88: 1027-1037, 2001 doi:10.1006/anbo.2001.1541, available online at http://www.idealibrary.com on Annals of Botany

The Variability of Organs Differentially Involved in Pollination, and Correlations of Traits in Genisteae (Legun inosae: Papilionoideae) JAVIER HERRERA*

Departamento de Biología Vegetal y Ecología, Universidad de Sevilla, Apartado 1095, E-41080 Sevilla, Received: 7 June 2001

Returned for revision: 18 June 2001

Accepted: 21 August 2001

Spain

Published electronically: 24 October 2001

Size differences in organs involved to varying degrees in pollen exchange (standard and keel petals, calyx, pedicel and inflorescence stem diameter), together with correlations among these traits, were investigated at both the intraspecific and interspecific levels in populations of 15 southern Spanish species of the papilionoid legume tribe Genisteae. Characteristics less related to pollination (e.g. pedicel length and stem diameter) were roughly twice as variable between individuals as petal or calyx dimensions. The coefficients of variation were in the range 16-19 % vs. 7-10 %, respectively. Standard and keel petal size varied in parallel, not only in conspecific individuals but also across species and genera, suggesting a tight developmental relationship between these organs and a low potential for evolutionary modification. The remaining characteristics covaried in different ways within and between species and thus appeared much less mutually constrained (particularly so for calyx size). Multivariate analyses of variance revealed significant population divergence in five out of six species. Overall, 28 % of all variation in flower size occurred among populations and 72 % within populations, but the proportion of variance accounted for by population membership varied extensively across species (from 1 to 64 %) and was highest in two selfing annuals (Lupinus). Outcrossing perennials may show either high (Stauracanthus, Genista) or low (Cytisus) population differentiation. In addition to contributing to the increasing amount of data on quantitative floral variation in plants, results of this study may help understand how pleiotropic effects limit the evolutionary modification of some flower traits more than others in diverse plant groups. © 2001 Annals of Botany Company Key words: Calyx, corolla, Genisteae, legumes, shrub, population divergence, Mediterranean, pollination, individual variation.

INTRODUCTION Studies of phenotypic variability in plants have favoured vegetative traits (reviewed in Venable, 1984) over floral traits. Apart from the spectacular and infrequent floral polymorphisms exhibited by some species, flowers have for a long time been regarded as relatively invariant organs with little intraspecific variability (Grant, 1949; Stebbins, 1974), and for many years the only information available on their quantitative variation was the size range of petals, sepals, etc. recorded by systematists from herbarium specimens. More recent studies on intraspecific variation in flowers have focused on character displacement and pollination ecotypes (e.g. Armbruster, 1985, 1991; Robertson and Wyatt, 1990; Andersson and Widen, 1993; Johnson, 1997). Studying plant species in which floral variations are marked, easily diagnosable externally, and already documented is understandable when the goal is evidencing variability and phenotypic selection, but this strategy has probably led to variable, widely distributed, and/or ecologically diverse species being over-represented in the literature on floral variability. With the increasing interest in the evolution of correlated traits and phenotypic integration in plants (e.g. Berg, 1960; Conner and Via, 1993; Conner and Sterlin, 1995; Cresswell, 1998; C.M. Herrera, 1995; Wilson, 1995; Armbruster et al., * E-mail [email protected]

0305-7364/01/121027+

11 $35.00/00

1999; Wolfe and Krstolic, 1999), new data on intraspecific variation of floral traits have become available. Since the focus of such studies is not documenting phenotypic variability per se but the relationships among traits, spectacular variation in floral phenotypes is no longer a prerequisite. Consequently, this type of research is perhaps more likely to provide an unbiased assessment of floral variation in the angiosperms. Unfortunately, however, phenotypic integration studies often deal with annuals and short-lived herbaceous species (which are amenable to cultivation and investigation of trait heritabilities), and this again might bias the information on quantitative trait variation. In this paper variation and covariation of floral traits are investigated in a set of plants that, in contrast to those in papers cited above, are phylogenetically related. Information on how patterns of phenotypic correlation are affected by phylogenetic relatedness is scarce both in animals (Riska, 1985; Lofsvold, 1986; Cheverud, 1989; Cowley and Atchley, 1990) and plants (see, however, Armbruster, 1988; Mazer and Hultgard, 1993; Armbruster et al., 1994; Torres, 2000). One goal of this study was to attempt to determine whether the pattern of trait integration was consistent below and above the species level. This is of interest because congruence or incongruence of the respective sets of phenotypic correlations can facilitate conclusions on how constrained the evolution of these traits has been during the diversification of this plant group. Although, in several cases, more than one population was sampled per species, the study was not © 2001 Annals of Botany Company

1028

Herrera-Floral Variation in Genisteae

designed specifically to investigate among-population covariance of traits. Rather, the goal was to contribute to the growing pool of data on floral variation with a study of relatively long-lived, perennial shrubs which, given the difficulty involved in quantifying their genetic variation, will probably always be under-represented in the literature on individual variation. The study group (tribe Genisteae, Leguminosae) was selected because it is species-rich, taxa are common components of Mediterranean shrublands, and as yet there is no information on how variable the species are. As in other investigations of floral trait integration in plants (Cheverud, 1988; Conner and Via, 1993; Waitt and Levin, 1993, 1998; Conner, 1997), the assumption is implicit for the rest of this study that the variability observed largely reflects heritable variation.

