Species interactions in three Lemnaceae species growing along a gradient of zinc pollution

Abstract Duckweeds (Lemnaceae) are increasingly studied for their potential for phytoremediation of heavy‐metal polluted water bodies. A prerequisite for metal removal, however, is the tolerance of the organism to the pollutant, e.g., the metal zinc (Zn). Duckweeds have been shown to differ in their tolerances to Zn; however, despite them most commonly co‐occurring with other species, there is a lack of research concerning the effect of species interactions on Zn tolerance. Here, we tested whether the presence of a second species influenced the growth rate of the three duckweed species Lemna minor, Lemna gibba, and Lemna turionifera. We used four different Zn concentrations in a replicated microcosm experiment under sterile conditions, either growing the species in isolation or in a two‐species mixture. The response to Zn differed between species, but all three species showed a high tolerance to Zn, with low levels of Zn even increasing the growth rates. The growth rates of the individual species were influenced by the identity of the competing species, but this was independent of the Zn concentration. Our results suggest that species interactions should be considered in future research with duckweeds and that several duckweed species have high tolerance to metal pollution, making them candidates for phytoremediation efforts.


| INTRODUC TI ON
Duckweeds are small, floating aquatic plants with a simplistic morphology. Individual duckweeds consist of single fronds with zero to multiple roots attached to the bottom surface. Duckweeds can flower; however, they rarely do so (Hicks, 1932). Their main reproduction strategy consists of asexual budding, where several daughter fronds bud and then detach from the mother frond (Laird & Barks, 2018). An individual frond can produce up to a couple dozen of daughter fronds over its life, which is only a few weeks short.
Duckweeds are found around the world in slow-flowing freshwater systems, when suitable anchoring possibilities are present (Landolt, 1986). Their fast reproduction cycle and their short life span make duckweeds useful model organisms for research in ecology and evolution (Laird & Barks, 2018). Duckweeds can tolerate high levels of nitrogen, phosphorus, and heavy metals, and different species can have different responses to temperature, light, nutrients, and toxicants (Landolt, 1996). In nature, duckweed species frequently coexist (Landolt, 1986).
Zn is an essential trace element for plant growth, but elevated concentrations inhibit growth and can lead to chlorosis. Therefore, elevated Zn levels are phytotoxic (Rout & Das, 2009). Zn is a commonly used building material and through run-off from roofs, galvanized items, and pipes it finds its way into waters and sediments, leading to Zn pollution in urbanized areas (AWEL, 2006) In the face of pollution of water systems, Lemnaceae are studied as potential organisms for phytoremediation (Liu et al., 2021). For example, one species of duckweed, Lemna minor L. (common duckweed), has shown to be a good accumulator of heavy metals such as cadmium, selenium, and copper (Zayed et al., 1998). Several studies have shown metal accumulation in different duckweed species , which depended both on the species (Cardwell et al., 2002) and on the metal (Gaur et al., 1994). A prerequisite for metal accumulation, however, is the tolerance of a species to elevated levels of heavy metals. Duckweed species differ in their tolerance to Zn: L. minor was shown to tolerate Zn concentrations above 100 mg/L but the gibbous duckweed Lemna gibba only tolerated concentrations up to 10 mg/L . Zn tolerance of other duckweed species such as Lemna turionifera (red duckweed) was, to our knowledge, never investigated.
Additionally, there is a lack of research concerning the influence of species interactions on duckweed resistance to metal pollution.
Previous studies suggest that species interactions in duckweeds can influence growth rates (Clatworthy & Harper, 1962;Gopal & Goel, 1993;Peeters et al., 2016). Here, we hypothesized that the presence of a second species in a mixed setting could increase heavy metal tolerance because of facilitation. The stress-gradient hypothesis predicts that interactions among plants are context-dependent, shifting from competition to facilitation as environmental stress increases (Callaway, 1995). At high Zn concentrations, if the more tolerant duckweed species accumulate Zn present in the medium, this could facilitate the persistence or even growth of the co-occurring, less heavy-metal tolerant species. However, at low levels of Zn concentration, competition for nutrients could override any facilitative mechanisms, leading to a negative effect of the presence of a second species.
To test our hypothesis, we grew three Lemnaceae species in isolation and in two-species pairings along a zinc sulfate (ZnSO 4 ) concentration gradient (0, 0.45, 1.82, 11.35 mg/L Zn). We chose to use a gradient that is relevant in terms of Zn pollution in Europe (1.3 mg/L on average, Zhou et al., 2020). The lowest concentration used was below the European average, the second lowest just above but below the indicator level (5 mg/L), and finally, the highest concentration level exceeded that found in a waterbody near a mining area in Turkey (7.23 mg/L, Sasmaz et al., 2015). Thus, the highest concentration used would represent a heavily polluted waterbody, where phytoremediation could be applied. We measured the Zn tolerance of three duckweeds species L. minor, L. gibba, and L. turionifera over 17 days in replicated microcosms under sterile and controlled conditions.

