Major Gene Effects During Weed Evolution: Phenotypic Characters Cosegregate With Alleles at the Ray Floret Locus in Senecio vulgaris L. (Asteraceae) H. P. Comes Ray floret length was used as a codominant marker to test for associations with phenotypic characters in two F2 progenies (F2A and F2B) derived from crosses between the nonweedy, late-developing Senecio vulgaris subsp. denticulatus and its weedy, early developing derivative subsp. vulgaris var. vulgaris. Variance analytical tests between the three marker genotype classes indicated that the chromosomal region associated with the ray floret locus controlled more than 10% of the phenotypic variation of 9 of the 17 characters analyzed in F2B. Particularly pronounced associations (r 2 5 28–37%) were found between the marker alleles and leaf number, pedicel length, and number of involucral bracts. In addition, formal linkage analysis between the ray floret locus and the previously postulated dominant gene affecting early development was positive (r 5 0.15 6 0.02). Estimates of dominance relationships at individual quantitative trait loci (QTLs) situated near the ray floret locus indicated directional dominance for a reduction of vegetative and reproductive structures, because the heterozygote marker class tended to be more similar to the weedy derivative, var. vulgaris. This may be interpreted as evidence for past directional selection acting on these characters. Taken together, it is suggested that a single chromosomal region within S. vulgaris exhibits a substantial proportion of nonadditive gene effects, contrasting strongly with the traditional additivepolygenic model of the genetic basis of evolutionary divergence in the wild. From the Institut für Spezielle Botanik und Botanischer Garten, Johannes Gutenberg-Universität Mainz, Bentzelweg 9, 55099 Mainz, Germany. I wish to acknowledge Dr. Kirsten Wolff, Dr. Richard J. Abbott, and Dr. Andrew J. Lowe (St. Andrews University, Scotland) for critically reading the manuscript, Prof. Joachim W. Kadereit (Mainz University, Germany) for constructive comments throughout the study, and two anonymous reviewers for their critical comments on an earlier draft of this article. I am also grateful to Dr. François Bretagnolle ( Neuchâtel University, Switzerland) for detailed statistical advice. This research was supported by an LGFG doctoral grant from the Land Baden-Württemberg, Germany. Journal of Heredity 1998;89:54–61; 0022-1503/98/$5.00 54 Understanding the genetic basis of plant speciation is a primary goal in evolutionary biology. Because the success of taxa to become reproductively isolated from each other often depends on the amount of genetic differences acquired (Mayr 1963), the characterization of the genetic basis of phenotypic differences between ecological races or between related species is fundamental for the study of adaptive phenotypic evolution. In particular, the number of genes controlling a trait, the magnitude of their effects, and the genomic organization of such genes are critical in understanding the rapidity and mode of evolutionary change ( Fisher 1958; Hill 1982; Kimura 1983; Mitchell-Olds and Rutledge 1986). The point has been made many times that evolutionary divergence often involves selection acting on quantitative polygenic systems (Charlesworth 1984; Coyne and Lande 1985). Empirical support for multifactorial evolution comes for example from analyses of the genetic architecture of mating system traits that mainly consists of a rather large number of genes with relatively small, primarily additive ef- fects ( Fenster and Ritland 1994, and references therein). Most recently, however, genome mapping in Mimulus has revealed genes of large effect controlling floral traits associated with reproductive isolation ( Bradshaw et al. 1995). Similarly, major genes with large allelic effects influencing components of fitness such as flowering time are being increasingly hypothesized as playing a central role in natural plant populations (Arabidopsis, MitchellOlds 1996; Senecio, Comes and Kadereit 1996). Although the genetics of phenotypic differences between ecological races or between related species has received extensive attention from botanists for several decades (Orr and Coyne 1992), surprisingly little investigation has been devoted to the genetic basis of phenotypic changes surrounding the origin of weeds. As already pointed out by Stebbins (1965), valid data can be obtained by the genetic analysis of synanthropic taxa and their nonweedy progenitors. Since the synanthropic component is likely to have evolved only recently, that is, only after man-made habitats have become avail- Figure 1. F2 frequency distribution (from two progenies pooled) for time from germination to bud formation and fitted underlying normal distributions, with respective means designated by arrows. Overlap between the two fitted distributions is 1.5% and represents the probability of phenotype misclassification given a discrimination point of approximately 42 days. Probable outliers, excluded from the analysis, are indicated by asterisks. See Comes and Kadereit (1996) for more details. able, the amount of evolutionary divergence following differentiation should be comparatively low in these model systems. Therefore their genetic analysis makes it more feasible to identify changes that occurred during the process of differentiation and not subsequent to it (Macnair and Cumbes 1989). In this article I focus on the genetics of a synanthropic subspecies in the genus Senecio L. sect. Senecio (Asteraceae) and its nearest nonweedy relative. The