* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
Download Tolerance assessment of Cistus ladanifer to serpentine soils by
Survey
Document related concepts
Plant defense against herbivory wikipedia , lookup
History of botany wikipedia , lookup
Evolutionary history of plants wikipedia , lookup
Plant stress measurement wikipedia , lookup
Ornamental bulbous plant wikipedia , lookup
Plant breeding wikipedia , lookup
Plant morphology wikipedia , lookup
Plant ecology wikipedia , lookup
Plant reproduction wikipedia , lookup
Plant physiology wikipedia , lookup
Plant nutrition wikipedia , lookup
Sustainable landscaping wikipedia , lookup
Glossary of plant morphology wikipedia , lookup
Base-cation saturation ratio wikipedia , lookup
Transcript
Plant Ecology 143: 51–66, 1999. © 1999 Kluwer Academic Publishers. Printed in the Netherlands. 51 Tolerance assessment of Cistus ladanifer to serpentine soils by developmental stability analysis Concepcion L. Alados1 , Teresa Navarro2 & Baltasar Cabezudo2 1 Instituto Pirenaico de Ecologı́a, CSIC. Avda. Montañana, 177. Aptdo 202, 50080 Zaragoza, Spain; 2 Departmento de Biologı́a Vegetal, Universidad de Málaga, Aptdo. 59. 29080 Málaga, Spain Received 14 November 1997; accepted in revised form 28 December 1998 Key words: Cistus ladanifer, Developmental instability, Fluctuating asymmetry, Serpentine, Translational asymmetry Abstract Developmental instability is the result of random environmental perturbations during development. Its absence (developmental stability) depends on an organism’s ability to buffer environmental disturbances. Both genotype and environment influence the phenotypic expression of developmental instability and it is susceptible to selection pressure. We studied developmental instability (as indicated by increased within-individual asymmetry of repeated traits) in vegetative and reproductive structures of three populations of Cistus ladanifer L. living in different soil substrates (serpentine, siliceous and contact zone) to detect tolerance to serpentine soils. Serpentine soils, characterized by high concentrations of heavy metals (Ni, Cr, and Co), low levels of Ca/Mg ratio and high water deficit, can adversely affect plant performance. In this study we demonstrated that asymmetry and within-plant variance were higher in the contact zone population than either the silica or serpentine populations, proving the adaptation of C. ladanifer to serpentine soils. Within-population estimates of developmental instability were concordant for both vegetative and reproductive traits. There was little or no within-individual correlation among estimates of developmental instability based on different structures, i.e., plants that had highly asymmetric leaves always had high developmental instability in translational symmetry. Radial asymmetry of petals was negatively correlated with petal size, especially in silica soil plants, providing evidence of selection for symmetric and large petals. While leaf size was positively correlated with absolute fluctuating asymmetry, suggesting selection for small or intermediate size leaves. Serpentine soils presented the largest foliar and floral traits, as well as shoot elongation, while silica soil plants had the smallest scores. On the contrary, aboveground plant biomass was larger in silica soil plants, while the contact zone plants had the lowest biomass. Introduction The phenotypic variation of a character has genetic, environmental and random components. Developmental ‘noise’ or instability represents the cumulative effects of small, random developmental perturbations of environmental origin (Waddington 1957). It is reflected by exaggerated intra-individual variation in repeated traits and patterns. Developmental stability, its opposite, is the ability of an organism to buffer environmental disturbances encountered during development (Mather 1953; Thoday 1955). Indeed, developmental stability requires the suppression of both genetic and environmental variation (Leary et al. 1992). An apparent confusion between developmental instability and plasticity may appear in plants. Plasticity is the ability of an organism to alter its morphology in response to changes in environmental conditions, that is, it is constant in sign (Bradshaw 1965), and it is controlled by a different genetic system than developmental stability, as demonstrated in Arabidopsis thaliana (L.) Heynh. by Bagchi & Iyama (1983). In principle, any phenotypic trait can be used to measure developmental instability, provided one knows a priori what phenotype should be produced in the absence of stress. Bilateral symmetry is one 52 such developmentally invariant trait. Both sides of a bilateral symmetrical character are assumed to be controlled by the same genes. Random deviations from bilateral symmetry, i.e., fluctuating asymmetry (Sumner & Huestis 1921; Ludwig 1932; Van Valen 1962; Palmer & Strobeck 1986), reflects the inability of an organism to control environmental perturbations during development, and increases under genetic or environmental stress (Mather 1953; Thoday 1958; Valentine & Soulé 1973; Parsons 1992). Developmental instability has the advantage that as a nonspecific measure of developmental disturbances, it can be used as an early sign of anthropogenic impacts on animals and plants (Zakharov et al. 1987; Parsons 1992; Graham et al. 1993; Freeman et al. 1993). Moreover, fluctuating asymmetry has an optimum reference value (zero), while other indices of phenotypic quality such as body size do not (Watson & Thornhill 1994). Not all developmental instability characters are equally sensitive. Stability will depend on the ability of the organism to buffer development against environmental perturbations (degree of canalization according to Waddington 1942) and characters not directly related to individual fitness are expected to display low developmental stability compared with characters important to fitness (Leary et al. 1985; Clarke 1995). In order for developmental instability to be consistently useful in environmental diagnosis, we must perform an integrated analysis of several phenotypic traits (Watson & Thornhill 1994; Swaddle 1997). Indeed, complex traits whose structure depends upon the coordination of several mechanisms have higher potential for integration, and thus buffering. Therefore, they may more accurately represent overall instabilities. Graham et al. (1993) and Freeman et al. (1993) have argued that bilateral symmetry is not the only useful within-individual developmentally invariant trait to detect developmental stability in organisms. For example, radial, spiral and translational (longitudinal) symmetries are within-plant measures of variation (Paxman 1956; Graham 1992; Graham et al. 1993; Freeman et al. 1993; Alados et al. 1994, 1998; Escós et al. 1995, 1997, Sherry & Lord 1996a) and have proven reliable and often superior alternatives to bilateral symmetry. Many plants present radial symmetry in flowers, leaves or fruits and, their within-trait variance may be used as a manifestation of radial asymmetry (Graham et al. 1993). The modular nature of plants allows the examination of acropetal growth, where, for example, internode length varies regularly with node order, and represents a form of transla- tional symmetry with scale (Freeman et al. 1993), and node order represents the scale of measurement. That is, they are invariant with multiplicative changes of scale (Schroeder 1991). The phenotypic expression of developmental instability reflects the inability of the organism to buffer environmental disturbances during development. The genetic basis of developmental stability is still under discussion (Clarke 1993; Markow & Clarke 1997; Møller & Thornhill 1997). Several authors have demonstrated variation in the phenotypic expression of developmental instability between lines and populations (Sakai & Shimamoto 1965, in Nicotiana tabacum L.; Bagchi et al. 1989, in Tectona grandis L; Barrett & Harder, 1992, in genus Eichhornia Kth.). Maynard Smith et al. (1985) hypothesized that recently evolving phenotypes display more instability than well-established ones. Later studies presented evidence to support this theory in the study of hybrid populations (Graham 1992; Freeman et al. 1995), in the evolution of pesticide resistance (Clarke 1993), and in gynodioecious species (Alados et al. 1998). Serpentine areas are important sources of metals, especially nickel (Ni), chromium (Cr) and cobalt (Co). The adverse effect of heavy metal is enhanced by the low levels of calcium (Ca) in relation to magnesium (Mg), the lack of organic matter and the poor physical texture of the soil. This soil structure restricts soil depth, reducing soil penetration and water content and contributing to water stress in plants. As a result, serpentine areas have several endemic species adapted to high