Genetic Basis of Speed of Development in Senecio Vulgaris L

Home , Senecio, Senecio vernalis, Senecio vulgaris

Heredity 77 (1996) 544—554 Received 19 January 1996

Genetic basis of speed of development in Senecio vulgaris L. var. vulgaris, S. vulgaris ssp. denticulatus (O.F. Muell.) P.D. Sell, and Senecio vernalis Waldst. & Kit.

HANS PETER COMES* & JOACHIM W. KADEREIT Institut für Spezielle Botanik u. Botanischer Garten, Johannes Gutenberg-Universitat Mainz, Bentze/weg 9, 55099 Mainz, Germany

Thegenetic basis of differences in speed of development from germination to first bud formation was investigated in Senecio vulgaris var. vulgaris and S. vulgaris ssp. denticulatus, and also in S. vernalis sampled from Israel and Germany. In the case of S. vulgaris, F2 segregation analysis and the recovery of very late and very early lines from extreme F2 phenotypes showed that differences can be explained by a single major gene model, whereas segregation data from F2 and backcross progenies in S. vernalis are not incompatible with a digenic model of inheritance. Senecio vernalis from Israel and S. vulgaris var. vulgaris reached the different developmental stages in a substantially shorter time than did C. European S. vemalis and S. vulgaris ssp. denticulatus. In both species the increased speed of development was achieved through drastic abbreviation of the rosette stage of development rather than through an acceleration of the entire developmental programme. It is suggested that the genes controlling speed of development in S. vulgaris and S. vemalis may be homologous to major heterochronic genes identified from mutants of Arabidopsis thaliana (e.g. early flowering [elfl—3]).

Keywords:earlyflowering mutants, genetic basis, life history trait, Senecio vemalis, Senecio vulgaris, speed of development.

Introduction required for development from germination to first flower bud formation, anthesis, and fruiting. Infor- Speedof development is an important variable in mation concerning the genetic control of these flow- theories of life history evolution (Sibly & Antono- ering characters is rapidly increasing in well-defined vics, 1992), as it is generally assumed to affect total experimental mutant systems, particularly Arabidop- lifetime fecundity (Lewontin, 1965; MacArthur & sis thaliana (L.) Heynh. (Koorneef et aL, 1991; Sung Wilson, 1967). At the macroevolutionary level, varia- et aL, 1992; Zagotta et at., 1992; Bernier et at., 1993), tion in developmental timing in a derivative taxon and many species of agronomic importance, for relative to its progenitor is commonly referred to as example maize (e.g. Freeling et at., 1992; see also heterochrony (Diggle, 1992). A full understanding of Bernier, 1988 for a review). In addition, other such major evolutionary changes requires an under- cereals with genomes suitably mapped at the standing of the genetic basis of phenotypic differ- molecular level have recently been shown to provide entiation. A knowledge of the inheritance of speed excellent material for RFLP marker studies of major of development is also of value in testing key genes controlling flowering time and developmental assumptions of genetic models of life history varia- rate (Triticum aestivum, Law et al., 1993; Hordeum tion (Anderson & King, 1970; Lande, 1982; Charles- vulgare, Laurie et at., 1994). In contrast, the genetics worth, 1984, 1990). of developmental timing in most wild plant taxa is Developmental timing in plant species can be poorly known (Murfet, 1977; Bernier, 1988). More- objectively determined by the number of days over, only very few studies of natural plant populations have provided evidence of the type and *Correspondence: School of Biological and Medical Sciences, direction of evolutionary change in developmental University of St Andrews, St Andrews, Fife KYI6 9TH, U.K. timing by merging qualitative or quantitative genetic

