typically lead to hybrid extinction (Wolf et al., 2001).
Ascertaining fitness of the Conyza hybrid and the potential
ecological displacement of other taxa in different environments
(vicariation) would require further experimentation and is not
within the scope of this study. The supposition that FH
1 hybrids
are highly fit compared to the Conyza parents is provisional,
pending proper field assessments of fitness. Nevertheless, we
believe that several hybrid traits would serve as ecological
advantages, at least under current agroecosystems. The FH
1
rosettes may be competitive by generating more and denser
leaves than RS2 and by producing wider rosettes than PS0 (Table
5), thus accumulating greater biomass than either parent (Fig.
2). In addition, FH
1 plants possessed leaves with higher trichome
densities than RS2, a trait that may reduce herbicide uptake and
hinder herbicide efficacy. This could be significant because the
mechanism of glyphosate resistance in C. canadensis was
attributed to an altered cellular distribution, resulting in
reduced herbicide translocation (Feng et al., 2004; Koger and
Reddy, 2005; Dinelli et al., 2006). Furthermore, FH
1 plants
produced multiple branches, which may facilitate recovery
from herbicide applications through meristematic regrowth
(Table 5).
We therefore suggest that the Conyza hybrid may be well
fitted to agroecosystems, because evaluation of the FH
1
evidenced heterosis and overdominance for glyphosate resistance compared to RS2 and PS0 (Table 2; Fig. 2). The FH
1
developed an equal number, if not more, achenes than either
parent (Table 5); this reproductive behavior could compensate
for the estimated depression in FH
1 seed viability (Fig. 3).
Achene adaptation for dispersion by wind (anemochory), the
capacity to overwinter (therophyte), and a prolific achene
production make C. canadensis extremely competitive under
no-tillage agroecosystems. However, the species is poorly
adapted to crop production systems where tillage is used as a
management strategy (Brown and Whitwell, 1988). Hence, the
larger FH
1 achenes may provide an ecological advantage over C.
canadensis by allowing Conyza hybrid seeds to emerge from
deeper soil profiles in tilled agroecosystems (Table 5).
According to Barton (2001), reasons for higher hybrid
fitness include the (1) occurrence of diverse phenotypes
through transgression that may have ecological advantages
under particular environments, (2) reconstruction of an
‘‘ancestral linkage’’ that was competitive in the past and is
probably adapted to present environments, and (3) coalescence
of previously disjunct gene sets that may have a positive
impact on fitness. Hybridization may therefore explain some
cases of niche differentiation and provide the raw materials for
the adaptation of novel weeds to the environment (Rieseberg et
al., 1999). It is important to acknowledge that the transgressive
fitness with respect to glyphosate resistance resulted from the
combination of phenotypes of the two Conyza populations
herein studied; a different fitness may be observed in hybrids
originating from crosses between other Conyza species.
Implication of hybridization on herbicide resistance
management—The impact of hybridization on glyphosate
resistance management would depend on the native introgression levels of the R allele and the stability, fertility, and fitness
of the Conyza hybrid with respect to RS2 and PS0 . Herein, we
demonstrated that the Conyza hybrid is fertile and competitive
C ONYZA
671
(Figs. 2 and 3). In addition, successful R allele transfer
occurred from C. canadensis to C. ramosissima through the FH
1
in backcrosses (Table 6). Furthermore, the FH
2 demonstrated
many phenotypic traits that may facilitate the adaptation to
different agroecological niches. These circumstances could
potentially complicate the management of glyphosate resistant
weed populations in glyphosate resistant crops and the
containment of glyphosate resistance genes within these
agroecosystems.