concentrations of heavy metals and generally adverse edaphic conditions (Proctor 1971; Brooks 1987; Proctor & Woodell 1975; Roberts & Proctor 1992; Arianoutsou et al. 1993, Freitas & Mooney 1996). Serpentine soil provides a distinct environment, which has persisted for millennia, allowing a specialized and characteristic flora to evolve. It is postulated that directional selection reduces developmental homeostasis during the evolutionary response (Maynard Smith et al. 1985; Watson & Thornhill 1994). That is, recently evolving phenotypes display more instability (Levin 1970; Maynard Smith et al. 1985) than well established ones, but, when the rate of evolution stabilizes, readaptation of the regulatory mechanisms maintain the trait at a new state. If adaptation to serpentine soil takes place, we expect plants living in serpentine substrate to be developmentally more stable than plants living in the contact zone. Indeed, silica plants should also display developmental stability. However a narrow contact zone should be less developmentally stable because col- 53 onizing plants can come from either serpentine or silica adapted populations, producing a preponderant number of non-adapted individuals. In this study we used not only fluctuating asymmetry, but also radial asymmetry, translational symmetry, and statistical noise in allometric relations to detect random intra-individual variability during development (Freeman et al. 1993; Alados et al 1994, 1998; Escós et al. 1995, 1997). Not only we did look at homeostasis disruption, but also at trait size and biomass in Cistus ladanifer (Cistaceae). We wanted to know whether adaptation to serpentine soils had occurred so that we could select stress tolerant ecotypes to use in future restoration of contaminated soils. Methods Study area and species C. ladanifer is an important component of Mediterranean matorral scrub, distributed in SW Europe (France, Spain and Portugal) and N Africa (Morocco and Algeria). It is a calcifuga species that grows on various soil and rock types, with preference for siliceous and ultrabasic soils. The study area is located in Sierra Parda de Tolox (Málaga province), in the southwest of the Iberian Peninsula (latitude 36◦380 N, longitude 4◦ 560 E) with an area of 30 km 2 . This area has serpentine soils (ultrabasic rocks of plutonic origin with serpentines and peridotites), siliceous soils (schists, gneises and pizarras), and contact zones (around 5 m wide) with materials from both substrates. The climate is Mediterranean with an average annual rainfall between 800 and 1000 cc. The mean annual temperature ranges between 10 to 15 ◦ C, but in summer mean temperatures can reach 25 ◦ C. The plant community is Mediterranean degraded matorral scrub dominated by Cistus ladanifer, an erect evergreen shrub, 275–290 cm high. In this area C. ladanifer forms extensive and densely aggregated populations with isolated individuals scattered in the periphery of the clump. The nomenclature of taxa and syntaxa used follows Muñoz Garmendia & Navarro (1993). The pheno-morphological behavior of C. ladanifer (Cabezudo et al. 1992; Talavera et al. 1993) is characterized by seasonal vegetative growth at the apex of the last season’s shoot, or from shorter basal axillary brachyblasts, between April and September. Flowerbuds develop in the axils of the upper leaves of the last season’s dolicoblasts between the first-half of April to the beginning of May and flowering occurs from the end of April to mid May. Data collection We selected three locations with similar sclerophilous Mediterranean vegetation communities, slope exposure, inclination, altitude and general appearance, differing only in soil characteristics (Table 1). The first area was located on serpentine soils, the second in siliceous substrate, and the third on the contact zone. Soil samples were collected from the three locations, air dried and sieved with a 2 mm screen. Concentrations of exchangeable ions in soils were determined with extracts of 1 M ammonium acetate (soil: extractant ratio of 1:10 for 1 h). In order to evaluate the plant absorption response to the different soil concentrations we analyzed ions (Ca2+ and Mg2+ ) and Ni2+ concentrations in C. ladanifer leaves. For this, leaves were oven dried at 60 ◦ C for 48 h and wet ashed with 2 mL of concentrated HNO3 in Teflon pressure vessels for 7 h. Concentrations of Ca2+ and Mg2+ were determined by flame AAS (Perkin-Elmer 380) and concentrations of Ni2+ were determined by graphite furnaces AAS (Perkin-Elmer 5100). C. ladanifer densities were similar in the three areas: 2.1 ind m−2 in serpentine soils and 2.7 ind m−2 in the silica and contact zones. A total of 120 C. ladanifer branches were collected, 40 per plot, at the end of July 1996, when shoots stopped growing and annual shoot lignification began. In order to separate the soil types from other non-controlled ecological variables we performed four transects per soil type, collecting 10 plants per transect. We chose plants of similar size to reduce variability in the analysis. We performed a randomized branch sampling, by randomly selecting one well-developed 3-year-old branch, from a specific orientation. Later, in the laboratory, we measured internode lengths from one developed annual shoot per branch. We also selected six leaves located in the same position (7th or 8th node counting from the base of the annual shoot, corresponding to L7 and L8 of Figure 1) from six different shoots per plant. Because flowers develop on the previous year’s stem, before the current shoot finishes elongation, we collected flowers at the beginning of June, from different individuals than those we took leaves and branches from. Six flowers per plant were selected from 96 individuals, 32 per soil type and 8 plants per transect. Floral and leaf asymmetries, internode lengths, and 54 Table 1. Concentrations of Ni2+ , Mg2+ , Ca2+ and Ca/Mg ratio Cistus ladanifer leaves in serpentine, siliceous and contact soils. Exchangeable concentrations in soils were obtained from 1 M ammonium acetate soil extracts. (Mean ± sd, n), n.d. not detected. Means with the same letters indicate no significant differences at the 0.05 level, Bonferroni test. Serpentine Contact Siliceous 1.19 ± 0.01 (2) 0.11 ± 0.02 (3) 3.53 ± 0.08 (3) 0.03 ± 0.01 (3) 0.15 ± 0.01 (2) 0.33 ± 0.02 (2) 0.84 ± 0.06 (2) 0.40 ± 0.06 (2) n.d.(2) 0.39 ± 0.00 (2) 0.20 ± 0.08 (2) 2.05 ± 0.08 (2) 10.10 ± 2.01 (5)a 4.85 ± 1.31 (9)a 3.24 ± 0.78 (9)a 1.63 ± 0.72 (9)a 7.92 ± 0.83 (4)a 2.43 ± 0.82 (8)b 1.72 ± 0.58 (8)b 1.54 ± 0.70 (8)a 3.75 ± 2.00 (5)b 4.08 ± 0.53 (6)a 2.48 ± 0.44 (6)ab 1.70 ± 0.44 (6)a Soil Ni2+ µg/g Ca2+ mg/g Mg2+ mg/g Ca/Mg Leaves Ni2+ µg/g Ca2+ mg/g Mg2+ mg/g Ca/Mg floral and leaf allometries were all examined to detect consistency in developmental stability. To reduce measurement error, measurements were performed in the laboratory by the same person, with an electronic caliper recording to the nearest 0.01 mm. Additionally, dried plant biomass was estimated for one 3-year-old branch from 36 different plants, 12 from each soil type. Plants were oven dried at 70 ◦ C for 48 h and weighed with a precision balance. Data measurement and analysis (a) Fluctuating asymmetry The traditional way to assess developmental instability is via fluctuating asymmetry. The leaves are 40–120×10–21 mm in size and are malacophyllous, sessile or subsessile with linear-lanceolate shape, and with 14–16 months old. We examined the asymmetry of leaves by measuring the width on each side of the midrib at the mid point of the leaf blade. Absolute fluctuating asymmetry (AFA) was calculated as the unsigned left (L) minus right (R) difference. Relative fluctuating asymmetry (RFA) was calculated as the absolute value of left (L) minus right (R) divided by the average (L + R)/2, correspond to index 2 of Palmer & Strobeck (1986). This scaling procedure has been used by many authors to correct for possible associations between asymmetry and leaf size. The validity of fluctuating asymmetry interpretation depends, according to Palmer & Strobeck (1986, 1992), on an absence of directional asymmetry (skew), anti-symmetry (bimodality or platykurtosis), and a normal distribution for (L − R) with mean zero. (b) Translational symmetry In addition to examining fluctuating asymmetry, we also examined the allometric relationship between the internode lengths and the node order (Alados et al. 1994, 1998; Escós et al. 1995, 1997). Its organization patterns and architectural diagrams are shown in Figure 1. C. ladanifer is a phanerophyte species with acropetal branch-shedding, and with two branch types: glabrous long shoots (dolicoblasts) and glabrous and very viscid short shoots (brachyblasts). The scaling relationship has previously been applied to plant developmental stability (Paxman 1956; Freeman et al. 1993; Sherry & Lord 1996a) although none of these earlier studies examined internode lengths per se. The relationship between internode length and node order follow a self–similar (or self-affine) sequence, invariant with multiplicative changes of scale (Schroeder 1991), where internode order (counted from the stem base) is interpreted as a scaling factor (Freeman et al. 1993). The relation between internode length (L) and node order (N), starting from the shoot base to its apex, fits the general equation. L = kN a e−bN , (1) where e is the natural base, and k, a, and b are fitted constants. 