El Campillo (4), El Cafiuelo (5), El Picacho (6), El Portil (7), El Rocio (8), El Rompido (9), Hinojos (10), La Vibora (11), Mazag6n (12), Olivares (13), Puerto Gabs (14), Punta Umbria (15), Valdelagrana (16), Valverde del Camino (17), and Venta del Culebrin (18). Study taxa

Genisteae are medium- to large-sized, leafless or with only minute deciduous leaves, often spiny shrubs which bloom massively in winter or early spring. Most species have yellow nectarless flowers which are obligately pollinated by bees, although in a few exceptional taxa (e.g. Retama) nectar is present. In the perennial taxa studied to date, openpollinated flowers have shown a low proportion of fruit set (Herrera, 1987; Parker, 1997; Rodriguez-Riafio et al., 1999a). Pollinator exclusion leads to complete reproductive failure, seemingly as a result of late-acting self-incompatibility (Gibbs and Bianchi, 1999). Only the annual Lupinus species have breeding systems dominated by selfing and exhibit high rates of fruit and seed production in the absence of pollinators (Arroyo, 1981; Pazy, 1984; Karoly, 1992, 1994; Herrera, 1999). Both classic (Bisby, 1981) and DNA systematic studies (Badr et al., 1994) indicate that Lupinus is a quite divergent genus within the Genisteae. Since they often look alike, members of the Genisteae have the reputation of being a `difficult' group (Bisby, 1981), particularly regarding the systematic delimitation of Genista and Cytisus. Nomenclature used in the present study follows a recent systematic treatment of Iberian Genisteae (Talavera et al., 1999) that records 112 species as being present in Spain and Portugal, with 44 within the area of this study. The 15 species studied here (Table 1) generally formed dense populations and were locally very abundant, even dominant,

MATERIALS AND METHODS Study area

Research was carried out in western Andalucia (southern Spain), across the lower Guadalquivir River valley and bordering mountain ranges (Cadiz, Huelva and Sevilla provinces). This area has a typical Mediterranean climate with hot summers and mild, wet winters. Average annual precipitation ranges from 500 to 1000 mm. A variety of sclerophyllous scrub types can be found in the region, from vegetation on stable sand dunes near the Atlantic coast, to scrub growing in the understorey of pine and evergreen-oak forests in the mountains. Plants were sampled from 18 localities distributed over an area of approx. 10 000 km2 , the most distant sites being 180 km apart. Locality names (each followed by a numerical code) were: Almonte (1), Aznalcázar (2), Bormujos (3),

TABLE 1. Identity of taxa, sampling localities, and average values (in 10

-1

mm) for the metric variables considered in

this study Sample sizes

Mean(CV)

Localities

Flowers

Plants

Populations

Adenocarpus telonensis Calicotome villosa Cytisus arboreus subsp baeticus C. grandifiorus C. scoparius

6 11 11 8,9,17 18 4,5,9 4,6,17 1,5,10,12 3,13 7,12,16 10,12,17 14

60 51 45 177 60 60 60 195 108 179 180 60

20 17 15 59 20 20 20 45 36 60 60 20

1 1 1 3 1 1 1 4 2 3 3 1

7. 6(12) 295 (18) 16. 7 (16) 65 .4 (16) 15 . 9 (23) 111 .2 (13) 17. 4 (15) 89 .0 (20) 12. 8 (14) 94 . 5 (17) 9.2(18) t 5 .9 (12) 12 . 5 (29) . . 33 1 (19) 24 4 (21) 40.6 (17) 47 .6 (9) 12.4 (13) 17 .7 (24) 6.4 (14) 26 .4 (27) 8 .2 (24) 44 . 1 (20)

91 .4(9) 29.9 (11) 47.2 (10) 53.0 (11) 52.0(8) 835 (12) 30.6 (14) 80.7(4) 104.3(6) 39.6(7) 106-9(11) 90.8 (9)

15

60

20

1

10.2 (17)

47 .7 (11)

109.8 (9)

2,10,12 10

187 24

55 12

3 1

9.9 (15) -

49 .9 (18) -

119.8 (9) 120.3 (13)

Genista hirsuta*

G. triacanthos* Lupinus angustifolius L. hispanicus Retama monosperma Stauracanthusgenistoides Teline linifolia Ulex argenteus

U. australis U. eriocladus

Stem

Pedicel

Calyx

Species

Keel 126 . 7(5) 118 . 2(6) 171 . 7(8) 202 . 5 (6) 202 .4(5) 117 . 8 (7) 77 . 6(8) 123 . 7(6) 1435 (2) 78 .4(7) 94 . 9 (10) 111 . 3 (5) 104 . 7(8) 112 . 2(8) 110 . 9 (12)

Standard 159 . 2(6) 144 . 6 (10) 169 . 9(9) 200 . 7(7) 208 . 2(5) 101 . 3(9) 60 . 6 (10) 133 . 6(7) 156 . 0(3) 105 . 7(7) 114 . 1 (10) 135 . 1 (6) 126 . 1 (9)

135-6(9) 134 . 1 (13)

All are perennial shrubs except for Lupinus species which are annuals. If more than one population was studied all plants were pooled to compute mean values and coefficients of variation (in %) for each species. -, missing data. See Materials and Methods for locality codes. *Three populations were sampled to measure standard length only. Means for the remaining variables pertain to 20 plants from one population. t Flowers lack a measurable pedicel.

Herrera-Floral Variation in Genisteae

components of scrub with the exception of the Lupinus species which are relatively unimportant components of the vegetation. The latter taxa were included to assess the effect of their breeding system, predominantly selfing, on intraspecific variability. Except for the widespread Cytisus scoparius, most species sampled are Iberian or IberianMoroccan endemics with a very narrow distribution on a world scale (Greuter et al., 1989; Talavera et al., 1999). In addition to morphological and pollination biology surveys of this tribe (Lopez et al., 1998, 1999, 2000; RodriguezRiafio et al., 1999b), previous studies reporting individualbased variation in Genisteae have dealt with pollinator visitation rates (Parker, 1997), flower size (Baonza and Malo, 1997), and ovule and seed numbers per fruit (Herrera, 1999). Trait measurements

Although I found no indication that petals might undergo size changes with flower age, only freshly opened flowers were used for the study. In order to include traits exhibiting a variety of effects on flower function, I adopted the strategy of choosing organs at increasing distances (both spatially and anatomically) from the anthers and stigma. The attributes measured were: total length of the right-handside petal in the keel; total length of the standard; distance from the base of the calyx to the tip of its longest tooth; pedicel length; and the diameter of the branch to which the flower was attached (at the point of pedicel insertion). The latter attribute will be referred to by the broad term `stem' because study species presented a variety of inflorescence types (in some species flowers were borne singly, in others they were on axillary clusters or racemes, and still others were on brachyblasts), so a more precise term would not have been generally applicable. Note also that although the keel and standard belong to the same floral whorl, the former is in direct contact with the pollen exchanging organs (i.e. anthers and stigma) whereas the latter is not. Thus, and according to the distance criterion, the keel is more directly involved in pollination than is the standard, which in turn is more involved than the calyx, etc.