544 1996The Genetica! Society of Great Britain. GENETICS OF SPEED OF DEVELOPMENT IN SENEC/O 545 information with hypotheses of progenitor-derivative rare taxon from the coasts of W. Europe and the relationships (Stephanomeria, Gottlieb, 1973, 1982; mountains of S. Spain and Sicily (Kadereit, 1984a, Mimulus, Macnair & Cumbes, 1989; Fenster & and unpublished results), where it occurs in both Ritland, 1994; Fenster et a!., 1995). A general natural and disturbed habitats (Comes, 1995a; conclusion gained from the studies cited above is Kadereit, unpublished results). Morphologically, var. that there is wide variation in the genetics of devel- vulgar/s and ssp. denticulatus are normally distin- opmental timing, and consequently there is no single guished by the presence of ray fiorets in the latter. type of genetic system —eitherMendelian inherit- Presence vs. absence of ray fiorets is governed by ance involving only a few genes with major effects, alternative alleles at a single locus (Trow, 1912). The or polygenic inheritance —thatunderlies the two taxa differ dramatically in two life history phenotypic differences. characters, with var. vulgar/s normally exhibiting no In this paper we examine the genetic basis of pronounced seed dormancy and up to twice the intraspecific differences in speed of development speed of development of ssp. denticulatus (Kadereit, from germination to first reproduction in two closely 1984a; Comes, 1995a). The difference in seed related species of Senecio L. (Compositae), S. dormancy appears to be regulated by a single gene vulgaris L. and S. vemalis Waldst. & Kit. Substantial with two alleles in ssp. denticulatus from Jersey differences were analysed between S. vulgaris var. (Kadereit, 1984a). Presence of seed dormancy was vulgaris and S. vulgaris ssp. denticulatus (O.F. Muell.) recessive to the absence of seed dormancy in British P. D. Sell on the one hand, and between populations var. vulgar/s. Seed dormancy, however, has also been of S. vemalis from Israel and Germany on the other shown to occur in var. vulgar/s from the Mediterra- hand. An almost twofold difference in speed of nean, where the winter-rain climate does not permit development has been observed in the former case the formation of more than one generation per year (Kadereit, 1984a, and unpublished results; Comes, (Ren & Abbott, 1991; Comes, 1995b). Seed 1995a), and a difference of almost identical magni- dormancy here appears to be regulated digenically tude was reported more recently in material of S. (Comes, 1995b). vernalis from different geographical regions (Comes, On the basis of morphological/phenological, 1995b). geographical and ecological considerations it has The genetic basis of the large difference in speed been argued that ssp. denticulatus should be of development in both S. vulgaris and S. vemalis is regarded as the progenitor of the weedy var. the subject of the present analysis. In the case of the vulgar/s, or as close to such (Kadereit, 1984b). On former taxon, it is conceivable that its polyploid the assumption that the differentiation of var. genomic constitution may lead to difficulties in inter- vulgar/s coincided with its geographical/ecological preting patterns of segregation. However, there is separation from ssp. denticulatus, var. vulgar/s may well-documented evidence in S. vulgaris of both have arisen in late glacial or early postglacial times. disomic inheritance at the ray floret locus (Trow, At that time ssp. denticulatus, which was probably 1912) and fixed heterozygosity at a number of distributed in Mediterranean coastal habitats during isozyme loci (AAT-3,c.EST-1,PGI-2; Ashton & the glacial period, reacted to climatic changes by Abbott, 1992), thus justifying the assumption that retreating into higher altitudes in the Mediterranean the species contains a diploidized genome. area, and north along the coasts, i.e. into those areas where it can be found today. At that time many new habitats would have become available for var. Material and methods vulgar/s. Senecio vulgar/s Senec/overnal/s Senecio vulgar/s (common groundsel, 2n =4x=40) consists of three taxonomic units, namely ssp. Seneciovemalis (spring groundsel, 2n =2x=20)is vulgans var. vulgaris, ssp. vulgar/s var. hibernicus widely distributed in C. to E. Europe, S. W. Asia, Syme, and ssp. denticulatus. Subspecies vulgar/s var. Siberia, S. Russia, and Afghanistan (Alexander, vulgar/s and ssp. denticulatus were the two taxa of S. 1979). The expansion of its range into C. Europe in vulgaris investigated in the present study. the last 200 years has been well documented (Hegi, Var. vulgar/s is considered to be native to W. 1987). In C. Europe S. vemalis mostly behaves as a Eurasia (Kadereit, 1984b), but is now distributed winter annual and is polymorphic for seed more or less worldwide, mainly as a ruderal and dormancy, which is regulated by a recessive allele of agricultural weed. In contrast, ssp. denticulatus is a a single gene (Comes, 1995b). In contrast, material

