Present address: Mediterranean Institute for Advanced Studies (IMEDEA, CSIC-UIB), C/Miguel Marquès 21, 07190 Esporles, Mallorca (Illes Balears), Spain.
Local adaptation of the pondweed Potamogeton pectinatus to contrasting substrate types mediated by changes in propagule provisioning
Article first published online: 21 NOV 2003
Journal of Ecology
Volume 91, Issue 6, pages 1081–1092, December 2003
How to Cite
Hangelbroek, H. H., Santamaría, L. and De Boer, T. (2003), Local adaptation of the pondweed Potamogeton pectinatus to contrasting substrate types mediated by changes in propagule provisioning. Journal of Ecology, 91: 1081–1092. doi: 10.1046/j.1365-2745.2003.00835.x
- Issue published online: 21 NOV 2003
- Article first published online: 21 NOV 2003
- Received 28 February 2003 revision accepted 31 July 2003
- fennel pondweed;
- nutrient availability;
- phenotypic plasticity;
- propagule size
- Top of page
- Materials and methods
- Supplementary material
- 1We studied local adaptation to substrate type within a population of the clonal aquatic macrophyte Potamogeton pectinatus and the role that genotypic variation in propagule-provisioning plays therein.
- 2P. pectinatus reproduces mainly by means of subterranean asexual propagules (tubers), whose survival and sprouting success depends on the interaction of factors such as the size of tubers, substrate type and predation risk by Bewick's swans.
- 3We studied a population of P. pectinatus in which genotypes producing large tubers predominate at the sandy shore and those producing small tubers at the clay-rich shore. Clonal lines originating from different shores were grown on a sandy and a clay-rich substrate in a common-garden. Plants from all clonal lines were grown from tubers of a comparable size range, but the various clones from within each shore differed in the average size of tubers they are genetically determined to produce.
- 4The performance of all clones was much lower on sandy substrate than on clay-rich substrate, indicating that the former is a stressful (nutrient-poor) environment. The reaction norms of morphological traits varied significantly among clones, revealing genetic variation in phenotypic plasticity. However, these differences were not related to our correlates of fitness (total tuber biomass, tuber size and tuber number). We found no evidence of local adaptation independent of genotypic tuber size. Instead, tuber size mediated local adaptation: clones producing larger tubers had a higher fitness in sandy substrate, while clones producing smaller tubers had a higher fitness in clay-rich substrate.
- 5Our results imply that diversifying selection for tuber size takes place between the two substrate types and confirms the importance of tuber-size provisioning for local adaptation to substrate heterogeneity.
- Top of page
- Materials and methods
- Supplementary material
Local adaptation of populations to specific environmental conditions is a well-known phenomenon (e.g. Van Tienderen & Van der Toorn 1991; Jordan 1992; Nagy & Rice 1997; McKay et al. 2001). In such populations, genotypes are genetically specialized to their local environment, where they show enhanced fitness. Fitness of these genotypes in non-native environments is, however, suboptimal. Specialization may be expected to occur when contrasting phenotypes show enhanced fitness in contrasting environments and intermediate static phenotypes cannot evolve due to developmental constraints (Van Tienderen 1991) or when intermediate phenotypes have lower fitness in both environments.
In heterogeneous environments, local adaptation to particular environmental conditions may also occur within populations, on a much smaller geographical scale. For this to happen contrasting selection pressures must be strong (Van Tienderen 1997) and gene flow must be low (Slatkin 1985), which is not common within populations. Nevertheless, local adaptation has been reported at small geographical scales in response to abiotic factors such as heavy metals (Bradshaw 1960; Jain & Bradshaw 1966), wind (Jain & Bradshaw 1966) and elevation (Galen et al. 1991), or to biotic factors such as interspecific competition (Prati & Schmid 2000) and herbivory (Sork et al. 1993). In heterogeneous environments selection may also lead to generalist genotypes that have high levels of phenotypic plasticity. Plastic genotypes may perform well in all the various environments yet be accompanied by costs of plasticity (Bradshaw 1965). Phenotypic plasticity does not necessarily preclude local adaptation; it is possible that local adaptation involves contrasting plastic responses (i.e. changes in the reaction norms) among genotypes from the populations or subpopulations growing in contrasting environments. Yet one must also bear in mind that phenotypic plasticity is not automatically adaptive. Non-adaptive plasticity may result from reduced growth under low-resource conditions (‘inevitable plasticity’, Sultan 1995), from random variation in traits that have no fitness effects (‘neutral plasticity’, Alpert & Simms 2002) or from correlation with selected traits. Last of all, plastic responses may also involve major reorganization of the relationships among traits, thus influencing the way in which phenotypic integration is maintained across environments (Schlichting 1986, 1989). Of particular importance are changes in the relative contribution of each trait to fitness from one environment to another (e.g. Schlichting 1989; Pigliucci et al. 1995).
Many aspects of substrate type may affect the environment as experienced by plants. These include organic content, redox potential, particle density and nutrient availability. In particular, the effect of nutrient availability on clonal plants has received increasing attention (De Kroon & Knops 1990; Schmid & Bazzaz 1992; De Kroon & Hutchings 1995; Arredondo & Johnson 1999; Dong & Alaten 1999; Fransen et al. 1999). Variation in response to substrate conditions in plants has been reported for, for instance, biomass allocation, shoot biomass, leaf : area ratio and total rhizome length (Lotz & Blom 1986; Idestam-Almquist & Kautsky 1995). If such traits increase fitness in particular environments and the trait shows heritable variation, local adaptation may eventually occur. We address whether the clonal pondweed Potamogeton pectinatus L. shows within-population local adaptation to contrasting substrate types. A sandy and a clay-rich shore within the same lake lay only 55–225 m apart and neutral markers showed no neutral genetic differentiation between plants originating from the two shores (Φct < 0.001, P = 0.73, Hangelbroek et al. 2002). This indicated un-restricted gene flow and the plants are therefore considered to be part of the same population.