1029

From November 1999 to March 2000, flowering branches from a number of plants chosen at random from each site were collected, labelled, placed in plastic bags and taken to the laboratory to be measured. Twenty plants were usually sampled per population unless insufficient individuals were available (Table 1). Initially, five flowers were measured at random per plant, but it was soon apparent that there was little variation within individuals so only three flowers per plant were studied thereafter. Flower parts were dissected, gently pressed between two glass slides, placed under a binocular microscope, and measured with digital calipers. According to the manufacturer, the resolution of the instrument was 0 . 01 mm, accuracy 0 . 03 mm, and repeatability 0 . 01 mm. Conservatively, however, measurements were rounded to the nearest 0 . 1 mm, well above the instrument's capacity. Additionally, a test of repeatability was performed with the study material in which 15 standards of the species with the smallest flowers (Genista triacanthos) were measured twice. This yielded a nearly perfect correlation among the two measurement sets (r = 0 . 993, P < 0 . 001), implying that low repeatability would not be responsible for any spurious pattern in the data (e.g. increased error variance with decreasing organ size). All measurements were performed by the author. For each trait, the mean of the three flowers was used as a plant estimate for that characteristic. Although metric differences found in this study were usually small in absolute terms (notably so among conspecific populations), there are several reasons why they should not be dismissed. First, it must be kept in mind that it is often the square of these differences that determines biologically relevant variability (e.g. the ability to attract pollinators will depend more on petal area than length). Second, floral traits are often so tightly correlated that size gains in one characteristic mean a parallel variation in every floral organ, not just those considered in the study (e.g. androecium and gynoecium). To illustrate this, the fresh mass of 30 Genista hirsuta flowers was determined and their standards were measured (Fig. 1). As expected, the variables were correlated closely (r = 0 . 973, P < 0. 001), but whereas the size of the standard ranged between 8 . 9 and 12 . 3 mm,

FIG. 1. Flower mass and standard petal size va ri abili ty in Genista hirsuta, and the relationship plant means (A) and individual flowers (B).

be tween

these va ri ables based upon individual

Herrera-Floral Variation in Genisteae

1030

flower mass va ri ed between 11 . 3 and 23 . 8 mg. In other words, an increase of just 3 mm in the length of the standard impli ed that the flower doubled in mass. This is relevant from a reproductive allocation standpoint, since small (but measurable) differences in the size of a flower part become amp lified when whole flower mass is considered.

RESULTS Summa ry statistics for the species studied are shown in Table 1. The size of the flowers (as standard length) ranged from 6 (Genista triacanthos) to 20 mm (Cytisus scoparius). Species means for standard length were distributed symmetri cally in the sample, with a mean of 13 . 9 mm and a median of 13 . 5 mm. As regards the vari abi li ty of the characte ri stics studied, the coefficients of va ri ation (CV) were smallest for the keel (in the range 2-12 %, mean 7 %), sli ghtly larger for the standard (3-13 %, mean 8 %) and the calyx (4-14 ~~ mean 10 %), and notably larger for the pedicel (9-27 ~~ mean 19 %) and the diameter of the stem (12-24 %, mean 16 %). On average (Fig. 2), pedicel and stem va ri abilities were statistically undistinguishable (Wilcoxon signed-rank test, Z = 0 . 9, n = 13, P = 0 . 3), whereas the calyx was significantly more vari able than the standard (Z = 2 . 3, n = 15, P = 0 . 02), which in turn was more va ri able than the keel (Z = 3 . 1, n = 15, P = 0 . 002). The hypothesis of decreasing vari ability of an organ with increasing involvement in pollination is supported. The va ri ability of flower size among conspecific individuals was greater in small-flowered than in large-flowered taxa, as demonstrated by the inverse relationship between the mean and the CV of keel size (rs = -0 . 587 9 P = 0 . 02;

FIG. 2. Box-whisker plot of the coefficient of va riation for the traits studied. The ho rizontal li ne re presents the mean, the box encompasses ± 1 s.e., and the vertical line delimits the non-out lier minimum and maximum values. Sample sizes (n) = 15 (calyx, keel, standard), 14 (stem) and 13 (pedicel) species. Va ri able means with the same letter a re not significantly diffe rent at P = 0 . 05 (Wilcoxon signed-rank test).

Spearman rank correlation). For standard size the equivalent relationship was only marginally significant (rs = -0 .492, P = 0 . 06; for both tests, n = 15 species). Given the repeatability and accuracy of measurements (see Materi als and Methods), this is unlikely to be an artefact of greater error variance in species with smaller flowers.

Phenotypic correlations

These are reported at two levels: within each species (i.e. how traits correlate among conspecific individuals) and among species. Interspecifically (Table 2A), the pattern of trait correlation was distinct: species with longer standards

2. Phenotypic correlations in Genisteae. A, Interspecific Spearman rank correlation coefficients as calculated from Table 1 (species-based), with significant correlations appearing in bold. B, Intraspecific (plant-based) correlations. C, Sample sizes for matrices A and B

T ABLE

Stem

Pedicel

Calyx

Keel

Standard

-0•14 -0•05

0•87***

-

6(0-44) 2(0 . 33) 1 (0•37)

8(0-52) 8(0-58)

13(0-76)

-

A Stem

Pedicel

Calyx Keel Standard

-

0. 43 -0-19 0•65* 0. 52

0.0 0•74** 0•82***

B Stem

Pedicel

Calyx Keel Standard

-

3(0. 34) 4(0-33) 3(0. 23) 4(0. 25)

C Stem

-

Pedicel

Calyx Keel

13 14 14

13 13

15

-

Standard

14

13

15

15

-

In matrix B two estimates of trait correlation are given in each cell: the number of taxa in which the correlation was found to be significant and (in parentheses) the Pearson correlation coefficient averaged over species. Means higher than the overall mean correlation (0 .42) are in bold type.