The Genetical Society of Great Britain, Heredity, 77, 544—554. 546 H. P. COMES&J. W. KADEREIT from the E. Mediterranean appears to be fixed for data were pooled. This analysis also revealed that seed dormancy, and develops substantially faster neither block (i.e. local environmental) factors nor from germination to first reproduction (Comes, family x block effects (i.e. gene-environment interac- 1995b). tions) were significant (data not shown). The Although there is no doubt that the C. European material was scored daily for the number of days populations of S. vernalis are of introduced status, it from (1) germination to bud formation, (2) germina- is not known whether this material is directly tion to first anthesis, and (3) germination to first derived from populations in the E. Mediterranean. fruiting. It seems likely, however, that E. Mediterranean To investigate the underlying genetic model populations of S. vernalis lie within the range the further, a simple one-generation divergent selection species occupied before man started to alter the experiment was carried out. Nine individuals from 'natural' distribution of the species. In this respect, the F2 were selected for extreme low and high therefore, C. European populations may be consid- phenotypic values for the traits under study. Selec- ered as derived, relative to E. Mediterranean popu- ted plants were selfed and used for the production lations. Truly native populations of S. vernalis in E. of the same number of F3 families of, on average, 26 Europe may ecologically be very similar to the intro- individuals (range 8—40). Between April and August duced C. European material. 1991, the F3s were raised in a fully randomized outdoor bed together with selfed progeny of two unselected individuals of each parental stock (16—36 Crosses plants per family). In a second experiment, 10 plants of each parental Crosseswere made between S. vulgaris var. vulgaris taxon were grown in a fully randomized design (v) and S. vulgaris ssp. denticulatus (d) using material under a natural light regime in an unheated green- from the Botanic Garden, Heidelberg University house and scored for both stem height and number (Germany), and from Jersey (Channel Islands; of leaves at first bud formation, first anthesis, and dunes, Les Quennevais, N. of La Pulente, F. Le first fruiting. Sueur, May, 1986), respectively. One individual of v Crosses were also made between S. vernalis from was used as the female parent, and two individuals C. Europe (c) and from Israel (i). F1 progeny was of d as male parents. One plant each from the two produced by two reciprocal crosses between material F1 families obtained was used for the production by collected as rosettes west of Grenzhof/Eppelheim self-pollination of two F2 families, A and B. near Heidelberg (topographic map 1: 25000, No. Because S. vulgaris has small capitula (width 6517, Mannheim-SUdost, RZ 3469.70, HZ 5476.25), <6 mm) and is extremely self-fertile, all F1 hybrids and material grown from seed collected in Israel were produced using Alexander's (1979) emascula- (Jerusalem; ruderal, near Sederot Levi Eshkol/ tion technique, and were verified by the presence of Nablus Road, H. P. Comes, 12 April, 1990). Further short, stubby ray fiorets, i.e. they were heterozygous F1 families were raised at the same time (i.e. F1 A: for parental, alternative alleles at the ray fioret locus c x i; F1B, F1C: i x c). These F1 individuals were (Trow, 1912). The degree of residual selfing on cross-pollinated at random, and used as male emasculated plants in the absence of cross pollen parents for the production of each of two separate was low. Of a total of 160 v xd F1 plants raised in families of F2, and backcross progenies (BC1: c x F1; this study only six (3.8 per cent) proved to be BC2: i x F1), respectively. As most material of S. rayless, apparently resulting from accidental self- vernalis is self-incompatible, no emasculation of fertilization of the female parent (v). female parents was necessary; however, before cross- Between July and October 1990, the parental taxa pollination, pollen of the female parents was washed (N =93and 129 for ssp. denticulatus and var. off the capitula. Between April and August 1992, the vulgaris, respectively), the F1 hybrids (N 151), and parental taxa (N =64and 73 for S. vernalis from C. the two F2 families (N =128 and 220 for A and B, Europe and Israel, respectively), their F1s (N =65) respectively) were cultivated in a single experiment and F2s (N =65),and the backcross to the C. Euro- in order to reduce the environmental variance. pean parent (N= 116) were cultivated in 8cm Plants were grown in 10 cm plastic pots under plastic pots in a completely randomized outdoor outdoor conditions using six randomized blocks. bed. No BC2 seed germinated. The material was Two-way analysis of variance (ANOVA) comparison scored as described above for the two S. vulgaris revealed that the two F2 families were not signifi- experiments. Within the F1, F2, and BC1 progenies, cantly (P<0.01) different from each other, so their no significant differences between families (of either

The Genetical Society of Great Britain, Heredity, 77, 544—554. GENETICS OF SPEED OF DEVELOPMENT IN SENEC/Q 547 reciprocal or nonreciprocal types) were observed difference in variance is likely to represent a scale and their data were pooled in all subsequent effect. analyses. The frequency distribution of the F2 shows a separation of 'early' and 'late' developing plants in a bimodal fashion, but there is also an undefined zone Statisticalprocedures of overlap. In a normal probability plot of the overall distribution the section below 40 days proved Therelatively large F2 and BC1 segregation populations produced for S. vulgaris and S. vernalis, respec- to be satisfactorily linear. On the assumption that tively, were examined for evidence of separation into these plants were all early developing, a normal normally distributed series differing in developmen- curve truncated at 40 days was fitted onto the lower tal timing. To superimpose normal curves onto each part of the histogram and extrapolated into the zone putative series we adopted a simplified maximum of overlap of both putative F2 types. Truncated likelihood solution of truncated normal distributions distributions were then iteratively fitted to the right (Cohen, 1959). The expected frequencies were side of the antimode (41—42 days), thereby account- computed from the estimated mean and standard ing for the portion of overlapping plants from the deviation, summed over series, and compared with early developing subgroup. Because the late observed frequencies of the entire distribution by developing series was obviously right-tailed, both the chi-square analysis with n—6 d.f., where n is the arcsin of the data was taken and three probable number of pairwise comparisons (Bliss, 1967; pp. outliers (* in Fig. 1), identified by exploratory data 162—167). Truncation points and transformations analysis (MINITAB, 1991), were rejected to improve were chosen so as to produce reasonably normal normality. The best fit between expected frequen- distributions and to maximize the agreement cies, summed over both series, and observed between observed and expected frequencies. Fitted frequencies resulted in placing an upper truncation curves were then used to predict the number of point at 43 days (X4= 30.23,0.7>.P>0.6). Figure 1 plants belonging to each series, and the estimates, in shows that the expected frequencies for the 'early' turn, were subjected to Mendelian analyses in order and 'late' series (smooth lines) are very close to the to test the adequacy of a simple genetic model. observed F2 data. It is important to note that locat- ing upper boundaries at 44 or 45 days, and no other change, gave a similarly good agreement to the Results above result, but including the outliers resulted either in a less satisfactory fit (= 47.88,46.48, Notunexpectedly, the phenological traits (time to 0.5 >P> 0.1, for 43 and 44 days, respectively) or bud formation, first anthesis, and first fruiting) had even in a significant discrepancy (5= 65.57, very similar frequency distributions and significantly P<0.05, for 45 days). high positive phenotypic correlations with each other According to our best fit, the total sample of 345 in all parental generations and their progeny (e.g. observed F2 plants (without outliers) can be for S. vulgaris F2B: r=0.94—0.98, N 203; for S. subsumed in two normal curves, one including an vemalis BC1: r =0.91—0.96, N 105). Accordingly, expected number of 267 early, and the other an one of these traits was considered accurate enough expected number of 80 late developing plants. Thus to be used as an indicator of general developmental taken over both fitted curves (early and late) only timing, and only the recordings for time to bud two more plants were expected than were observed. formation, as the environmentally least affected trait The early/late ratio found is not significantly (Comes, unpublished results), are presented here. different from a Mendelian 3:1 ratio (x= 0.70, 0.5 >P>0.4) and is best explained by postulating a single gene with a dominant allele for expression of Seneciovulgaris early development present in var. vulgaris and a Thenumber of days for development from germina- recessive allele for late development at the locus in tion to bud formation for the parental taxa, the F1, ssp. denticulatus. Similarly to the F1, however, the and the pooled F2 families is shown in Fig. 1. mean of the early developing F2 series is shifted Subspecies denticulatus was substantially more vari- towards higher values than those of var. vulgaris able than var. vulgaris (CV =0.21,0.11, respect- (Fig. 1), indicating an incomplete rather than a ively). This difference in variance could be removed strictly dominant pattern of inheritance of early/late or substantially lowered by arcsin (1/(x + 1)) or free development. The gene postulated is likely to regu- (Smith, 1976) transformations. Accordingly, the late general developmental timing, owing to the

