Contrasting Effects of Plant-Soil Feedbacks on Growth and Morphology of Two Clonal Plants

Aim Soil abiotic and biotic conditions are often spatially variable, challenging plants with a heterogeneous environment consisting of favorable and unfavorable patches of soil. Many stoloniferous clonal plants can escape from unfavorable patches by elongating stolon internodes, but aggregate in favorable ones through shortening stolon internodes. However, whether these plants can use their stolons to respond to plant-soil feedbacks (PSFs) is largely unknown. Methods


Introduction
Plants can change properties of the soil where they grow, and these changes can in turn in uence the performance of other conspeci c or heterospeci c plants either positively or negatively (Bever 1994;Ehrenfeld et al. 2005;van der Putten et al. 2013). These interactions, also known as plant-soil feedbacks, can facilitate subsequent plant performance through soil nutrient accumulation (Berendse 1990;Chapman et al. 2006) and symbiotic mutualist development (Klironomos 2002;van der Putten et al. 2016), or inhibit it through soil nutrient depletion (Berendse 1994) and pathogen accumulation (van der Putten et al. 2016). However, the negative feedbacks are more commonly observed in nature (Kulmatiski et al. 2008). A large number of plants can propagate vegetatively by producing connected genetically identical individuals (ramets) via clonal growth, and these clonal plants are predominant in many local communities (Klimes et al. 1997;Song and Dong 2002;Klimešová et al. 2018). As offspring ramets of clonal plants are commonly produced near their mother ramets, plant-soil feedbacks may affect both the performance of mother and offspring ramets (Cartenì et al. 2012;D'Hertefeldt and van der Putten 1998;van der Stoel et al. 2002).
Many clonal plants can produce rhizomes or stolons of strong plasticity in response to environmental variation in terms of their internode length (de Kroon and Hutchings 1995;Xie et al. 2014). Plasticity of stolon and rhizome internode (hereafter also referred to as "internodes" for brevity) length allows clonal plants to have a higher opportunity to encounter resource-rich patches, and consequently a greater ability to adapt to heterogeneous environments (de Kroon and Hutchings 1995;Gao et al. 2012;Roiloa et al. 2014;van Kleunen and Fischer 2001;Xue et al. 2018c;Ye et al. 2006). In general, clonal plants would produce shorter internodes to utilize adequate resources in favorable habitats, while longer ones can help to "escape" from unfavorable patches (Benedek et al. 2017;de Kroon and Schieving 1990;Dong 1993;Si et al. 2020;Oborny and Cain 1997;Ye et al. 2006). This morphological response in clonal plants has been extensively linked to external environmental factors such as nutrient, water and light availabilities (Dong 1993;Hagiwara et al. 2010;Wijesinghe et al. 2001). However, it is still unclear whether plant-soil feedbacks are involved in such growth strategies. For example, if a clonal plant experiences negative plant-soil feedbacks, it may produce longer internodes to facilitate its offspring ramets to escape from the local stressful environment. By contrast, if a clonal plant experiences positive plant-soil feedbacks, it may produce shorter internodes to maximize use of local soil resources.
When different ramets of the same clone are located in patches of contrasting resource quality, ramets growing in resource-rich patches can translocate nutrients, water and carbohydrates to the ramets growing in resource-poor patches via the connection of internodes (Alpert 1991;Herben et al. 1994;Song et al. 2013;Xue et al. 2020;Yu et al. 2004). The ability to translocate resource among interconnected clonal plants, also known as clonal integration, can bene t certain section(s) of clonal plants experiencing various environmental stresses, including submergence (Luo et al. 2014;Xiao et al. 2010), sand burial and wind erosion (Yu et al. 2004), salinity (Hester et al. 1994Xiao et al. 2011) andshading (Li et al. 2018;Stuefer et al. 1994;Xu et al. 2010). Similarly, clonal integration may also bene t clonal plants by reducing effects of negative plant-soil feedbacks or by enhancing the effects of positive plant-soil feedbacks (D'Hertefeldt and van der Putten 1998). However, connection between ramets may also increase the risks of spreading harmful soil microbes (e.g. pathogens) rapidly across connected clones (D'Hertefeldt and van der Putten 1998;Vannier et al. 2018), which may suppress the growth of the clonal plants. Therefore, the response of a clonal plant to plant-soil feedbacks may also depend on the positive and negative effects of biotic plant-soil feedbacks of other connected clonal plants via physiological integration.
Here, we conducted an experiment to examine the effects of biotic plant-soil feedbacks in two wellstudied clonal plants, Hydrocotyle vulgaris and Glechoma longituba Liu et al. 2017). We rst grew either Hydrocotyle vulgaris or Glechoma longituba clonal plants separately in mesocosms for three months to condition bulk soil. In the feedback phase, we grew connected mother and daughter ramets of each of the two species in soil inoculated with the unsterilized or sterilized soil conditioned by conspeci cs. Speci cally, we tested three hypotheses, for both clonal species: (1) daughter or mother ramets grown in bulk soil inoculated with sterilized (conspeci c-conditioned) soil will show greater growth compared with those grown in bulk soil inoculated with unsterilized (conspeci c-conditioned) soil, due to accumulation of species-speci c soil-borne pathogens in the unsterilized soil (i.e. negative conspeci c plant-soil feedbacks); (2) daughter or mother ramets grown in bulk soil inoculated with unsterilized soil will grow better when connected to ramets growing in bulk soil inoculated with sterilized than with unsterilized soil, due to clonal integration; (3) daughter or mother ramets grown in bulk soil inoculated with unsterilized soil will show greater internode growth to show escape response from these unsterilized soil patches.