two focal taxa are S. vulgaris L. subsp. denticulatus (O. F. Muell.) P. D. Sell and S. vulgaris L. subsp. vulgaris var. vulgaris ( both 2n 5 4x 5 40). Subspecies denticulatus is a winter annual herb restricted to Atlantic maritime and Mediterranean montane areas in Europe ( Kadereit 1984b; Perring and Sell 1968) that shows pronounced seed dormancy ( Kadereit 1984b) and normally produces an outer ring of ray florets up to 3 mm long, whereas var. vulgaris is a cosmopolitan weed lacking ray florets. Occasionally seed dormancy can also be found in Mediterranean var. vulgaris (Ren and Abbott 1991; Comes 1995a) but otherwise is absent from this taxon. Based on the natural geographical distribution of subsp. denticulatus and its specialized maritime habitat (at least in western Europe; Comes 1995b), it has been assumed that subsp. denticulatus is the predecessor of the weedy var. vulgaris, or close to such ( Kadereit 1984a,b). Both taxa are predominantly self-pollinating and form hybrids exhibiting up to 73% seed fertility (Comes 1994). Data from previous genetic studies (e.g., Hull 1974; Trow 1912) have established that the presence or absence of ray florets in S. vulgaris is controlled by a single, incompletely dominant locus, with the longrayed subsp. denticulatus homozygous for the dominant R allele, the discoid var. vulgaris homozygous for the recessive r allele, and their short-rayed F1 hybrids heterozygous for both alleles. More recently it was reported that taxon differences in speed of development can also be attributed to hereditary factors (Comes and Kadereit 1996). In these studies crosses between the early developing var. vulgaris and the late developing subsp. denticulatus produced an F2 that had an early:late ratio not significantly different from that expected under a single gene model of inheritance, with early development dominant over late. Accurate discrimination between early and late F2 phenotypic classes was possible due to a small overlap between two underlying normal distributions fitted to the data ( Figure 1). Another character which is known to show discrete disomic and dominant inheritance in crosses between the taxa is the absence of seed dormancy ( Kadereit 1984b). Taken together, it thus appears that the initial ecological shift from the weedy to the nonweedy condition in S. vulgaris is due to selection for (1) loss of seed dormancy, (2) early flowering, and (3) loss of ray florets, with each of the three traits being controlled by a single gene. Given the simple nature of intraspecific capitulum variation in S. vulgaris and its suitability for serving as a marker polymorphism (e.g., Abbott 1986; Abbott et al. 1992), I examine here the association patterns between the polymorphism for capitulum type and a number of quantitative characters, analyzing F2 progenies of a cross between subsp. denticulatus and var. vulgaris. However, in contrast to previous investigations into quantitative variation associated with capitulum type in S. vulgaris (reviewed by Abbott 1986; Oxford et al. 1996), capitulum type is scored here as a quantitative gene effect via the expression of ray floret length, and not as a qualitative, dominant gene effect via the presence [RR, Rr] or absence [rr] of ray florets. This approach has the advantage of utilizing the ray floret gene as a codominant marker locus, with genotypes [TrTr], [TrTn], and [TnTn] for long-rayed, shortrayed, and discoid plants, respectively (see Hull 1974 for allele designation); it thus allows complete classification of genotypes in F2 progenies, and consequently permits examination of dominance relationships at individual quantitative trait Comes • Cosegregation in Senecio vulgaris 55 loci (QTLs) putatively linked to the ray floret marker. Explicitly the following questions have been addressed. (1) Are there significant associations between the ray floret locus and character variation distinguishing the weedy derivative and its nonweedy progenitor? (2) If significant associations exist, what is the genetic architecture of the suite of characters, that is, are genes of small additive effect involved or ones of large, nonadditive effect? (3) What is the linkage relationship between the previously identified major gene controlling speed of development and the ray floret locus? Materials and Methods Plant Materials One population from each taxon was chosen for study and collected for achenes: (1) a long-established winter annual population of subsp. denticulatus on Jersey (Channel Islands, Les Quennevais, North of La Pulente), restricted to carbonate-rich and nutrient-poor soils of open, stabilized dunes and open turf (Comes 1995b); (2) a population of var. vulgaris in the Botanic Garden, Heidelberg University, Germany. Crossing Design and Characters Scored Two F2 progenies (A and B) were generated by self-pollinating two F1 half-sibs from crosses between a single plant of var. vulgaris (used as female parent) and two plants of subsp. denticulatus. Further F1 progeny were produced at the same time. All crosses were produced by Alexander’s (1979) emasculation technique, which has been shown to reduce the degree of residual selfing to less than 4% (Comes and Kadereit 1996). F1 hybrids were verified by the presence of short, stubby ray florets. Parental stocks of subsp. denticulatus and var. vulgaris were maintained by random selfing of individuals within each taxon. F2 progenies were grown simultaneously with parental and F1 samples. Plants were grown in 10 cm plastic pots under outdoor conditions using six randomized blocks. Two-way analysis of variance (ANOVA) comparison