| Experimental set up
The three species occupy a similar ecological niche. They float on the water surface and are very similar in terms of their morphology. In particular, non-gibbous L. gibba and L. minor fronds cannot be distinguished by eye (De Lange & Pieterse, 1973) but the differentiation between L. minor and L. turionifera can be equally challenging (Senevirathna et al., 2021). Therefore, we separated the species using a floating ring ( Figure S4). Thus, we could overcome previous limitations (Clatworthy & Harper, 1962) and investigate the competition between these closely related species. There were six different compositions (three isolated, three mixed), four Zn concentration levels, and every treatment was replicated three times, thus in total, there were 72 bottles. At the beginning of the experiment, we added 5-7 fronds to the 250 ml bottles according to the study design ( Figure 1). In the isolated setting, the 5-7 fronds were added to the inner area (inside the plastic ring). No fronds were added to the outer area. In the mixed setting, 5-7 fronds from one species were added to the inner area and 5-7 fronds from a second species were added to the outer area. We could thus individually track the populations that were either growing alone in a culture bottle or with a population of a competing species (n = 108). The inside/outside position of the two respective species was kept for the three replicas but changed depending on the concentration and the specific species composition. So was the position the same for 0 mg/L Zn and 1.82 mg/L Zn, and for 0.45 mg/L Zn and 11.35 mg/L Zn but this differed between the two mixed settings ( Figure 1). The bottles were kept in an incubator (Appendix Figure S4) at 20°C with a light regime of 16/8h light/dark for 17 days. During the experiment, the species competed for the same nutrients in an uncrowded culture, but there was no competition for light.
The number of fronds was counted eight times over the course of 17 days. Completely white (dead) fronds were not included in the total number of fronds. During the experiment, the species got mixed in a small number of bottles. Due to their morphological similarity, the species could not be reliably distinguished anymore and consequently any data from these bottles post the mixing event were excluded.

| Data analysis
Four populations were excluded from all analyses because they de- Subsequently, population growth rates for each individual population (n = 104) were calculated as ln(N 2 /N 1 )/(t 1 − t 2 ). We used the initial population abundance at the start of the experiment (t 1 , 20.07.2021) and the final time point after 17 days (t 2 , 6.8.2021) to calculate total growth rates. In the mixed setting, we calculated growth rates for each species separately. Because most other studies examining the influence of Zn on duckweed growth were conducted over less than 10 days, we additionally calculated initial growth rate comparing the first day of the experiment (t 1 ) with day 8 (t 2 , 28.7.2021). Initial growth rates represent growth rates for when we can be confident that nutrients were not limiting but nutrients were likely not limiting throughout the experiment due to the combination of small populations sizes and the nutrient-rich medium used in the experiment.
Linear models with abundance over time and total growth rate as response terms and position as explanatory variable showed that there was, on average, no significant effect of position (inside vs. outside area, Appendix Figure S1). Due to the study design (Figure 1), position was confounded with the Zn × composition interaction; however, this interaction was never significant (see Appendix Table S1).
Total and initial growth rates were then used as response variables in the linear models. We did analyses of variances (ANOVAS) for each species separately but also the full linear models (see Appendix). The treatment variables were species identity (L. gibba