Materials And Methods
The species Hydrocotyle vulgaris L. (Araliaceae) is a perennial clonal herb that commonly occurs in moist habitats such as bogs, valleys, dune grasslands and moorlands (Dong 1995). It can produce creeping stems with many nodes; each node has the potential to develop into a ramet consisting of a single leaf and adventitious roots (Dong et al. 2015). This species exhibits rapid clonal reproduction and high morphological plasticity (Dong et al. 2015;Dong 1995). Glechoma longituba (Nakai) Kupr. (Lamiaceae) is a perennial clonal herb that can produce ramets connected by stolons. Each ramet has two single leaves originating from a stolon node and some adventitious roots (Liao et al. 2003). The plant is widely distributed in grasslands and forests, on roadsides or by creeks (Liao et al. 2003).

Soil conditioning phase
To create bulk soil, we collected soil from a barren hill in Taizhou, Zhejiang Province, China, air-dried and being sieved (2-cm-mesh) and then homogeneously mixed with river sand at a 1:1 volume ratio. The bulk soil was used to ll eight pots (1.5 L), each with 3.5 kg, and grown with plants for soil conditioning. Before lling the pots, we placed a piece of non-woven ber at the bottom of each pot to avoid soil from running out the pots.
On 28 June 2019, we collected ramets of both H. vulgaris and G. Longituba in the campus of Taizhou University in Jiaojiang District, Taizhou, Zhejiang Province, China. We planted these ramets individually in cells (6.5 cm × 6.5 cm × 6.0 cm) on the seedling plates lled with sterilized potting soil (Hebei Dewoduo Fertilizer Co., LTD, Hengshui, China). Su cient water was supplied daily. Ten days after transplantation, we selected 20 similar-sized ramets of each species and planted them into the pots. The remaining cultivated ramets were allowed to propagate vegetatively until the start of treatments in the feedback phase as described below.
For each species, we planted ve ramets in each pot ( Fig. 1) and grown for a period of 12 weeks to condition the soil. Each plant species had four replicates, making a total of 8 pots. We replaced dead ramets during the rst week of the treatment. All pots were watered daily to promote plant growth.
After 12 weeks, plants from each pot were harvested and large roots were removed from the soil by hand before being homogenized. After which, the soil from each pot was divided into two equal parts. One part of the soil was sterilized by autoclaving at 121 ℃ for 120 minutes, and the other part was not sterilized. In total, 16 soil samples were created at the end of the conditioning phase (2 species × 2 sterilization treatments × 4 replicates).