revealed that the two F2 progenies (A and B) were significantly (P , .01) different from each other for a number of traits studied, so their data were not pooled. This analysis also revealed that block (i.e., local environmental) factors and family 3 block effects (i.e., gene-environment interactions) were generally insignificant (data not shown). If not indicated otherwise, only the recordings for 56 The Journal of Heredity 1998:89(1) Table 1. Description of traits analyzed in Senecio, with code designations appearing in Tables 2 and 3 No. Trait Code Measured at anthesis of first capitulum 1. Stem height (cm) 2. No. of leaves on main stem 3. Pedicel length of first capitulum (mm) STH NL PL Measured at fruiting of first capitulum 4. No. of lateral branches from main stem 5. Longest internode length on main stem 6. Longest leaf length (cm) 7. Longest leaf width (cm) 8. Longest leaf length:width ratio 9. Longest leaf area (cm2) NB LIN LL WL SL AL Measured on a capitulum of an early branch 10. No. of outer involucral bracts 11. Capitulum length (mm) 12. Capitulum width (mm) 13. Capitulum length : width ratio 14. Ligule length (mm) 15. No. of disc florets 16. No. of florets NOIB LCAP WCAP SCAP LRAY NDISC NFLORET Scored on an isolated capitulum of an early branch 17. No. of achenes 18. Percentage seed set NACH SET the parental and F2 progenies are reported here. A total of 18 quantitative characters were measured for each plant in all progenies. These characters are listed in detail in Table 1. Attention was principally focused on vegetative, floral, and reproductive characters that contribute conspicuously to intraspecific differences in certain annual and winter annual species of sect. Senecio (Comes 1995a). No attempt was made to include all characters that are taxonomically important to distinguish the parental taxa. Measurements of leaf area were made using a HAFF Planimeter No. 315 E. Floral measurements and counts were made on a single capitulum of an early flowering branch, stored in 70% ethanol until usage. Most measurements were made using caliper rules or calipers and are accurate to 0.2 mm. Ray floret length was measured from the apex of the ligule to the distal part of the floral tube using either an ocular micrometer to within 0.04 mm, or millimeter paper to within 0.2 mm (only subsp. denticulatus). Statistical Analysis The two F2 populations segregated for the ray floret marker. Goodness-of-fit to a codominant ratio of 1:2:1, expected for an F2, was tested by chi-square analysis. Associations between alleles at the ray floret locus and all scored quantitative traits were detected by comparing variation between allelic classes with variation within classes, using single-factor ANOVA F statistics. As the number of comparisons made between allelic classes increases in the ANOVA, so too does the probability that a type I statistical error will occur. In order to eliminate this bias, sequential Bonferroni tests (Rice 1989) were conducted to determine table-wide statistical significance (a 5 0.05) for the 17 hypothesis tests for each character within each of the two F2 populations. Nonparametric statistics ( Kruskal–Wallis test, Wilcoxon’s two-sample test), which do not depend on the assumption of homogeneous variances (Sokal and Rohlf 1981) were also computed, but were consistent with the results of the ANOVAs given here. To determine the magnitude of the phenotypic effect of QTLs that are putatively linked to the ray floret marker, the proportion of phenotypic variance explained by the marker locus genotype classes was computed by using the general linear model (GLM) procedure of SAS ( VMS version of SAS release 6.07; SAS Institute Inc. 1988). This fraction is indicated by the r2 value of the variance-analytical regression model. Comparisons between the parental taxa, as well as the mean F2 phenotypic values of the three genotype classes were made using Tukey’s test (Steel and Torrie 1980), evaluated at a significance level of 1%. From the means of homozygous and heterozygous F2 genotype classes, the ratio of dominant (d) to additive (a) gene effects at a putative QTL was estimated with a statistic modified after Stuber (1989): A/B 2 [(A/A 1 B/B)/2] 5 (1 2 2r) d/a (B/B 2 A/A)/2 where A and B are the alleles at the ray floret locus and r is the recombination frequency between the latter and the QTL. Estimates of this ratio, which describe the degree to which the heterozygous genotype class (A/B) resembles either the greater ( B/B) or lesser (A/A) homozygote class (d/a . 0 or d/a , 0), can be regarded as only approximate, because information on r is generally lacking. Though an increase in r (maximum 0.5) would certainly cause a downward bias of estimates, it is assumed that significant differences between genotype classes are indicative of small r values (e.g., Edwards et al. 1987). The recombination frequency between the ray floret marker and the previously identified gene controlling speed of development was calculated using an F2 maximum likelihood algorithm published by Allard (1956), and their map distance, presented in centiMorgans (cM), was derived employing the Kosambi function ( Kosambi 1944). All computations were accomplished using the computer package Linkage-1 version 3.50, by Suiter et al. (1983). Results Segregation of the Morphological Marker Locus The observed bimodal phenotypic F2 distribution for ray floret length clearly reflected segregation at the ray floret locus ( Figure 2). Based on variation among contemporaneous subsp. denticulatus plants, radiate F2 individuals with a ray floret length shorter