| RE SULTS
Over all concentrations and compositions, we found that L. turionifera had the highest total growth rate, followed by L. minor and L. gibba (Appendix Figures S2, Appendix Figures S3). However, initial growth rate was greatest for L. minor (Appendix Figures S2, Appendix Figures   S3). All three species showed an overall high tolerance to Zn (Figure 2).
L. turionifera was not influenced by Zn pollution. In contrast, the addition of ZnSO 4 to the experimental cultures significantly influenced L. gibba and L. minor but only at the beginning of the experiment The setting (isolated vs. mixed) only significantly influenced L. gibba growth rates (Table 1), specifically, L. gibba profited from growing alone and showed significantly lower growth rates (initial and total) when paired with either L. minor or L. turionifera (also mirrored TA B L E 1 Summary of the Type 3 ANOVAs showing the influence of the setting (isolated vs. mixed), the Zn concentration (factorial), and the composition (isolated, pairing with species 1, pairing with species 2) on the three study species Significant (<.05) and near-significant (<.06) p-values are in bold. The linear mixed models for setting and composition included position (outer vs. inner) as random factor, the linear mixed model for concentration included composition as random factor. Within species, interaction terms were never significant and thus not shown here (but see Appendix Table S1). numDF: numerator degrees of freedom, dDF: denominator degrees of freedom. 2 0 .J u l 2 1 .J u l 2 2 .J u l 2 3 .J u l 2 4 .J u l 2 5 .J u l 2 6 .J u l 2 7 .J u l 2 8 .J u l 2 9 .J u l 3 0 .J u l 3 1 .J u l 0 1 .A u g 0 2 .A u g 0 3 .A u g 0 4 .A u g 0 5 .A u g 0 6 .A u g 2 0 .J u l 2 1 .J u l 2 2 .J u l 2 3 .J u l 2 4 .J u l 2 5 .J u l 2 6 .J u l 2 7 .J u l 2 8 .J u l 2 9 .J u l 3 0 .J u l 3 1 .J u l 0 1 .A u g 0 2 .A u g 0 3 .A u g 0 4 .A u g 0 5 .A u g 0 6 .A u g 2 0 .J u l 2 1 .J u l 2 2 .J u l 2 3 .J u l 2 4 .J u l 2 5 .J u l 2 6 .J u l 2 7 .J u l 2 8 .J u l 2 9 .J u l 3 0 .J u l 3 1 .J u l 0 1 .A u g 0 2 .A u g 0 3 .A u g 0 4 .A u g 0 5 .A u g 0 6 .A u g  Figure 4). The positive and negative effects of the three species on each other are summarized in Table 2.

| Zn tolerance was high in all three studied species
Contrary to expectations, growth rates were not highest in the treatments without the addition of Zn. Low levels of Zn even increased growth rates for two of the species studied (L. gibba and L. minor).
For L. minor, the results partially confirm previous work. Jayasri and Suthindhiran (2017) showed that L. minor increased biomass yield in fronds treated with a lower concentration (0.5 mg/L) of Zn by 30% compared to the control. However, in their 4-day experiment 10 mg/L had little effect on growth (Jayasri & Suthindhiran, F I G U R E 3 Growth rates across the entire time series (total growth rate, a) and growth rates during the first 8 days of the experiment (initial growth rates, b) of three species separately and in the two settings across the Zn gradient. Orange: isolated setting, blue: mixed setting. Shown are means and associated standard errors across replicates and, in the mixed setting, across the two pairings 2017). Only 20 mg/L started inhibiting growth of L. minor. This contrasts our results that found that 11.35 mg/L reduced growth rates of L. minor (significantly so in comparison with 1.82 mg/L). However, our findings also suggest that at lower concentrations Zn promotes the growth of L. minor, but inhibits its growth at concentrations higher than 10 mg/L. Lahive, O'Callaghan, et al. (2011) reported that L. minor tolerated Zn concentration above 100 mg/L, but specific biomass growth rate was reduced significantly at low concentrations of 3 mg/L (reduction by 20%). Whether some of these discrepancies could also have been due to different culture mediums used The duration of our experiment exceeded that of previous experiments, which were conducted over 4-7 days (e.g., Jayasri & Suthindhiran, 2017;Lahive, O' Halloran, et al., 2011;Megateli et al., F I G U R E 4 (a) Growth rates across the entire time series and (b) for the first 8 days of the experiment (initial growth rates) of the three species for each composition separately. Means and associated standard errors across all Zn treatments (including no Zn) are presented. For initial growth rate, the interaction between composition and concentration was not significant, but for total growth rate it was (Appendix Table S1). Dashed lines are drawn to visualize the difference between the isolated setting and the two different pairings    (Landolt, 1986).