Herrera-Floral Variation in Genisteae

also had longer keels and flower pedicels, with these three characteristics mirroring each other closely. Moreover, keel length also correlated with stem diameter. The size of the calyx, on the other hand, varied independently of that of the other organs. (Non-normality in the distribution of species means required the use of rank correlations in Table 2A). In contrast to the interspecific pattern, depicting a single pattern of intraspecific trait association applicable to the members of the Genisteae studied was not straightforward, because phenotypic correlations did not need to be (and were not) identical in all taxa. I approached this problem by constructing the intraspecific correlation matrix for each taxon, and then counting the number of times (i.e. species) in which a particular pair of traits appeared significantly correlated (Table 213; in contrast to species means, plant means were normally distributed so Pearson correlation coefficients were used). To facilitate comparison of the intraspecific pattern of phenotypic correlations with the interspecific one, values of the Pearson coefficient averaged across species are given in Table 213; but note that these coefficents are not amenable to statistical testing because of obvious non-independence of correlations within any matrix. The most common correlation was that between standard and keel dimensions, which proved significant in 86 % of species. Next were those between the calyx and the

1031

standard, the calyx and keel, and the calyx and pedicel, which were all significant in roughly 50 % of cases. The remaining correlations occurred in smaller proportions ranging from 7-28 % (again, statistical testing on these proportions is not advisable for the same reason of nonindependence of correlations). Figure 3 shows scatter plots of the trait correlations for Ulex australis, which is reasonably representative of the major phenotypic correlations observed in the Genisteae sample as a whole.

Variation among populations

Multivariate analyses of the effect of geographical origin on intraspecific differences were performed in six taxa (Table 3). In five of these, a multivariate analysis of variance (MANOVA) model including stem, pedicel, calyx, keel and standard as predictor variables detected significant among-population variability. The effect was particularly strong in two annual Lupinus species. In general, population heterogeneity was not due to a single trait or group of traits; on the contrary, virtually any characteristic might be population-specific (Table 3, univariate statistics). Often, only some characteristics showed significant heterogeneity among populations, although in two instances (Lupinus hispanicus and Stauracanthus genistoides) all five traits studied did so.

FIG. 3. An example (Ulex austrahs) of the usual intraspecific pattern of trait cor relation found in the Genisteae. Each plot in the scatter plot matrix repre sents the relationship among two vari ables, with dots denoting individual plants. Significant (P < 0 .05) correlations are distinguished by thicker frame li nes. Histogram on top of each column shows the frequency dist ri bution of the corre sponding va ri able. The bivariate confidence elli pse for each relationship is also displayed.

F 4.7 28 . 0 13 . 8 4.2 6.9 1 .7

Willk's Lambda

0 . 48 0 . 05 0 . 30 0 . 52 0 . 36 0 . 72 10,104 10,76 5,30 10,106 10,106 10,96

d.f. *** *** *** *** *** ns

P ** *** ** * * ns

Stem

*** P < 0 . 001; ** P < 0 .01; * P < 0.05; ns, not significant. A separate analysis was performed on each of six taxa. Significances for individual variables are also reported.

Cytisus grandifiorus Lupinus angustifolius L. hispanicus Retama monosperma Stauracanthus genistoides Ulex australes

Species

Multivariate tests

* ns

***

ns *** ***

Pedicel

ns *** ** ** ***

ns

***

ns

Keel *** ns *** ns

Calyx

Univariate tests for variable:

TABLE 3. Results of multivariate analyses of variance (MANOVAs) aimed at detecting among population variability

*

***

ns *** *** ns

Standard

Herrera-Floral Variation in Genisteae T

ABLE

1033

4. Population effects and statistics describing flower size variation (as standard length) in populations of eight Genisteae species Standard length (0 . 1 mm)

Population Species Cytisus grandifiorus

%VAR

No.

n

Range

Mean

CV(%)

1

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24

20 20 19 20 20 18 20 19 20 15 15 15 20 16 20 20 20 20 20 20 20 15 20 20

173-236 177-216 176-224 83-112 84-121 91-118 47-67 56-69 57-76 129-145 112-140 133-151 114-138 153-164 145-162 89-118 83-124 95-118 106-145 97-132 87-119 107-160 121-165 120-159

203 197 203 97 102 105 55 62 65 138 126 143 129 159 154 104 107 106 123 115 104 128 138 139

8 5 7 8 10 7 10 6 8 3 6 4 4 2 3 6 9 6 8 8 8 11 8 8

Genista hirsuta

14

G. triacanthos

51

Lupinus angustifolius

64

L. hispanicus

46

Retama monosperma

1

Stauracanthus genistoides

51

Ulex australis

17

%VAR is the percentage of total among-plant variance accounted for by population membership. Ranges, means, sample sizes, etc. refer to individual plants. Code numbers identifying populations are those used in Fig. 4.

Phenotypic variance in flower size (taken as standard length) was further investigated in a somewhat larger sample of taxa (Table 4). ANOVA, which included species (fixed factor) and population within species (random factor) as predictor variables for flower size, revealed a distinct (and scarcely surprising) effect of species identity (MS = 98518, d.f. = 7, F = 1252 .7, P < 0 . 001), along with a highly significant effect of population (MS = 647, d.f. = 16, F = 8 . 2, P < 0 . 001 for the error term; MS = 78, d.f. = 428; R Z = 0 . 954 for the whole model). This model can be assessed visually in Fig. 4A. Nevertheless, to consider just the individual-based variability that occurs among species and/or populations can be misleading because plants within a population also varied significantly. Fig. 413 illustrates the extensive underlying variation that occurred within populations. Overall, 72 % of all variance in the size of flowers occurred within populations, whereas 28 % was accounted for by population membership. Note, however, that species varied greatly in this regard, and the proportion of total phenotypic variance expressed at the population level ranged from 1 to 64 % (Table 4). The hypothesis that intraspecific variability was related to habit was tested by an analysis of variance (using data in Table 4) which included habit (annual or perennial) and species identity as predictors of the coefficient of variation. This test is unavoidably very imperfect because the annuals were in a single genus (Lupinus), therefore the analysis actually lumped together generic membership and habit as a single factor. With this caution, ANOVA evidenced a

highly significant effect due to habit (MS = 80 . 2, d.f. = 1, F= 38 . 6, P < 0 . 001), and no effect due to species within habit groups (MS = 2 . 6, d.f. = 6, F = 1 . 3, P = 0 . 3; for the error term MS = 2 . 1, d.f. = 16; the coefficents of variation used for this analysis were approximately normal according to a Kolmogorov-Smirnov one-sample test, with d = 0 . 148, P = 0 . 61; Table 4). The average within-population coefficient of variation for six annual populations was 3 . 3 %, whereas for 18 populations of perennials it was 7 . 8 %. DISCUSSION Trait variability