Feedback phase
Each of the 16 soil samples from the conditioning phase was homogenized with the sterilized bulk soil (autoclaving at 121 ℃ for 120 minutes) at a ratio of 1:9 (W:W), resulting in eight soil samples inoculated with the sterilized soils conditioned by either of the two plant species and another eight soil samples inoculated with unsterilized soils conditioned by either of the two plant species. Each of these 16 soil samples were treated as experimental "blocks" and further divided to ll two pots (12 cm in diameter and 10 cm in height) and two polyvinyl chloride (PVC) trays (100 cm long × 15 cm wide × 8 cm high; Fig. 1). Four treatment combinations for each plant species were created using these lled pots and trays containing soil inoculated with sterilized or unsterilized soil conditioned by conspeci cs: 2 pot soil inoculum treatments (sterilized, unsterilized) × 2 tray soil inoculum treatments (sterilized, unsterilized).
Each of these soil treatment combinations were replicated four times, with each set of pot and trays containing sterilized or unsterilized soil inoculum originating from the same soil sample obtained from the end of the conditioning phase.
For each of the two clonal plants, we selected 16 similar-sized ramets (mother ramets) that had developed a primary stolon with two nodes (potential offspring ramets), from the cultivated ramets reserved for the feedback phase. Each mother ramet was planted in a pot containing either unsterilized or sterilized soils conditioned by conspeci cs (Fig. 1), and its connected potential daughter ramets by the stolon were grown in a tray containing either unsterilized or sterilized soils conditioned by conspeci cs (Fig. 1). Any further primary stolons produced by the mother ramets in the pots and secondary stolons (i.e., side-branches developing from the primary stolon) produced by ramets in the trays were not allowed to root and left trailing from the pots and trays. The feedback phase was maintained for 12 weeks in the greenhouse. We replaced dead ramets during the rst week of the experiment. During this experiment, the daily mean temperature in the greenhouse was 23 °C. All pots were watered every two days.
At the end of the experiment, we harvested leaves, stolons and roots for mother (in pots) and daughter ramets (in trays) separately. We also counted number of ramets and measured total stolon length. Internode length was calculated by using total stolon length divided by the number of nodes along the stolons. All plant materials were oven-dried at 70 °C for at least 48 hr and weighed.

Data analysis
We used a linear mixed-effects model to analyze each of the measured variables (i.e., total mass, root mass, stolon mass, leaf mass, number of ramets and internode length). In this model, inoculum treatment (inoculated with the unsterilized vs. sterilized conspeci c-conditioned soil) of the mother ramets, inoculum treatment of the daughter ramets and their interaction were included as xed factors. Pot identity in the conditioning phase was included as a random factor in order to account for the nonindependence of the inoculum. We performed separate linear mixed-effects models for measurements of the daughter ramets, the mother ramets and the whole clone (daughter + mother ramets). In these analyses, signi cant local effects of inoculum treatment of the daughter (or mother) ramets on the growth of the daughter (or mother) ramets indicate signi cant plant-soil feedback effects; while signi cant nonlocal effects of inoculum treatment of the daughter (or mother) ramets on the growth of the mother (or daughter) ramets indicate clonal integration effects (Stuefer et al. 1994). When a signi cant effect was detected, Tukey tests were used for post-hoc comparisons.
All analyses were performed with R (version 3.4.4; http://www.r-project.org) in RStudio (version 1.1.423; http://rstudio.org). Linear mixed-effects models were tted with the nlme package (version 3.1-128; Pinheiro et al., 2016); Post-hoc comparisons were made using the glht function in the multcomp package (version 1.4-15). All data were checked graphically for normality and homogeneity of variance. Total mass, root mass, leaf mass and internode length of the whole clone of H. vulgaris and total mass of the whole clone of G. longituba were log-transformed before analysis.

Effects on growth and morphology of H. vulgaris
In H. vulgaris, biomass (total mass, root mass, stolon mass and leaf mass) of the daughter ramets was signi cantly greater when they were grown in trays containing sterilized conspeci c-conditioned soil (LS and SS) than when they were grown in trays containing unsterilized conspeci c-conditioned soil (LL and SL; Fig. 2A-D; Table 1A), indicating that there was a signi cant negative plant-soil feedback effect on the growth of the daughter ramets of H. vulgaris. However, the inoculum treatment of the mother ramet did not in uence the growth of the daughter ramets, indicating that there was no integration effect on the daughter ramets. Internode length of the daughter ramets was also greater when they were grown in trays containing sterilized conspeci c-conditioned soil (LS and SS) than when they were grown in trays containing unsterilized conspeci c-conditioned soil (LL and SL; Fig. 2F; Table 1A), indicating that daughter ramets of H. vulgaris did not show escape strategy in response to the negative plant-soil feedbacks. There was no treatment effect on the ramet number of the daughter ramets of H. vulgaris ( Fig.  2E; Table 1A).
The inoculum treatment of the mother ramet or daughter ramets did not in uence the growth or morphology of the mother ramet of H. vulgaris ( Fig. 2; Table 1B), indicating that there was no signi cant effect of plant-soil feedback or integration on the mother ramet.
The growth and morphology of the whole-clone showed a similar pattern as those of its daughter ramets, except for a signi cant interactive effect on root mass of H. vulgaris ( Fig. 2; Table 1).