Absent from this hypothesis, however, are the natural
hybridization frequencies in Conyza. Native hybridization
frequencies in the European Conyza approximate 50 plants in
thousands of individuals, although 60% of the resultant hybrids
were infertile (Thébaud and Abbott, 1995). These hybridization
frequencies are completely dependent on the ecological
circumstances under which plants developed. However, we
would expect higher hybridization frequencies than those
reported in Europe since the parents evaluated in this study are
compatible and the resultant Conyza hybrid demonstrated
negligible postzygotic reproduction barriers. Even low initial
hybrid frequencies within the population could increase in time
because Conyza plants can produce more than 240 000 achenes
per growing season, mostly viable, which are capable of
dissemination to 30 m in 16 km/h wind (Muenscher, 1935;
Dauer et al., 2006; Nandula et al., 2006).
Interspecific hybridization may therefore affect the extinction rates of weeds in the environment and serve as a
mechanism for the dissemination of transgenic and herbicide
resistance genes (Owen and Zelaya, 2005). Farmers should
consider the potential for hybridization between weeds when
developing programs aimed at managing herbicide-resistant
weeds. Production systems that depend on a single herbicidal
chemistry for weed control should be combined with
alternative management tactics, thus mitigating the evolution
of herbicide resistance and maintaining the sustainability of
current agroecosystems. Further research is needed to monitor
native gene flow levels between weeds and ascertain the
potential for dissemination of herbicide resistance through
introgressive hybridization.
LITERATURE CITED
ABBOTT, R. J. 1992. Plant invasions, interspecific hybridization and the
evolution of new plant taxa. Trends in Ecology and Evolution 7: 401–
405.
ANAMTHAWAT-JÓNSSON, K. 2001. Molecular cytogenetics of introgressive
hybridization in plants. Methods in Cell Science 23: 139–148.
ANONYMOUS. 2004. Crop protection reference. C & P Press, New York,
New York, USA.
AOSA. 2003. Rules for testing seeds. Association of Official Seed
Analysts, Las Cruces, New Mexico, USA.
BALFOURIER, F., G. CHARMET, AND C. RAVEL. 1998. Genetic differentiation
within and between natural populations of perennial and annual
ryegrass (Lolium perenne and L. rigidum). Heredity 81: 100–110.
BARTON, N. H. 2001. The role of hybridization in evolution. Molecular
Ecology 10: 551–568.
BRADSHAW, L. D., S. R. PADGETTE, S. L. KIMBALL, AND B. H. WELLS. 1997.
Perspectives on glyphosate resistance. Weed Technology 11: 189–
198.
BROWN, S. M., AND T. WHITWELL. 1988. Influence of tillage on horseweed,
Conyza canadensis. Weed Technology 2: 269–270.
CAVAN, G., P. BISS, AND S. R. MOSS. 1998. Herbicide resistance and gene
flow in wild-oats (Avena fatua and Avena sterilis ssp. ludoviciana).
Annals of Applied Biology 133: 207–217.
CHAMBERS, J. M., W. S. CLEVELAND, B. KLEINER, AND P. A. TUKEY. 1983.
672
A MERICAN J OURNAL
Graphical methods for data analysis. Duxbury, Boston, Massachusetts, USA.
COMAI, L. 2000. Genetic and epigenetic interactions in allopolyploid
plants. Plant Molecular Biology 43: 387–399.
CONOVER, W. J. 1999. Practical nonparametric statistics. John Wiley, New
York, New York, USA.
CRONQUIST, A. 1980. Vascular flora of the southeastern United States, vol.
I. Asteraceae. University of North Carolina Press, Chapel Hill, North
Carolina, USA.
DARBYSHIRE, S. J. 1990. Conyza ramosissima, another new weed in the
Ottawa District. Trail & Landscape 24: 94–98.
DAUER, J. T., D. A. MORTENSEN, AND R. HUMSTON. 2006. Controlled
experiments to predict horseweed (Conyza canadensis) dispersal
distances. Weed Science 54: 484–489.
DINELLI, G., I. MAROTTI, A. BONETTI, M. MINELLI, P. CATIZONE, AND J.