than 2 mm were classified as short rayed, while those having a length of 2 mm or longer were considered long rayed. The data of 80 discoid [TnTn] individuals, together with 171 short-rayed [TrTn] and 79 long-rayed [TrTr] individuals adequately fitted their expected codominant ratio of 1:2:1 (for F2A, x2[2] 5 1.02, .75 . P . .5; for F2B, x2[2] 5 0.01, P 5 .995; pooled data, x2[2] 5 0.44, .9 . P . .75). Although there still remains an unknown amount of inaccuracy in identifying the genotype of radiate F2 individuals, these results provide evidence for the utility of ray floret length as a codominant morphological marker. Associations Between the Ray Floret Locus and Quantitative Traits The parental values for the traits listed in Table 1 are shown in Table 2, and the corresponding F2A and F2B values of the three genotype classes [TnTn, TrTn, and TrTr], together with the essential results of the cosegregational analysis, are summarized in Table 3A and B, respectively. Figure 2. Frequency distributions for ray floret length in S. vulgaris subsp. denticulatus and F1 and F2 progenies derived from a cross with the rayless var. vulgaris ( F2 data from two progenies pooled). Of those 17 traits examined in both F2 progenies, 8 traits within F2A (47.1%) and 15 within F2B (88.2%) showed statistically significant associations with the marker locus as indicated by the sequential Bonferroni procedure. All associations significant in the F2A progeny were also present in the larger F2B population. These included associations of the marker with leaf number, pedicel length, leaf length, and all floral characters except capitulum width. The proportion of the phenotypic variation explained by the marker locus (r2) was generally lower in the F2A than in the F2B progeny. It is likely, however, that lower predictability of marker genotypes in the F2A population was more influenced by statistical chance effects due to small sample size rather than by an unknown biological effect peculiar to that progeny. In the following, therefore, all results reported will refer to the F2B population, if not stated otherwise. With respect to leaf length, leaf area, pedicel length, and number of outer involucral bracts, the proportion of phenotyp- Comes • Cosegregation in Senecio vulgaris 57 Table 2. Number of observations (N), mean values (X), and standard errors (SE) analyzed in S. vulgaris subsp. denticulatus and S. vulgaris var. vulgaris grown together with their F 2; means sharing the same superscript are not significantly different (P , .01, Tukey’s test) Subsp. denticulatus Var. vulgaris Trait N Vegetative STH NL PL NB LIN LL WL SL AL 131 9.7b 130 18.9b 124 4.6a 130 11.7b 130 22.3b 122 5.8b 122 2.1b 122 2.8b 121 4.8b Floral NOIB LCAP WCAP SCAP NDISC NFLORET Reproductive NACH SET X 99 14.9b 98 7.95a 99 4.32a 98 1.84a 99 58.3a 99 58.3a 127 28.3a 127 46.1b SE N X SE 0.24 0.17 0.21 0.39 0.44 0.07 0.04 0.03 0.12 68 68 67 55 56 44 44 44 44 23.9a 30.3a 4.1a 15.5a 27.9a 8.7a 2.6a 3.5a 6.6a 0.76 0.43 0.19 0.45 0.92 0.19 0.06 0.10 0.25 0.25 0.048 0.030 0.015 0.76 0.76 62 21.0a 61 7.71a 62 4.43a 61 1.74b 62 52.2b 61 62.3a 0.64 0.074 0.038 0.019 0.91 1.01 1.03 1.59 46 30.7a 46 50.8a 1.92 3.12 ic variance explained by the morphological marker ranged from 17–37%. For all other vegetative traits analyzed, variation in marker locus genotypes accounted for much less of the phenotypic variation present. Although significant differences between F2B genotype classes were observed for stem length, number of branches, internode length, and leaf width, only 2–7% of the total phenotypic F2B variation could be traced back to variation at the ray floret locus. The cosegregation of the marker with capitulum dimensions and (disc-) floret number still explained 7–20% of the corresponding variances. Marker prediction in absolute seed set (only significant in F2B) was poor, with an r2 value of 5%. With the exception of branch number, variation in the discoid class [TnTn] was generally in the direction of var. vulgaris. For example, discoid F2B plants possessed on average 20.0 leaves, while the homozygous radiate individuals showed on average 27.5 leaves—a difference which equaled 65.8% of the difference in the parental controls. Similarly the mean difference in leaf length associated with the two homozygous classes was 2.4 cm in both F2 progenies, corresponding to 82.8% of the parental difference. For a large number of characters in each F2 progeny, the heterozygous radiate [TrTn] marker class did not differ significantly (P , .01) from the discoid one [TnTn]. These included: leaf num- 58 The Journal of Heredity 1998:89(1) ber, pedicel length, bract number, capitulum length and shape, and floret number. Dominance Relationships at QTLs For those traits found to be significantly associated with alleles at the marker locus, a wide range of gene effects were observed in both F2 progenies ( Table 3). Approximately 25% ( F2A) and 40% ( F2B) of these characters had ratios of dominance to additive effects of 0.55 or less, indicating additive or only partially dominant gene action ( Edwards et al. 1987). A comparatively large proportion of traits, 62% in F2A and 47% in F2B, were partially dominant or dominant, with values ranging between 0.55 and 1.25. Only a small fraction of traits, 13% in both F2A and F2B, exhibited ratios exceeding 1.25, implying overdominance/underdominance (i.e., heterosis) for chromosomal regions associated with the ray floret locus. In each F2 progeny, (partial) dominance or heterosis was generally in the direction of the discoid individuals, that is, in the direction of