(American Journal of Botany. 2007;94:660-673.)
© 2007 Botanical Society of America, Inc.
Transfer of glyphosate resistance: evidence of hybridization in Conyza (Asteraceae)1
Ian A. Zelaya4,
Micheal D. K. Owen and
Mark J. VanGessel
2Iowa State University, Department of Agronomy, 2104 Agronomy Hall, Ames, Iowa 50011-1011 USA;
3University of Delaware, Department of Plant and Soil Sciences, Research and Education Center, 16684 County Seat Highway, Georgetown, Delaware 19947-9575 USA
Received for publication May 14, 2006.
Accepted for publication February 11, 2007.
ABSTRACT
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 = 18) and C. ramosissima (2n = 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 (
) demonstrated an intermediate phenotype between the parents but superior resistance to glyphosate compared to the resistant C. canadensis parent. Inheritance of glyphosate resistance in the selfed
(
) followed the partially dominant nuclear, single-gene model;
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 x 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
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 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 = 18) and C. ramosissima Cronq. (2n = 18) by monitoring transmission of the allele conferring resistance to the herbicide N-phosphonomethyl glycine (glyphosate).
View this table:
[in this window]
[in a new window]
|
Table 1. Cases of confirmed transfer of herbicide resistance gene(s) through introgressive hybridization in weeds.
|
|
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 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 (
) 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 (
) 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 (
) 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
at the rate of 2.0 kg acid equivalent (AE) glyphosate/ha (Zelaya et al., 2004
). Complete specimen sets of
, 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
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, 35° : 25°C day : night, 70–90% relative humidity (RH), and 600 µmol·m–2·s–1 photosynthetic photon flux density (PPFD) conditions. The RS2,
, and hybrid progenies (explained next) were grown in a greenhouse set to 25°–35°C and 50–80% RH diurnal conditions and 20°–25°C and 50% RH nocturnal conditions; natural light was supplemented with 600–1000 µmol·m–2·s–1 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
were resistant and 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
population had never been exposed to glyphosate. Furthermore, treatment of the progeny of selfed
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 (
x RS2, RS2 x
) 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
parent plants were manually removed with forceps (emasculation). Upon stigma protrusion of the remaining
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
and RS2 serving as the pollen receptor and pollen donor parent, respectively. Non-emasculated capitula in
plants were removed to prevent autogamy. Approximately 50 capitula per
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
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
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 (
) frequencies. One
plant per family was then permitted to self-pollinate in isolation, and segregation of the R allele was monitored in the resultant generation (
); 297 individual
plants were evaluated in the segregation analysis to develop a genetic model. Finally, introgression of the R allele was confirmed by backcrossing one
plant per family to the original RS2 (
) and
(
) parents. The parents used in backcrosses originated from florets of the original
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
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
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 each of the 10 parents (RS2 and
) or families (
,
,
, and
) were germinated (n = 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 (Moore, 1985
). Seed viability experiments were repeated in time 1 wk after the initial assessment (n = 2000). For postzygotic reproductive barriers, hybrid inviability was tested by assessing viability of
seeds, hybrid sterility tested on the
seed, and hybrid breakdown on the viability of
and
seeds.
Characterization of interspecific hybrids
Ten randomly selected RS2 and
plants or one randomly selected
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 = 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 ÷ leaf width), leaf dentation, adaxial leaf trichomes, leaf area, leaf mass, and the specific leaf area (SLA; leaf area ÷ 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 post-anthesis but prior to seed physiological maturity and included capitula length, capitula width, capitula shape (capitula length ÷ 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 = 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% = non-injured; 100% = completely necrotic). Plants were then cut at the soil surface, placed in paper bags, and dried at 80°C 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 (LSD
=0.05) 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), |
50| (the absolute difference between two estimated GR50 values), and 
2 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 (
2) of the data were normalized by natural-log transformation (Shapiro and Wilk, 1965
). ANOVA was done on individual quantitative traits considering taxon and family nested within taxon as fixed and random effects, respectively. Fisher's LSD=0.05 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 2 into the different ANOVA components for traits that converged to Shapiro-Wilk's normality assumption; these 2 components were then used to calculate the intraclass correlation coefficient (t):
| (1) |
where
and
corresponded to the 2 between and within taxon, respectively.