BARNES. 2006. Physiological and molecular insight on the mechanisms of resistance to glyphosate in Conyza canadensis (L.) Cronq.
biotypes. Pesticide Biochemistry and Physiology 86: 30–41.
ELLSTRAND, N. C. 2003. Current knowledge of gene flow in plants:
implications for transgene flow. Philosophical Transactions of the
Royal Society of London, B, Biological Sciences 358: 1163–1170.
FENG, P. C. C., M. TRAN, T. CHIU, R. D. SAMMONS, G. R. HECK, AND C. A.
CAJACOB. 2004. Investigations into glyphosate-resistant horseweed
(Conyza canadensis): retention, uptake, translocation, and metabolism. Weed Science 52: 498–505.
FRANSSEN, A. S., D. Z. SKINNER, K. AL-KHATIB, M. J. HORAK, AND P. A.
KULAKOW. 2001. Interspecific hybridization and gene flow of ALS
resistance in Amaranthus species. Weed Science 49: 598–606.
FRITZ, R. S., C. M. NICHOLS-ORIANS, AND S. J. BRUNSFELD. 1994.
Interspecific hybridization of plants and resistance to herbivores:
hypothesis, genetics, and variable responses in a diverse herbivore
community. Oecologia 97: 106–117.
GUADAGNUOLO, R., J. CLEGG, AND N. C. ELLSTRAND. 2006. Relative fitness
of transgenic vs. non-transgenic maize 3 teosinte hybrids: a field
evaluation. Ecological Applications 16: 1967–1974.
HARRING, T., J. C. STREIBIG, AND S. HUSTED. 1998. Accumulation of
shikimic acid: a technique for screening glyphosate efficacy. Journal
of Agricultural and Food Chemistry 46: 4406–4412.
HEAP, I. 2006. The international survey of herbicide resistant weeds
[online]. Website http://www.weedscience.com/ [accessed 15 December 2006].
HOLLÄNDER-CZYTKO, H., AND N. AMRHEIN. 1983. Subcellular compartmentation of shikimic acid and phenylalanine in buckwheat cell
suspension cultures grown in the presence of shikimate pathway
inhibitors. Plant Science Letters 29: 89–96.
JASIENIUK, M. 1985. Constraints on the evolution of glyphosate resistance
in weeds. Resistant Pest Management Newsletter 7: 31–32.
KNOBLOCH, I. W. 1972. Intergeneric hybridization in flowering plants.
Taxon 21: 97–103.
KOGER, C. H., AND K. N. REDDY. 2005. Role of absorption and
translocation in the mechanism of glyphosate resistance in horseweed
(Conyza canadensis). Weed Science 53: 84–89.
KWON, Y. W., AND D. S. KIM. 2001. Herbicide-resistant geneticallymodified crop: its risks with an emphasis on gene flow. Weed Biology
and Management 1: 42–52.
LAÍNZ, M. 2001. Conyza blakei (Cabrera) Cabrera (Compositae), ¿especie
peninsular? Anales Jardı́n Botánico de Madrid 59: 352–353.
LÉGÈRE, A. 2005. Risks and consequences of gene flow from herbicideresistant crops: canola (Brassica napus L.) as a case study. Pest
Management Science 61: 292–300.
MARDIA, K. V. 1970. Measures of multivariate skewness and kurtosis with
applications. Biometrika 57: 519–530.
MCCLINTOCK, D., AND J. B. MARSHALL. 1988. On Conyza sumatrensis
(Retz) E. Walker and certain hybrids in the genus. Watsonia 17: 172–
173.
MICHAUX, A. 1803. Flora Boreali-Americana, vol. II. Parisiis et
Argentorati, Apud fratres Levrault, Paris, France.
MOORE, R. P. 1985. Handbook on tetrazolium testing. International Seed
Testing Association, Zürich, Switzerland.
OF
B OTANY
[Vol. 94
MUENSCHER, W. C. 1935. Weeds. Macmillan, New York, New York, USA.