var. vulgaris ( Tables 2 and 3). Of those traits significantly associated with the marker locus, only branch number in F2B showed overdominance in the direction of subsp. denticulatus. Pronounced additivity (ø d/a ø 0) was found for leaf length, width, and area ( both latter traits only significant in F2B). For all traits consistently associated with the ray floret locus, gene action did not vary markedly among progenies. Linkage Relationship Between the Ray Floret Locus and the Major Gene for Speed of Development A previous genetic study (Comes and Kadereit 1996) involving the present cross indicated that speed of development in S. vulgaris is controlled by a major gene ( here arbitrarily called [G/g]) with the trait expressions ‘‘early’’ and ‘‘late,’’ and with homozygous recessive plants exhibiting late development. A total of 329 F2 individuals were classified for ray floret length, based on the criteria given above, and for the time to bud formation, given a discrimination point of 42 days between early and late developing plants and assuming a 1.5% probability of early/late misclassification ( Figure 1). With parental genotypes of var. vulgaris and subsp. denticulatus designated as [GGTnTn] and [ggTrTr], respectively, and assuming independent assortment of the speed of development locus and the ray floret locus, the F2 was expected to provide six phenotypic classes in a Mendelian ratio of 3:6:3:1:2:1. A chi-square contingency test revealed a highly significant deviation from the Mendelian expectation (pooled data F2A and F2B, x2[2] 5 137.2, P , .001). However, the recombination frequency between the pair of loci remained rather high, r 5 0.15 6 0.02 (15.37 cM 6 2.10), suggesting only ‘‘loose’’ linkage and the possibility of recombination events between the ray floret locus and the major gene controlling speed of development. Discussion Association of Alleles at the Ray Floret Locus With Quantitative Characters Determining the genetic basis of phenotypic changes during plant speciation is critical both to understanding how and at what rate species respond evolutionarily to changing environmental conditions and to evaluating alternative genetic models of speciation. To do this requires the availability of adequate model systems of comparatively young geologic age, allowing direct insights into the genetic changes involved. Taxa that have recently evolved into weeds and their nearest nonweedy relatives provide such excellent experimental material. In this study, the genetic analysis of the nonweedy S. vulgaris subsp. denticulatus and its weedy derivative var. vulgaris revealed that a significant portion of the vegetative and floral differences between the two taxa can be explained by the segregation of the ray floret locus. With respect to the larger F2B progeny and using the sequential Bonferroni procedure for declaring a marker effect significant, this major gene, that is, the chromosomal region associated with it, controlled more than 10% of the phenotypic variation of 9 of the 17 characters analyzed ( Table 3B). However, an important caveat must be introduced at this point. Throughout this article, significant associations are interpreted as indicating linkage between the ray floret marker and individual QTLs affecting the trait, but given the linkage relationship between the ray floret marker and the major gene controlling speed of development (15.37 cM 6 2.10), pleiotropic or developmental (epistatic) effects of each of the two major genes on all quantitative traits measured may not be excluded a priori. However, in a preceding study (Comes 1994) it was found that variation in none of these traits is controlled by a single gene locus, so pleiotropic effects of the major genes would be highly unlikely to create the association patterns ob- Table 3. Number of observations (N), mean values (X), and standard errors (SE), F values, results of Tukey’s test, and fraction of phenotypic variation explained by the ray floret locus (r2) for characters analyzed in discoid [TnTn], heterozygous radiate [TrTn], and homozygous radiate [TrTr] individuals of two F2 Senecio progenies F 2A (Panel A) and F 2B (Panel B) Discoid [TnTn] Short-rayed [TrTn] N X SE Vegetative STH NL PL NB LIN LL WL SL AL 28 28 27 27 27 25 25 25 25 17.1a 20.0b 6.7a 13.9a 28.9a 7.2b 2.2a 3.4a 5.9a 0.89 0.60 0.49 0.47 1.11 0.25 0.13 0.18 0.49 Floral NOIB LCAP WCAP SCAP NDISC NFLORET 27 26 26 26 27 27 14.7b 8.52a 4.63a 1.85a 66.9a 66.9b Reproductive NACH SET 21 21 Vegetative STH NL PL NB LIN LL WL SL AL N Long-rayed [TrTr] ø d/a X SE N X SE F value P value r2 66 66 61 63 63 56 56 56 53 18.2a 20.7b 6.4a 13.5a 28.5a 7.7ab 2.3a 3.3a 6.5a 0.69 0.46 0.30 0.36 0.89 0.16 0.06 0.06 0.30 27 27 27 23 23 19 19 19 19 20.6a 26.9a 3.3b 12.5a 27.2a 8.4a 2.4a 3.5a 6.6a 1.25 0.89 0.35 0.51 1.50 0.25 0.08 0.12 0.31 2.79 29.70 20.49 1.89 0.41 5.78 0.73 0.79 0.72 0.0637 ,0.0001* ,0.0001* 0.1537 0.6701 0.0044* 0.4888 0.4604 0.4937 0.05 0.33 0.28 0.03 0.01 0.11 0.02 0.02 0.02 20.37 20.80 0.82 0.43 0.53 20.17 0.00 23.00 0.71 v v v d v v — v d 0.65 0.110 0.078 0.035 2.39 2.39 66 66 66 66 66 66 15.2b 8.36a 4.72a 1.78a 58.4b 68.9b 0.48 0.077 0.049 0.022 3.79 1.20 27 27 27 27 27 27 20.4a 7.70b 4.79a 1.62b 67.1a 79.0a 1.21 0.118 0.090 0.034 2.52 2.74 14.93 14.43 0.96 12.13 9.10 9.25 ,0.0001* ,0.0001* 0.3879 ,0.0001* 0.0003* 0.0002* 0.20 0.20 0.02 0.17 0.13 0.14 20.83 0.61 0.13 0.40 286.00 20.67 v v d v v v 18.7a 28.0a 3.57 5.11 55 55 27.1a 39.4a 2.54 3.46 18 18 31.4a 38.8a 4.89 5.53 2.45 1.69 0.0899 0.1883 0.05 0.04 0.32 1.11 d d 51 51 46 52 52 47 47 47 47 19.9b 20.0b 7.6a 13.8ab 32.0a 7.3c 2.4b 