55 instability (a decline in the accuracy of the curve fitting). Thus, one minus the coefficient of determination R 2 , the standard error of the regression (Syx ), and the standard error of the parameters a and b (Sa and Sb ), are good estimators of developmental instability or homeostasis disruption. Regressions were performed separately for each plant and the resulting R 2 , Syx, and, Sa or Sb were then analyzed. Because the number of internodes varies among shoots, we used R 2 adjusted to the mean square sum. Figure 1. (a) Organization pattern and architectural diagram of C. ladanifer (according to Scarrone and Leeuwenberg models) corresponding to a three year old branched system: current season shoot (I), former season shoots (II, III, IV), sylleptic dolicoblasts (sd), single flowers from last season shoot (f). (b) Architectural unit of C. ladanifer: axillary buds for the next flower formation (d), scale node from current season shoot (s). L1 to L10 are internode length. This equation has two components. The first component (kN a ) corresponds to the allometric relationship between internode length and node order (Escós et al. 1997). The second component (e−bN ) represents the inhibition mechanism of flower formation (Meinhardt 1984). The parameter values of this equation do not have intrinsic biological meaning. It is simply a convenient function to express individual differences. Taking the logarithm of both sides of the equation gives: ln L = ln k + a ln N − b N (2) The constants can be obtained from a regression analysis. Under stress, the impact of random perturbations increases, leading to enhanced developmental (c) Radial symmetry The flower is actinomorphic and solitary and among the largest in the Spanish flora (Talavera et al. 1993). The single flowers are flat, with radial symmetry in five planes of reflectional symmetry, and with a dish-like appearance during pollination (Faegri & Pijl 1971). Flowers are white and shortly pedicellated, with five petals of 38–46 mm size, and basal bracts. The calyx bears 3 sepals 13–19 mm long. The variance in petal or sepal length affects the radial symmetry of the flower. We used the standard deviation of petal and sepal length as a measure of developmental instability instead of the coefficient of variation, since the latter measure declines with trait size when developmental instability is independent of trait size (Soulé 1982). Another way of estimating radial asymmetry in floral traits is by performing a nested ANOVA with trait size nested within flowers, flowers nested within individuals, and individuals nested within transects. The nested analysis yielded estimates of variance due to differences between floral traits (petals or sepals length or width) within flowers, or radial asymmetry. (d) Allometric error Leaf or petal length (L) and width (W) is related by a power law, as also is calix allometry (see details in: Alados et al. 1998). La _ W b. Thus, ln L and ln W are linearly related, and we can write β1 ln L + β2 ln W = 1 + , where is an error term. From the same plant structure (leaf, petal or sepal) we measured n repeated units to obtain the matrix x11 x12 . . . x1n X= x21 x22 . . . x2n where row 1 is length and row 2 is width. 56 In matrix form, + βX = 1. Allometry error is calculated as: 0 /n − 1 AL = (n − β X 10 )/n − 1 being β transposed equal to β 0 = (XX0 )−1 X10 . Statistical analysis Measurement error was estimated by re-measuring the same plant after all the measurements had been taken once. Repeatability of morphological measures of the same individuals was estimated by a two way nested ANOVA where leaf side petals or sepals were nested within individuals. Individual was the random effect factor and within error variance, the measurement error (Sokal & Rohlf 1980; Merilä & Björklund, 1995). Measurement error was estimated from 30 individuals measured twice and then calculated as the proportion of within-measurement MS (mean square) to the among-individual MS (fraction of trait size variation due to measurement error) and, to the individual within-trait MS (fraction of developmental instability variation due to measurement error) (Appendix1). Measurement error in foliar and floral structures was insignificant; values ranged between 0.002 for half leaf width and 0.04 for petal length and width (Appendix 1). The proportion of variance in organ length or width due to measurement error ranged between 0.56% for sepal width to 0.02% of petal width. Finally, the proportion of within individual variation (developmental instability) due to measurement error was close to 1% for the less precise measurements of sepal width and half leaf width. Fluctuating asymmetry refers to randomly distributed differences between right and left sides in bilateral characters. Tests for fluctuating asymmetry depend on an absence of directional asymmetry (systematic biases towards greater development of one side, i.e., skewness), antisymmetry (systematic nondirectional deviations from bilateral symmetry, i.e., bimodality or platykurtosis) and, according to Palmer & Strobeck (1986) a normal distribution of the unsigned difference. To detect confounding factors, a series of preliminary tests were performed. We first did a mixed model ANOVA with sides and soil type as fixed effect factors and three level nested analysis, with transect nested in soil, individuals nested in transect, and leaves nested within individuals as an error term. Side effects represented the directional asymmetry, soil effects represented size and shape variations, and the interaction term represented the antisymmetry. Inter-individual variation was represented by the individual within transect term. The error term referred to the intra-individual variation between leaves in a plant and represents the overall intra-individual variation in half leaf width including not only AFA but also interleaf variability. Estimating the variance components (after Sokal & Rohlf 1981) completed the nested model ANOVA. Normality tests were then done on the distribution of signed (L − R) differences to test skewness and kurtosis. Correlations between the magnitude of AFA (L − R) and character size (L + R)/2 were also investigated to prevent size scaling. Variation in fluctuating asymmetry due to soil type was investigated by three levels nested analysis of variance with soil type as the fixed effect factor, transect nested in soil and individuals nested in transect. The error term represented the among-leaf withinindividual variance. In the same way we compared floral trait size and within flower variance, where the error term represented the among-flower withinindividual variance. Within trait variance was also calculated from the error term of three nested ANOVAs, one per soil type, where trait sizes were nested within flowers, flowers nested within individuals, and individuals nested in transects. A two-level nested analysis of covariance, with transects nested within soil types and, the naperian logarithm of the base stem diameter as covariate, was performed to measure the effect of soil type on curve fitting estimators: R 2 , Syx , Sa and, Sb ; and on the regression parameters: a, b and ln k. Comparisons between soil types were assessed by a two-tailed F test with sequential Bonferroni correction on the P values, to protect against group-wide type errors associated with multiple tests (Rice 1989). When intermediate level mean squares were not significant we pooled the implicated source of variation and calculated F values from pooled mean squares. All variables were tested for normality. When the variables were not normally distributed we transformed the data as suggested by Sokal & Rholf (1981). 