The traits studied here can be assigned to one of two groups, one with within-species coefficients of variation of around 7 % (keel, standard and calyx) and the other in which CVs approach 20 % (stem diameter and pedicel length). These values are comparable to those observed for floral and vegetative traits in previous studies. For example, the average CV of flower size reported by Wolfe and Krstolic (1999) for 14 phylogenetically unrelated species with bilaterally symmetrical flowers was 9 %, whereas Armbruster et al. (1999) reported a CV of 6 . 9 % for floral traits, and 14 % for vegetative traits in a number of unrelated taxa with specialized pollination systems. Results of the present study support the hypothesis that vegetative characteristics show higher phenotypic variance than reproductive traits and, within the latter, variability is inversely related to involvement in pollen exchange.

1034

Herrera-Floral Variation in Genisteae

Population (species)

Plant no.

FIG. 4. A, Flower size variations in species and populations of Genisteae. Means ± 1 s.e. are given for each of 24 populations in eight taxa. Uppercase letters identify species (A, Cytisus grandifiorus; B, Genista hirsute; C, G. triacanthos; D, Lupinus angustifolius; E, L. hispanicus; F, Retama monosperma; G, Stauracanthus genistoides; H, Ulex australis). Within species, population means are presented in the order in which data were taken. Numbers identifying populations are those used in Table 4. B, Flower size variation in plants of Lupinus angustifolius from four different populations. Vertical lines span the range of variation within individuals (with n = 3 flowers). The horizontal dash across each line represents the mean for that plant.

Pedicels are usually regarded as being part of the flower, but the intraspecific variability of pedicel length found in this study (mean CV = 19 %) was similar to that of a vegetative character. Intuitively, one might think that pedicel length needs to be flexible enough to position the flower in the `best' possible aspect vis-a-vis pollinators, but note that the variability reported here occurred between individual plants. Since there seems to be no reason for the `best' aspect to differ from one plant to another, there must be an alternative explanation for the high variability in pedicel length (perhaps in a probable relationship between plant vigour, inflorescence size, and pedicel length). Unfortunately, other studies of trait integration in plants have omitted the pedicel, so it is difficult to compare the variability detected here with that of other plant species. However, C.M. Herrera (1990) reported that pedicel length was the most variable character in Viola cazorlensis, with a CV of 21 % (furthermore, a positive relationship existed between pedicel length and components of reproductive success in Viola). In order to pollinate papilionaceous flowers of a given size a bee visitor needs both skill and a minimal strength which is, to a great extent, determined by body mass. Consequently, flower size may determine pollinator diversity in Genisteae just as tube length does in plants with fused petals. This is supported by the fact that large-flowered species in the study area are worked efficiently only by large bees such as Bombus and Xylocopa species (less so by Apis mellifera), whereas small-flowered Genisteae (including the nectar-producing Retama species) are visited by a considerably diverse array of small-, medium-, and large-sized hymenoptera (Herrera, 1988; P.E. Gibbs, pers. comm.). Thus, the trend of decreasing intraspecific variability with increasing flower size reported here could result from a gradient of pollinator selection on flower size, and is interpreted as supporting Fenster's (1991) idea that specialization in pollination relationships results in decreased phenotypic variance in flowers. Ideally, however, a proper

test of this hypothesis in the Genisteae would need to control for the effect of gene ri c ownership by performing intrageneri c contrasts of floral va ri ability and visitor diversity. Phenotypic correlations

Compa ri ng the patterns of trait correlation that appear within and among species can provide some insights into how conservative is the developmental architecture of flowers in the Genisteae sampled. The compa ri son may also help to understand why evolutiona ry diversification has affected some floral traits but not others. Since there were differences as well as simila ri ties in the intraspecific and interspecific patterns, these are depicted graphically in Fig. 5 to facilitate compari sons. The correlation between the size of the standard and that of the keel held at both levels of analysis. However, the function of these petals differs markedly in papilionaceous flowers (pollinator attraction vs. pollen exchange, respectively), and given that most Genisteae rely on crosspollination for fruit set, selection for increased standard petal size relative to that of the keel might have operated in at least some species or genera. Apparently, however, decoupling of their va ri abilities has not occurred to any great extent (i.e. the ratio of standard to keel length is approximately constant). This suggests that the developmental relationship between keel and standard is so tight that evolutiona ry decoupling of their sizes has become largely unfeasible. With regard to calyx size, however, the intraspecific and interspecific patterns are dramatically different. Within species, the calyx cova ried with both the keel and the standard, whereas across species no correlation of the calyx with petals could be found. I hypothesize that selective pressures related to pollination have fostered morphological diversification of the calyx in the Genisteae, and that this has been possible precisely because the developmental link with the corolla is not ve ry tight. That pol lination has had a

Herrera-Floral Variation in Genisteae Intraspecific

1035

Interspecific

Stem

FIG. 5. A summa ry of intra- and interspecific phenotypic cor relations in the Genisteae, as deri ved from data in Table 2. Line thickness is roughly propor tional to correlation intensity. In the intraspecific pattern, only cor relations above the overa ll mean corre lation are shown.