3.3a 6.1c 0.82 0.52 0.41 0.46 0.92 0.19 0.08 0.15 0.29 103 103 101 103 102 93 93 93 93 21.4ab 21.5b 7.0a 14.5a 32.4a 8.5b 2.6ab 3.3a 7.5b 0.55 0.35 0.25 0.25 0.70 0.15 0.05 0.06 0.22 52 52 49 48 48 32 32 32 32 23.7a 27.5a 3.6b 12.6b 27.8b 9.7a 2.8a 3.6a 8.9a 0.78 0.58 0.33 0.41 0.96 0.22 0.08 0.11 0.41 5.87 60.65 37.33 7.45 7.73 28.44 6.74 1.88 16.78 0.0035* ,0.0001* ,0.0001* 0.0009* 0.0007* ,0.0001* 0.0069* 0.1534 ,0.0001* 0.05 0.37 0.28 0.07 0.07 0.25 0.07 0.02 0.17 20.21 20.60 0.70 2.17 1.19 0.00 0.00 21.00 0.00 v v v d v — — v — Floral NOIB LCAP WCAP SCAP NDISC NFLORET 52 51 51 51 51 51 14.4b 8.50a 4.54b 1.88a 66.7a 66.7b 0.47 0.093 0.062 0.026 1.38 1.38 105 105 105 105 105 105 14.6b 8.45a 4.64ab 1.84a 57.2b 67.3b 0.36 0.056 0.040 0.017 0.92 1.03 52 52 52 52 52 52 20.8a 8.00b 4.84a 1.66b 67.6a 79.3a 0.73 0.093 0.051 0.023 1.57 1.70 45.26 11.70 7.38 25.47 25.44 24.69 ,0.0001* ,0.0001* 0.0009* ,0.0001* ,0.0001* ,0.0001* 0.31 0.10 0.07 0.20 0.20 0.19 20.94 0.80 20.33 0.64 222.11 0.91 v v v v v v Reproductive NACH SET 50 50 21.7b 31.0a 2.63 3.62 88 88 26.2ab 38.3a 2.01 2.78 41 41 34.2a 44.6a 3.28 3.86 4.83 3.24 0.0091* 0.0405 0.05 0.04 20.28 0.07 v d Trait Da Panel A Panel B a Direction of the dominant effect, that is, if the heterozygote marker class resembles var. vulgaris (v) or subsp. denticulatus (d). Group means sharing the same superscript are not significantly (P , .01) different. P values from one-way ANOVAs found to be significant at a table-wide a 5 0.05 with sequential Bonferroni adjustments (Rice 1989) are indicated by asterisks, otherwise values are nonsignificant. Also given is the approximate ratio of dominance to additive effects (ø d/a). served. Nonetheless, it is feasible that a significant portion of the variance in several quantitative traits may be influenced by developmental constraints imposed by the epistatic interaction of alleles at each of the two major genes with different regions of the genome, and this possibility requires further investigation. It has to be pointed out that assaying the ray floret gene as a codominant marker is clearly advantageous compared to its use as a qualitative, dominant-recessive one with the character states ‘‘radiate’’ and ‘‘discoid.’’ Subsuming the heterozygote marker class into the dominant class of ‘‘radiate’’ individuals normally would have resulted in an underestimate of the r2 value of the regression model and its overall significance, with not a single comparison in the F2A, and only 47% rather than 70% of the comparisons showing P values of less than .001 in the F2B (Comes HP, unpublished data). Limited statistical resolution of dominant marker-based predictability of phenotypic trait expression is most likely due to the fact that the direction of effects of many QTLs analyzed here do not match with the a priori assumption that the presence of a single dominant denticulatus allele [R] at the ray floret locus is already associated with a denticulatus-like phenotype. This is in- ferred from the above comparisons of discoid [TnTn] and heterozygous radiate [TrTn] marker classes, revealing nonsignificant (P , .01) differences for several vegetative and floral traits ( Table 3). Since presence versus absence of ray florets in various Asteraceae has often been claimed to have a simple genetic basis, the results presented here can be contrasted with other recent investigations into the suite of characters associated either with capitulum variation within populations or the evolutionary shift from the radiate to the discoid condition among closely related taxa. Thus British populations of S. vulgaris often contain two ca- Comes • Cosegregation in Senecio vulgaris 59 pitulum morphs (radiate and discoid) in addition to a rare short radiate one, a situation most likely resulting from introgressive hybridization involving S. squalidus L. (e.g., Abbott 1992). Extensive genecological research has demonstrated that differences between the radiate and discoid morphs do exist for various life-history traits, such as germination behavior, flowering time, and reproductive capacity (reviewed by Abbott 1986; Theaker and Briggs 1992). In a recent article Oxford et al. (1996) provide cosegregational evidence suggesting that the reasons for these differences are complex and may vary across populations. The different degrees of associations found among the three populations studied appear to reflect various stages in an evolutionary scenario that includes (1) introgression of an S. squalidus linkage group containing the radiate allele, among others; (2) its subsequent disruption by chromosomal rearrangements over time; and (3) the persistence of resulting chance associations because of low levels of outcrossing. At this point it is certainly premature to speculate on any structural similarities or even evolutionary relationships between the gene cluster identified by Oxford et al. (1996) and the hypothesized linkage group described in the present article. Major gene associations of capitulum type with quantitative characters have also been reported in an interspecific cross of the progenitor-derivative pair Layia glandulosa ( Hook.) Hook. & Arn. and L. discoidea Keck ( Ford and Gottlieb 1989). Although radiate and discoid F2 plants were found to differ significantly in a number of floral traits, they did not differ consistently in vegetative ones. Moreover, in a series of backcross progenies with L. discoidea as recurrent parent, radiate and discoid plants only differed slightly in pappus length, but not in other traits. In Layia, however, complete classification of marker genotypes in hybrid