Several studies on P. pectinatus have reported the existence of different ecotypes (Van Wijk et al. 1988; Vermaat & Hootsmans 1994), while others revealed high levels of phenotypic plasticity (Van Wijk 1988; Idestam-Almquist & Kautsky 1995; Pilon & Santamaría 2001) and yet others uncovered neutral genetic differentiation between populations (Mader et al. 1998; King et al. 2002) and high genetic variation within populations (Hangelbroek et al. 2002). These findings are in agreement with the concept that common plant species, such as P. pectinatus, possess both high phenotypic plasticity and large genetic variation (Bradshaw 1984; Bazzaz 1986). We consider whether static ecotypes, high phenotypic plasticity or a combination of both (i.e. ecotypes possessing distinct plastic responses) occur within a single population.
A previous study by Santamaría & Rodríguez-Gironés (2002) revealed a potential relationship between local adaptation and genotypic variation in propagule size. Santamaría & Rodríguez-Gironés (2002) showed that P. pectinatus clones from the sandy shore produced larger tubers (below-ground asexual propagules) than those from the clay-rich shore, following growth under standardized conditions. In a follow-up experiment by Hangelbroek, Santamaría and Rodríguez-Gironés (unpublished manuscript) this difference in tuber size production was shown to have a large genetic basis (i.e. a broad-sense heritability estimate of H2 = 1.01). Their experiment was based on 15 clonal lines grown for three generations under standardized, common-garden conditions. In the second and third year of cultivation, only tubers of a comparable size range were planted for each clone (i.e. no differences in initial tuber sizes existed between clones). In both years, the size of the tubers produced by the plants at the end of the growth season differed significantly between clones, even though they were grown from tubers of comparable sizes. The slope of the regression between the produced tuber size of the second and third generation was interpreted as a measure of trans-generational trait repeatability and under the assumption of negligible environmental effects, as a measure of broad-sense heritability (Dohm 2002).
Santamaría & Rodríguez-Gironés (2002) argued that this genotypic difference in propagule provisioning may be the result of two separate factors. First, higher foraging pressure by Bewick's swans on tubers in the sandy sites may lead to stronger selection for deeply buried tubers that have a higher probability of escaping foraging. Secondly, deeper burial depths and higher sprout mortality in sandy sites promote larger tuber sizes, because sprout survival increases with size and decreases with depth (see also Spencer 1987). However, the potential costs of propagule provisioning, or benefits of local adaptation to substrate type during the growth phase were not addressed by Santamaría & Rodríguez-Gironés (2002). In sandy and clay-rich areas, tuber provisioning is likely to involve different energetic and functional costs as a result of differential nutrient availability. Under nutrient limitation, plants are likely to experience two contrasting effects On the one hand, a decreased investment in photosynthetic tissue and its enzymatic machinery (which are typically costly in terms of nutrients) will result in a decreased supply of carbohydrates to newly growing tubers (i.e. an increased cost of tuber biomass production, e.g. Saulnier & Reekie 1995). On the other hand, the decreased demand of carbohydrates for growth (as the latter is nutrient-limited) will result in an increased allocation to carbohydrate storage (i.e. a decreased cost of tuber biomass production). The relative importance of the two effects will be modulated by the number of tubers produced per plant, which may in turn be limited by plant size (i.e. by the numbers of apical meristems in the rhizome). Different costs of propagule provisioning in sandy vs. clay-rich substrate may interfere with the selection pressure favouring large-tuber-producing clones in sandy substrate postulated by Santamaría & Rodríguez-Gironés (2002), thus constraining or promoting the effects of diversifying selection for tuber size between the two substrate types.
We used an experimental set-up aimed at dissecting the relative contribution of propagule provisioning as opposed to other morphological, biomass and allocation traits, to the response to substrate type of P. pectinatus genotypes from the sandy and clay-rich subpopulations. The following questions were specifically addressed: (a) How do vegetative and (asexual) reproductive traits respond to the different substrate types? Do trait relationships (phenotypic integration) differ between substrate types, revealing changes in the determinants of fitness? (b) Has local adaptation to substrate type taken place within this population? If so, is it related to adaptive static differences or to differences in plastic responses (i.e. reaction norms) of the traits analysed? (c) Is local adaptation to substrate type mediated by genotypic variation in propagule provisioning? If so, is it consistent with the patterns reported from the field population, i.e. do plants making larger tubers perform better in the sandy sites?
Materials and methods
- Top of page
- Materials and methods
- Supplementary material
species and study system
Potamogeton pectinatus (Potamogetonaceae) is a clonal submerged angiosperm with a wide geographical distribution, ranging from the subtropics to the subarctic (Casper & Krausch 1980; Wiegleb & Kaplan 1998). In temperate regions it has a pseudo-annual life cycle, i.e. plants senesce every autumn, surviving exclusively by means of asexual propagules (tubers) formed at the apex of the underground rhizomes (van Wijk 1988). No dormant tuber bank is formed; all tubers that survive the winter start a new life cycle in spring. Seed production also takes place but local recruitment is low; hence, seeds are generally thought to contribute to population re-establishment following disturbances or to long distance dispersal by waterfowl (Van Wijk 1989a). However, Hangelbroek et al. (2002) detected high clonal diversity within the population studied here (number of genets/number of ramets = 0.76), suggesting that seedling recruitment may be sufficient to maintain high levels of genotypic diversity.