57 Table 2. Partitioning analysis of variance with two fixed effect factors (sides and soil type) and three level nested analysis, with transect nested within soil (T ⊂ S), individuals nested within transect (I ⊂ T ), and leaves nested within individuals as the error term. Dependent variable is half leaf width. ∗∗∗ P < 0.001, ∗∗ P < 0.01, ∗ P < 0.05, no asterisks indicates P > 0.05, s 2 is the variance components. Source MS df F s2 0.22 544.24 0.19 3.68 4.53 0.49 (1, 1202) (2, 1202) (2, 1200) (18, 216) (216, 1200) 1200 0.44 1110.1∗∗∗ 0.39 0.81 9.23∗ 0.0% 79.2% 0.2% 0.2% 11.8% 8.6% Half leaf width Sides Soil (S) Sides × Soil (T ⊂ S) (I ⊂ T ) Error Table 3. Partitioning analysis of variance with the three level nested analysis, with soil as fixed effect, transept nested in soil (T ⊂ S), individuals nested within transept (I ⊂ T ) and, leaves nested within individuals as the error term. Dependent variable is AFA. ∗∗∗ P < 0.001, ∗∗ P < 0.01, ∗ P < 0.05, no asterisks indicates P > 0.05, s 2 is the variance components. Source MS df F s2 17.87∗∗∗ 1.54 1.24∗ 7.4% 1.2% 3.6% 87.8% Absolute fluctuating asymmetry Soil (S) (T ⊂ S) (I ⊂ T ) Error 1.677 0.178 0.115 0.092 (2, 9) (9, 708) (108, 600) 600 Results (a) Fluctuating asymmetry Signed (L − R) averaged for all leaf data (n = 720) gave a mean ± se of −0.024 ± 0.02, which is not significantly different from zero (t-test =1.33, NS). The skewness of the distribution was −0.029 (t-test=0.32 NS) and the kurtosis 0.404 (t-test=2.21, P < 0.05). That is, the distribution was leptokurtic. The mixed model ANOVA which held the side of the leaf and soil type as fixed effects (Table 2) showed that leaf side explained an insignificant amount of the total variance, indicating an absence of directional asymmetry, i.e. skewness. Thus, one side of the leaf was not consistently larger than the opposite. Leaf side with soil interaction, which represents the antisymmetry, was not significant. Soil type significantly affected the half leaf width, which represents the size and shape variation in leaves, and explained 79.2% of the variance. The transect within soil nested factor was also not significant, indicating no transect effect. The individual within transect term was significant due to the great inter-individual variability. Finally, intra-individual variation, represented by the error term, absorbed 8.6% of the explained variance, and was responsible for the non-directional asymmetry plus the intra-individual variance in half leaf width. That is, this term included the variation due to fluctuating asymmetry plus the variation between leaves within individual. Size scaling appeared unnecessary as the correlation between the magnitude of absolute fluctuating asymmetry and character size is very low, although significant (r = 0.157, n = 720, P < 0.05). Multiple analysis of variance with absolute fluctuating asymmetry and average half leaf width as dependent variables was significant (Wilks’ lambda = 0.03, F2,707 = 13960.8, P < 0.001), with a significant effect of soil type (Wilks’ lambda = 0.57, F4,1414 = 117.3, P < 0.001). Posterior analyses were performed separately for each variable. In order to determine the effect of soil type on absolute (L − R) fluctuating asymmetry of leaves we performed a three levels nested analysis of variance with soil type as a fixed effect factor. The partitioning nested analysis of variance presented in Table 3 revealed that soil type accounts for 7.4% of variance, significantly affecting the absolute fluctuating asymmetry (AFA) of C. ladanifer leaves. Average absolute fluctuating asymmetry increased significantly for plants living in contact and in serpentine areas in comparison with plants living in the silica zone (Table 4). Relative fluctuating asymmetry (RFA) was analyzed by the same procedure. Average relative fluctuating asymmetry increased significantly for plants living in the contact zone (P < 0.05, Bonferroni adjustment), although the overall difference was not significant (F2,9 = 1.91, NS). The decline in significance of relative fluctuating asymmetry in comparison with absolute fluctuating asymmetry was due to the relationship between soil type and half leaf width. Half leaf width in silica soil was smaller than in serpentine or in contact areas (Table 4, F2,717 = 252.46, P < 0.001). Therefore, once asymmetry was adjusted for leaf size, differences among soil types declined. 58 Table 4. Means ± se (n) of absolute fluctuating asymmetry (AFA), relative fluctuating asymmetry (RFA), half leaf width (HLW), and leaf allometry error (ALH) for C. ladanifer inhabiting silica, serpentine and contact zones. Means with the same letters indicate no significant differences at the 0.05 level, Bonferroni test. Trait AFA RFA HLW ALH Silica Contact Serpentines 0.29 ± 0.02 (240)a 0.46 ± 0.02 (240)b 0.06 ± 0.03 (240)a 0.07 ± 0.03 (240)b 10.20 ± 0.03 (240)a 0.0044 ± 0.0007 (40)a 13.55 ± 0.03 (240)b 0.0038 ± 0.0007 (40)a 0.40 ± 0.02 (240)b 0.06 ± 0.03 (240)a 14.16 ± 0.03 (240)c 0.0037 ± 0.0007 (40)a Table 5. Means ± se (n) and F values of the two level nested analysis of covariance, with transect nested in treatments and naperian logarithm of the base stem diameter as covariate. Dependent variables are the scale asymmetry parameters of the relation between internode length and node 2 ), and equation parameters a, b and ln k, are from the leaf arrangement order (Syx , Sa , Sb and Radj equation L = k · N a · e−bN . ∗∗∗ P < 0.001, ∗∗ P < 0.01, ∗ P < 0.05, no asterisks indicates P > 0.05. Means with the same letters show no significant differences at the 0.05 level, Bonferroni test. Syx Sa Sb 2 Radj a -b ln k Silica Contact Serpentine F(2, 116) 0.678±0.035 (40)ab 0.854±0.043 (40)a 0.183±0.014 (40)a 0.462±0.040 (40)ab 1.613±0.242 (40)a −0.247±0.085 (40)a 1.388±0.193 (40)a 0.790±0.033 (40)a 1.188±0.040 (40)b 0.296±0.013 (40)b 0.382±0.038 (40)a 2.395±0.226 (40)ab −0.531±0.079 (40)b 1.763±0.180 (40)a 0.567±0.035 (40)b 1.011±0.042 (40)c 0.288±0.014 (40)b 0.582±0.040 (40) b 2.615±0.238 (40) b −0.819±0.084 (40)c 3.156±0.190 (40)b 10.64∗∗∗ 15.33∗∗∗ 19.12∗∗∗ 6.39∗∗ 4.24∗ 10.10∗∗∗ 22.20∗∗∗ (b) Translational symmetry The relation between internode length (L) and node order (N) is given by L = 8.18 N 2,21 e−0.53N , averaged over all individuals, and represents the translational symmetry of internode length. The parameter values of the relationships between internode length and node order are presented in Table 5, for each soil type separately. The intercept of Equation (1) represents the starting conditions in the relationship between internode length and node order. A larger intercept indicates larger internode length at the shoot base. In this study we observed that the intercept changed significantly within the soil substrate. Plants living in serpentine soils presented the larger intercept. Parameter a represents the rate of internode enlargement with node order, which changes significantly with soil type and increases in the serpentine zone. Plants living in silica soil show a lower a, significantly different from plants living in serpentines. The parameter b, representing the inhibition mechanism of growth patterns due to flower formation, describes internode shrinking as we go up the stem. In consequence the larger the absolute value of b, the faster internode shortening occurs. The parameter b varied significantly between soil types, reaching its lowest value in serpentinophytes, and the highest in siliceous substrata plants (Table 5). A clear picture of the morphological aspect that produces the variations in the parameter values averaged per soil type is presented in Figure 2. This figure shows that shoots produced by plants living in silica grow slower during the first internodes, but as node order increases, the decline in internode length with order is less steep than in serpentinophytes, where it drops drastically after a few internodes. The number of internodes per annual shoot varied significantly between populations (F2,117 = 6.16, P < 0.01). The silica population presented more internodes (11.1 ± 0.4, n = 40) than serpentinophytes (9.2 ± 0.4, n = 40), while the contact population was intermediate (9.9 ± 0.4, n = 40). In spite of this, the total length achieved at the end of the growth period was significantly larger in serpentine plants than in either silica or contact zone plants 59 and Sb , respectively, indicating that variables are normally distributed. The partitioning nested analysis of variance revealed that soil type explains 9.8% of the variance in R 2 . Soil differences accounted for 17.1%, 18.6% and 16.7% of variance in Syx , Sa and Sb , respectively. The results, presented in Table 5, show a lower value for R 2 , and a higher Syx , indicating lower developmental stability, in the contact zone. The best fit was observed in the serpentine substrate. Sa and Sb also were the worst fit for plants from the contact substrate. Comparisons between soil types were significantly different. There were no differences among the transects within soil types. Comparisons between soil types were assessed using a Bonferroni correction. Above ground biomass of 3 year-old branches varied significantly between soil types (F2,33 = 3.22, P = 0.05). Contact zone plants had lower biomass (29.47 ± 2.4 g, n = 12) than serpentine plants (33.50 ± 2.4 g, n = 12), while C. ladanifer inhabiting siliceous soil presented the largest biomass (38.30 ± 2.4 g, n = 12). Figure 2. Scatterplots of