|Home | About | Journals | Submit | Contact Us | Français|
Intraspecific variation in flooding tolerance is the basic pre-condition for adaptive flooding tolerance to evolve, and flooding-induced shoot elongation is an important trait that enables plants to survive shallow, prolonged flooding. Here an investigation was conducted to determine to what extent variation in flooding-induced leaf elongation exists among and within populations of the wetland species Rumex palustris, and whether the magnitude of elongation can be linked to habitat characteristics.
Offspring of eight genotypes collected in each of 12 populations from different sites (ranging from river mudflats with dynamic flooding regimes to areas with stagnant water) were submerged, and petioles, laminas and roots were harvested separately to measure traits related to elongation and plant growth.
We found strong elongation of petioles upon submergence, and both among- and within-population variation in this trait, not only in final length, but also in the timing of the elongation response. However, the variation in elongation responses could not be linked to habitat type.
Spatio-temporal variation in the duration and depth of flooding in combination with a presumably weak selection against flooding-induced elongation may have contributed to the maintenance of large genetic variation in flooding-related traits among and within populations.
Flooding is a common and widespread natural phenomenon, and has severe negative effects on growth of terrestrial plants, mainly due to the slow gas diffusion in water as compared with air (reviewed by, amongst others, Visser and Voesenek, 2004; Bailey-Serres and Voesenek, 2008). The reduced entry of oxygen into the plant when submerged leads to inhibition of aerobic respiration, and thereby to shortage of ATP. Ultimately, this may result in death of cells in non-adapted plants (Jackson and Armstrong, 1999). Flooding often reduces light intensity and changes the light spectrum (Holmes and Klein, 1987), and, moreover, imposes strongly reduced CO2 availability (Mommer et al., 2006a), causing low photosynthesis and thus low carbohydrate gain. Therefore, flooding is likely to be a strong selection pressure in habitats where it occurs frequently, such as in river floodplains, especially during the growing season.
Species that are regularly subjected to flooding have evolved a range of physiological and morphological adaptations to survive. Glycolysis and fermentation supply the flooded, anaerobic plants with energy to maintain basic metabolic processes (reviewed by Sauter, 2000). Aerenchyma formation improves transport of gases within the plant (Colmer, 2003), whereas newly formed narrower and thinner leaves facilitate gas exchange between the shoot and water when submerged (Mommer et al., 2005) and thereby increase survival (Mommer et al., 2006b). Finally, vertically orientated petioles (Voesenek and Blom, 1989a) and stimulated stem or petiole elongation (Ridge, 1987; Voesenek et al., 1993; Kende et al., 1998) may eventually bring leaves to the water surface to re-establish gas exchange with the atmosphere.
It is well known that plant species differ widely in their ability to express such flooding-induced traits and hence in their flooding tolerance. This leads to a zonation of species differing in flooding tolerance along flooding gradients (van Eck et al., 2004; Voesenek et al., 2004). Artificial selection has shown that flooding-induced elongation is a selectable trait in crop species and that rice varieties differing in elongation response differ in performance depending on the flooding regime (Jackson and Ram, 2003). However, little is known about naturally occurring intraspecific variation in expression of traits needed for flooding tolerance, even though this variation forms the basic pre-condition for adaptive flooding tolerance to evolve. Previous studies indicated that populations from habitats differing in flooding regimes may display different morphological, photosynthetic and metabolic characters (Keeley, 1979; Lessmann et al., 1997; Lenssen et al., 2004; Huber et al., 2009), suggesting that there is a potential for population differentiation in these flooding-induced traits. Surprisingly, few data are available on naturally occurring intraspecific variation in flooding-induced shoot elongation, a trait that is one of the most important characteristics of wetland plants in habitats where they frequently become shallowly submerged (Voesenek et al., 2004).