| Mixing two species can have contrasting effects on the focal species
Based on the stress-gradient hypothesis (e.g., Callaway et al., 2002), we hypothesized that in a more stressful environment we would observe more facilitative interactions while in the medium without Zn pollution we would observe more competitive interactions. We did not find evidence for this hypothesis, since the levels of Zn pollution we used in our experiment were not high enough to impact duckweed growth rates. Instead, we found that species interactions were highly dependent on focal species identity and competitor identity. For example, the effect of the setting on growth rate was only consistent for one of the three species, L. gibba. However, contrary to expectations, it did not profit from growing in the presence of a second species that could have helped it to accumulate Zn. Instead, L. gibba grew the best alone. Previous research with L.
gibba competing against a different duckweed species, the greater duckweed Spirodela polyrhiza, showed that L. gibba was highly competitive due to its gibbous fronds which could overgrow S. polyrhiza (Clatworthy & Harper, 1962). However, in our experiment there was no physical contact between the species and L. gibba only produced non-gibbous fronds. The outcome of species pairings could, therefore, be dependent on frond morphology, which in turn is dependent on the environmental conditions. Gibbosity in L. gibba depends on the environmental conditions, and may be restricted to optimal growth conditions, including very high nutrient availability (Vaughan & Baker, 1994). Under our conditions in the experiment, L. gibba may have had a disadvantage due to the lack of gibbous fronds and the inability to overgrow the competitor.
For the other two species, growth performance depended on the pairing, but there was no evidence of facilitation. Instead, we observed strong competition between L. minor and L. turionifera, with L. minor outcompeting L. turionifera. To the best of our knowledge, this is the first study investigating L. turionifera growth rates in the presence or absence of a competitor. Recently, it has been found that some populations previously thought to be L. minor were in fact L. turionifera (Senevirathna et al., 2021). It is thus possible that L. turionifera has a wider geographic distribution than previously thought and is, therefore, a good candidate for phytoremediation in many regions of the world.
For L. minor, we found that, even though it profited from growing with L. turionifera, the presence of L. gibba had a negative impact on its growth. Simultaneously, L. gibba was negatively influenced by the presence of L. minor. Thus, this pairing had a negative effect on both interacting species, which is surprising given that they frequently coexist in nature. Due to their morphological similarity, L. gibba and L. minor have rarely been used in competition experiments. In instances where they did compete in experiments the analysis was not completed (Clatworthy & Harper, 1962;Wołek, 1972), or the two species were even grouped together (Peeters et al., 2016). As an exception, Rejmánková (1975) found that in gibbous form, L. gibba was always the stronger competitor, overgrowing L. minor (Rejmánková, 1975).
Duckweeds have great potential for heavy metal removal in wastewaters (Abdel-Gawad et al., 2020) but many questions remain, in particular in terms of how polycultures of multiple species may improve efficiency. Our study is a first assessment of the interaction between competition and tolerance to Zn pollution for a subset of Lemnaceae species. We do acknowledge that there might be large variation among ecotypes (genotypes) from the same species (Ziegler et al., 2015). Thus, a next step should be to test the accumulation of Zn not only for species in an isolated setting across different metal concentrations (Lahive, O'Callaghan, et al., 2011) but also in a mixed setting and, importantly, in a natural environment with duckweed populations representing also natural genetic variation.
In addition, due to their fast growth, it is possible that duckweed species will evolve in response to their competitor, especially when there is intraspecific genetic diversity present (Hart et al., 2019). For future phytoremediation efforts, species mixing could be interesting, but the effects of a second species need to be evaluated first as they cannot be readily predicted from the performance in individual cultures.

ACK N OWLED G M ENTS
We thank Prof. Walter Lämmler for his efforts in maintaining the Landolt duckweed collection and for providing us with the strains used in this experiment. Thanks to Owen Petchey and the whole Petchey group for advice and Yves Choffat for technical support in the lab. Sofia Vámos provided helpful feedback on the manuscript.
We are grateful for the support of the URPP Global Change and Biodiversity of the University of Zurich. This study was funded by the University of Zurich (Forschungskredit 31259 awarded to S.J.V.M.).

CO N FLI C T O F I NTE R E S T
The authors declare no conflict of interest.