Phenotypic intermediacy of the
, as it related to the RS2 and
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 (
2) analysis (Mardia, 1970
). PROC PRINCOMP was invoked to test whether the data covariance matrix was singular. If rejected, squared Mahalanobis distances (
) from the centroid were computed for chi-square (2) quantile–quantile (Q–Q) comparisons of multivariate normal data; 75% confidence intervals were constructed from standard deviation () 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 = 0.33; P = 0.98;
= 0.82) and shikimic acid (F = 1.36; P = 0.15;
= 0.92) confirmed suitability of the log-logistic model for describing plant response as a function of increasing glyphosate rates.
was killed at 0.85 kg AE of glyphosate/ha (Figs. 1 and 2). Compared to
, 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
and RS2, a statistically significant |50| value of 1.60 (Fobs = 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
(Table 2). Untreated Conyza plants contained 7–14 µmol of shikimic acid/g dry mass, which increased sigmoidally to a maximum of 150 µmol 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).
View this table:
[in this window]
[in a new window]
|
Table 2. Response of Conyza canadensis (RS2), C. ramosissima (), the interspecific hybrid (), and the hybrid progeny () 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.
|
|
Levels of interspecific hybridization
Pistillate florets in emasculated
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 environments, approximately 60% of the total achenes on
plants produced viable seeds (Fig. 3). Therefore, we estimated that the emasculation method was 98% effective at preventing self-fertilization in
.
View this table:
[in this window]
[in a new window]
|
Table 3. Frequencies of hybridization in the Conyza canadensis x 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 () parents evaluated. Crosses between C. canadensis (RS2) and 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 (); data represent the sum of the 10 generated families (see Materials and Methods).
|
|
View larger version (20K):
[in this window]
[in a new window]
|
Fig. 3. Proportion of germinated (gray bars), dormant (dark gray bars), and nonviable (white bars) seeds among the Conyza canadensis (RS2), C. ramosissima (), interspecific hybrid (), hybrid progeny (), and the backcross to RS2 () or (). Bars represent the mean of two experiments comprised of 100 seeds per each of the 10 parents (RS2 and ) or 10 families (, , , and ) generated in the no-pollen competition studies (n = 2000); extensions above bars designate the standard error associated with individual means (M). Letters above bars represent minimum statistical differences according to Fisher's LSD for comparisons within germinated (LSD=0.05 = 12%), dormant (LSD=0.05 = 7%), and nonviable (LSD=0.05 = 14%) seeds.
|
|
Interspecific hybridization under pollen competition conditions ranged from 0% to 7% in the RS2 to
crosses and from 0% to 9% in the reciprocal
to RS2 crosses, although five in 10
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 conditions when RS2 served as pollen donor to
(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 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
and
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,
, and
(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 (2) was accounted for differences between the studied Conyza taxa. Furthermore, we assumed that this 2 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).
View this table:
[in this window]
[in a new window]
|
Table 4. Partitioning of the total phenotypic variance (2) 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 2 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 (), and the interspecific hybrid ().
|
|
Phenotype of
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
. Furthermore, both RS2 leaf dimorphisms were longer and wider and possessed more dentations than
(Table 5). In contrast,
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,
produced longer achenes (Table 5).
View this table:
[in this window]
[in a new window]
|
Table 5. Comparison of quantitative traits among Conyza canadensis (RS2), C. ramosissima (), and the interspecific hybrid (). Values represent the mean of five observations per each of 10 sampled plants and assessments repeated in time (n = 100); numbers in parentheses designate the standard error associated with means (M). A one-tailed sign test of intermediate vs. non-intermediate was used to test the hypothesis of phenotypic intermediacy.
|
|
To investigate whether the phenotype of
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
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
was intermediate to both parents (D = 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 (z+ = 3.27; P = 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
represented an intermediate between both parents.
Intermediacy of
Multivariate normality analysis determined that only 17 of the 27 evaluated morphometric traits converged to Mardia's statistics (2 = 328.2; P = 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 ( = 0.004; F = 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 2, respectively. Taxa clustered in discrete sections within the canonical graph;
(6.8) and the
(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
parents (
= 226.5; F = 630.2; P < 0.0001), and approximately half and one-fourth that square distance was estimated for comparisons between the
and RS2 (
= 107.5; F = 299.1; P < 0.0001) and
and
(
= 53.7; F = 149.4; P < 0.0001), respectively. Hierarchical clustering formed a dendrogram with a common root node for the taxon (Fig. 4). The
and
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
phenotype, CDF analysis and UPGMA strongly emphasized that the
was more similar to
than to RS2.