the non-linear, best-fit equation L=4.01N 1.61 e−0.25N for plants living in silica, L=5.83N 2.39 e−0.53N from transition and, L=23.48N 2.61 e−0.82N from serpentine soil. L is internode length and N is node order. Equation parameters were obtained by least square regression analyses of the linearized equation (ln L = ln k + a ln N − bN ). (P < 0.001 for Bonferroni test, F2,117 = 31.13, P < 0.001). Average ± se shoot length in cm. over all the shoots measured was 26.85 ± 0.57 (40) for serpentinophytes, 21.21 ± 0.57 (40) for silica substrate plants, and 21.45 ± 0.57 (40) in the contact zone plants. To determine how soil type affects developmental stability of internode length, we performed a nested analysis of covariance of the adjusted coefficient of determination, R 2 , standard error of the regression, Syx , standard error of the parameter a, Sa , and standard error of the parameter b, Sb , as dependent variables. A Kolmogorov-Smirnov one sample test for the normal distribution gave a maximum difference of D = 0.08, D = 0.03, D = 0.06, D = 0.07, lower than the critical value 0.094 for P < 0.01, for R 2 , Syx , Sa , (c) Radial symmetry Average petal and sepal length and width were normally distributed (maximal difference of the Kolmogorov–Smirnov normality test are: 0.04, 0.03, 0.06, 0.03 respectively; P < 0.01). Table 6 shows the variation of floral trait size with soil characteristics. Results of the three level nested analysis of variance show that petal and sepal length (F2,9 = 19.33; F2,489 = 227.98) and petal and sepal width (F2,9 = 38.50; and F2,489 = 330.84), are significantly larger in serpentine soils (P < 0.001) than in silica or contact soils. Before we performed the analysis of variance, we tested for normality of floral symmetry variables. Results of the Kolmogorov–Smirnov normality test gave a maximal difference of D = 0.09, for standard deviation of petal length (SDLP); D = 0.10, for standard deviation of petal width (SDWP); D = 0.11, for standard deviation of sepal length (SDLS), D = 0.07 for standard deviation of sepal width (SDWS). Because each variable was larger than the critical value D = 0.043, for P < 0.01, we log transformed the data to meet the assumptions of analysis of variance. Correlation analyses between floral asymmetry and trait size revealed that this relationship changed depending on whether all the flowers are analyzed together or were separated by soil type. While a low negative correlation was found between petal length and the standard deviation of petal length (r = −0.16) 60 Table 6. Means ± sd (n) of average floral trait per flower, and floral asymmetry. MS represents mean squares of the nested analysis of variance for the within flower variation term. Means with different letters indicate significant difference at the 0.05 level, sequential Bonferroni test; g1 and g2 are the skewness and kurtosis statistics. ∗ Indicates the statistic is significantly different from 0 at the 0.05 level. Trait Silica Mean ± sd (n) g1 Av. petal length 33.79±3.71(192)a −0.41∗ Av. petal with 28.95±3.16(192)a −0.30∗ Av. sepal length 10.02±0.96(192)a −0.12 Av. sepal with 8.11±0.81(192)a 0.40∗ 0.97∗ Sd. petal length 1.44±0.79(192)a a Sd. petal with 1.90±0.99(192) 1.32∗ a Sd sepal length 0.55±0.35(192) 2.20∗ a Sd. sepal with 0.84±0.47(192) 1.01∗ a MS petal length 2.75 MS petal with 4.60a MS sepal length 0.43a MS sepal with 0.92a g2 Contact Mean ± sd (n) g1 g2 0.35 38.44±4.64(192)b 0.54∗ −0.02 b 0.35 35.07±4.70(192) −0.15 −0.26 −0.57 11.17±0.85(192)b −0.09 −0.35 −0.36 9.17±1.04(192)b 0.65∗ 1.19∗ 0.70∗ 1.84±0.96(192)b 1.31∗ 3.58∗ ∗ b ∗ 1.75 2.67±1.43(192) 1.51 4.28∗ ∗ b ∗ 9.88 0.71±0.43(192) 1.08 1.32∗ ∗ a ∗ 1.05 0.85±0.48(192) 0.74 0.52 4.37b 9.20b 0.68b 0.95a for all the 576 flowers together. Separated analyses showed that plants living in silica soil presented a correlation coefficient: r = −0.31, P < 0.001, n = 192; while, serpentinophytes presented no such correlation (r = −0.05, NS), and contact soil plants presented intermediate values (r = −0.18, P < 0.05). Analyses of floral asymmetry, measured by the standard deviation of petal or sepal length or width, are presented in Table 7. Multiple analysis of variance including all the floral asymmetry variables and average floral traits revealed a significant effect of soil type (Wilks’ lambda = 0.34, F20,1128 = 78.62, P < 0.001). Results of the analysis are presented for average floral traits and floral asymmetry separately. The percent of the variance explained by soil type effects ranged from 10% in petal length, to 0% in sepal width. No significant differences in transects nested within soil types were observed. In consequence, treatment effects were tested over the error term plus the variance components due to the interaction term. Within flower variance was significantly influenced by soil type for petal and sepal length and petal width. Variation among flowers within individuals was lower than variation among individuals within transects (see MSE vs MSI ⊂T in Table 7). Mean and se values of floral asymmetry are presented in Table 6. Floral asymmetry measured by standard deviation of floral traits or by mean square of the nested ANOVA for the within flower variation term give the same results. Plants living in the contact zone had higher develop- Serpentine Mean ± sd (n) 41.83±3.93(192)c 40.65±4.88(192)c 11.42±1.11(192)c 10.01±0.91(192)c 1.29±0.78(192)a 2.11±1.08(192)a 0.77±0.54(192)b 0.90±0.48(192)a 2.28a 5.61a 0.87c 1.04a g1 g2 0.15 −0.23 0.25 −0.02 0.43∗ 1.22∗ 0.14 −0.25 1.65∗ 3.55∗ ∗ 1.00 0.86∗ ∗ 2.19 8.31∗ ∗ 0.68 0.30 mental instability in petal length and width than those from silica and serpentine zones. Sepals behaved more like leaves, tending towards increasing instability in serpentinophytes, even presenting higher sepal length asymmetry there than in the contact zone. Sepal width asymmetry was similar in the three populations. No significant difference between transects nested within treatments appeared in any case. (d) Allometric error Errors in the allometric relationship among length and width of leaves, petals and sepals were not normally distributed (Kolmogorov–Smirnov test D = 0.18, 0.14, P < 0.01 for ALP and ALS, respectively). Consequently, a logarithmic transformation was applied. No significant effect of soil type was observed on the naperian logarithm of leaf allometric error (F2,107 = 0.02, NS). The partitioning analysis of variance revealed that petal allometric error, which explained 4.7% of the variance, varied significantly with soil type. That is, plants in the contact zone presented a significantly higher petal allometric error (1.6 × 10−4 ± 0.1 × 10−4 ) than those in serpentine (0.6 × 10−4 ± 0.1 × 10−4 ). Sepal allometric error explained only 0.8% and did not present any significant variation. A Pearson correlation matrix of comparisons among developmental instability measurements (Table 8) is presented for floral structures separately from 61 Table 7. Partitioning analysis of variance of the three level nested analysis, with transect nested in soil (T ⊂ S), individuals nested with transect and (I ⊂ T ), flowers nested on individuals as error term. Dependent variables are petal and sepal asymmetries measured as the standard deviation of petal length (SDLP), petal width (SDWP), sepal length (SDLS) and sepal width (SDWS), and petal and sepal allometry error (ALP, and ALS respectively) after their logarithmic transformation. ∗∗∗ P < 0.001, ∗∗ P < 0.01, ∗ P < 0.05, no asterisks indicates P > 0.05, s 2 is the variance components. Source SDLP MS df 3.58 0.38 0.25 0.13 Source SDLS MS df 1.38 0.27 0.14 0.10 s2 (2, 489) 25.82∗∗∗ 9.7% (9, 84) 1.53 1.6% (84, 480) 1.83∗ 10.8% 480 77.9% Soil (S) (T ⊂ S) (I ⊂ T ) Error Soil (T ⊂ S) (I ⊂ T ) Error F F s2 (2, 489) 12.87∗∗∗ 4.8% (9, 84) 1.90 2.3% (84, 480) 1.38∗ 5.5% 480 87.4% SDWP MS df 3.44 0.20 0.23 0.15 (2, 489) 22.56∗∗∗ (9, 84) 0.87 (84, 480) 1.55∗ 480 SDWS MS df 0.11 0.24 0.19 0.10 s2 F (2, 489) (9, 84) (84, 480) 480 foliar structures and stem growth patterns because floral measurements were performed on different individuals. The results revealed that no correlations existed between different organs. Discussion Serpentine soils, characterized by impoverished substrate with high concentrations of heavy metals, in particular Ni, provide an excellent opportunity to study the tolerance of plants to metalipherous substrate. Soil showed a clear increase in Ni2+ and Mg2+ and a decrease in Ca2+ from siliceous soil to serpentine soils (table 1). The high concentration of Ni2+ in serpentine soil lead to a high concentration of Ni2+ in leaves of this soil type, reaching critical levels observed in other moderately tolerant species (Marschner 1986). Serpentine soils had high amounts of Mg2+ and low Ca2+ , with a Ca/Mg ratio lower than one. That the Ca/Mg ratio in leaves is higher than one indicates that the plants