role in the diversification of calyx morphology in this group is supported by the correlation between calyx type and the type of reward offered to pollinators (i.e. the few nectariferous species have cylindrical, relatively long calices while in nectarless species the calyx is invariably campanulate and short; Talavera et al., 1999). Lack of petal-calyx covari ation across species would thus denote evolutiona ry decoupling of these flower parts brought about by circumstances related to pollination. Lastly, significant correlations existed across species between vegetative and floral traits (Fig. 5) which were unexpected from a `correlation pleiades' (Berg, 1960) standpoint. The tendency of taxa with thick stems to have larger flowers and longer pedicels probably relates to the way in which flowers are organized within plants (i.e. inflorescence size and architecture) but, because inflorescence characteristics were not recorded in the present study, this source of interspecific variation must remain largely unexplained. Primack (1987) suggested that a physiological continuity exists in plants between seeds, fruits, flowers, inflorescences, and even leaves. This functional continuity may open a way to phenotypic correlations between vegetative and reproductive characteristics. Variations between populations

In a previous study (Herrera, 1999), I have shown that populations of Genisteae differ in the number of ovules per ovary and in the number of seeds per pod. Results reported here demonstrate that between-population phenotypic differentiation also occurs in the size of flower parts. As could be predicted from their predominantly autogamous breeding system, the two annuals studied showed relatively high heterogeneity among populations, which is probably the result of restricted gene flow and founder effects (Stebbins, 1957; Levin and Kerster, 1974; Levin, 1988; Heywood, 1991). More surprisingly, however, outcrossing perennials also showed overall significant population divergence in floral characteristics and, in two cases at least, the amount of population divergence was comparable to that of annuals.

In this study populations were relatively close (in the order of 10 to 100 km), and the environment is notably uniform throughout the area, at least with regard to climate. Indeed, most taxa sampled are narrow endemics (Greuter et al., 1989; Talavera et al., 1999) with all populations being located within a few hundred kilometres of one another, and often much closer. Yet relative spatial proximity did not preclude phenotypic divergence. For example, even though the study populations of Stauracanthus genistoides were on average just 30 km apart, population means differed significantly for every trait. Moreover, population membership explained half of the variance in flower size both in Stauracanthus and in Genista triacanthos. This sort of intraspecific variation is more likely to represent random variation due to restricted gene flow (Levin, 1988; Armbruster and Schwaegerle, 1996) rather than ecotypic divergence, but it was surprising that morphometric differentiation occurred among populations of outcrossing shrubs over such short distances. Further field research would be required to demonstrate whether or not this variation is selectively neutral. On the other hand, there were also species in which up to 99 % of phenotypic variance occurred within populations, so anticipating the extent of population divergence solely from breeding systems is not justified in this group. This study has not addressed the possibility that conspecific populations differed in the pattern of trait covariation; this will be dealt with in a separate paper. Mechanisms that can give rise to among-population covariance of traits have been reviewed by Armbruster and Schwaegerle (1996; see also Endler, 1995), and it would be worth checking whether some of these (e.g. drift covariance, selective covariance) have operated in the Genisteae and have resulted in altered patterns of variation between populations.

ACKNOWLEDGEMENTS Salvador Talavera shared valuable expertise on the biology, systematics and distribution of Genisteae. Thanks are also due to Carlos M. Herrera for suggestions regarding data

1036

Herrera-Floral Variation in Genisteae

analysis. Comments from W.S. Armbruster and P.E. Gibbs greatly improved the manuscript. This study was funded by Plan Andaluz de Investigación (Junta de Andalucía), and by grant BOS2000-0328 of the Spanish Dirección General de Enseñanza Superior a Investigación Científica.

LITERATURE CITED Andersson S, Widen B. 1993. Pol li nator-mediator selection on floral traits in a synthetic population of Senecio integrifolius (Asteraceae). Oikas 66: 72-79. Armbruster WS. 1985. Patterns of character divergence and the evolution of reproductive ecotypes of Dalechampia scandens (Euphorbiaceae). Evolution 39: 733-752. Armbruster WS. 1988. Multilevel comparative analysis of the morphology, function, and evolution of Dalechampia blossoms. Ecology 69: 1746-1761. Armbruster WS. 1991. Mul tilevel analysis of morphometric data from natural popula tions: insights into ontogene tic, gene tic, and selective correlati ons in Dalechampia scandens. Evolution 45: 1229-1244. Armbruster WS, Schwaegerle KE. 1996. Causes of covaria tion of phenotypic traits among popula ti ons. Journal of Evolutionary Biology 9: 261-276. Armbruster WS, Edwards ME, Debevec EM. 1994. Floral character displacement generates assemblage structu re of western Austra lian tri ggerplants (Stylidium). Ecology 75: 315-329. Armbruster WS, Distilio VS, Tuxill JD, Flores TC. 1999. Covariance and decoupling of floral and vegetative traits in nine neotropical plants: A re-evaluation of Berg's correlation-pleiades concept. American Journal of Botany 86: 39-55. Arroyo MTK. 1981. Breeding systems and pollination biology in Leguminosae. In: Polhill RM, Raven PH, eds. Advances in legume systematics, part 1. Kew: Royal Botanic Gardens, 723-769. Badr A, Martin W, Jensen U. 1994. Chloroplast DNA restriction site polymorphism in Genisteae (Leguminosae) suggests a common origin for European and American lupines. Plant Systematics and Evolution 193: 1-4. Baonza J, Malo JE. 1997. Floral size variability of Cytisus scoparius along an altitudinal gradient. Lagascalia 19: 845-850. Berg RL. 1960. The ecological significance of correlation pleiades. Evolution 14: 171-180. Bisby FA. 1981. Tribe 32. Genisteae (Adans.) Benth. (1865). In: Polhill RM, Raven PH, eds. Advances in legume systematics, part 1. Kew: Royal Botanic Gardens, 409-425. Cheverud JM. 1988. A comparison of genetic and phenotypic correlations. Evolution 42: 958-968. Cheverud JM. 1989. A comparative analysis of morphological variation patterns in the papionins. Evolution 43: 1737-1747. Comer JK. 1997. Floral evolution in wild radish: The roles of pollinators, natural selection, and genetic correlations among traits. International Journal of Plant Sciences 158: 5108-S 120. Comer JK, Sterling A. 1995. Testing hypotheses of functional relationships: A comparative survey of correlation patterns among floral traits in five insect-pollinated plants. American Journal of Botany 82: 1399-1406. Comer JK, Via S. 1993. Patterns of phenotypic and genetic correlations among morphological and life-history traits in wild radish, Raphanus raphanistrum. Evolution 47: 704-711. Cowley DE, Atchley WR.1990. Development and quantitative genetics of correlation structure among body parts of Drosophila melanogaster. American Naturalist 135: 242-268. Cresswell JE. 1998. Stabilizing selection and the structural variability of flowers within species. Annals of Botany 81: 463-473. Endler JA. 1995. Multiple-trait coevolution and environmental gradients in guppies. Trends in Ecology and Evolution 10: 22-29. Fenster CB. 1991. Selection on floral morphology by hummingbirds. Biotropica 23: 98-101. Gibbs PE, Bianchi MB. 1999. Does late-acting self-incompatibility (LSI) show family clustering? Two more species of Bigno -

