Ecotypes or phenotypic plasticity—The aquatic and terrestrial forms of Helosciadium repens (Apiaceae)

Abstract Morphological and ecological differences of two forms of Helosciadium repens are known and described in the literature: aquatic and terrestrial. However, their taxonomic status is currently unknown. The question whether they are genotypically adapted to specific environmental conditions or are those differences a result of phenotypic plasticity is addressed in this study. SSR and ISSR data were used to uncover genotypic differences. Data from drought stress experiments (system water content and relative water content of leaves) were used to evaluate the response to water as an environmental factor. The stomatal index of both forms grown under different water treatments was analyzed. The principal component analysis of the ISSR data revealed no clustering that would correspond with ecotypes. The diversity parameters of the SSR data showed no significant differences. The aquatic populations showed a tendency toward heterozygosity, while the terrestrial ones showed a bias toward homozygosity. Both forms responded similarly to the changes in water availability, with newly produced leaves after drought stress that were better adapted to repeated drought stress. Stomatal indices were higher in plants from aquatic habitats, but these differences disappeared when the plants were grown in soil. The observed responses indicate that the differences between forms are due to phenotypic plasticity.

The hydrophytic populations or aquatic form (hereafter Aqu) can be occasionally found in Southern Germany, Bavaria (pers. observation). The aquatic form tends to exhibit vegetative growth only and does not produce flowers (Casper & Krausch, 1981;Schossau 2000 cited in Hacker et al., 2003;NLWKN, 2011). They can grow leaves up to 40 cm in length (Casper & Krausch, 1981), can colonize waterbodies up to a depth of 60 cm, and their stolons can grow up to a length of 150 cm (Voightländer & Mohr, 2008). They stay immotile due to their roots anchored on driftwood, tree roots, or other aquatic vegetation. The plants do not root in the substrate (pers. observation).
There is scarce information on the two different manifestations in the literature. However, when mentioned, authors address both appearances as different forms of the species, and do not specify what the word "forms" means in the corresponding context (Casper & Krausch, 1981;Hacker et al., 2003;NLWKN, 2011;Voightländer & Mohr, 2008). Whether they are genotypically adapted to specific environmental conditions or a result of phenotypic plasticity is thus still unknown. T. Herden, M. Bönisch, & N. Friesen (unpublished data) analyzed 27 populations of H. repens in Germany with SSRs and found only low levels of variation within the analysed markers. There we found no genetically based separation into a Terr or Aqu cluster, suggesting differences due to phenotypic plasticity. However, our sample set was not aimed to address the taxonomic status of both forms. The ecotype hypothesis cannot be excluded based only on these results. Markers may fail to detect quantitative variation for adaptively important traits (Bekessy, Ennos, Burgman, Newton, & Ades, 2003;McKay & Latta, 2002).
If both forms appear to be ecotypes, it can have consequences on the conservation management. Ex situ conservation management for aquatic forms needs to be adapted as well as conservation at the natural sites. Additionally, this information might be interesting for plant breeders as both ecotypes may harbor specific traits of interest.
This comparison study aimed to answer the question of the taxonomical status of both forms by using simple sequence repeats (SSR) and intersimple sequence repeats (ISSR) data on a balanced sample set.
Additionally, the adaptation of both forms to drought stress was studied by measuring the relative water content (RWC) of leaves, F I G U R E 1 Terrestrial and aquatic forms of Helosciadium repens (a) terrestrial form at the natural site, (b) aquatic form at the natural site, (c) leaf of terrestrial form, (d) inflorescence of the terrestrial form, and (e) leaf of the aquatic form system water content, and water loss during drought stress conditions. The stomatal index (SI) was measured for different water treatment levels. A small scale experiment was set up, to determine whether H. repens is capable of self-fertilization.

| SSR analysis
SSR or microsatellites are short stretches of repeated short nucleotide motifs. These motifs typically consist of mono-, di-, and tri-nucleotides, but even longer ones can be found. The repetitions of the motifs are mainly <100 base pairs (bp) long and can be found in all genomes (Tautz, 1989). They can show side-specific length variation because of the occurrence of different numbers of repeat units (Morgante & Olivieri, 1993). Most of these length differences are caused by the slippage effect during replication and accumulate over time (Tautz & Schlötterer, 1994). Using the polymerase chain reaction (PCR), with specific primer pairs flanking a specific microsatellite, it is possible to amplify and measure the exact bp length of a microsatellite. SSR markers are considered to be a reliable system for diversity studies as they are codominant and multiallelic (Baldwin, Pither-Joyce, Wright, Chen, & McCallum, 2012;Fu, Kong, Yingxiong, & Cameron, 2005;Geethanjali, Anitha Rukmani, & Rajakumar, 2018;Park, Lee, & Kim, 2009;Yasodha et al., 2018). They are neutral markers and are thus usually not subjected to natural selection (Holderegger, Kamm, & Gugerli, 2006;Kimura, 1983