on serpentine soils actively take up Ca2+ . Amounts of Ni2+ , Ca2+ and Mg2+ were intermediate in soils from the contact zone. While leaves from this zone maintained a Ca/Mg ratio similar to plants on silica and serpentine soils. Actual levels of both Ca2+ and Mg2+ in leaves from the contact zone were lower than in plants from the other two zones. The lower amounts in the contact zone may due to Ni2+ , a divalent cation that competes with other divalent cations 9.27% 0% 7.7% 83.4% s2 F 1.09 1.26 1.80∗ 0% 0.86% 11.7% 87.9% ALP MS df 2.1 10−6 3.8 10−7 2.8 10−7 1.5 10−7 (2, 489) 13.07∗∗∗ 4.7% (9, 84) 1.35 1.1% (84, 480) 1.83∗ 11.4% 480 82.7% ALS MS df 1.2 10−6 4.5 10−7 5.2 10−7 4.8 10−7 (2, 489) (9,564) (84, 480) 480 F F 2.6 0.92 1.1 s2 s2 0.8% 0% 1.6% 97.9% (Ca2+ or Mg2+ ) and induces a deficiency of these elements. Apparently, there was been no adaptation in the contact zone for selective uptake of these elements in the presence of Ni2+ . This result is concordant with the ecology of C. ladanifer, which prefers acid soils with low Ca2+ concentrations, where can successfully compete with other species. Environmental stress has been correlated with abnormal traits in several plant species (Grant 1956, 1975; Huether 1969; Barret & Harder 1992; Kozlov et al. 1996). However patterns of instability may result from complex interactions between the genotype and the environment (Ellstrand & Mitchell 1988; Barrett & Harder 1992), providing evidence that several vegetative and reproductive traits are required to assess developmental stability as a stress detector. In this study we observed that plants living in the contact zone presented larger absolute fluctuating asymmetry of leaves (relative fluctuating asymmetry was not significant), larger radial asymmetry in petal length and width, and larger translational symmetry of internode length in the contact area than in silica or serpentine soils. This is probably because directional selection has ceased in serpentine areas where regulatory homeostatic mechanisms have restored developmental stability in vegetative and reproductive traits. However, in the contact zone, destabilization occurs because colonizing plants can come from either serpentine or silica adapted populations (Talavera et al. 1993). Talavera et al. (1993) demonstrated that C. ladanifer has a 62 Table 8. Pearson correlation matrix of comparisons between the developmental instability measurements used in this study: absolute fluctuating asymmetry (AFA), leaf allometry 2 ), standard deviaerror (ALH), translatory asymmetry parameters (Syx , Sa , Sb and Radj tion of petal length (SDLP), petal width (SDWP), sepal length (SDLS) and sepal width (SDWS), and petal and sepal allometry error (ALP, and ALS respectively) ∗∗∗ P < 0.001, ∗∗ P < 0.01, ∗P < 0.05, no asterisks indicates P > 0.05. AFA ALH 2 Radj Syx Sb Sa SDLP SDWP ALP SDLS SDWS ALS AFA ALH 1 −0.043 0.068 1 −0.118 −0.030 0.083 0.056 0.144 0.003 0.076 2 Radj Syx Sb Sa −0.806∗∗∗ −0.141 −0.518∗∗∗ 1 0.273∗∗ 0.710∗∗∗ 1 0.868∗∗∗ 1 1 SDLP SDWP ALP SDLS SDWS ALS 1 0.683∗∗∗ 0.800∗∗∗ 0.067 0.068 −0.086 1 0.488∗∗∗ 0.129 0.106 −0.021 1 −0.058 −0.018 0.034 1 0.202∗ 0.368∗∗∗ 1 0.338∗∗ 1 gametophytic mechanism of incompatibility together with a low pollinator mobility that results in declining plant fecundity with increased nearest neighbor distance. This self-incompatibility limits gene exchange, ultimately resulting in the selection of serpentine and silica ecotypes. Evidence supporting the restoration of stability after a change in environmental conditions has been found by McKenzie & O’Farrel (1993). They reported that, after resistance was widespread, the continued use of pesticides selected for a modifier allele in Lucilia cuprina blowfly that restored the relative fitness of resistant phenotypes in the absence of pesticides (McKenzie et al. 1982). In the absence of modifier phenotypes, resistant strains are less successful than susceptible strains when pesticides were absent, while in the presence of the modifiers success is similar (McKenzie 1994). Differences in values of bilateral traits are usually small (often <1–5% of the total variation of a given trait; Merilä & Björklund, 1995). To prevent differences in values of bilateral traits due to measurement error, Palmer & Strobeck (1986) and Merilä & Björklund (1995) pointed out the necessity of controlling measurement error in fluctuating asymmetry studies. In the latter study, the percentage of variance explained by floral or foliar trait asymmetry ranged between 5 and 10%. In our study, measurement error accounted for only 1% of the developmental instability index in the worst case. This, together with the fact that measurement error was evenly distributed between the different treatments, led us to discard the possibility of a reduction in our accuracy due to measurement error. Symmetry occurs because organs develop mechanisms to buffer environmental disturbances during development (Mather 1953; Thoday 1955). As a result, leptokurtosis in bilaterally symmetrical organs should be more pronounced since selection acts to minimize asymmetry. Leptokurtic asymmetries are found in several studies (Harvey & Walsh 1993; Polak 1993). According to Leung & Forbes (1997), leptokurtosis could also arise when developmental noise is normally distributed and so should not be excluded from asymmetry studies. We observed leptokurtosis in floral and foliar asymmetries, proving the existence of a selection mechanism for symmetrical flowers and leaves. Constancy of floral traits within taxa is well documented, but deteriorates near the limit of ecological tolerance (Stebbins 1951; Bradshaw 1965). In consequence, we may expect floral structures to be under high homeostatic control (Sherry & Lord 1996b), and as a result, to be reliable indicators of developmen- 63 tal homeostasis and of phenotypic quality. Petals are a major visual attractant for animal pollinators, and petal size and symmetry are means of competing for pollen vectors (Bell 1985; Willson 1990; Møller & Eriksson 1994; Møller 1995a; Cronk & Möller 1997). In concordance with that, we found a negative correlation between asymmetry and petal size, especially in silica plants, indicating a directional selection towards large and symmetrical flowers. However, leaf width asymmetry was positively related to mean leaf width, as also observed by Møller (1995b) in Ulmus glabra Hudson, and Kozlov et al. (1996) in Betula pubescens Ehrh. The higher fluctuating asymmetry of larger leaves may be due to the larger cost of the bigger traits. As a result, leaf size will be under stabilizing selection with an optimal leaf size that may depend on environmental factors such as resource availability. Traditionally, studies of developmental instability have been based on fluctuating asymmetry and number of phenodeviantes (Markow 1994). In this study, we prove that other traits can indicate developmental instability, i.e., variation in translational symmetry is concordant with other within plant variations (see also Alados et al. 1994, 1998; Escós et al. 1995, 1997). Growth rhythm varies among individuals and with developmental stage (Comte 1993). In this study we also demonstrate that variation in growth patterns may be an indicator of environmental disturbance. Thus, serpentinophytes grow faster at the beginning of the growing season and produce fewer internodes, while the silica soil population has a sustainable and slower growing period but with a larger number of internodes. After internode enlargement reaches its maximum, the production of sylleptic branches takes place. In general, foliar and floral traits are larger in serpentine plants both for vegetative (annual shoot length and leaf width) and reproductive (petal and sepal length and width) structures, whereas silica populations show lower floral and foliar trait size, and the contact population is intermediate. Phenological delay between individuals of the same species is also an indicator of low plant vigor (Borchert 1976; Ng 1979). The phenological phase of vegetative growth occurs at the same time for the serpentine and siliceous soil populations, and with a delay of around 15 days in the contact zone. A similar trend was observed during flowering, with individuals growing in the contact zone flowering later than the others. The higher developmental stability of C. ladanifer inhabiting serpentine substrate together with its rapid growth at the beginning of the growing season, just before the sylleptic extension for brachyblast formation, may suggest that serpentine plants show a competitive strategy based on the highest shoot vigor related to syllepsis (Champagnat 1950). The competition between the growth of the main shoots and developement of sylleptic axillary branches (extension of main shoot), as well as the brachiblast and floral stalk formation