NANDULA, V. K., T. W. EUBANK, D. H. POSTON, C. H. KOGER, AND K. N.
REDDY. 2006. Factors affecting germination of horseweed (Conyza
canadensis). Weed Science 54: 898–902.
NATALI, L., T. GIORDANI, E. POLIZZI, C. PUGLIESI, M. FAMBRINI, AND A.
CAVALLINI. 1998. Genomic alterations in the interspecific hybrid
Helianthus annuus 3 Helianthus tuberosus. Theoretical and Applied
Genetics 97: 1240–1247.
NESOM, G. L. 1990. Further definition of Conyza (Asteraceae: Astereae).
Phytologia 68: 229–233.
NOYES, R. D. 2000. Biogeographical and evolutionary insights on
Erigeron and allies (Asteraceae) from ITS sequence data. Plant
Systematics and Evolution 220: 93–114.
OWEN, M. D. K., AND I. A. ZELAYA. 2005. Herbicide-resistant crops and
weed resistance to herbicides. Pest Management Science 61: 301–
311.
PRUSKI, J. F., AND G. SANCHO. 2006. Conyza sumatrensis var. leiotheca
(Compositae: Astereae), a new combination for a common neotropical weed. Novon 16: 96–101.
REICHMAN, J. R., L. S. WATRUD, E. H. LEE, C. A. BURDICK, M. A. BOLLMAN,
M. J. S TORM, G. A. K ING , AND C. M ALLORY -S MITH . 2006.
Establishment of transgenic herbicide-resistant creeping bentgrass
(Agrostis stolonifera L.) in nonagronomic habitats. Molecular
Ecology 15: 4243–4255.
RIESEBERG, L. H. 1997. Hybrid origins of plant species. Annual Review of
Ecology and Systematics 28: 359–389.
RIESEBERG, L. H., M. A. ARCHER, AND R. K. WAYNE. 1999. Transgressive
segregation, adaptation and speciation. Heredity 83: 363–372.
RIESEBERG, L. H., S. J. E. BAIRD, AND K. A. GARDNER. 2000. Hybridization,
introgression, and linkage evolution. Plant Molecular Biology 42:
205–224.
SAS. 2000. SAS procedures guide, version 8. SAS Institute, Cary, North
Carolina, USA.
SCHARDL, C. L., AND K. D. CRAVEN. 2003. Interspecific hybridization in
plant-associated fungi and oomycetes: a review. Molecular Ecology
12: 2861–2873.
SHAPIRO, S. S., AND M. B. WILK. 1965. An analysis of variance test for
normality. Biometrika 52: 591–611.
SHONTZ, J. P., AND H. J. OOSTING. 1970. Factors affecting interaction and
distribution of Haplopappus divaricatus and Conyza canadensis in
North Carolina old fields. Ecology 51: 780–793.
ŠÍDA, O. 2003. Conyza triloba, new to Europe, and Conyza bonariensis,
new to the Czech Republic. Preslia 75: 249–254.
STACE, C. A. 1975. Hybridization and the flora of the British Isles.
Academic Press, London, UK.
STEINRÜCKEN, H. C., AND N. AMRHEIN. 1980. The herbicide glyphosate is a
potent inhibitor of 5-enolpyruvyl-shikimic acid-3-phosphate synthase.
Biochemical and Biophysical Research Communications 94: 1207–
1212.
STEYERMARK, J. A. 1963. Flora of Missouri. Iowa State University Press,
Ames, Iowa, USA.
TABASHNIK, B. E. 1991. Determining the mode of inheritance of pesticide
resistance with backcross experiments. Journal of Economic
Entomology 84: 703–712.
TEAL, P. E. A., AND A. OOSTENDORP. 1995. Effect of interspecific
hybridization between Heliothis virescens and H. subflexa (Lepidoptera: Noctuidae) on sex pheromone production by females. Journal of
Insect Physiology 41: 519–525.