The Genetical Society of Great Britain, Heredity, 77, 544—554. 548 H. P. COMES & J. W. KADEREIT

S. vulgaris 30 N = 129

15 C, U- S. vulgaris subsp. denticulatus 10 N = 93 5

0 15 25 35 45 55 65 75

Fl 40 N= 151

20 C, LL P

0 1015 25 35 45 55 65 75

30 F2 1 Observed frequencies for the 25 Fig. N = 348 time to bud formation in Senecio 20 vulgaris ssp. denticulatus, S. vulgaris var. vulgaris, and their F1 and F2 gener- C,15 ations. Smooth lines indicate expected

10 F2 frequencies according to two truncated normal distributions with days and arcsin (1/(days + 1)) as the metameter for the 'early' and 'late' F2 series, respectively. Means are desig- 15 25 35 45 55 65 nated by arrows. Note that three probable outliers (*) were omitted from the DAYS FROM GERMINATION TO BUD FORMATION analysis.

extremely high phenotypic correlations of time to (± 0.080).The strong F2—F3 regression clearly bud formation with time to first anthesis and first reflects genetic segregation between six F2 parents fruiting (see above). with 'early' and three parents with 'late' develop- The one-generation selection experiment offers ment. An estimate of narrow-sense heritability (h2), additional evidence for a simple genetic basis of calculated from the F2—F3 slope and adjusted for speed of development in S. vulgaris. Figure 2 shows inbreeding (Smith & Kinman, 1965), equalled 0.46, the means of the nine F3 families for time to bud thus indicating a high heritability (additive genetic formation plotted against the value of their selected variance) of developmental timing. More import- F2 parent; the control values of ssp. denticulatus and antly, however, after a single generation of selfing, var. vulgaris are also represented. The regression all F3 families derived from extreme F2 individuals between the F2 parent and the F3 offspring values, came near the control values of the parental taxa shown in Fig. 2, has a slope of 0.687 (±0.044) (Fig. 2 and Table 1). This is expected if there are whereas the entire data set has a slope of 0.567 only a single or few loci involved. Furthermore, a

The Genetical Society of Great Britain, Heredity, 77, 544—554. GENETICS OF SPEED OF DEVELOPMENT IN SENECIO549

60 Fig. 2 Regression of mean phenotypic values of time to bud formation for Ui 1.3 nine F3 families (E1—6, L1—3) against L2 the F2 parent plant value. F2 parents 50 S LI were selected for extreme phenotypic dent2. values from an F2 of 348 plants. The z

Cl) slope of the best fitted line equals 'S. 0.687 (± 0.044), corresponding to an 0u- 40 adjusted heritability estimate of 0.46. Parental control values of Senecio vulgaris ssp. denticulatus (dent) and S. vulgaris var. vulgaris (vulg) are also 30 presented, but were not included in 20 30 40 50 60 the regression analysis. DAYS FROM GERMINATION TO BUD FORMATION (PARENT)