niaceae with LSL• Dolichandra cynanchoides and Tabebuia nodosa. Annals of Botany 84: 449-457. Grant V. 1949. Pollination systems as isolating mechanisms in angiosperms. Evolution 3: 82-97. Greuter W, Burdet HM, Long G. 1989. Med-checklist vol 4. Dicotyledones (Lauraceae-Rhamnaceae). Geneva: Editions des Conservatoire et Jardin botaniques de la Ville de Geneve. Herrera CM. 1990. The adaptedness of the floral phenotype in a relict endemic, hawkmoth-pollinated violet. 1. Reproductive correlates of floral variation. Biological Journal of the Linnean Society 40: 263-274. Herrera CM. 1995. Floral traits and plant adaptation to insect pollinators: a devil's advocate approach. In: Lloyd DG, Barrett SCH, eds. Floral biology: studies on floral evolution in animalpollinated plants. New York: Chapman and Hall, 65-88. Herrera J. 1987. Flower and fruit biology in southern Spanish Mediterranean shrublands. Annals of the Missouri Botanical Garden 74: 69-78. Herrera J. 1988. Pollination relationships in southern Spanish Mediterranean shrublands. Journal of Ecology 76: 274-287. Herrera J. 1999. Fecundity above the species level: ovule number and brood size in the Genisteae (Fabaceae:Papilionoideae). International Journal of Plant Sciences 160: 887-896. Heywood JS. 1991. Spatial analysis of genetic variation in plant populations. Annual Review of Ecology and Systematics 22: 335-355. Johnson SD. 1997. Pollination ecotypes of Satyrium hallackii (Orehidaceae) in South Africa. Botanical Journal of the Linnean Society 123: 225-235. Karoly K. 1992. Pollinator limitation in the facultatively autogamous annual, Lupinus nanus (Leguminosae). American Journal of Botany 79: 49-56. Karoly K. 1994. Inbreeding effects on mating system traits for two species of Lupinus (Leguminosae). American Journal of Botany 81: 1538-1544. Levin DA. 1988. Local differentiation and the breeding structure of plant populations. In: Gottlieb LD, Jain SK, eds. Plant evolutionary biology. London: Chapman and Hall, 305-329. Levin DA, Kerster HW. 1974. Gene flow in seed plants. Evolutionary Biology 7: 139-220. Lofsvold D. 1986. Quantitative genetics of morphological differentiation in Peromyscus. L Tests of the homogeneity of genetic covariance structure among species and subspecies. Evolution 40: 559-573. Lopez J, Devesa JA, Ruiz T, Ortega-Olivencia A. 1998. Seedling morphology in Genisteae (Fabaceae) from south-west Spain. Botanical Journal of the Linnean Society 127: 229-250. Lopez J, Devesa JA, Ruiz T, Ortega-Olivencia A. 1999. Seed germination in Genisteae (Fabaceae) from south-west Spain. Phyton 39: 107-129. Lopez J, Devesa JA, Ruiz T, Ortega-Olivencia A. 2000. Production and morphology of fruit and seeds in Genisteae (Fabaceae) of southwest Spain. Botanical Journal of the Linnean Society 132: 97-120. Mazer SJ, Hultgard UM. 1993. Variation and covariation among floral traits within and among four species of northern European Primula (Primulaceae). American Journal of Botany 80: 474-485. Parker IM. 1997. Pollinator limitation of Cytisus scoparius (Scotch broom), an invasive exotic shrub. Ecology 78: 1457-1470. Pazy B. 1984. Insect induced self-pollination. Plant Systematics and Evolution 144: 315-320. Primack RB. 1987. Relationships among flowers, fruits, and seeds. Annual Review of Ecology and Systematics 18: 409-430. Riska B. 1985. Group size factors and geographic variation of morphometric correlations. Evolution 39: 792-803. Robertson JL, Wyatt R. 1990. Evidence for pollination ecotypes in the yellow-fringed orchid, Platanthera ciliaris. Evolution 44: 121-133. Rodriguez-Riaño T, Ortega-Olivencia A, Devesa JA. 1999a. Types of androecium in the Fabaceae of SW Europe. Annals of Botany 83: 109-116.

Rodriguez-Riaño T, Ortega-Olivencia A, Devesa JA. 1999b. Reproductive biology in two Genisteae (Papilionoideae) endemic of the western Mediterranean region: Cytisus striatus and Retama sphaerocarpa. Canadian Journal of Botany 77: 809-820.

Herrera-Floral Variation in Genisteae Stebbins GL. 1957. Self fertilization and population variability in the higher plants. American Naturalist 91: 337-354. Stebbins GL. 1974. Flowering plants: evolution above the species level. Cambridge, Massachusetts: Belknap. Talavera S, Aedo C, Castroviejo S, Romero Zareo C, Sáaez K, Salgueiro FJ, Velayos M. 1999. Flora Iberica, vol VII (I), Leguminosae (partim). Madrid: Real Jardin Botánico CSIC. Torres C. 2000. Pollen size evolution: correlation between pollen volume and pistil length in Asteraceae. Sexual Plant Reproduction 12:365-370.