THÉBAUD, C., AND R. J. ABBOTT. 1995. Characterization of invasive Conyza
species (Asteraceae) in Europe: quantitative trait and isoenzyme
analysis. American Journal of Botany 82: 360–368.
THÉBAUD, C., A. C. FINZI, L. AFFRE, M. DEBUSSCHE, AND J. ESCARRE. 1996.
Assessing why two introduced Conyza differ in their ability to invade
Mediterranean old fields. Ecology 77: 791–804.
THOMPSON, B. 1984. Canonical correlation analysis: uses and interpretation. Sage Publications, Thousand Oaks, California, USA.
TONSOR, S. J. 1990. Spatial patterns of differentiation for gene flow in
Plantago lanceolata. Evolution 44: 1373–1378.
TRANEL, P. J., J. J. WASSOM, M. R. JESCHKE, AND A. L. RAYBURN. 2002.
April 2007]
Z ELAYA
ET AL .—H YBRIDIZATION IN
Transmission of herbicide resistance from a monoecious to a
dioecious weedy Amaranthus species. Theoretical and Applied
Genetics 105: 674–679.
TRUCCO, F., M. R. JESCHKE, A. L. RAYBURN, AND P. J. TRANEL. 2005.
Promiscuity in weedy amaranths: high frequency of female tall
waterhemp (Amaranthus tuberculatus) 3 smooth pigweed (A.
hybridus) hybridization under field conditions. Weed Science 53:
46–54.
VANGESSEL, M. J. 2001. Glyphosate-resistant horseweed from Delaware.
Weed Science 49: 703–705.
VERLOOVE, F., AND V. BOULLET. 2001. Conyza bonariensis en Conyza
sumatrensis: recent ingeburgerd in België? Dumortiera 77: 2–8.
VINCENT, G., AND M. CAPPADOCIA. 1987. Interspecific hybridization
between common ragweed (Ambrosia artemisiifolia) and giant
ragweed (A. trifida). Weed Science 35: 633–636.
VOSS, E. G. 1996. Michigan flora, part III, Dicots concluded. Cranbrook
Institute of Science, Bloomfield Hills, Michigan, USA.
WATRUD, L. S., E. H. LEE, A. FAIRBROTHER, C. BURDICK, J. R. REICHMAN, M.
BOLLMAN, M. STORM, G. KING, AND P. K. VAN DE WATER. 2004.
Evidence for landscape-level, pollen-mediated gene flow from
C ONYZA
673
genetically modified creeping bentgrass with CP4 EPSPS as a
marker. Proceedings of the National Academy of Sciences, USA 101:
14533–14538.
WEAVER, S. E. 2001. The biology of Canadian weeds. 115. Conyza
canadensis. Canadian Journal of Plant Science 81: 867–875.
WETZEL, D. K., M. J. HORAK, D. Z. SKINNER, AND P. A. KULAKOW. 1999.
Transferal of herbicide resistance traits from Amaranthus palmeri to
Amaranthus rudis. Weed Science 47: 538–543.
WILSON, P. 1992. On inferring hybridity from morphological intermediacy.
Taxon 41: 11–23.
WOLF, D. E., N. TAKEBAYASHI, AND L. H. RIESEBERG. 2001. Predicting the
risk of extinction through hybridization. Conservation Biology 15:
1039–1053.
ZELAYA, I. A., AND M. D. K. OWEN. 2006. Interspecific hybridization in
Asteraceae: case studies in Ambrosia, Conyza and Helianthus.
Proceedings of the Weed Science Society of America Meeting 46:
(abstract).
ZELAYA, I. A., M. D. K. OWEN, AND M. J. VANGESSEL. 2004. Inheritance of
evolved glyphosate resistance in Conyza canadensis (L.) Cronq.
Theoretical and Applied Genetics 110: 58–70.