| ISSR analysis
Intersimple sequence repeats (ISSR) are regions between microsatellite loci. In a PCR, only one primer containing an SSR motif is used, which amplifies multiple fragments with various length (Reddy, Sarla, & Siddiq, 2002;Zietkiewicz, Rafalski, & Labuda, 1994). Only regions between adjacent, inversely oriented SSRs are thus amplified (Zietkiewicz et al., 1994). Usually, the PCR products are visualized on an agarose gel, and the banding pattern is transformed into a binary matrix. Every band is treated as a single trait. By analyzing the matrix, kinship relations can be computed. Polymorphism can be detected due to mismatches in the priming site (changes in the SSR where the primer binds) or differences in length of the amplified sequences (Zietkiewicz et al., 1994). This method has been widely used for decades in population genetic studies and studies to characterize genetic divergence among species (Andiego et al., 2019;Kumar, Mishra, Singh, & Sundaresan, 2014;Reddy et al., 2002;Schlotteröer, Amos, & Tautz, 1991;Zietkiewicz et al., 1994).

| Self-fertilization test
Plants from two populations that were currently available (nine individuals from 9R and nine from a population from Austria) were potted in trays. These were then isolated from potential pollinators using transparent plastic hoods with Drosophila impermeable mesh for airflow. One control from each population was potted outside of the isolation hoods. The isolated individuals were pollinated by hand with their pollen. At the end of their vegetation period, the seeds were collected. Seeds were drawn randomly for germination tests.

| Dry stress experiment
Stolons from 15 Terr plants (population 9R) were potted in 10 × 10 cm pots (the stolon was approximately 5 cm long with two leaves).
For substrate, 173 g of "Einheitserde Special" (Einheitserdewerke Werkverband e.V., Sinntal-Altengronau, Germany) was used. Plants were grown for three weeks in a greenhouse to ensure that they have rooted successfully. During that time, all pots stood in trays filled with water to ensure that they were watered to their maximum water capacity. They were treated with extra light using one unit of the KIND LED L600 grow light (Santa Rosa, CA), until the start of the experiment.
All plants were weighted (system water content (SWC) = weight of soil, pot, and plant) just before they were put into a climatic chamber (maximum run time for every run: 20 days, day temperature: 33°C; night temperature: 22°C; light: 14 hr; dark: 10 hr; rel. humidity >80%).
The pots were weighed daily during the runs. During the experiment, the pots were not watered. To ensure that all plants grew under the same condition, the pots' positions in the chamber rotated every day.
If plants lost all their leaves due to wilting, they were taken out of the chamber and watered immediately to their maximum water capacity, to prevent the loss of study material.
All plants recovered during a recuperation period of three weeks in the same greenhouse conditions as mentioned above. The experiment was then repeated with the same plants to assess potential adaptation.
The same experiment was conducted with plants from Aqu populations (population 16R, 22R, 24R), which were grown in soil for a time period of one year. For that, five individuals were collected at three natural sites (with the maximum distance between each sample) and cuttings were used in the experiment. For every plant, the results were statistically evaluated by one-way analysis of variance (ANOVA), using the software R (R Core Team, 2017).
At the beginning of each run, one leaf from every plant was used to measure the RWC. For that, the weight (W) of a freshly harvested leaf was measured and put in a 50-ml centrifuge tube with 5 ml of distilled water for rehydration. As Arndt, Irawan, and Sanders (2015) already indicated, rehydration by floating leads to erroneous RWC estimates. Therefore, the leaves were put petiole first in the distilled water, making sure that the water level did not reach the lowest pair of leaflets. They were rehydrated for three hours in darkness under room temperature conditions, and the turgid weight (TW) was measured afterward. All leaves were left in a dry chamber with 10% relative air humidity overnight and weighted afterward to measure the dry weight (DW). The RWC was calculated using the formula of Weatherley (1950). The measurement was also carried out with leaves that exhibit a complete loss of turgor pressure. The system water loss (SWL) was Tests for significance were made with the geom_signif function using the R package ggplot2, and plots were drawn using the function ggplot from the R package ggplot2 (Wickham & Chang, 2018).