produces an inhibition of elongation of subsequent internodes (Borchert 1976; Remphrey & Powel 1985). This inhibition is more drastic in the serpentine populations than in the contact zone. Plants living in silica substrate present a more conservative strategy with a less pronounced growth rhythm, due to reduced competition between main shoot growth and syllepsis. Individual branches of single plants respond to local conditions (Stebbins 1963), allocating limiting resources to the organs located in better environmental conditions. As a result, we might expect high levels of within individual variation in developmental instability measurements. For example, branches in full light are larger and more branched than the corresponding branches in relative shade (Hallé et al. 1978; Jones & Harper 1987). In addition, light favors new branch formation (Thiébaut et al. 1985; Pickett & Kempf 1980), whereas shade favors branch death (Willson 1990). In spite of this, we observed that variation among measurements within individuals was lower than variation between individuals within soil type. Additionally, different traits are under different selection pressures and development, resulting in different responses to the changing environment. For example, coastal Teucrium lusitanicum Schreb populations exhibit a conservative strategy with high homeostatic vegetative structures and low stability in reproductive organs. Inland populations, by contrast, show a colonizing strategy, resulting in larger stability of reproductive organs that enhances the colonization of new territories (Alados et al. 1998). Indeed, we observed that the asymmetry of different characters varied independently among structures, providing no support for the existence of organism-wide developmental stability, as noted previously (Van Valen 1962; Evans & Marshall 1996; Leung & Forbes 1997). However developmental instability values for different traits are concordant at the population level, as observed by other authors (Evans & Marshall 1996; Sherry & Lord 1996a; Leary et al. 1985). Correlations between developmental instability of several traits were probably limited by morphological compensation and functional integration of traits. Floral instabilities were positively correlated with each 64 other at the level of petals and sepals, separately. At the vegetative structure level, only curve fitting estimators of the stem growth pattern equation were correlated with each other. Other authors also observed foliar instabilities correlated with each other (Sakai & Shimamoto 1965; Bagchi et al. 1989), although foliar instability was not significantly correlated with floral instabilities (Sakai & Shimamoto 1965; Evans & Marshall 1996). When fitness depends upon the phenotypic characteristic of a particular trait, organisms should buffer the effects of environmental disturbances on the development of those traits (Fowler & Whitlock 1994). Natural selection should act to minimize phenotypic variation in traits that are functionally important to an organism. However, the way this occurs will differ in animals and plants. Whilst animals may avoid environmental disturbances by evolved physiological and behavioral mechanisms, plants resist environmental perturbations by modifying their structures. In consequence, plants present highly variable structures that change in response to light interception, soil moisture or wind exposure. The difficulty involved in providing perfect symmetrical traits may be determined by the coefficient of variation of each developmental instability measurement. We observed that floral asymmetry presents a lower coefficient of variation (0.58 for standard deviation of petal length, and 0.55 for standard deviation of petal width) than leaf asymmetry (0.84), with curve fitting accuracy of the internode length equation showing a lower coefficient of variation (0.34 for Syx , or 0.29 for Sa ). Sherry & Lord (1996b) also demonstrated, in genus Clarkia Pursh. that floral asymmetry values were about half those of leaves. In summary we conclude that asymmetry and within-plant variance were higher in the contact zone populations than in either silica or serpentine populations, proving the adaptation of C. ladanifer to ultrabasic serpentine soils. Probably the competitive abilities of C. ladanifer inhabiting acidified soils with low Ca2+ concentrations allowed the plant to perform adequately in serpentine soils in spite of high Ni2+ concentrations. Acknowledgements We thank H. Freitas and C. Nabais for performing chemical analyses in soil and plant samples. We are grateful to J. Emlen, H. Freitas, S. Talavera and M. A. Quesada for their valuable suggestions and comments on earlier drafts of the manuscript. We also thank J. Toro for measuring plant samples. This study was partially supported by Ministerio de Educación y Cultura (PSPGC) into the cooperative program between Spain and Portugal, project HP96-42. Appendix 1 Appendix 1. Mean square (MS) measurement error and the proportion of within-measurement MS to the among-individual MS (variation due to trait size) and, to the within-trait MS (variation due to developmental instability, DI). Trait MS error Error as % of Trait size DI Half leaf width Petal length Petal width Sepal length 0.002 0.04 0.04 0.006 0.02% 0.08% 0.02% 0.1% 0.95% 0.5% 0.25% 0.32% References Alados, C. L., Escós, J. & Emlen, J. M. 1994. Scale asymmetry: a tool to detect developmental instability under the fractal geometry scope. Pp. 25–36 In: Novak, M. M. (ed.), Fractals in the Natural and Applied Sciences. Elsevier Science B.V., North-Holland. Alados, C. L., Navarro, T., Cabezudo, B., Emlen, J. M. & Freeman C. D. 1998. Developmental instability in gynodioecicus Teucrium lusitanicum. Evol. Ecol. 12: 21–34. Arianoutsou, M., Rundel, P. W. & Berry, W. L. 1993. Serpentine endemics as biological indicators of soil elemental concentrations. Pp. 179–189 In: Markert. B. (ed.), Plants as Biomonitors. VCH Weinheim, New York. Bagchi S., & Iyama, S. 1983. Radiation induced developmental instability in Arabidopsis thaliana. Theor. Appl. Genetic 65: 85–92. Bagchi S., Sharma, V. P. & Grupta, P. K. 1989. Developmental instability in leaves of Tectona grandis. Silvae Genetica 38: 1–6. Barrett, S. C. H., & Harder, L. D. 1992. Floral variation in Eichhornia paniculata (Spreng.) Solms (Pontederiaceae) II. Effects of development and environment on the formation of selfing flowers. J. Evol. Biol. 5: 83–107. Bell, G. 1985. On the function of flowers. Proc. Roy. Soc. London B224: 223–265. Bradshaw A. D. 1965. Evolutionary significance of phenotypic plasticity in plants. Adv. Genet. 31: 115–155. Borchert, R. 1976. Differences in shoot growth patterns between juvenile and adult trees and their interpretation based on systems analysis of trees. Acta Horticulturae 56: 123–130. Brooks, R. R. 1987. Serpentine and its vegetation. Dioscorides, Portland, Oregon. Cabezudo, B., Navarro, T., Pérez Latorre, A.V., Nieto-Caldera J. M. & Orshan, G. 1992. Estudios fenomorfológicos en la vegetación del sur de España. I. Cistus L. Acta Bot. Malacitana 16: 229–239. 65 Champagnat, P. 1950. Corrélations d ’inhibition chez les plantes. Année Biologique 26(12): 1–18. Clarke, G. M. 1993. The genetic basis of developmental stability. I. Relationships between stability, heterozygosity and genomic coadaptation. Genetica 89: 15–23. Clarke, G. M. 1995. Relationships between developmental stability and fitness: application for conservation biology. Cons. Biol. 9: 18–24. Comte, L. 1993. Rythmes de croissance et structures spatiales périodiques d’ arbres tropicaux. Example de cinq espèces de fôret Equatriale. Thése de Doctorat, Univ. Montpellier II. Montpellier 449 p. Cronk, Q. & Möller M. 1997. Genetics of floral symmetry revealed. Trends Ecol. Evol. 12: 85–86. Ellstrand, N. C. & Mitchell, R. J. 1988. Spatial and temporal patterns of floral inconstancy in plants and populations of Ipomopsis aggregata (Polemoniaceae). Bot. Gazette 149: 209–212. Escós, J., Alados, C. L. & Emlen, J.M. 1995. Fractal structures and fractal functions as disease indicators. Oikos 74: 310–314. Escós, J., Alados, C. L. & Emlen, J. M. 1997. Grazing impact on plant fractal architecture and fitness of a Mediterranean shrub (Anthyllis cytisoides L.). Funct. Ecol. 11: 66–78. Evans, A. S. & Marshall, M. 1996. Developmental instability in Brassica campestris (Cruciferae) – Fluctuating asymmetry of foliar and floral traits. J. Evol. Biol. 9: 717–736. Faegri, K & Pijl, Van der. 1971. The principles of pollination ecology. Pergamon Press, Headington Hill Hall, Oxford 291, p.736. Fowler, K., & Whitlock, M. C. 1994. Fluctuating asymmetry does not increase with moderate inbreeding in Drosophila melanogaster. Heredity 73: 373–376. Freeman, D. C., Graham, J. H. & Emlen, J. M. 1993. Developmental stability in plants: symmetries, stress and epigenesis. Genetica 89: 97–119. Freeman, D. C., Graham, J. H., Byrd, D. W., McArthur, E. D., & Turner, W. A. 1995. Narrow hybrid zone between two subspecies of big sagebrush, Artemisia tridentata (Asteraceae). III. Developmental instability. Am. J. Bot. 82: 1144–1152. Freitas, H. & Mooney, H. 1996. Effects of water stress and soil texture on the performance of two Bromus hordeaceus ecotypes from sandstone and serpentine soils. Acta Oecologica 17: 307–317. Graham, J. H. 1992. Genomic coadaptation and developmental stability in hybrid zones. Acta Zool. Fennica 192: 121–132. Graham, J. H., Freeman, D. C. & Emlen J. M. 1993. Developmental stability: A sensitive indicator of population under stress. Pp. 136–158. In: Landis, W.G., Hughes J. S. & Lewis, M. A. (eds.) Environmental Toxicology and Risk Assessment. ASTM STP 1179, American Society for Testing and Materials, Philadelphia. Grant, V. 1956. The genetic structure of races and species in Gilia. Adv. Genetic 8: 55–87. Grant, V. 1975. Genetics of flowering plants. Columbia University Press, New York. Hallé, F., Oldeman, R. A. & P. B. Tomlinson. 