progenies, and consequently estimates of gene effects at individual QTLs putatively linked to the morphological marker, cannot be straightforwardly obtained because variation in capitulum type is controlled in a complex manner by two epistatically interacting genes ( Ford and Gottlieb 1990). Amount and Direction of Dominance at QTLs According to theoretical models ( Broadhurst 1979; Fisher 1958; Mather and Jinks 1982), natural selection is expected to promote dominance of favorable alleles, 60 The Journal of Heredity 1998:89(1) masking the less fit, supposedly intermediate phenotype of the heterozygote. As expected, in the case of significant (P , .01) differences between the means of discoid [TnTn] and homozygous radiate [TrTr] F2 plants, all discoid plants varied in the direction of S. vulgaris var. vulgaris, while all radiate ones varied in the direction of subsp. denticulatus ( Table 3). The more interesting finding was that in both F2 progenies the heterozygote marker class had a large number of characters more similar to those of the fast developing derivative var. vulgaris than to those of the slow developing progenitor subsp. denticulatus. These characters included leaf number, pedicel length, internode length, bract number, capitulum length and shape, and total floret number. All these traits exhibited high dominance:additive ratios (0.61 # d/a # 1.35) at individual QTLs. Accordingly these results may indicate that rapid development together with the reduction of vegetative and reproductive structures had been under (strong) directional selection in the past, probably enabling the derivative taxon to utilize alternative, weedy habitats. That directional selection may not have been acting directly on these structures but on the pleiotropic effects of the postulated gene for speed of development seems rather unlikely, as pointed out above. In an automatically self-fertilizing species like S. vulgaris, the observation that dominant gene action is predominant may be somewhat surprising, since fixation probabilities for recessive and dominant mutations are likely to be similar under inbreeding. It should be kept in mind, however, that the populations studied here occur in routinely—naturally or humanly—disturbed habitats. For this reason, it can be assumed that these populations are permanently confronted with drift, founder events, etc. Under such a setting, recessive mutations are more likely to be lost both on a local and a wider scale unless the effect of strong inbreeding is combined with selection (Caballero et al. 1991; Caballero and Hill 1992). Conclusions The present study does not support the assumption that ecotypic differentiation or speciation in the wild often involves selection in quantitative polygenic systems encompassing the joint effect of a large number of unlinked loci, each with a relatively small, additive effect on the phenotype (e.g., Charlesworth 1984; Coyne and Lande 1985). Instead, quantitative characters in S. vulgaris such as leaf number, pedicel length, and bract number were shown to be controlled by chromosomal regions with comparatively large, nonadditive effects (i.e., with r2 values of 28–37%; and [ø d/a] values of between 0.60 and 0.94). In considering that these regions together with the major gene for speed of development are presumably linked to the ray floret gene, a single chromosomal region within S. vulgaris seems likely to exhibit a substantial proportion of gene effects that contribute considerably to the identity of the weedy var. vulgaris. A similar pattern is seen with subspecies of Plantago major L., where in particular the Pgm-1 marker locus forms a gene complex with a substantial number of loci affecting ecologically important characters (van Dijk 1984). In addition, extensive molecular QTL studies have provided evidence that a few chromosomal regions of large and often dominant effect are responsible for key morphological differences between maize and its wild progenitor teosinte ( Doebley and Stec 1993; Doebley et al. 1995). If gene complexes encompassing genes of large nonadditive effects are an important feature of the genetic architecture not only of recently evolved domesticated but also natural plant species, then the practice of ignoring such major genetic coadaptation in quantitative genetic models of speciation (e.g., Barton and Turelli 1989; Charlesworth 1984; Coyne and Lande 1985) must be questioned. References Abbott RJ, 1986. Life history variation associated with the polymorphism for capitulum type and outcrossing rate in Senecio vulgaris L. Heredity 56:381–391. Abbott RJ, 1992. Plant invasions, interspecific hybridization and the evolution of new plant taxa. Trends Ecol Evol 7:401–405. Abbott RJ, Ashton PA, and Forbes DG, 1992. Introgressive origin of the radiate groundsel, Senecio vulgaris L. var. hibernicus Syme: Aat-3 evidence. Heredity 68:425– 435. Alexander JCM, 1979. The Mediterranean species of Senecio sections Senecio and Delphinifolius. Note R Bot Garden Edinburgh 37:387–428. Allard RW, 1956. Formulas and tables to facilitate the calculation of recombination values in heredity. Hilgardia 24:235–278. Barton NH and Turelli M, 1989. Evolutionary quantitative genetics: how little do we know? Annu Rev Genet 23:337–370. Bradshaw HD Jr, Wilbert SM, Otto KG, and Schemske RW, 1995. Genetic mapping of floral traits associated with reproductive isolation in monkeyflowers (Mimulus). Nature 376:762–765. Broadhurst PL, 1979. The experimental approach to behavioral evolution. In: Theoretical advances in behav- iour genetics (Royce JR and Mos LP, eds). Alphen aan