Such phenotypic plasticity, i.e. the ability to change the phenotype in response to the environment, is an important feature of plants. These responses may result in increased plant performance in heterogeneous conditions, and have been the focus of research for the last decades (Bradshaw, 1965; Schlichting, 1986; Sultan, 2000; Schmitt et al., 2003). However, the degree of genetic variation in plasticity and selection of plasticity in natural habitats still remain largely unresolved (van Kleunen and Fischer, 2005). So far, only a limited number of studies have attempted to study the variation among populations in traits that are assumed to be important during flooding (Huber et al., 2009). It can be hypothesized that, comparably with the responses to shade (Dudley and Schmitt, 1995; Petit and Thompson, 1998; Donohue et al., 2000; Maloof et al., 2001; von Wettberg et al., 2008), the potential of shoot elongation in response to flooding will differ depending on the environmental conditions characterizing a given habitat. In the few experimental studies on variation among populations in flooding-related traits, typically a small number of genotypes were used, or no significant differences were found. For example, Keeley (1979) reported that plants of the hardwood species Nyssa sylvatica that originated from floodplains were similar to upland plants in drained conditions, but very similar to swamp plants in flooded experimental conditions, in terms of root morphology, metabolism, oxygen transport and nutrient uptake. Lenssen and co-workers (2004) compared lakeside and landside genotypes of the clonal species Ranunculus reptans and found that carbohydrate use efficiency did not differ between microhabitats, and also that genotypes from the lakeside had, unexpectedly, lower root aerenchyma content than those from landside habitats. Here we investigate whether flooding-induced shoot elongation in Rumex palustris depends on the original microhabitat conditions.
The wetland species R. palustris is an annual or biennial plant that occurs in river floodplains characterized by periodic flooding events, or areas with rather stagnant water. When completely submerged, R. palustris plants position their leaves from rather horizontal to almost vertical and elongate their petioles rapidly to restore contact with the atmosphere (Voesenek et al., 1990; Cox et al., 2003). So far, research has focused on flooding-induced petiole elongation of R. palustris originating from one single population (reviewed in Peeters et al., 2002). However, R. palustris lives in a variety of semi-aquatic habitats where frequency, duration and depth of flooding can differ widely (Nabben et al., 1999; Vervuren et al., 2003). Genetic variation in the expression of flooding-related traits and, more particularly, expression of enhanced leaf elongation can be expected as a result of natural selection, since the flooding-induced leaf elongation is crucial for survival in specific habitats characterized by shallow and prolonged temporal flooding (Voesenek et al., 2004).
It is unclear whether genetic variation exists in flooding-induced leaf elongation, and whether variation in the flooding regime has led to population differentiation. Using seeds collected in 12 populations of R. palustris, characterized by a range of flooding regimes, the within- and among-population variation in flooding-induced leaf elongation was investigated. We specifically tested whether flooding-induced leaf elongation depended on habitat characteristics (river floodplains with fluctuating water levels vs. areas with relatively stagnant water). Two parts of leaves, the petiole and lamina, contribute to total leaf elongation (Voesenek and Blom, 1989b), so we also investigated whether variation of leaf elongation was due to responses of the petiole or the lamina, or a combination of both. Depending on the type of flooding regime, selection on immediate and strong response vs. slow elongation response may take place (Voesenek et al., 2004). Both responses may eventually lead to the same final length. It is thus expected that variation among populations may be expressed not only in total leaf length but also as variation in kinetics, i.e. the magnitude or the timing of this growth response, or both. Since elongation responses may have an optimal developmental window in time, with different elongation potential of leaves differing in developmental stage (Groeneveld and Voesenek, 2003), the response of older and younger leaves was compared. Furthermore we examined, as a plant fitness estimate, plant biomass development.
Rumex palustris Sm. is mainly found in mudflats in the river floodplain, but also in areas characterized by standing water. Seeds typically germinate in non-flooded conditions in the autumn. Plants survive winter as vegetative rosettes, restart growing in the spring, flower in the summer and set seeds in the autumn. Long-lasting and/or deep spring and/or summer floods during the growing/flowering season can prevent plants from completing their life cycle in the same year, and may delay flowering until the next year.