Venable DL. 1984. Using intraspecific variation to study the ecological significance and evolution of plant life-histories. In: Dirzo R,

1037

Sarukhan J, eds. Perspectives on plant population ecology. Massachusetts, Sinauer, 166-187. Waitt DE, Levin DA. 1993. Phenotypic integration and plastic correlations in Phlox drummondii (Polemoniaceae). American Journal of Botany 80: 1224-1233. Waitt DE, Levin DA. 1998. Genetic and phenotypic correlations in plants: a botanical test of Cheverud's conjecture. Heredity 80: 310-319.

Wilson P. 1995. Selection for pollination success and the mechanical fit of Impatiens flowers around bumblebee bodies. Biological Journal of the Linnean Society 55: 355-383. Wolfe LM, Krstolic JL. 1999. Floral symmetry and its influence on variance in flower size. American Naturalist 154: 484-488.

The Variability of Organs Differentially Involved in ...

respectively. Standard and keel petal size varied in parallel, not only in conspecific individuals but also across species .... massively in winter or early spring.

214KB Sizes 0 Downloads 210 Views

Recommend Documents

anatomical variability of organs via principal factor ...
degrees of freedom in the transformation, the bias from the reference data is minimized, in terms of organ shape, size and position. ..... Mapping (ICBM).

The cerebellum is involved in predicting the sensory consequences of ...
parametrically varied degrees of discrepancy between the predicted and ... Using this robotic interface, computer controlled ... monitor [1] or internal forward model (Fig. 1) has been .... age 24 years; age range 8 years) gave informed consent for t

An Audit of the Processes Involved in Identifying ... - Semantic Scholar
The Commission for Racial Equality (Special Educational Needs. Assessment in Strathclyde: Report of a Formal Investigation,. CRE, London, 1996) highlighted ...

Interdecadal variability in the Gulf of Maine zooplankton ...
each cruise, the sample mesh is cut in sections representing. 10 nautical miles and ...... Our proposal that the community shift pattern in the Gulf of. Maine zooplankton ... sessments and Forecasts for Ecosystems (CAFÉ) in the. Gulf of Maine ...

is the left hippocampal formation involved in the ...
(2) head direction, which is processed in the head direction cell system and (3) place, which is processed at ..... was recorded using a magnetic tracking system.

Interdecadal variability in the Gulf of Maine zooplankton ...
J. W. Jossi: Narragansett Laboratory, 28 Tarzwell Drive,. Narragansett, RI 02882 ... C. finmarchicus has yet to be linked with fish recruitment, it does explain some of ...... Impact of climate variability on the recovery of endan- gered North Atlant

Physiological Condition Differentially Affects the ...
Mar 30, 2010 - Thermal data loggers (resolution ... physiological, behavioral, and environmental data were log10 transformed to reduce ...... cessful recovery of the physiological status of coho salmon on board a ... Elsevier, San Diego, CA.

Physiological Condition Differentially Affects the ...
Mar 30, 2010 - solely on endogenous energy reserves, as sockeye salmon cease .... of clean river water passing through the individual's mouth and gills.

Physiological Condition Differentially Affects the ...
Mar 30, 2010 - physiological stress, are reproductively advanced, are ill pre- .... Wireless) were used with three-element or four-element Yagi ...... technology.

The Intraparietal Cortex: Subregions Involved in ... - Semantic Scholar
the Spike2 software (Cambridge Electronics Design; Cam- bridge, U.K.). Saccades and reaching movements were made to a video (touch-sensitive) screen ...... Boussaoud D, Ungerleider LG, Desimone R (1990) Pathways for mo- tion analysis: cortical connec

Simulating Frontotemporal Pathways Involved in ... - eScholarship
experimental manipulations. The current ..... Using this simulated manipulation of frequency and context, we .... IEEE International Joint Conference on Neural ...

Recruitment of an Area Involved in Eye ... - Semantic Scholar
Jun 19, 2009 - Even young children and uneducated adults readily conceive of numbers as forming an internal spatial continuum or “mental number line” (6). Merely perceiving an Arabic digit suf- fices to elicit a spatial bias in both attentional o

Distinct Hematopoietic Stem Cell Subtypes Are Differentially ...
Mar 4, 2010 - dilutions of whole bone marrow—myeloid-biased (My-bi), lym- phoid-biased ...... files were generated with GCOS 1.0 software (Affymetrix).

Overlapping and distinct brain regions involved in estimating the ...
Overlapping and distinct brain regions involved in esti ... l position of numerical and nonnumerical magnitudes.pdf. Overlapping and distinct brain regions ...

Simulating Frontotemporal Pathways Involved in ...
storing new words should be addressed by a quick automatic process. .... interpretation is generated online by adding neurons between the TL and LIFG, and .... to the same degree interactions between several brain regions that have been ...

P3 explain the security risks and protection mechanisms involved in ...
P3 explain the security risks and protection mechanisms involved in website performance.pdf. P3 explain the security risks and protection mechanisms involved ...

P3 explain the security risks and protection mechanisms involved in ...
virus can have the risk of simply causing your internet connection fail or as ... Facebook has to legally comply with the Data Protection Act. of 1998 ... P3 explain the security risks and protection mechanisms involved in website performance.pdf.

DNA protection mechanisms are not involved in the ...
system (Bio-Rad) at 6.5 V/cm2 for 24 h at 16°C, with a linear pulse ramp of 15±70 s and a switching angle of 120°C. The gels were stained with water containing ...

Automatic segmentation of the thoracic organs for ...
and another with the lungs, the heart and the rest soft tissues is achieved by ..... These scans can detect smaller lung tumors than a conventional CT scan and the ex- amination takes only a few minutes. • With bronchoscopy, a careful examination o

mesoscale variability of organic matter composition in ...
Concentrations are presented as bovine serum albumin (BSA) equivalents. Carbohydrates were analysed according to Gerchacov and Hatcher (1972) and ...