Table I Parent values, means and SE of time to bud = 3.18,P<0.05; h2 =0.60).On the assumption formation for nine F3 families and controls of Senecio that these latter plants were homozygous dominant vu/garis ssp. denticulatus (dent) and S. vulgaris var. vu/ga ris at the major locus, this observation may suggest that (vuig) additional minor genes (modifiers) affecting developmental timing were present as well. The failure to N Family Parent value Mean SE detect residual heritability among the 'late' F3s could 23 31.8 0.54 21 depend on the small number of families raised. vu/gl A distinct difference in growth pattern was found vulg2 23 31.0 0.55 16 El 23 32.6 0.74 23 between early flowering var. vulgaris and late flower- E2 25 36.6 1.13 14 ing ssp. denticulatus (Fig. 3). It was obvious that E3 27 38.9 0.52 40 whereas the latter undergoes a distinct rosette stage E4 26 35.8 0.52 37 consisting of a large number of leaves, followed by E5 27 37.6 0.43 29 moderate internode elongation of the axis after bud E6 22 34.6 0.69 29 formation, height increment is greatly enhanced Li 48 50.3 1.86 8 after bud formation in var. vulgaris, which is a L2 50 52.7 0.84 20 sparsely leaved plant with less close spacing of the L3 46 52.5 0.69 28 basal nodes. Figure 3 also illustrates that rates of dentl 44 49.9 0.76 26 leaf production do not differ between progenitor dent2 60 47.1 0.92 36 and derivative, indicating that early flowering in var. vulgaris does not arise as a simple consequence of its E, L designate F3 families selected for early and late faster growth, but instead is achieved by the abbre- development, respectively. N indicates the number of viation of the rosette stage of development. individuals analysed per family. Seneclo vernal/s regression slope within the cluster of the 'late' F3 The number of days for development from germina- families alone (L1—3) was not significantly different tion to bud formation for the parental taxa and the from zero (b =0.050±0.644,r2 (adj) =0.0,t =0.08, F1, F2, and BC1 progenies is shown in Fig. 4. The P =0.95),so there was no residual heritability left results are similar to those found in S. vulgaris, with for speed of development (h2 =0.033).This result is the late flowering plants of S. vemalis from C. entirely consistent with the hypothesis that the Europe being much more variable for the character respective 'late' F2 parents were homozygous reces- observed than the early flowering plants from Israel sive for the postulated gene controlling speed of (CV =0.20,0.12, respectively). The same explana- development. In contrast, a significant regression tion as given above also applies here. slope within the cluster of 'early' F3s (Ei—6) indi- The frequency distribution of the BC1 appears to cated that there was still a heritable component in fall broadly into two groups, namely those individ- the degree of developmental timing shown by 'early' uals that are predominantly like the F1, and those F2 parents(b =0.900±0.283,r2(adj) =0.65, that show a developmental timing as in S. vernalis The Genetical Society of Great Britain, Heredity, 77, 544—554. 550 H. P. COMES & J. W. KADEREIT

60-

U) E >a) C) (a 35 0) 0 a,30 .0 Fig. 3 Increment of both stem height E 25 E (solid symbols) and number of leaves a) Ct) 20 z (open symbols) in Senecio vulgaris ssp. 15 denticulatus (squares) and S. vulgaris 10- var. vulgaris (triangles) following germination. Measurements were made at the time of first budding, first 10 20 30 40 50 60 70 80 90 anthesis, and first fruiting. The data presented are based on the mean value Days after germination of 10 plants of each taxon.

fromC. Europe. However, following a similar calcu- speed of development in S. vernalis, and therefore lation and transformation pattern as was carried out should be re-evaluated with subsequent derived for the F2 in S. vulgaris, the overall distribution of generations to test the predicted digenic genotype of the BC1 did not easily fit into two normal distribu- individual BC1 plants. Nevertheless, the model does tions truncated at any reasonable point on each side show that the F1 and the 'early' BC1 series should of the zone of overlap between the two hypothesized resemble one another closely (see above). It also classes. Figure 4 presents two typical normal predicts the scarcity of 'late' genotypes observed in distributions found, one for the 'early' individuals the F2 as such genotypes would segregate at a rate of truncated at 52 days, and the other for the 'late' only 1/16. However, if the behaviour of the two individuals truncated at 66 days, fitted to the overall hypothesized BC1 distributions is taken to discrim- distribution of the 116 BC1 plants. Despite the fact inate 'early' and 'late' classes of the present F2 data that the observed distribution appears far from two set, any confident classification of comparatively normal distributions, there is formal agreement 'late' F2 individuals is prevented because their when one probable outlier (* in Fig. 4) is excluded phenotypic values lie within the small range of (= 90.78,0.25>P>0.1). Hence, the BC1 seemed overlap between the two BC1 distributions (Fig. 4). to segregate into two subpopulations, with an esti- There is a large discrepancy for growth pattern mated number of 81 'early' and 37 'late' individuals, between 'early' S. vernalis from Israel and 'late' S. and with three plants expected but not observed. vernalis from Germany (Fig. 5). Stem elongation, These data, in fact, are not inconsistent with a 3:1 though rapid after bud formation in both accessions, ratio expected for two independent loci acting domi- is initiated from a large and densely leafy rosette in nantly (x =2.54,0.2>P>0.1). With such a model, German material, but follows a nonrosette stage of the allelic constitution of the Israeli and C. Euro- growth in Israeli material. As already described pean parental material could be denoted as AABB above for S. vulgaris, developmental rates in terms andaabb, respectively, and the BC1 would then of leaf production are virtually identical in the two consist of four genotype classes (AaBb, Aabb, aaBb S. vernalis accessions, thus indicating that early flow- and aabb) with identical 'early' phenotypes for the ering of the Israeli material is achieved through a first three. Because the means of both the F1 and the drastic abbreviation of the rosette stage of develop- early developing BC1 series are shifted towards ment rather than through an acceleration of the higher values in comparison to that of S. vernalis entire developmental programme. from Israel (Fig. 4), early development appears to be only incompletely dominant over late development, thus paralleling the situation found in S. vulgaris. Discussion Because of the generally low significances in the Attemptsto understand and explore the evolution of data presented above, this model may not be taken life history characters (Sibly & Antonovics, 1992) as an exact representation of the genetics underlying have utilized both single-locus or two-locus (Ander-

The Genetical Society of Great Britain, Heredity, 77, 544—554. GENETICS OF SPEED OF DEVELOPMENT IN SENECIO 551

S. vernalis (Israel) 20 N = 73

S. vernalls (Germany) J10 N=64

0-ml .hil ,1, 15 25 35 45 55 65 75 85 95 105 115 125

C U.