Seeds of R. palustris were collected from eight randomly selected mother plants from each of 12 natural populations in The Netherlands (Fig. 1, Table 1) at least 1 year prior to the experiment, and directly used in this experiment. The 12 populations were found in two habitat categories. The first category encompasses river floodplains where plants were flooded due to periodic rising river water level (up to 4 m). Populations from the second category were subjected mainly to stagnant water, which may rise slightly (10–20 cm) after local heavy rainfalls. Although R. palustris is largely self-pollinating (X. Chen, pers. obs.), wind pollination also takes place in open fields, and the seed families collected from different mother plants within a population were thus assumed to differ genetically and are therefore referred to as genotypes for the remainder of the manuscript. Genetic variation therefore refers to differences in trait expression among seeds collected from the different genotypes. The seeds were germinated on filter paper moistened with tap water (each genotype in a separate Petri dish) in a germination cabinet for 10 d [12 h light, 20 µmol m−2 s−1 photosynthetic photon flux density (PPFD), 25 °C, and 12 h dark, 10 °C]. Homogeneously germinated seedlings were transplanted singly into plastic pots (300 mL), containing a mixture of sieved potting soil and sand (1 : 1, v:v), and grown until the fifth oldest leaf emerged (Banga et al., 1997; for 20 d on average) in a growth chamber [20 °C, 60 % relative humidity, 16 h light: 250 µmol m−2 s−1 PPFD (sodium lamps Philips SON-T plus 600 W and fluorescent light TDL Reflex 36 W/840R, Philips, Eindhoven, The Netherlands)]. After transplanting, pots were covered with transparent plastic foil for 3 d, after which the foil was removed. All plants were watered with tap water frequently. Each plant received 20 mL of nutrient solution, containing 0·8 mm KNO3, 0·6 mm Ca(NO3)2, 0·3 mm MgSO4, 0·2 mm KH2PO4, 40 µm Fe-EDTA and micronutrients 10 d after transplanting.
Plants were completely submerged for 17 d in ten interconnected opaque polyethylene basins (336 L; water column height 70 cm) in a growth chamber (20 °C, 60 % relative humidity, 16 h light: 130 µmol m−2 s−1 PPFD). The basins were filled with tap water (pH 7·4, alkalinity 2·7 mEq L−1) which was filtered and circulated at a flow rate of 1·2 L min−1 to prevent algal growth. Superficial algal growth on pots assigned to the submergence treatment was removed before experiments started. Drained control plants were kept under the same light conditions on tables in the same growth chamber as the submerged plants and were watered regularly. As plasticity strongly depends on developmental stage, plants were subjected to treatments when their fifth oldest leaf had grown to a size of 5–10 mm. Four plants per genotype were selected for homogeneity with respect to the developmental stage of the fifth oldest leaf, and randomly allocated to the treatments. Since the offspring of different genotypes varied in growth rate, treatments started for the different genotypes at different time points. The total time between the first and the last groups of plants being subjected to treatments was 6 d.
Petiole and lamina length from the third oldest leaf onward of both drained and submerged plants were measured to the nearest millimetre on day 0 and day 7 of the experiments using a ruler. Submerged plants were measured on day 7 without de-submergence. At the harvest on day 17, shoots were dissected into individual leaves (petiole and lamina separated) and soil was washed from the roots. Digital pictures of the petioles and laminas were taken, and lengths of petioles and laminas and leaf area were measured from these pictures using a PC-based system equipped with KS400 image analysis software (Carl Zeiss Vision 3·0, Oberkochen, Germany). Petioles, laminas, roots and lateral shoots were dried at 70 °C for 72 h, after which dry weights were measured.
Data were analysed by means of three-way mixed model nested analysis of variance (ANOVA), with treatments and habitats being the main effects and populations nested within habitats and genotypes nested within populations. Treatments and habitats were treated as fixed effects, and populations and genotypes as random factors. Petiole and lamina lengths of the third to the fifth oldest leaves were analyeed by using repeated measures ANOVA, with the leaf plasticity (i.e. the position of the leaf in the rosette) being treated as the repeated factor. The kinetics of elongation were analysed by using repeated measures ANOVA, with the timing of measurements (first 7 d and last 10 d) being treated as the repeated factor. As the rosette leaves were in different developmental stages and had finished a different proportion of their natural growth process, the latter analyses were performed for all leaves separately. In both repeated measures ANOVA treatments, habitat and populations were the main effects, with treatments being treated as a fixed effect, and habitats, populations within habitats and genotypes within populations as random factors. Correlation analyses were used to test whether petiole and lamina plasticity were correlated. Plasticity was calculated as the percentage of petiole and lamina length in submerged as compared with drained conditions. Population means were used for this analysis. All data were analysed with SAS (ver. 9.1, SAS Institute Inc., Cary, NC, USA).