The studied population is situated in the Babbelaar, a former river branch of the Lauwerszee estuary in the Netherlands that became part of Lake Lauwersmeer following its closure in 1969 (Fig. 1). A deep-water gully, approximately 55–225 m wide, separates the population into two non-connected beds of P. pectinatus, which occupy shores of contrasting substrate type (Nolet et al. 2001). The depth of the water gully prevents plants from growing across the gully; however, water-mediated dispersal of seeds, dislodged tubers or other plant fragments may take place. Every autumn, Bewick's swans forage on P. pectinatus upon arrival from their migratory flight from the tundra, before turning to other available food sources (Beekman et al. 1991). Swans consume on average 39% of the tuber bank available in autumn, and show preferential consumption and lower giving-up thresholds (thus resulting in increased tuber predation) in sandy than in clay-rich substrate (Nolet et al. 2001).
sampling and cultivation of clonal lines
We used a selection of clonal lines obtained as described in Santamaría & Rodríguez-Gironés (2002): in April 1997 tubers were collected from the two subpopulations occupying the two shores of the study population (Fig. 1). Within each subpopulation sampling took place in either two or three sites approximately 200 m apart. Sites number one and two, located on the clay-rich shore, contained, respectively, 36 and 17% clay (i.e. percentage of substrate particles < 63 µm, Santamaría & Rodríguez-Gironés 2002). Sites three, four and five, located on the sandy shore, contained only 8–9% clay (Santamaría & Rodríguez-Gironés 2002). At each site, 18 tubers were collected from nine random sampling points chosen on a 24-point 1 m × 1 m grid (Fig. 2). At each sampling point, the largest and smallest tuber present in a standard sample of substrate (12 cores of 7 cm ∅ and 30 cm length, making a total volume of 13.8 L) were selected for cultivation. The tubers were kept at 4 °C to continue their hibernation period until the beginning of spring in May 1997.
To obtain clonal replicates of the tubers and to minimize the influence of potential carry-over effects from the different maternal environments, the tubers collected from the field were then grown in outdoor, common-garden conditions for a complete season (May–October 1997). Each tuber was grown in a 5.5-L pot containing a mixture of commercial potting clay and river sand (1 : 3 dry weight) placed in 1 m3 tanks filled with tap water. The mixture ratio of clay and sand was chosen because it had proven to be a successful mixture for clonal propagation of P. pectinatus clones originating from diverse environments (Pilon & Santamaría 2001) and was intermediate between the two substrate-types used later. The tubers produced were then harvested and individually weighed (fresh weight, fw), before hibernation at 4 °C until May 1998. In order to preserve the tuber stock for various experiments, tuber dry weights (dw) were estimated from fresh- to dry-weight regressions fitted on a subsample of the harvested tubers (dw = 0.34 × fw, R2 = 0.95, n = 60).
To study the effect of genetically based differences in tuber size (hereafter referred to as genotypic-tuber-size), clones that had produced different average tuber sizes under common garden conditions were selected. Clones from both subpopulations were assigned to one of three genotypic-tuber-size classes according to whether they had produced small (average size 6–12 mg dw), medium (16–19 mg dw) or large (23–29 mg dw) tubers. To be able to distinguish between potential effects related to the substrate of origin of the clones and their genotypic-tuber-size, we selected clones for the three different genotypic-tuber-sizes within each subpopulation (one clone per genotypic-tuber-size per sampling site, making a total of 3 × (2 + 3 sites) = 15 clones for the whole experiment; Figs 1 and 2). This means that the experimental design does not reflect the actual frequencies of genotypic-tuber-size within each subpopulation, but a factorial combination of origin and genotypic-tuber-size.
Amplified fragment length polymorphism (AFLP) analysis was carried out to test whether lines did indeed represent 15 different clones, according to Vos et al. (1995). All clones were distinguished from one another with the usage of the primer combination: EcoRI + ACC/MseI + CTT.
The size of the planted tubers (initial tuber size, n = 16 per clone) was standardized to a comparable size range for all clones (average ± SE = 15 ± 10 mg dw), to minimize the influence of (non-genetic) phenotypic maternal effects. The initial tuber sizes within each clonal line were selected as evenly as possible across the range to have comparable sizes and variation in sizes used between clones, despite their different genotypic-tuber-sizes (see insets Fig. 2) this was possible because of the relative abundance of small tubers within all clonal lines (as tuber size distribution is right-skewed) and the large tuber stocks available for the experiment. We had tested whether initial tuber size indeed varied comparably between the different genotypic-tuber-size classes by conducting an ancova designed as for all other traits (as below) with initial tuber size as the dependent variable. Initial tuber size did not vary significantly between genotypic-tuber-size classes, subpopulations or treatments, or any of their two-way interactions, although it was significantly affected by the random factor ‘clone’. Hence, to further account for the potential effects of this remaining variation in initial tuber size (within the standardized range reported above), we included it as a covariate in the statistical analyses.
At the end of May, the selected tubers were pre-sprouted in trays filled with sand and placed in an outdoor tank filled with local ground water. After a week, eight sprouted tubers per clone per substrate treatment were randomly selected and transferred to 5.5-L pots containing a substrate mixture with either a high or a low clay content (making a total of 2 substrate treatments × 15 clones × 8 replicates = 240 pots for the complete experiment). The high-clay treatment was achieved by mixing commercial potting clay and washed aquarium sand in a dry-weight ratio of clay : sand equals 1 : 2, resulting in 36% clay particles as indicated by Malvern analysis, and the low clay by a 1 : 20 mixture (7% clay particles). Hereafter these treatments are referred to as ‘clay-rich’ and ‘sandy’. Laboratory analysis confirmed that the clay-rich mixture had a much higher nutrient content (1.6-fold more P, 4.1-fold more N and 4.6-fold more K; Table 1). Carbon was not included in the nutrient analysis as C-uptake is from the water-column by above-ground plant parts (Van Wijk 1989b). The substrate mixture in the pots was covered with a 2-cm layer of washed sand to minimize leakage of nutrients into the water-column and the resulting algal growth. The size of the pots was large enough to prevent nutrient limitation in the clay-rich substrate (Rodríguez-Gironés et al. 2003) and did not play a role in the sandy substrate treatment, as only a small fraction of the pot surface was occupied by the plants (as a result of nutrient-limitation). The pots were distributed over 12 tanks (0.9 × 1.1 m2, water depth 0.55 m) filled with local groundwater and situated in an outdoor common-garden facility at Heteren (the Netherlands). Four of these tanks were randomly assigned to each of the three genotypic-tuber-size classes. Within each genotypic-tuber-size class, plants from the different clones (five clones per tank, belonging to two subpopulations) were randomly assigned to the two different substrate treatments and distributed over the corresponding four tanks (20 pots per tank), i.e. following a split-block design with two fixed factors randomized within a random factor. Both the 12 tanks and the various clone × treatment combinations within tanks were randomly interspersed to avoid position effects. Water was added whenever necessary, and algal growth was controlled by adding waterfleas (Daphnia) at the onset of the experiment.