0 15 25 35 45 55 65 15 85 95 105 115 125

BC1 N 116

S U.

0 15 25 35 45 55 65 75 85 95 105 115 125

F2 Fig. 4 Observed frequencies for the 10 N=85 time to bud formation in Senecio vernalis from Germany, S. vemalis from Israel, and their F1, BC1 (F1 x S. vernalis, Germany) and F2 generations. Smooth lines indicate expected BC1 frequencies according to two truncated normal distributions with days and arcsin (1/(days+1)) as the metameter for the 'early' and 'late' BC1 series, respectively. Means are designated by 15 25 35 45 55 65 75 85 95 105 115 125 arrows. Note that one probable outlier (*) was omitted from the analysis. DAYSFROM GERMINATION TO BUD FORMATION

The Genetical Society of Great Britain, Heredity, 77, 544—554. 552 H. P. COMES & J. W. KADEREIT

40 mountainousS. European habitats. In S. vernalis, fast genotypes with seed dormancy from Israel and E C) slow genotypes without seed dormancy from C.

C) Europe can be regarded as being well adapted to 12) their respective climates. These are a strict and E reasonably reliable winter-rain climate in Israel and 12) 0D a mild subatlantic climate without severe frosts in S.W. Germany. 0 10 20 30 40 5060 70 80 90 100 110 If, in S. vulgaris, the fast-developing ruderal var. vulgaris is accepted as being derived from the slow- Days after germination developing ssp. denticulatus then var. vu/guns may be Fig. 5 Increment of both stem height (solid symbols) and regarded as being fixed for an 'early flowering' number of leaves (open symbols) in Senecio vernalis from mutant allele relative to ssp. denticulatus, which is Germany (squares) and Israel (triangles) following germi- fixed for a 'late flowering' allele. In S. vernalis, nation. Measurements were made at the time of first however, the direction of evolutionary change is budding, first anthesis, and first fruiting. The data more ambiguous between slow-developing material presented are based on the mean value of 60 and 69 from C. Europe and fast-developing material from plants from Germany and Israel, respectively. Israel. Nevertheless, by abbreviating the rosette stage of development, and showing a marked reduc- tion in leaf number, the early phenotypes of S. son & King, 1970) and quantitative (e.g. Lande, vulgaris and S. vemalis resemble three 'early flower- 1982) genetic models. Quantitative genetic models ing' mutants (elf 1—3) recently described from Arabi- have been used more commonly because (1) life dopsis thaliana by Zagotta et al. (1992). Elf 1—3 are history characters, based on their usually continuous stage transition and probably loss-of-function muta- variation, have been assumed or shown to be inher- tions, which is in accordance with their recessive ited polygenically (Lande, 1982; Charlesworth, 1984, mode of inheritance (see below). It might be 1990; Stearns, 1992); and (2) when the nature and conceivable that the major gene regulating speed of number of genes underlying the observed phenotype development in S. vulgaris and one of the two genes are unknown, a quantitative genetic approach has that act similarly in S. vernalis are homologous to been considered useful for some defined purpose one of the elf genes of A. thaliana. (Smith, 1991). It is noteworthy that a rosette stage is involved in The present study has shown that the life history virtuallyall cases where major genes which character 'speed of development' (i.e. time from dramatically influence speed of development have germination until first bud formation, anthesis, and been identified (annual vs. biennial races in Me/ilotus fruiting) in two closely related Senecio species, S. a/ba Med. [Clarke, 1935], Beta vulganis L. [Abegg, vu/ga risandS. vernalis, is inherited by genes with 19361, Hyoscyamus niger L. [Sinnott, 19601; annual large effects. However, whereas an almost twofold Lactuca graminifolia Michx. vs. biennial L. canaden- difference in this character between S. vulgaris var. sis L. [Whitaker, 19441; length of rosette growth vulgaris and ssp. denticulatus is largely governed by duration in Carthamus fiavescens Spreng. and C. one major gene, a difference of almost identical tinctorius L. [Imrie & Knowles, 19701). Accordingly, magnitude appears to be inherited digenically in S. all of these examples of natural species are candi- vernalis from Israel and Germany. dates for the action of genes like elfl—3. However, in In combination with the absence or presence of contrast to the action of elf mutant alleles in A. seed dormancy, which has been shown previously to thaliana (Zagotta et at., 1992) the 'early flowering' be inherited monogenically in both species (Kader- alleles of S. vulgaris var. vulgaris and S. vemalis are eit, 1984a; Comes, 1995b), the dramatic difference dominant in expression. Bernier et al. (1993) in speed of development between var. vulgaris and recently concluded that most 'late flowering' mutant ssp. denticulatus within S. vulgaris results in very alleles (not explicitly stage transition mutations) are divergent life histories. Whereas var. vulgaris typi- recessive, but Freeling et at. (1992) observed that cally occurs in ruderal habitats and can produce up most heterochronic mutants that retard developmen- to three more or less synchronized and successive tal stage transitions are dominant. The two taxa generations per year, at least in the mild climate of investigated here would fit the conclusion by Bernier S.W. Germany (Vack, 1992), ssp. denticulatus is a et at. (1993) when the alleles of ssp. denticu/atus and winter annual of coastal W. and N.W. European and C. European S. vemalis are interpreted as 'late flow-