| Stomatal index
To estimate the SI, nail polish impressions from the epidermis were made (as described in Miller & Ashby, 1968) from plants cultivated ex situ in the Botanical Garden of Osnabrueck, Germany. Ten impressions from the upper surface were made from all leaflet pairs of a leaf, to test whether there are significant differences between each leaflet pair.
The same was done for the lower leaf surface. Pictures of impressions were made using a transmitted light microscope under 400× magnification. Stomata counts (SC) and epidermis cell counts (EC) were quantified (guard cells were treated as a part of the stomatal apparatus).
The observed surface area was measured (A), and the stomatal density (SCD), as well as the epidermal cell density (ECD), was calculated.
Three pictures were taken from every leaflet pair, and the quantifications of the SC and EC were averaged. The SI was calculated for every leaflet pair using the equation from Salisbury (1928).
The SI was calculated for two different water treatment levels for

| SSR analysis
There were no significant differences in the numbers of MLG, SLG, alleles, allelic richness, rare alleles, and private alleles between Terr and Aqu plants (Figure 2a-d,g,h). As T. Herden, M. Bönisch, & N.
Friesen (unpublished data) showed, there is no genetically based separation into a Terr or Aqu cluster. However, there were significant differences (p < .01) in the F-and F is -Indices between both forms.

| ISSR analysis
Only eight out of 26 tested ISSR markers produced evaluable polymorphic bands. A total of 108 bands were amplified out of which 64 were polymorphic, and 42 were monomorphic bands (Table S1).  (Table S1).

| Self-fertilization test
There was an evident difference in the number of seeds between the isolated and their control pots. However, due to high humidity in the isolated trays, some of the inflorescences and infructescences started to rot. Therefore, a test for statistical significance was not possible. Nevertheless, the isolated plants produced seeds when fertilized with their pollen. Randomly selected seeds were able to germinate. given in Figure S1a,b). Only plant VI had an even water loss which was comparable to the linear regression ( Figure S1a).
The new leaves that grew back during the recovery period were smaller and stiffer.
In the second run, all plants endured the scheduled time of 20 days without a complete loss of leaves (Figure 4a). The plants F I G U R E 3 Principal component analysis of the ISSR data of eight terrestrial and seven aquatic populations. Blue = Aq=aquatic population, orange = Terr=terrestrial populations; Lab IDs = first digits including the letter R (see Table 1  showed signs of withering, after an average SWL of 55% (39%-65%). The control pot had an SWL of 38%. Overall, the SWC was significantly higher during the second run. The adjusted R 2 s were higher than 0.98, except for plants II, IV, and V (>0.97) ( Figure S1a).
The relationship of the variables during the second run almost fits a linear regression in all investigated plants ( Figure S1a  3.75%) of water.  The RWCs between leaves at the start of the run one (with full turgor pressure) and those at the end of the run one (complete loss of turgor pressure) differed significantly (Figure 5b). The leaves lost on average 23.95% (lowest: 7.33%; highest: 49%) of water. In run two, the RWCs were again significantly different when comparing the beginning and the end of the run. The leaves lost on average 2.51%
The RWC at the start of both runs was significantly different, comparing both conditions (Aqu and Terr). On average, the differences were 1.28% in the first run and 4.3% in the second run. At the end of both runs, the RWCs in both conditions were not significantly different anymore (p < .001, data not shown).

| Stomatal index
There were no significant differences between the different leaflet pairs in a leaf (Figure 6a,b). In all conditions (Aqu, A-T, T-Wet, and Terr), the SI of the upper surface was significantly lower than the SI from the lower surface (p < .001) (Figure 6c-f). On the upper surfaces, the SI was significantly higher for Aqu than all other conditions with different levels of significance (Figure 6g). There were no significant differences between conditions A-T, T-Wet, and Terr.
On the lower surfaces, the SI of Aqu was significantly higher (with different levels of significance) in comparison with the SI of plants grown under other conditions (Figure 6h). There was a significant The ratio (SI upper/SI lower) between the upper and lower surfaces for each condition was analyzed (Figure 6i). The only significant difference was detected between A-T and T-Wet (.01 < p < .05).

| General observations
Five cuttings from every Aqu populations were potted and the rest grown in small trays with water. All plants, in the trays with water and the pots, build inflorescences and infructescences.

| D ISCUSS I ON
Two main results derived from this study: (a) The analyses of the SSR and ISSR data showed similar outcomes and no significant separation into ecotypes; (b) the differences in morphological characters of the two forms faded when plants were grown under the same conditions.