Total leaf length varied between leaves that were at a different developmental stage at the onset of submergence and among populations (Figs 2 and and3).3). Both petioles and laminas contributed to total leaf length, but to a different extent mainly depending on leaf age and treatment. The relative contribution of petioles of control and older leaves was much less (approx. 15 %) than that of submerged and younger leaves (approx. 50 %; Fig. 3).
When the plants were submerged, both petioles and laminas responded by elongation growth (Table 2). Overall, submergence hardly affected lamina length (Table 3). However, there was a significant lamina position × treatment interaction, indicating that the response to treatments differed between leaf developmental stages, with younger laminas being increasingly more responsive (Table 2). In contrast, petioles of submerged plants were up to 6-fold longer than those of drained plants, with younger petioles responding significantly more to submergence than older petioles (Tables 2 and and3).3). Therefore, petiole elongation contributed much more to total leaf elongation than did lamina elongation. Final lengths of the submerged young leaves were 2·5 times longer than those of older leaves. Although young leaves displayed stronger elongation than old leaves, the timing of the response in old and young leaves was remarkably similar, being much more pronounced in the first 7 d of submergence than in the last 10 d (Table 4 and Fig. 5). As expected, lamina length was positively correlated with petiole length (third oldest leaf, r = 0·802, P = 0·002; fourth oldest leaf, r = 0·817, P = 0·001; fifth oldest leaf, r = 0·631, P = 0·028), indicating that plants which had longer petioles under water often also had longer laminas.
Populations differed significantly in the response of petiole length to submergence treatments (Table 3). In some populations, petioles of the fifth oldest leaves of submerged plants were just 50 mm longer than those of the drained plants, whereas in another population, the petioles of submerged plants were almost 90 mm longer than the petioles of drained plants. In contrast to this submergence-induced variation, petioles of drained plants of these populations, and also of all other populations, had similar lengths (Fig. 4).
Not only the leaf lengths differed after submergence, but the timing of elongation was also different among populations (Fig. 5 and Table 4). Some populations (e.g. populations 2 and 4) had a much stronger increment in petiole elongation in the first 7 d of submergence than did other populations, and just a modest increment in the last 10 d. In contrast, other populations (e.g. population 7) displayed a rather small increment in the first 7 d, but were among the fastest growers in the last 10 d. There was great variation among genotypes within populations in both lamina and petiole length and responses (Tables 3 and and44).
Habitat types did not significantly affect flooding-induced elongation responses, indicating that, in contrast to our expectations, plants from areas with stagnant water levels were not generally less responsive to flooding than plants originating from floodplains (Tables 3 and and4).4). For example, populations 12 and 9 in habitat category 2 with stagnant water had a higher increase in petiole lengths than populations 5 and 6 in habitat category 1 with fluctuating water levels.
As a result of submergence, both total dry weight and tap root weight decreased, and the relative contribution of the tap root to the total biomass became less (Table 2 and Fig. 6A, B). Submerged plants had, on average, a 2·5-fold larger specific leaf area (SLA) and a 3-fold larger specific petiole length (SPL; Table 2 and Fig. 6C, D), indicating that submerged plants had thinner and less dense laminas and invested less biomass per unit petiole length than drained plants.
Remarkably, unlike the significant variation in leaf lengths among populations, there was neither a population effect nor a population × treatment effect on any of these other morphological and growth-related traits (Fig. 6). All populations showed rather similar changes in biomass, SLA and SPL when submerged. On the other hand, comparably with lamina and petiole lengths, there was large variation among genotypes within populations (Fig. 6).
Flooding is a widespread environmental stress that can have severe effects on plant survival in natural habitats. Species growing in habitats which are often flooded have evolved a set of specific adaptations (Bailey-Serres and Voesenek, 2008). These adaptations can be fixed, such as constitutively high aerenchyma content in the roots (Visser et al., 1996), or plastic, such as inducible shoot elongation (Voesenek et al., 1993). The latter can bring the tip of the shoot above the water surface, thereby restoring contact with the air and increasing chances of survival (Pierik et al., 2009). We tested whether populations of the same species originating from habitats with different environmental conditions have evolved different responses, whether these responses depend on the developmental stage and to what extent genetic variation is maintained within populations. Our results are among the first studies providing evidence for the existence of a great genetic variation in flooding-induced leaf elongation among natural populations. They clearly show that although R. palustris generally elongates its leaves in response to submergence, the degree of elongation can differ considerably among populations, among genotypes originating from the same population, among leaves produced by a single plant and also between two functionally different parts of the leaf.