|Clay-rich treatment||Sandy treatment|
|Dry weight ratio: clay : sand||1 : 2||1 : 20|
|Soluble P ppm†||8.56||5.28|
Plants were harvested at the end of the growth season (beginning of October) to ensure that full potential asexual reproduction had taken place, but early enough to recover all vegetative plant material (shoots, rhizomes and roots). A single plant emerges from each tuber producing a single branching rhizome along which multiple shoots are produced. After measuring several morphological traits (the number of nodes and total length of both the rhizome and the longest shoot), plants were separated into above-ground, below-ground and tuber fractions for biomass determination (dry weight, after 48 h at 70 °C). In large rhizomes, morphological variables were estimated on a subsample, and total length recalculated using a regression of length vs. dry weight. Tubers were counted and their individual (fresh) weights measured. Tuber dry weights were estimated from fresh- to dry-weight regressions based on a subsample of tubers and carried out for each substrate treatment separately (sandy substrate: dw = 0.30 × fw, R2 = 0.94, n = 210; clay-rich substrate: dw = 0.35 × fw, R2 = 0.95, n = 210). Reproductive allocation was estimated as the ratio between tuber and total biomass (in dw) and rhizome thickness as rhizome mass per cm (mg dw cm−1).
As yearly recruitment depends almost exclusively on tuber production, total tuber biomass (as dry weight plant−1) was used as the main fitness surrogate. In addition, we also considered both tuber number and tuber size, as both affect different components of fitness (i.e. the number of asexual propagules vs. their potential for survival and growth) but often are negatively correlated (i.e. there is a size–number trade-off, e.g. Santamaría & Rodríguez-Gironés 2002).
All variables were analysed by means of mixed-models ancovas using the General Linear Models module of Statistica 5.5 (Statsoft 1999). The experimental unit was a single pot (i.e. plant). Sites were pooled within subpopulations, resulting in six clones from the clay-rich subpopulation and nine from the sandy subpopulation. The model included genotypic-tuber-size, subpopulation and substrate treatment as fixed factors and clone and tank as random factors. The random factor tank was nested within genotypic-tuber-size; hence, the effect of the latter was estimated from an error term equivalent to the tank × genotypic-tuber-size interaction (i.e. 1 tank = 1 replicate). Subpopulation and substrate treatment were nested within random factor tank, i.e. this part of the design is equivalent to a split-block anova (Steel & Torrie 1981). Random factor clone was nested within the interaction between subpopulation and genotypic-tuber-size, and initial tuber size was included as a covariate. Note that for simplicity, F-ratios for random factor tank are not shown in Table 2 as they do not result in interpretable tests of hypotheses. All variables were transformed (square root, arcsin√ or log 10(x + 1)) to assure homoscedasticity and normality of residuals. Individual tuber weights were log 10(x + 1) transformed before averaging within each pot, as the original data were strongly right skewed (see Table 2). The three-way interaction between subpopulation, genotypic-tuber-size and substrate treatment was not significant for any trait and was therefore left out of the analyses.
|Substrate treatment (ST)||Sub- population (SP)||Genotypic tuber size (GTS)||Clone (C)||ST × SP||ST × GTS||ST × C||SP × GTS||Covariate: Initial tuber size|
|Initial tuber size||4.22||3.86||3.86||5.31**||0.83||0.75||0.81||0.17|
|Total tuber biomass (mg dw)*||1173***||7.30*||0.79||1.33||0.27||10.99**||1.11||0.36||16.80***|
|Average tuber size (mg dw)‡||544***||0.01||4.79*||8.44***||0.73||13.89**||1.22||1.69||14.30***|
|Biomass yield and allocation|
|Vegetative biomass (mg dw)*||329***||1.45||0.17||0.98||0.72||4.39*||4.16***||0.92||11.71***|
|Shoot to root ratio (in dw)*||320***||2.79||1.43||1.46||3.21||0.26||2.09*||0.75||2.77|
|Asexual reproductive allocation (%, dw)§||139***||0.87||1.51||3.26*||2.86||0.26||2.92**||0.84||0.42|
|Rhizome length (cm)*||338***||0.22||0.10||1.74||0.18||4.41*||2.59**||0.16||11.97***|
|Rhizome internode length (cm)*||2.97||6.11*||1.00||0.81||1.93||2.17||4.20***||1.60||2.12|
|Rhizome thickness (dw/cm)*||46.95***||< 0.01||0.04||4.07*||0.37||1.29||2.20*||< 0.01||4.04*|
|Shoot length (cm)*||22.52***||1.32||0.66||1.05||0.26||1.46||3.06***||0.07||9.15**|
|Shoot internode length (cm)*||0.86||1.73||1.80||1.08||< 0.01||0.55||4.09***||0.51||0.16|
Phenotypic plasticity of those traits for which substrate × genotypic-tuber-size interactions were significant were subsequently analysed using one-way anovas. For this purpose, the average phenotypic plasticity of each trait was calculated for each clone according to Cheplick (1995):
where X̄ is the clone's average in either the clay-rich or sandy treatment. PPC is thus the percentage change from the clay-rich treatment to the sandy treatment. In the anovas, genotypic-tuber-size was entered a fixed factor.
Phenotypic integration (Schlichting 1986; Pigliucci & Marlow 2001; Relyea 2001) between nine phenotypic traits was measured separately for the two substrate treatments. Pearson correlations between pairs of traits were calculated, using the average values of each clone (Statistica 5.5 1999). All traits were transformed (as above) to assure normality of residuals and linearity of relationships. The relationships were visualized in correlation networks, where changes in the pattern of the trait integration between treatments indicate differential relationships between traits at different environments (Schlichting 1986). In addition, the correlation networks were used to identify changes in the relationships between fitness and non-fitness traits at different environments.