de Rijn: Sijthoff and Noordhoff; 43–95. of mating system divergence in the yellow monkeyflower species complex. Heredity 73:422–435. Caballero A and Hill WG, 1992. Effects of partial inbreeding on fixation rates and variation of mutant genes. Genetics 131:493–507. Fisher RA 1958. The genetical theory of natural selection. New York: Dover. Caballero A, Keightley PD, and Hill WG, 1991. Strategies for increasing fixation probabilities of recessive mutations. Genet Res 58:129–138. Charlesworth B, 1984. The evolutionary genetics of life histories. In: Evolutionary ecology (Shorrocks B, ed). Oxford: Oxford University Press; 117–133. Comes HP, 1994. Genetische Analyse phänologischer und morphologischer Unterschiede innerhalb des nahverwandten Sippenpaares Senecio vulgaris L. und S. vernalis Waldst & Kit (Asteraceae) (PhD dissertation). Heidelberg: Ruprecht-Karls-Universität. Comes HP, 1995a. Genecological and isozyme studies in Senecio vernalis Waldst. & Kit. and S. vulgaris L. var. vulgaris from Central Europe and Israel. Flora 190:1–24. Comes HP, 1995b. Senecio vulgaris L. subsp. denticulatus (OF Muell.) PD Sell and S. vulgaris subsp. vulgaris var. vulgaris on Jersey (Channel Islands). Watsonia 20: 205–227. Comes HP and Kadereit JW, 1996. Genetic basis of speed of development in Senecio vulgaris L. var. vulgaris and subsp. denticulatus (OF Muell.) PD Sell, and Senecio vernalis Waldst & Kit. Heredity 77:544–554. Coyne JA and Lande R, 1985. The genetic basis of species differences in plants. Am Nat 126:141–145. Ford VS and Gottlieb LD, 1989. Morphological evolution in Layia (Compositae): character recombination in hybrids between L. discoidea and L. glandulosa. Syst Bot 14:284–296. Perring FH and Sell PD, 1968. Critical supplement to the atlas of the British flora. London: T Nelson and Sons. Ford VS and Gottlieb LD, 1990. Genetic studies of floral evolution in Layia. Heredity 64:29–44. Rice WR, 1989. Analyzing tables of statistical tests. Evolution 43:223–225. Hill WG, 1982. Rates of change in quantitative traits from fixation of new mutations. Proc Natl Acad Sci 79: 142–145. SAS Institute Inc., 1988. SAS/STAT user’s guide. Cary, North Carolina: SAS Institute. Hull P, 1974. Self-fertilisation and the distribution of the radiate form of Senecio vulgaris L. in central Scotland. Watsonia 10:69–75. Kadereit JW, 1984a. The origin of Senecio vulgaris (Asteraceae). Plant Syst Evol 145:135–153. Kadereit JW, 1984b. Studies on the biology of Senecio vulgaris L ssp. denticulatus (OF Muell) PD Sell. New Phytol 9:681–689. Kimura M, 1983. The neutral theory of molecular evolution. Cambridge: Cambridge University Press. Kosambi DD, 1944. The estimation of map distances from recombination values. Ann Eugen 12:172–175. Mather K and Jinks JL, 1982. Biometrical genetics. London: Chapman and Hall. Mayr E, 1963. Animal species and evolution. Cambridge: Harvard University Press. Doebley J and Stec A, 1993. Inheritance of the morphological differences between maize and teosinte: comparison of results for two F2 populations. Genetics 134: 559–570. Macnair MR and Cumbes QL, 1989. The genetic architecture of interspecific variation in Mimulus. Genetics 122:211–222. Doebley J, Stec A, and Gustus C, 1995. Teosinte branched 1 and the origin of maize: evidence for epistasis and the evolution of dominance. Genetics 141: 333–346. Mitchell-Olds T, 1996. Genetic constraints on life-history evolution: quantitative-trait loci influencing growth and flowering in Arabidopsis thaliana. Evolution 50:140–145. Edwards MD, Stuber CW, and Wendel JW, 1987. Molecular-marker-facilitated investigations of quantitativetrait loci in maize. I. Numbers, genomic distribution and types of gene action. Genetics 116:113–125. Mitchell-Olds T and Rutledge JL, 1986. Quantitative genetics in natural plant populations: a review of the theory. Am Nat 127:379–402. Fenster CB and Ritland K, 1994. Quantitative genetics Oxford GS, Crawford TJ, and Pernyes K, 1996. Why are capitulum morphs associated with other characters in natural populations of Senecio vulgaris (groundsel)? Heredity 76:192–197. Orr HA and Coyne JA, 1992. The genetics of adaptation: a reassessment. Am Nat 140:725–742. Ren Z and Abbott RJ, 1991. Seed dormancy in Mediterranean Senecio vulgaris L. New Phytol 117:673–678. Sokal RR and Rohlf FJ, 1981. Biometry, 2nd ed. New York: WH Freeman. Stebbins GL, 1965. Colonizing species of the native California flora. In: Genetics of colonizing species ( Baker HG, ed). New York: Academic Press; 173–192. Steel RGD and Torrie JH, 1980. Principles and procedures of statistics. New York: McGraw-Hill. Stuber CW, 1989. Isozymes as markers for studying and manipulating quantitative traits. In: Isozymes in plant biology (Soltis D and Soltis P, eds). Portland, Oregon: Dioscorides Press; 206–220. Suiter KA, Wendel JF, and Case JS, 1983. LINKAGE-1: a PASCAL computer program for the detection and analysis of genetic linkage. J Hered 74:203–204. Theaker AJ and Briggs D, 1992. Genecological studies of groundsel (Senecio vulgaris L) III. Population variation and its maintenance in the University Botanic Garden, Cambridge. New Phytol 121:281–291. Trow AH, 1912. On the inheritance of certain characters in the common groundsel—Senecio vulgaris Linn— and its segregates. J Genet 2:239–276. van Dijk H, 1985. Genetic variability in Plantago species in relation to their ecology. 2. Quantitative characters and allozyme loci in P. major. Theor Appl Genet 68:43– 52. Received September 17, 1996 Accepted May 5, 1997 Corresponding Editor: Jonathan F. Wendel Comes • Cosegregation in Senecio vulgaris 61
© Copyright 2026 Paperzz