Plant species from riverine habitats (comparable with our populations in habitat category 1) with long-lasting shallow flooding are usually characterized by fast and strong elongation responses, whereas species from habitats with less frequent and short flooding (comparable with our populations in habitat category 2) often display delayed and less elongation growth (Voesenek and Blom, 1989a; Voesenek et al., 2004). We found that plants from 12 populations did show variation in the elongation response. However, this variation was not clearly linked to habitat characteristics, even though the two most responsive populations (2 and 4) were from riverine areas with strong fluctuations in water level, and the two least responsive populations (8 and 10) were not. Two of our populations (populations 11 and 12) may have developed just recently (i.e. in the last 10 years) in rural areas. The first seeds that established in these newly developed populations may have come from nearby populations with different habitat characteristics. Although flooding in these rural areas is infrequent and shallow, time may not have been long enough for these populations to have adapted to these local conditions. It would be interesting to sample plants from these two populations over time to see whether and in what time span selection shapes phenotypic traits.
Another explanation for the lack of correlation between elongation capacity and habitat type may be that the difference in selection pressure on the elongation response is not strong enough between habitats. The significant variation we found among populations may have been either due to random processes or caused by other, non-determined environmental variables imposing selection on elongation, such as shade (Schmitt et al., 2003; Weijschedé et al., 2006, 2008; Bell and Galloway, 2008) and herbivory (Cipollini 2004; McGuire and Agrawal, 2005; Valladares et al., 2007).
A third factor that may have limited variation in responses is the fact that in the present experiment all plants were fully submerged throughout the experiment. Longer term experiments covering a full growing season, possibly in outdoor conditions, that vary the flooding duration and water depth, and also take into account that plants in their natural habitat may be much larger than the juvenile stage as used in the present experiment, may further help to detect initially relatively small morphological differences. It is not uncommon for such small differences in response to lead ultimately to large variation in plant performance, particularly if the target resource is distributed heterogeneously, such as in dense canopies (light; Pierik et al., 2003) or, indeed, during flooding (oxygen and light; Laan and Blom, 1990). In addition, great variation within populations may have blurred variation among populations, which makes linking variation among populations to habitats even more difficult.
Potentially, large gene flow between populations could also have resulted in the low interpopulation variation that was found among the populations sampled. However, although Rumex seeds can be transported by waterfowl (Wongsriphuek et al., 2008) and floodwater, it is unlikely that this comprises a significant proportion of the seeds present in the population, even if populations are only a few kilometres from each other. Most seeds fall on the ground in the close vicinity of the mother plant. Similar reasoning holds for the dispersal of pollen, particularly since R. palustris is mainly a self-pollinator (X. Chen, pers. obs.).
A striking difference was found among populations in all three leaves tested in terms of how fast petioles elongated in early and later stages during submergence, but this difference did not affect final petiole length. Depending on the population, plants differed in their temporal elongation pattern of single leaves. For example, plants from population 9 achieved 98 % of their elongation in the third oldest petiole within the first 7 d, whereas the plants from population 11 reached only 74%. The fast elongators can benefit earlier from improved photosynthesis, whereas the slow ones invested less in elongation, and may benefit if flooding duration is short. Surprisingly, all plants and all leaves of different developmental stages elongated considerably more in the first 7 d after the onset of flooding than in the ten subsequent days, making leaf elongation, independently of the developmental stage of the leaf, overall a relatively fast response to submergence. Leaves of very young developmental stages which were just visible at the onset of treatments could thus not elongate for a longer period of time than developmentally older leaves which had already finished most of their development prior to submergence. It may be that growth slowed down in time because resources for elongation growth were depleted while flooding continued (Das et al., 2005), or because the petioles just reached their physical limits (e.g. in cell length or number).