- Top of page
- Materials and methods
- Supplementary material
The substrate treatment had highly significant effects on all measured traits except the internode lengths of shoots and rhizomes (Tables 2 and 3). All fitness-related traits (total tuber biomass, tuber number and tuber size), as well as vegetative biomass, reproductive allocation and shoot to root ratio, were significantly lower in the sandy treatment (Tables 2 and 3). Plants growing in sandy substrate had significantly shorter and thicker rhizomes, and shorter shoots, than those growing in clay-rich substrate (Tables 2 and 3). The two original subpopulations differed only in a few traits: total tuber biomass and rhizome internode length were significantly larger for clones from the clay-rich than from the sandy subpopulation (Table 2). The interaction of substrate and subpopulation had no significant effects on any of the traits measured (Table 2).
|Clay-rich substrate||Sandy substrate|
|Total tuber biomass (mg dw)||890.6 ± 110.4||121.7 ± 50.4|
|Tuber number||61.4 ± 9.1||14.3 ± 2.9|
|Average tuber size (mg dw)*||11.0 ± 2.1||6.4 ± 1.7|
|Biomass yield and allocation|
|Vegetative biomass (mg dw)||590.3 ± 120.5||131.9 ± 42.5|
|Shoot to root ratio (in dw)||2.5 ± 0.5||1.0 ± 0.2|
|Asexual reproductive allocation (%, dw)||60.2 ± 4.8||46.5 ± 5.3|
|Rhizome length (cm)||338.6 ± 74.8||83.5 ± 22.0|
|Rhizome internode length (cm)||3.1 ± 0.5||2.9 ± 0.6|
|Rhizome thickness (mg dw/cm)||0.5 ± 0.1||0.8 ± 0.2|
|Shoot length (cm)||16.5 ± 3.9||11.5 ± 2.9|
|Shoot internode length (cm)||1.6 ± 0.3||1.5 ± 0.4|
Genotypic-tuber-size had a significant effect on produced tuber size (Table 2). Significant increases in tuber size occurred from small through medium to large genotypic-tuber-size classes (P < 0.05, Tukey post hoc tests), confirming that tuber size had a genetic component. The interaction between substrate and genotypic-tuber-size was significant for all fitness-related traits (total tuber biomass, tuber number and tuber size), vegetative biomass and rhizome length (Table 2, Fig. 3). Total tuber biomass was comparable for all genotypic-tuber-size classes in the clay-rich treatment; however, in the sandy treatment the clones from the small class had a significantly lower total tuber biomass (P < 0.05, Tukey post hoc tests, Fig. 3a). Tuber number showed the opposite trend: it was comparable for all genotypic-tuber-size classes in the sandy treatment, while it was lower for the large than for the small size class in the clay-rich treatment (P < 0.05, Tukey post hoc tests, Fig. 3b). Tuber size varied significantly among all three genotypic-tuber-size classes in the sandy treatment, while in the clay-rich treatment it was comparable for the medium and large classes and smaller for the small class (P < 0.05, Tukey post hoc tests, Fig. 3c). Vegetative biomass was larger for the small and medium than for the large genotypic-tuber-size class in clay-rich substrate, but in the sandy substrate it decreased significantly from the largest to the smallest genotypic-tuber-size class (P < 0.05, Tukey post hoc tests, Fig. 3d). Rhizome length showed a comparable pattern, although the small and medium genotypic-tuber-size also differed significantly in the clay-rich treatment. In general, when interactions between substrate and genotypic-tuber-size occurred, plasticity decreased with increasing genotypic-tuber-size class (Fig. 3; one-way anova and Tukey post hoc tests on phenotypic plasticity, Table 4).
|d.f., Error d.f.||Genotypic-tuber-size (GTS) 2, 12|
|Phenotypic plasticity of:|
|Total tuber biomass||6.45*|
|Average tuber size||14.28***|
The random factor clone significantly affected tuber size, reproductive allocation and rhizome thickness (Table 2). The interaction between clone and substrate was significant for all morphological, biomass yield and allocation traits, but it was not significant for fitness-related traits (total tuber biomass, tuber number and tuber size, Table 2). Initial tuber size had significant effects on nearly all traits; only shoot to root ratio, asexual reproductive allocation, and rhizome and shoot internode lengths, were not affected (Table 2).
Phenotypic integration varied between substrate treatments (Fig. 4). Overall, the number of significant relationships was larger for the sandy than for the clay-rich treatment. Significant correlations of total tuber biomass with tuber number, vegetative biomass and reproductive allocation, of vegetative biomass with tuber size, tuber number and rhizome thickness, and of reproductive allocation with rhizome thickness and shoot to root ratio, occurred in sandy but not in clay-rich substrate. Significant correlations of tuber number with tuber size and reproductive allocation, and of vegetative biomass with reproductive allocation were, however, found in clay-rich but not in sandy substrate. Only two correlations, namely that of tuber size with total tuber biomass and with reproductive allocation, were significant in both treatments. Internode lengths of rhizomes and shoots were not correlated with any other traits in either treatment.
- Top of page
- Materials and methods
- Supplementary material
plastic responses to substrate type
Substrate composition had major effects on all clones, regardless of the subpopulation they originated from or the genotypic-tuber-size class they belonged to. Biomass yield and fitness-related traits (total tuber biomass, number and size) were much lower in sandy than in clay-rich substrate, indicating that the sandy substrate can be considered as stressful for all clones. Tuber number varied more between substrate treatments than did tuber size, supporting the view that selection acts on propagule size rather than on number, and thus that size is more stable than number (Smith & Fretwell 1974; Lloyd 1987; Vaughton & Ramsey 1998).