The Genetical Society of Great Britain, Heredity, 77, 544—554. GENETICS OF SPEED OF DEVELOPMENT IN SENECIQ 553 ering' mutant alleles, instead of interpreting their tion. Proc. Nat!. Acad. Sci. US.A., 66, 780—786. alternative alleles as 'early flowering' mutations as ASHTON, P. A. AND ABBOTF, R. .t. 1992. Isozyme evidence suggested by the phylogenetic relationships at least and the origin of Senecio vulgaris (Compositae). P1. Syst. in S. vulgaris. Thus the dominance relationships of Evol., 179, 167—174. the S. vulgarislS. vernalis 'early flowering' alleles are BERNIER, G. 1988. The control of floral evocation and incompatible with all former evidence when inter- morphogenesis. Ann. Rev. P1. Physiol. P1. Mo!. Biol., 39, 175—219. preted as 'early flowering' stage transition mutations. BERNIER, G., HAVELANGE, A., HOUSSA, C., PETITJEAN, A. In contrast to the examples summarized by Freeling AND LEJEuNE, I'. 1993. Physiological signals that induce et a!. (1992) and Bernier et al. (1993), however, the flowering. P1. Cell, 5, 1147—1155. cases discussed here do not represent mutant BLISS, C. i. 1967. Statistics in Biology, vol. 1. McGraw-Hill, systems of young age, but rather natural taxa which New York. have had ample time to undergo evolutionary transi- CHARLESWORTH, n. 1984. The evolutionary genetics of life tion of dominance relationships. histories. In: Shorrocks, B. (ed.) Evolutionary Ecology, Though aiming at a testable hypothesis, our pp. 117—133. Blackwell Scientific Publications, Oxford. suggestion that the major genes controlling speed of CHARLESWORTFI, n. 1990. Optimization models, quanti- development in S. vulgaris and S. vernalis are homol- tative genetics and mutation. Evolution, 44, 520—538. CLARKE, A. E. 1935. Inheritance of annual habit and mode ogous to major heterochronic elfl—3 genes in A. of pollination in an annual white sweet clover. J. Am. thaliana needs to be treated with caution. Even if Soc. Agric., 27, 492—496. allelic differences in gene action (i.e. dominance! COHEN, A. C., JR. 1959. Simplified estimators for the recessiveness) may be neglected in this connection, normal distribution when samples are singly censored the possibility of genetic heterogeneity cannot be or truncated. Technometrics, 1, 217—237. ruled out. Thus, the early developing Senecio types COMES, H. P. 1995a. Senecio vulgaris L. ssp. denticulatus could have mutations in non-elf genes conferring (O.F. Muell.) P.D. Sell and S. vulgaris ssp. vulgaris var. early development or, vice versa, there might be vulgaris on Jersey (Channel Islands). Watsonia, 20, other Arabidopsis mutations of a similar phenotype. 205—227. The examination of phenotypes, however, will not COMES, H. P. 1995b. Genecological and isozyme studies in help to confirm or refute the action of one of the elf Senecio vernalis Waldst. & Kit, and S. vulgaris L. var. vulgaris (Asteraceae) from Central Europe and Israel. loci in S. vulgaris and S. vernalis. In order to do that, Flora, 190, 1—24. it will be necessary to examine the gene itself. DIGGLE, P. K. 1992. Development and the evolution of plant reproductive characters. In: Wyatt, R. (ed.) Ecology and Evolution of Plant Reproduction, pp. Acknow'edgements 326—355. Chapman and Hall, New York, London. Wegratefully acknowledge valuable comments of FENSTER, C. B., DIGGLE, P. K., BARRETT, S. C. H. AND Drs R. J. Abbott and M. G. Ritchie (St Andrews RITLAND, K. 1995. The genetics of floral development differentiating two species of Mimulus (Scrophularia- University, Scotland) and very helpful corrections of ceae). Heredity, 74, 258—266. two anonymous referees on earlier drafts of this FENSTER, C. B. AND RITLAND, K. 1994. Quantitative genetics manuscript. We also wish to thank Dr Z. R. Sung of mating system divergence in the yellow monkey- (University of California, Berkeley) for kindly flower species complex. Heredity, 73, 422—435. drawing attention to a valuable reference, Miss A. FREELING, M., BERTRAND-GARCIA, R. AND SINHA, N. 1992. Berg for preparing Figs 2 and 5, and Miss P. Siegert Maize mutants and variants altering developmental who assisted in growing the S. vulgaris plants for time and their heterochronic interactions. BioEssays, 14, developmental analysis. This research was conducted 227—236. while H. P. C. was a recipient of an LGFG doctoral GOTTLIEB, L. D. 1973. Genetic differentiation, sympatric grant from the Land Baden-WUrttemberg, Germany. speciation and the origin of a diploid species of Stepha- nomeria. Am. J. Bot., 60, 545—553. GOTI'LIEB, L. 13. 1982. Does speciation facilitate the evolu- References tion of adaptation? In: Barigozzi, C. (ed.) Mechanisms of Speciation, pp. 176—190. Alan R. Liss, New York. ABEGG,F. A. 1936. A genetic factor for the annual habit in HEGI, o. 1987. illustrierte Flora von Mitteleuropa, Ed. 2, Bd. beets and linkage relationships. J. Agric. Res., 53, VI/4. Verlag Paul Parey, Berlin. 493—511. IMRIE, B. C. AND KNOWLES, P. F. 1970. Inheritance studies ALEXANDER, J. C. M. 1979. The Mediterranean species of in interspecific hybrids between Carthamus flavescens Senecio sections Senecio and Deiphinifolius. Notes R. and C. tinctorius. Crop Sci., 10, 349—352. Bot. Gard. Edinburgh, 37, 387—428. KADEREIT, ..w.1984a. Studies on the biology of Senecio ANDERSON, W. W. AND KING, C. E. 1970. Age-specific selec- vulgaris L. ssp. denticulatus (O.F. Muell.) P.D. Sell. New