| Genetic comparison
Both fingerprinting methods (SSR and ISSR) together portray the genetic diversity of the entire genomes of all investigated individuals.
Nevertheless, most populations can be genetically told apart from each other; both forms are not genetically differentiated (Figures 2   and 3). Therefore, a taxonomical division based on molecular data is not justified.
The only significant difference recovered from the genetic data was from the F-statistics (Figure 2e,f; Table S2). The heterozygote excess, revealed by a negative F is , can be caused by asexual propagation (Stoeckel et al., 2006). Four out of the seven Aqu populations exhibited negative F and F is values. These findings confirm the observations that these populations tend to grow clonally (Casper & Krausch, 1981;Schossau 2000cited in Hacker et al., 2003NLWKN, 2011). However, three of them have positive F-statistic values. A heterozygote deficiency (homozygote excess) is revealed by positive F is values and can be caused by self-fertilization. This is mainly the case in the Terr populations.

or in
Xanthium pungens Wallr. (Abeles, 1967). The plants in the tray were in contact with the bottom and were thus able to sustain upright leaves above the water surface. Due to fluctuations in the water level at the natural sites, the very similar conditions can occur possibly leading to infrequent flowering. Burmeier and Jensen (2008) observed that seeds were able to germinate even under water. Therefore, seed recruitment during low water seems possible and could explain the positive F is values.
However, these interpretations remain largely hypothetical and constitute a basis for further research.

| Drought stress
Both forms undoubtedly adapted between the first and the second runs ( Figure 4). This adaptation is also visible in the RWC values of the leaves ( Figure 5). During the second run, the RWC values of the leaves did not drop as much as in the first runs ( Figure 5). In some cases, the leaves even gained water and plants grew new leaves during the run (pers. observation).

| Stomatal index
There were no significant differences between forms ( Figure 6).

| CON CLUS ION
In general, neither molecular data nor the results from watermanipulating experiments alone can rule out the hypothesis of ecotypes. Molecular markers may fail to detect differences (Bekessy et al., 2003), and there could be other ecological factors in which the two forms behave differently. Billet, Genitoni, and Bozec (2018) analyzed aquatic and terrestrial morphotypes of Ludwigia grandiflora (Michx.) Greuter & Burdet and based on morphological traits they found that the terrestrial morphotype outcompetes the aquatic one. However, they did not perform molecular analyses; thus, the molecular basis of L. grandiflora adaptation remains unknown.
Ecotype hypotheses can be addressed only when morphology as well as genetic foundation studies is combined (McKay & Latta, 2002). In a study on Alternanthera philoxeroides (Mart.) Griseb., Geng et al. (2007) used molecular data (ISSR) and common garden experiments to test the ecotypes hypotheses for aquatic and terrestrial forms. Their data supported, however, the plasticity hypothesis.
For Coccothrinax argentata (Jacq.) L.H.Bailey, Davis, Lewis, Francisco-Ortega, and Zona (2007) found minute differences in the ISSR analysis between the mainland and insular populations. However, they found a great deal of plasticity in the traits included in the study that do not support a separation into different taxa. In Ageratina adenophora (Spreng.) R.M.King & H.Rob., the authors found evidence for phenotypic plasticity after checking 16 populations with ISSR and common garden experiments (Zhao, Yang, & Xi, 2012). Noel, Machon, and Porcher (2007) analyzed Ranunculus nodiflorus L. populations in France with microsatellites and common garden experiments. They found no genetic diversity and strong evidence favoring phenotypic plasticity.
Since our molecular data provide strong evidence against the ecotype hypothesis and the morphological differences disappeared during a simple drought stress experiment, the results can only lead to one explanation: phenotypic plasticity. Moreover, the drought stress experiment showed that plants that experienced drought stress performed better when subjected to drought stress again. This adaptive plasticity in this species enables it to endure short periods of drought stress and periods of water stress (Longa, 2019). It also gives the plants an advantage over competitors in zones of water fluctuations such as wet pastures and littoral zones, where this species naturally occurs. The ability of self-fertilization may benefit H. repens in environments where pollinators are scarce.

ACK N OWLED G M ENTS
The authors thank Dr. Anselm Krumbiegel for providing living material from the aquatic populations. The German Federal Ministry of Food and Agriculture (BMEL) financially supported this work through the Federal Office for Agriculture and Food (BLE) (grant number 2814BM110).

CO N FLI C T O F I NTE R E S T
None declared.

AUTH O R CO NTR I B UTI O N S
T.H. and N.F. conceived the ideas and designed methodology; N.F.

DATA AVA I L A B I L I T Y S TAT E M E N T
Data is available under https ://doi.org/10.5061/dryad.m0cfx pnzg.