Substrate types reflect different nutrient levels, and, on this basis, responses were comparable with those of terrestrial plants: above- and below-ground biomass, shoot length and reproductive allocation were all larger in nutrient rich (clay) conditions (Lotz & Blom 1986; Landhäusser et al. 1996; Mabry et al. 1997), probably resulting from inevitable rather than adaptive plasticity (Sultan 1995). Rhizome internode length did not differ between substrate types, similar to most reports for terrestrial rhizomatous plants (Dong & de Kroon 1994; Dong et al. 1996). Of the only two traits with plastic responses that could be interpreted as adaptive, increased allocation to roots in sandy substrate can be taken to indicate an increased investment in nutrient as opposed to light capture, with likely fitness benefits in the respective environments, whereas increased rhizome thickness (in contrast to decreased stolon thickness in terrestrial stoloniforous plants growing in nutrient-poor substrate, Price & Hutchings 1992) may reflect increased storage when C is in excess due to N and P deficiency. Indeed, vegetative biomass and rhizome thickness were positively correlated in the sandy but not in the clay-rich substrate, which indicates that biomass accumulation under nutrient limitation primarily involves carbohydrate storage in the rhizome. This also supports the view that rhizomes, in contrast to stolons, are more likely to serve as storage organs than as foraging devises (Dong & de Kroon 1994; Dong et al. 1996; Dong & Aalten 1999).
Phenotypic integration of traits was higher in sandy than in clay-rich substrate, most likely as a result of greater size dependence of traits in nutrient-limiting conditions. This agrees with the idea that stressful environments may promote enhanced phenotypic integration (Schlichting 1986). Phenotypic integration also varied qualitatively among treatments. Positive correlations between total tuber biomass (our surrogate of plant fitness) and vegetative biomass, reproductive allocation and tuber number in sandy substrate were absent in clay-rich substrate, where total tuber biomass correlated exclusively with tuber size. Phenotypic integration networks thus indicate an increased dependency of total tuber biomass on plant biomass and reproductive allocation in sandy substrate. The relationship between total tuber biomass and tuber number in sandy substrate may be attributed to meristem limitation, where reduced plant size results in shorter rhizomes with few apical tips available for tuber formation (as compared with larger plants with longer, well-branched rhizomes in clay-rich substrate).
local adaptation independent of propagule provisioning
We did not find any evidence of local adaptation through either static traits or plastic responses. First of all, non-significant substrate × subpopulation interactions for (asexual) fitness traits indicate a lack of home-versus-away differences that could be interpreted as local adaptation. The higher total tuber biomass of clones from the clay-rich subpopulation implies that it has a higher (asexual) fitness in both substrate types, a difference that (given our experimental protocol) most likely arises from genotypic differences between subpopulations. Although the grand-maternal nutrient environment might still have an effect on the performance of the clones, e.g. as in Wulff et al. (1999), the lack of subpopulation differences in total tuber biomass reported by Santamaría & Rodríguez-Gironés (2002) in the first generation after collection suggests that such effects are unlikely to account for the differences in total tuber biomass described here.
Secondly, we did not find any significant trait differences between subpopulations that can be interpreted as static traits of adaptive value in their local environments. The only significant difference, i.e. longer rhizome internodes in the clay-rich subpopulation, is unlikely to be of adaptive value as: (i) it does not result in local-versus-away differences in fitness traits (see above); (ii) network diagrams did not reveal any relationship between internode length and fitness traits (neither positive in clay-rich substrate, nor negative in sandy substrate); and (iii) previous studies reporting significant plastic variation in rhizome internode length as a response to substrate type (e.g. Dong et al. 1997) indicate that internodes become longer in resource-poor environments to facilitate foraging for nutrients (i.e. the opposite pattern to that found here).
Thirdly, the subpopulations did not respond differently in biomass yield and allocation or morphology to the substrate treatments (i.e. we found no significant substrate × subpopulation effects for these traits measured), indicating that they are not locally adapted through differential phenotypic plasticity to local substrate type. Local adaptation was not constrained by lack of genotypic variation onto which selection could act: nearly all morphological, biomass yield and allocation responses to substrate varied significantly among clones (significant clone × substrate interactions), indicating the existence of genotypic variation in phenotypic plasticity to substrate type within both subpopulations (similar to the responses of terrestrial plant species to a variety of ecological factors, e.g. Cheplick 1995; Skálováet al. 1997; Prati & Schmid 2000). However, these clonal differences were not accompanied by corresponding differences in fitness (non-significant clone × substrate interaction). This may indicate that variation in plasticity to substrate type is essentially neutral or that comparable fitness is achieved through varying combinations of plasticity in different traits (e.g. Sultan & Bazzaz 1993).
local adaptation through propagule provisioning
Our results indicate that, in the population under study, local adaptation to substrate type is mediated by genetically based changes in propagule provisioning. The effects of genotypic-tuber-size on fitness-related traits (total tuber biomass, tuber number and tuber size) depended highly on substrate type and were consistent with the differences in genotypic-tuber-size observed between subpopulations by Santamaría & Rodríguez-Gironés (2002), i.e. we found an increased fitness of large genotypic-tuber-size clones in sandy substrate, whereas small genotypic-tuber-size clones had an increased fitness in clay-rich substrate. Indeed, genotypic-tuber-size affected size-number allocation but not total tuber biomass in clay-rich substrate, while sandy substrate clones that produce larger tubers showed enhanced total tuber biomass without a detectable trade-off in terms of tuber number. These results are in contradiction with the expectation of increased costs of propagule provisioning in nutrient-poor conditions (Saulnier & Reekie 1995), which would result in decreased fitness of large genotypic-tuber-size clones in sandy substrate. Instead, the positive correlation between genotypic-tuber-size and total tuber biomass might be a consequence of the stimulating effect that a larger sink of C has on photosynthesis (Sweet & Wareing 1966; Herold 1980). This possibility is fully consistent with the meristem limitation in sandy substrate hypothesized above.
Our results are also consistent with the specialization hypothesis of Lortie & Aarssen (1996), which proposed that clones specialized to stressful environments show less plasticity in fitness traits than both generalists and genotypes specialized to non-stressful environments. In this case, sandy-substrate specialists with large genotypic-tuber-sizes were less plastic in biomass yield and fitness-related traits, while clay-rich substrate specialists with small genotypic-tuber-sizes were more plastic.