The Genetical Society of Great Britain, Heredity, 77, 544—554. c (Lc)

554 H. P. COMES & J. W. KADEREIT

Phytol., 97, 681—689. terranean Senecio vulgaris L. New Phytol., 117, 673—678. KADEREIT,i. w. 1984b. The origin of Senecio vulgaris SIBLY, R. AND ANTONOVICS, .j.1992.Life-history evolution. (Asteraceae). P1. Syst. Evol., 145, 135—153. In: Berry, R. J., Crawford, T. J. and Hewitt, G. M. (eds) KOORNEEF, M., HANHART, C. J. AND VAN DER VEEN, J. H. Genes in Ecology, pp. 87—122. Blackwell Scientific 1991. A genetic and physiological analysis of late flow- Publications, Oxford. ering mutants in Arabidopsis thaliana. Mo!. Gen. Genet., SINNOfF, E. w. 1960. Plant Morphogenesis. McGraw-Hill, 229, 57—66. New York. LANDE, R. 1982. A quantitative genetic theory of life SMITH, J. D. AND KINMAN, M. L. 1965. The use of parent- history evolution. Ecology, 63, 607—615. offspring regression as an estimator of heritability. Crop LAURIE, D. A., PRATCHETF, N., BEZANT, J. H. AND SNAPE, J. Sci., 5, 595—596. w. 1994. Genetic analysis of a photoperiod response SMITH, j.Ia.K. 1976. Data transformation in analysis of gene on the short arm of chromosome 2(2H) of variance. .1. Verb. Lear Verb. Behav., 15, 339—346. Hordeum vulgare (barley). Heredity, 72, 619—627. SMITH, R. E. 1991. Genetic and phenotypic aspects of life- LAW, C. N., DEAN, C. AND COUPLAND, G. 1993. Genes history evolution in animals. Adv. Ecol. Res., 21, controlling flowering and strategies for their isolation 63—120. and characterization. In: Jordan, B. R. (ed.) The Mole- STEARNS, S. C. 1992. The Evolution of Life Histories. Oxford cular Biology of Flowering, pp. 47—68. CAB Inter- University Press, Oxford, New York, Tokyo. national, Oxford. SUNG, Z. R., BELACHEW, A., SHUNONG, B. AND BERTRAND- LEWONTIN, R. C. 1965. Selection for colonizing ability. In: GARCIA, R. 1992. EMF, an Arabidopsis gene required for Baker, H. G. and Stebbins, G. L. (eds) The Genetics of vegetative shoot development. Science, 258, 1645—1650. Colonizing Species, pp. 77—91. Academic Press, New TROW, A. H. 1912. On the inheritance of certain characters York. in the Common Groundsel —Seneciovulgaris Linn. — MACARTHUR, R. H. AND WILSON, Ia. 0. 1967. The Theory of andits segregates. J. Genet., 2, 239—276. Island Biogeography. Princeton University Press, Prince- vACK, A. 1992. Demographische und Experimentelle Unter- ton, NJ. suchungen Saisonaler Variation am Beispiel von Senecio MACNAJR, M. R. AND CUMBES, o. .1989.The genetic archi- vulgaris L. var vulgaris —GenetischeDifferenzierung tecture of interspecific variation in Mimulus. Genetics, oder Modifikatorische Plastizität? Ph.D. Dissertation, 122, 211—222. Universität Heidelberg. MINITAB. 1991. MINITAB Reference Manual. Release 8. Data WHITAKER, T. H. 1944. The inheritance of certain charac- Tech Industries, Valley Forge, PA. ters in a cross of two American species of Lactuca. Bull. MURFET, I. C. 1977. Environmental interaction and the Torrey Bot. Club, 71, 347—355. genetics of flowering. Ann. Rev. P1. Physiol., 28, ZAGOfFA, M. T., SHANNON, S., JACOBS, C. AND MEEKS- 253—278. WAGNER, D. R. 1992. Early-flowering mutants of Arabi- REN, Z. AND ABBOTF, R. j.1991.Seed dormancy in Mcdi- dopsis thaliana. Aust. J. P1. Physiol., 19, 411—418.

The Genetical Society of Great Britain, Heredity, 77,544—554.