1978. Tropical trees and forests: an architectural analysis. Springer-Verlag, Berlin. Harvey, I. F. & Walsh, K. J. 1993. Fluctuating asymmetry and lifetime success are correlated in males of damselfly Coenagrion puella. Ecol. Entomol. 18: 198–202. Huether, C. A., Jr. 1969. Constancy of the pentamerous corolla phenotype in natural populations of Linanthus. Evolution 23: 572–588. Jones, M., & Harper, J. L. 1987. The influence of neighbours on the growth of trees. II. The fate of buds on long and short shoots in Betula pendula. Proc. Roy. Soc. London B232: 19–33. Kozlov, M. V., Wilsey, B. J., Koricheva, J. & Haukioja, E. 1996. Fluctuating asymmetry of birch leaves increases under pollution impact. J. Appl. Ecol. 33: 1489–1495. Leary, R. F., Allendorf, F.W. & Knudsen, K. L. 1985. Inheritance of meristic variation and the evolution of developmental stability in rainbow trout. Evolution 39: 308–314. Leary, R. F., Allendorf, F. W. & Knudsen, K. L. 1992. Genetic, environmental, and developmental causes of meristic variation in rainbow trout. Acta Zool. Fennica, 191: 79–95. Leung, B. & Forbes, M. R. 1997. Modelling fluctuating asymmetry in relation to stress and fitness. Oikos 78: 397–405. Levin, D. A. 1970. Developmental stability and evolution in peripheral isolates. Am. Nat. 104: 343–353. Ludwig, W. 1932. Das Rechts-Links Problem im Tierreich und beim Menschen. Springer-Verslag, Berlin, Germany. Markow, T. A. 1994. Contemporary issues in genetics and evolution. Developmental Instability: Its origins and evolutionary implications. Kluwer Academic Publishers, Dordrecht, The Netherlands. Markow, T. A. & Clarke, G. M. 1997. Meta-analysis of the heritability of developmental stability: a giant step backward. J. Evol. Biol. 10: 31–37. Marschner, H. 1986. Mineral nutrition of higher plants. Academic Press, London. Mather, K. 1953. Genetic control of stability in development. Herdity 7: 297–336. Maynard Smith, J., Burian, R., Kaufmann, S., Alberch, P., Campbell, J., Goodwin, B., Lande, R., Raup, D. & Wolpert, L. 1985. Developmental constraints and evolution. Quat. Rev. Biol 60: 265–287. McKenzie, J. A. 1994. Selection at the diazinon resistance locus in overwintering populations of Lucilia cuprina (the Australian sheep blowfly). Heredity 73: 57–64. McKenzie, J. A. & O’Farrell K., 1993. Modification of developmental instability and fitness: malathio-resistance in the Australian sheep blowfly, Lucilia cuprina. Genetica 89: 67–76. McKenzie, J. A., Whitten, M. J. & Adena, M. A. 1982. The effect of genetic background on the fitness of diazinon resistance genotypes of the Australian sheep blowfly, Lucilia cuprina. Heredity 49: 1–9. Meinhardt, H. 1984. Models of pattern formation and their application to plant development. Pp: 1–32. In: Barlow, P. W. & Carr, D. J. (eds) Positional controls in plant development. Cambridge University Press, Cambridge. Merilä, J., & Björklund, M. 1995. Fluctuating asymmetry and measurement error. Syst. Biol. 44: 97–101. Møller A. P., & Eriksson, M. 1994. Patterns of fluctuating asymmetry in flowers: Implications for sexual selection in plants. J. Evol. Biol. 7: 97–113. Møller A. P. 1995a. Bumblebee preference for symmetrical flowers. Proc. Nat. Acad. Sci. USA 92: 2288–2292. Møller A. P. 1995b. Leaf-mining insects and fluctuating asymmetry in elm Ulmus glabra leaves. J. Animal Ecol. 64: 697–707. Møller A. P. & Thornhill, R. 1997. A meta-analysis of the heritability of developmental stability. J. Evol. Biol. 10: 1–16. Muñoz Garmendia , F. & Navarro, C. 1993. Cistaceae. Pp. 319–436. In: Castroviejo, S, Aedo, C. Cirujano, S., Laínz, M., Montserrat, P., Morales, R., Muñoz Garmendia, F., Navarro, C., Paiva, J. & Soriano, C. (eds.). Flora Ibérica Vol. III. Real Jardín Botánico. C.S.I.C. Madrid. Ng, F. S. P. 1979. Growth rhythms in tropical juvenile trees. Bull. Soc. Botanique Fr. 126, Actulités botaniques 3: 139–149. 66 Palmer, A. R. & Strobeck, C. 1986. Fluctuating asymmetry: measurement, analysis, patterns. Annals Review Systematic 17: 391–421. Palmer, A. R. & Strobeck, C. 1992. Fluctuating asymmetry as a measure of developmental stability: Implications of non-normal distributions and power of statistical tests. Acta Zool. Fennica 191: 57–72. Parsons, P. A. 1992. Fluctuating asymmetry: a biological monitor of environmental and genomic stress. Heredity 68: 361–364. Paxman, G. J. 1956. Differentiation and stability in the development of Nicotiana rustica. Ann. Bot. 20: 331–347. Pickett, S. T. A. & Kempf, J. S. 1980. Branching patterns in forest shrubs and understory trees in relation to habitat. New Phytol. 86: 219–228. Polak, M. 1993. Parasites increase fluctuating asymmetry of male Drosophila nigrospiracula: implications for sexual selection. Genetica 89: 255–265. Proctor, J. 1971. The plant ecology of serpentine. II. Plant response to serpentine soils. J. Ecol. 59: 397–410. Proctor, J., & Woodell, S. R. J. 1975. The ecology of serpentine soils. Adv. Ecol. Res. 9: 256–347. Remphrey, W., & Powell, G. R. 1985. Crown architecture of Larix laricia saplings: shoot preformation and neoformation and their relationships to shoot vigour. Can. J. Bot. 62: 2181–2192. Rice, W. R. 1989. Analyzing tables of statistical tests. Evolution 43: 223–225. Roberts, B. A. & Proctor, J. 1992. The ecology of areas with serpentinized rocks a world view. Kluwer Academic Publishers, Dordrecht, The Netherlands. Sakai, K., & Shimamoto, Y. 1965. Developmental instability in leaves and flowers of Nicotiana tabacum. Genetics 51: 801–813. Schroeder, M. 1991. Fractals, chaos, power laws. Minutes from an infinite paradise. W.H. Freeman and Company, New York. Sherry, R. A. & Lord, E. M. 1996a. Developmental stability in leaves of Clarkia tembloriensis (Onagraceae) as related to population outcrossing rates and heterozygosity. Evolution, 50: 80–91. Sherry, R. A. & Lord, E. M. 1996b. Developmental stability in flowers of Clarkia tembloriensis (Onagraceae). J. Evol. Biol., 9: 911–930. Sokal, R. R. & Rohlf, F. J. 1981. Biometry, 2d ed., Freeman, San Francisco, CA. Soulé, M. E. 1982. Allomeric variation. 1. The theory and some consequences. Am. Nat. 120: 751–764. Stebbins, G. L., Jr. 1951. Natural selection annd the differentiation of angiosperm families. Evolution 5: 299–324. Stebbins, G. Y. J. 1963. Variation and evolution in Plants. Columbia Biological Series, 16, Columbia University Press, New York, 643 p. Summer, F. B. & Huestis, R. R. 1921. Bilateral asymmetry and its relation to certain problems of genetics. Genetics 6: 445–485. Swaddle, J. P. 1997. On the heritability of developmental stability. J. Evol. Biol. 10: 57–61. Talavera, S., Gibbs, P. E. & Herrera, J. 1993. Reproductive biology of Cistus ladanifer (Cistaceae). Plant Syst. Evol. 186: 123–134. Thiébaut, B., Cuguen, J. & Dupré, S. 1985. Architecture de jeunes hêtres, Fagus sylvatica. Can. J. Bot. 63: 2100–2110. Thoday, J. M. 1955. Balance, heterozygosity and developmental stability. Cold Spring Harbor Symposium on Quantitative Biology 22: 318–326. Thoday, J. M. 1958. Homeostasis in a selection experiment. Heredity 12, 401–415. Valentine, D. W., & Soulé, M. E. 1973. Effect of p,p0 -DDT on developmental stability of pectoral fin rays in the grunion, Leuresthes tenius. Fisheries Bull. 71: 920–921. Van Valen, L. 1962. A study of fluctuating asymmetry. Evolution 16: 125–142. Waddington, C. H. 1942. Canalization of development and the inheritance of acquired characters. Nature 150: 563–565. Waddington, C. H. 1957. The strategy of genes. Allen and Uniwin, London. Watson P. J. & Thornhill, R. 1994. Fluctuating asymmetry and sexual selection. Trends Ecol. Evol. 9: 21–25. Willson, M. F. 1990. Sexual selection in plants and animals. Trends Evol. Ecol. 5: 210–214. Zakharov, V. 1987. Animal asymmetry: populations phenogenetic approach. Nauka, Moscow.