The higher fitness of clones with large genotypic-tuber-sizes in sandy substrate may reinforce the selection pressure that favours large tubers in sandy sites, which results from the higher sprouting survival and reduced predation risk of deeply buried, large tubers (Santamaría & Rodríguez-Gironés 2002). In clay-rich substrate, on the other hand, predation risk is low and sprouting survival high; hence, the production of small, abundant tubers is optimal (Santamaría & Rodríguez-Gironés 2002) and selection pressure should favour clones with small genotypic-tuber-size. Our results thus indicate the presence of diversifying selection on tuber size, linked to substrate heterogeneity in our field population.
- Top of page
- Materials and methods
- Supplementary material
This study revealed that local adaptation to substrate type within the studied population of P. pectinatus was mediated by genetically determined differences in propagule provisioning. Our results show that clones producing larger tubers had a higher fitness in sandy substrate, while clones producing smaller tubers had a higher fitness in clay-rich substrate. This is consistent with the genotypic-tuber-size frequencies found in the field, where clones that produce large tubers predominate in the sandy shore while clones that produce small tubers predominate in the clay-rich shore. In contrast, local adaptation independent of genotypic-tuber-size did not occur through either static traits or differential plastic responses. Our results suggest that propagule provisioning is the only trait that has contrasting fitness effects on plants growing in different substrate types, reinforcing previous indications of the importance of tuber-size provisioning for adaptation to substrate heterogeneity (Santamaría & Rodríguez-Gironés 2002).
- Top of page
- Materials and methods
- Supplementary material
We would like to thank T. Dekkers, K. Swart and H. de Jong for their technical assistance. Furthermore, we would like to thank J. van Groenendael, N.J. Ouborg and three anonymous reviewers for their critical comments. This is publication 3220 of the Netherlands Institute of Ecology (NIOO-KNAW).
- Top of page
- Materials and methods
- Supplementary material
The following material is available from http://www.blackwellpublishing.com/products/journals/suppmat/JEC/JEC835/JEC835sm.htm
Appendix S1 Detailed results of the nested analyses of covariance.
- Top of page
- Materials and methods
- Supplementary material
- 2002) The relative advantages of plasticity and fixity in different environments: when is it good for a plant to adjust? Evolutionary Ecology, 16, 285–297. & (
- 1999) Root architecture and biomass allocation of three range grasses in response to nonuniform supply of nutrients and shoot defoliation. New Phytologist, 143, 373–385. & (
- 1986) Life history of colonizing plants: some demographic, genetic and physiological features. Ecology of Biological Invasions of North America and Hawaii (eds H.A.Mooney & J.A.Drake), pp. 96–110. Springer, Berlin. (
- 1991) Bewick's Swans Cygnus columbianus bewickii utilising the changing resource of Potamogeton pectinatus during autumn in the Netherlands. Wildfowl Suppl. 1, 238–248. , & (
- 1960) Population differentiation in Agrostis tenuis Sibth. populations in varied environments. New Phytologist, 59, 92–103. (
- 1965) Evolutionary significance of phenotypic plasticity in plants. Advances in Genetics, 13, 115–155. (
- 1984) Ecological significance of genetic variation between populations. Perspectives on Plant Population Ecology (eds R.Dirzo & J.Sarukhán), pp. 213–228. Sinauer Associates, Sunderland. (
- 1980) S∫ßwasserflora Von Mitteleuropa. Bd. 23. Fischer, Stuttgart. & (
- 1995) Genotypic variation and plasticity of clonal growth in relation to nutrient availability in Amphibromus scabrivalvis. Journal of Ecology, 83, 459–468. (
- 1995) Morphological plasticity in clonal plants – the foraging concept reconsidered. Journal of Ecology, 83, 143–152. & (
- 1990) Habitat exploration through morphological plasticity in two chalk grassland perennials. Oikos, 59, 39–49. & (
- 2002) Repeatability estimates do not always set an upper limit to heritability. Functional Ecology, 16, 273–280. (
- 1999) Clonal plasticity in response to rhizome severing and heterogeneous resource supply in the rhizomatous grass Psammochloa villosa in an inner mongolian dune, China. Plant Ecology, 141, 53–58. & (
- 1994) Plasticity in morphology and biomass allocation in Cynodon dactylon, a grass species forming stolons and rhizomes. Oikos, 70, 99–106. & (
- 1996) Morphological responses to nutrient availability in four clonal herbs. Vegetatio, 123, 183–192. , & (
- 1997) Clonal plasticity in response to nutrient availability in the pseudoannual herb, Trientalis europaea L. Plant Ecology, 131, 233–239. , & (
- 1999) Root morphological and physiological plasticity of perennial grass species and the exploitation of spatial and temporal heterogeneous nutrient patches. Plant and Soil, 211, 179–189. , & (
- 1991) Ecotypic divergence in alpine Polemonium viscosum– genetic structure, quantitative variation, and local adaptation. Evolution, 45, 1218–1228. , & (
- 2002) Clonal diversity and structure within a population of the pondweed Potamogeton pectinatus foraged by Bewick's swans. Molecular Ecology, 11, 2137–2150. , , & (
- 1980) Regulation of photosynthesis by sink activity – the missing link. New Phytologist, 86, 131–144. (
- 1995) Plastic responses in morphology of Potamogeton pectinatus L. to sediment and above-sediment conditions at two sites in the northern Baltic proper. Aquatic Botany, 52, 205–216. & (
- 1966) Evolution in closely adjacent plant populations. I. The evidence and its theoretical analysis. Heredity, 21, 407–441. & (
- 1992) Path analysis of local adaptation in two ecotypes of the annual plant Diodia teres Walt. (Rubiaceae). American Naturalist, 140, 149–165. (
- 2002) Population differentiation of Potamogeton pectinatus in the Baltic sea with reference to waterfowl dispersal. Molecular Ecology, 11, 1947–1956. , , & (