Environmental variation has been proposed to contribute to the creation and maintenance of genetic variation and influence macro-evolutionary patterns (Miner et al., 2005; Agrawal et al., 2006; Hirao and Kudo, 2008), but not all studies were able to confirm this hypothesis (Stratton and Bennington, 1998). Possibly the large variation within populations of R. palustris found in our study is maintained because floods are often unpredictable in essentially all natural habitats where we sampled. Several years may pass by without severe flooding, followed by one or more years with increased flooding duration and depth (Nabben et al., 1999; Vervuren et al., 2003). Plants of this biennial species that survive better in a given year produce more seeds than others. If a certain type of flood occurs in successive years, there is possibly a shift in the composition of genotypes, as the genotypes in favour become more and more abundant. However, since R. palustris forms a persistent seed bank (Voesenek and Blom, 1992), genotypes will not be wiped out in the short term. When the type of flooding changes in subsequent years, other genotypes may survive better and produce seeds. Therefore, different plants within a population may display a range of elongation responses; as a result, there are always some genotypes that survive a specific type of flooding and produce more seeds than others. Spatio-temporal heterogeneity in the timing and duration of submergence may thus contribute to the maintenance of genetic variation within populations in R. palustris. Possibly, the temporal heterogeneity in each of the habitats is larger than the spatial heterogeneity among habitats, given the fact that the variation within populations is great in all plant traits investigated and the lack of link between habitats and elongation. This intrapopulation variation may have increased population performance in different flooding regimes from year to year (cf. Eckhart et al., 2004).
In R. palustris leaves, the petiole gradually broadens into an ellipsoid lamina. Both petioles and laminas, therefore, contribute to total leaf length. Consequently, elongation of both can lead to positioning of the tip of leaves above the water surface. When the plant is completely submerged, even though both petiole and lamina elongated, shoot elongation under water proved to be mainly achieved by elongating petioles in all populations. One reason for this shift may be that biomass invested in petioles per unit length is less than half of that in laminas (data not shown). Also, investment in petioles will not include as much expensive photosynthetic machinery as in laminas. Therefore, elongation of petioles is less costly than that of laminas, and is equally capable of bringing the tip of the shoot above the water surface. Once plants reach out of the water, a higher porosity in the petiole than in the lamina is needed to facilitate internal aeration.
Despite the comparable pattern of elongation in time, young leaves elongated 7-fold more than old leaves, indicating that the amount of elongation achieved per unit of time depends on the developmental stage of the leaves. The ability and magnitude of responding to a specific environmental cue may strongly depend on the developmental stage of an organ, and plastic responses are characterized by an optimal developmental window in time (Voesenek and Blom, 1989a; Watson et al., 1995; Groenenveld and Voesenek, 2003). Different mechanisms may constrain the developmental window in which an organ can optimally respond to environmental cues. Resources for growth may be preferentially allocated to younger leaves (Barthélémy and Caraglio, 2007). Younger petioles may also be less constrained in their elongation response because lignification of the xylem has not been completed (Christiernin et al., 2005), the number of cells is not fixed and cell walls are not rigid yet. Investment in younger leaves will also have a longer term return of benefits, as younger leaves have a higher photosynthetic capacity and older leaves will shed earlier.
It was found that there is variation among populations in flooding-induced leaf elongation, not only in final length, but also in the timing of the elongation response, which was particularly clear in the strong elongation response of petioles of young leaves. This variation, however, did not correlate with the variation in flooding frequency and depth in the respective habitats. Spatio-temporal variation in the flooding regimes within habitats in subsequent years, in combination with a persistent seed bank, may have contributed to the maintenance of the relatively large variation in elongation responses which was found within populations.
We are grateful to Harry van de Steeg for collecting the seeds and for information on the habitat characteristics, Katarzyna Banach, Gerard Bögemann, José Broekmans, Hannie de Caluwe, Sara Gómez, Corien Jansen, Janneke van der Loop, Tamara van Mölken, Annemiek Smit-Tiekstra and Josef Stuefer for assistance in the greenhouse, Maarten Terlou for developing the image analysis program in KS400, MultiMedia for providing the schematic overview of The Netherlands, Ronald Pierik for useful comments on the set-up of the experiment and on the manuscript, and anonymous referees for their insightful comments on the manuscript. This is part of a project from the Centre for Wetland Ecology, a partnership of The Netherlands Institute of Ecology, Radboud University Nijmegen, Utrecht University and the University of Amsterdam.