Postzygotic reproductive barriers
Partial correlation estimates (r2 = 0.76; P = 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 = 0.14; P = 0.71) and multivariate (Wilks' = 0.99; P = 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 28°C 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 (LSD=0.05 = 12%) from the 57% mean viability estimated for
seeds (Fig. 3).
Evidence for hybrid inviability was apparent in the 23% increase in nonviable
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
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
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
zygotes.
The presence of a hybrid sterility reproductive barrier was disregarded because the
had embryos with approximately twice the level of viability of their progenitor (
) (Fig. 3). Estimates of
viability were significantly less than for either RS2 or
, an indication that some level of self-infertility remained within hybrids and evidence that the inviable
zygotes probably arose from genetic incompatibilities rather than the physical stress of emasculation (Fig. 3). Backcrosses of
to RS2 or
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 (
). Establishment of hybrid interspecific populations in the environment is often limited by postzygotic reproductive barriers (Barton, 2001
). While the germination of
seeds was less than that of the RS2 or
parents, the collective viability data established that the Conyza hybrid was fertile and probably capable of establishment in natural environments.
Transgressive segregation in
Most hybrid progenies demonstrate marked traits of transgression (Rieseberg et al., 1999
). The majority of
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
individuals. Transgressive segregation was also discernable in whole-plant rate responses to glyphosate. Compared to the biomass accumulated by RS2 (2 = 0.67),
(2 = 0.42), or the
(2 = 0.87),
rosettes demonstrated greater variance (2 = 1.17) in response to glyphosate. Similarly, greater variations in the level of endogenous shikimic acid accumulation and visual injury were observed in the
(data not shown). The
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
plants.
Phenotype of backcrosses
Visual observations confirmed that the progeny of
backcrosses to parents (
and
) were phenotypically similar to either C. canadensis or C. ramosissima. For example,
resembled RS2 at the rosette and cauline stages, except that the plant developed lateral branches below the center-main stem. In addition,
produced serrated and nonserrated leaves, while RS2 produced only the serrated dimorphism.
possessed a well-defined axis that was absent in
and generated slightly larger leaves and thicker stems than
.
was also precocious compared to
, completing the reproductive cycle in 2–3 mo.
Inheritance of glyphosate resistance
Application of 0.40 kg AE of glyphosate/ha to
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
or the selfed progeny of
, 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
).
Overdominance of
The
produced larger rosettes compared to
and more and denser (SLA) leaves than RS2 (Table 5). The difference in these two leaf parameters resulted in an apparent heterotic response of
recorded at the 10–12-cm diameter rosettes stage (Fig. 2). Compared to
and RS2, greater glyphosate rates were required to reduce biomass accumulation or cause mortality in
(Table 2; Figs. 1 and 2). This divergence in response to glyphosate was confirmed by |50| values for comparisons of
and RS2 (Fobs = 1.74; P < 0.01) or
and
(Fobs = 1.76; P < 0.01). Not only did
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 absorption into
plants (Table 5). The response of
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 ratios of
,
, and
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
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 2 values for all
families, substantiating appropriateness of the partially dominant monogenic model (Table 6). Concomitantly, homogeneity analysis (2 = 5.72; P = 0.77) confirmed that the combined
data converged to the 1 : 2 : 1 genetic model.
View this table:
[in this window]
[in a new window]
|
Table 6. Segregation for glyphosate resistance of the R allele in the progeny () of one interspecific hybrid plant () per family allowed to self-pollinate in isolation. The 10 families evaluated arose from a single plant isolated from the Conzya canadensis (RS2) to C. ramosissima () no-pollen competition cross. For back-crosses, served as pollen donor to RS2 () or ().
|
|
Backcrossing of
to RS2 (
) and subsequent progeny treatment with 2.0 kg AE of glyphosate/ha identified only R and IR phenotypes (Table 6). The observed segregation ratios among
families were homogenous (2 = 4.40; P = 0.88), and the collective data were consistent with the expected 1 : 1 (R : IR) ratio. Similarly, analysis of
backcrosses to
(
) resulted in homogenous IR and S segregation ratios (2 = 5.81; P = 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
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
mortality by comparing to that mortality expected as suggested by Tabashnik (1991)
: Y = WR (0.25) + WIR (0.50) + 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 </