- 1996) Rhizome growth of Calamagrostis canadensis in response to soil nutrients and bulk density. Canadian Journal of Plant Science, 76, 545–550. , , & (
- 1987) Selection of offspring size at independence and other size-versus-number strategies. American Naturalist, 129, 800–817. (
- 1996) The specialization hypothesis for phenotypic plasticity. International Journal of Plant Sciences, 157, 484–487. & (
- 1986) Plasticity in life history traits of Plantago major L. ssp. pleiosperma Pilger. Oecologia, 69, 25–30. & (
- 1997) Genotypic variation in Polygonum pensylvanicum: nutrient effects in plant growth and aphid infestation. Canadian Journal of Botany, 75, 546–551. , , & (
- 1998) Clonal diversity in the submerged macrophyte Potamogeton pectinatus L. inferred from nuclear and cytoplasmic variation. Aquatic Botany, 62, 147–160. , & (
- 2001) Local adaptation across a climatic gradient despite small effective population size in the rare sapphire rockcress. Proceedings of the Royal Society of London Series B – Biology Sciences, 268, 1715–1721. , , , , & (
- 1997) Local adaptation in two subspecies of an annual plant: implications for migration and gene flow. Evolution, 51, 1079–1089. & (
- 2001) Spatial variation in tuber depletion by swans explained by differences in net intake rates. Ecology, 82, 1655–1667. , , , , , et al. (
- 2001) Differentiation for flowering time and phenotypic integration in Arabidopsis thaliana in response to season length and vernalization. Oecologia, 127, 501–508. & (
- 1995) Reaction norms of Arabidopsis. I. Plasticity of characters and correlations across water, nutrient and light gradients. Journal of Evolutionary Biology, 8, 421–438. , & (
- 2001) Clonal variation in the thermal response of the submerged aquatic macrophyte Potamogeton pectinatus. Journal of Ecology, 90, 141–152. & (
- 2000) Genetic differentiation of life-history traits within populations of the clonal plant Ranunculus reptans. Oikos, 90, 442–456. & (
- 1992) The causes and developmental effects of integration and independence between different parts of Glechoma hederacea clones. Oikos, 63, 376–386. & (
- 2001) Morphological and behavioural plasticity of larval anurans in response to different predators. Ecology, 82, 523–540. (
- 2003) Asymmetric competition and the evolution of propagule size. Journal of Ecology, 91, 554–562. , & (
- 2002) Hiding from swans: optimal burial depth of Sago Pondweed tubers foraged by Bewick's swans. Journal of Ecology, 90, 303–315. & (
- 1995) Effect of reproduction on nitrogen allocation and carbon gain in Oenothera biennis. Journal of Ecology, 83, 23–29. & (
- 1986) The evolution of phenotypic plasticity in plants. Annual Review of Ecology and Systematics, 17, 667–693. (
- 1989) Phenotypic integration and environmental change. Bioscience, 39, 460–464. (
- 1992) Growth responses of rhizomatous plants to fertilizer application and interference. Oikos, 65, 13–24. & (
- 1997) Within population genetic differentiation in traits affecting clonal growth: Festuca rubra in a mountain grassland. Journal of Evolutionary Biology, 10, 383–406. , , , , , et al. (
- 1985) Gene flow in natural populations. Annual Review of Ecology and Systematics, 16, 393–430. (
- 1974) The optimal balance between size and number of offspring. American Naturalist, 108, 499–506. & (
- 1993) Evidence for local adaptation in closely adjacent subpopulations of northern red oak (Quercus rubra L.) expressed as resistance to leaf herbivores. American Naturalist, 142, 928–936. , & (
- 1987) Tuber size and planting depth influence growth of Potamogeton pectinatus L. American Midland Naturalist, 118, 77–84. (
- 1999) Statistica, version 5.5, Tulsa, Oklahoma. (
- 1981) Principles and Procedures of Statistics: A Biometrical Approach. McGraw-Hill, Singapore. & (
- 1995) Phenotypic plasticity and plant adaptation. Acta Botanica Neerlandica, 44, 363–383. (
- 1993) Phenotypic plasticity in Polygonum persicaria. III. The evolution of ecological breadth for nutrient environment. Evolution, 47, 1050–1071. & (
- 1966) Role of plant growth in regulating photosynthesis. Nature, 210, 77–79. & (
- 1991) Evolution of generalists and specialists in spatially heterogeneous environments. Evolution, 45, 1317–1331. (
- 1997) Generalists, specialists and the evolution of phenotypic plasticity in sympatric populations of distinct species. Evolution, 51, 1372–1380. (
- 1991) Genetic differentiation between populations of Plantago lanceolata. I. Local adaptation in three contrasting habitats. Journal of Ecology, 79, 27–42. & (
- 1988) Ecological studies on Potamogeton pectinatus L. I. General characters, biomass production and life cycles under field conditions. Aquatic Botany, 31, 211–251. (
- 1989a) Ecological studies on Potamogeton pectinatus L. III. Reproductive strategies and germination ecology. Aquatic Botany, 33, 87–112. (
- 1989b) Ecological studies on Potamogeton pectinatus L. V. Nutritional ecology, in vitro uptake of nutrients and growth limitation. Aquatic Botany, 35, 319–335. (
- 1988) Ecological studies on Potamogeton pectinatus L. II. Autecological characteristics, with emphasis on salt tolerance, intraspecific variation and isoenzyme patterns. Aquatic Botany, 32, 239–260. , & (
- 1998) Sources and consequences of seed mass variation in Banksia marginata (Proteaceae). Journal of Ecology, 86, 563–573. & (
- 1994) Intraspecific variation in Potamogeton Pectinatus L. a controlled laboratory experiment. Lake Veluwe, a Macrophyte-Dominated System under Eutrophication Stress (eds W.Van Vierssen, M.J.M.Hootsmans & J.E.Vermaat), pp. 26–39. Kluwer Academic, Dordrecht. & (
- 1995) AFLP: a new technique for DNA + fingerprinting. Nucleic Acids Research, 23, 4407–4414. , , , , , et al. (
- 1998) An account of the species of Potamogeton L. (Potamogetonaceae). Folia Geobotanica, 33, 241–316. & (
- 1999) Intraspecific variability and maternal effects in the response to nutrient addition in Chenopodium album. Canadian Journal of Botany, 77, 1150–1158. , , & (