Effects on growth and morphology of G. longituba
In G. longituba, the inoculum treatment of the daughter or mother ramets did not in uence the growth measures (total mass, root mass, stolon mass, leaf mass and number of ramtes) of the daughter ramets ( Fig. 3A-E; Table 2A), indicating that there was no signi cant effect of plant-soil feedback or integration on the daughter ramets of G. longituba. However, internode length of the daughter ramets of G. longituba was overall greater when the mother ramet was grown in pots containing unsterilized conspeci cconditioned soil (LL and LS) than when it was grown in pots containing sterilized conspeci c-conditioned soil (SL and SS; Fig. 3F; Table 2B), indicating a signi cant integration effect on the morphology of the daughter ramets.
There was no signi cant treatment effect on the growth or morphology of the mother ramets of G. longituba ( Fig. 3; Table 2B).
Total mass, root mass and internode length of the whole clone of G. longituba were signi cantly greater when the mother ramet was grown in pots containing unsterilized conspeci c-conditioned soil (LL and LS) than when it was grown in pots containing sterilized conspeci c-conditioned soil (SL and SS; Fig. 3A, B and F; Table 2C). Stolon mass and leaf mass of the whole clone of G. longituba was signi cantly greater when the daughter ramets were grown in trays containing sterilized conspeci c-conditioned soil (LS and SS) than when they were grown in trays containing unsterilized conspeci c-conditioned soil (LL and SL; Fig. 2C-D; Table 2C). Ramet number of the whole clone was not in uenced by the inoculum treatments of the mother or daughter ramets ( Fig. 2E; Table 2C).

Discussion
Our results showed that the daughter ramets rather than the mother ramet of H. vulgaris experienced negative biotic plant-soil feedbacks, but no biotic plant-soil feedbacks were found for either the daughter or mother ramets of G. longituba. Moreover, there was no evidence of facilitation by the H. vulgaris mother ramets grown in sterilized conspeci c-conditioned soil to their connected daughter ramets suffered from negative biotic plant-soil feedbacks. More importantly, the daughter ramets of H. vulgaris did not show "escape" responses in terms of internode length from the negative biotic plant-soil feedbacks. These results indicated that biotic plant-soil feedbacks not only differed among different species, but also differed in physically-connected ramets of the same clone. However, physiological integration and stolon plasticity may not be involved in the response of the two clonal plants to the biotic plant-soil feedbacks as what they usually do in response to abiotic stresses (Gruntman et al. 2017;Roiloa et al. 2010;Yu et al. 2004). further revealed that biotic plant-soil feedbacks may also differ between the connected ramets of the same clone in H. vulgaris. These differences may have important implications for the thriving of the clonal plants in nature as this variation may "blur" the negative feedbacks on the whole clone (Hart et al. 2016).
Although interconnected, the daughter ramets were suppressed by the negative biotic plant-soil feedbacks, but the mother ramet was not. This result was contrast to a previous study showing that soilborne pathogens had greatly in uenced the mother ramets of a rhizomatous clonal grass Carex arenaria, but the daughter ramets were optimally defended (D'Hertefeldt and van der Putten 1998). This was explained by the transportation of resources (physiological integration) from the mother ramets to support their connected daughter ramets, as reported in also many other studies (Lu et al. 2020;Song et al. 2013;Wang et al. 2017). In the present study, however, we did not observe a signi cant facilitation of the mother ramet of H. vulgaris to its connected daughter ramets. Therefore, our second hypothesis was not supported. We proposed that local plant defense may have played a major role in driving the distinct responses of the mother and daughter ramets to biotic plant-soil feedbacks (Stuefer et al. 2004). The daughter ramets were more vulnerable in facing species-speci c soil pathogens, compared to the mother ramet, despite that they had access to more resources in terms of nutrients and spaces.
While H. vulgaris daughter ramets were affected by negative biotic plant-soil feedbacks, it did not produce longer stolon internodes. This result thus does not support our hypothesis that clonal plants can elongate their inter-ramet distance to facilitate their escape from negative plant-soil feedbacks. Studies have shown that clonal plants can elongate their internodes under stressful conditions such as low light intensity (de Kroon and Hutchings 1995), low nutrient availability (Thomas and Hay 2008) and drought (Zhao et al. 2008). However, under stressful conditions, stolon internodes can be shortened rather than elongated compared to non-stressful conditions, as plants may switch their development from vegetative growth to owering (Navas and Garnier 2002;Stuefer et al. 2004;Ye et al. 2006;Yu and Dong 2003). This is likely the case in our study, as we found that stolon internode of H. vulgaris was shortened under the negative effects of plant-soil feedbacks. This result indicated that negative plant-soil feedbacks may less far-reaching than stolon length. Therefore, it may still not be a problem for this species to be ourishing in natural circumstances despite that they cannot escape negative plant-soil feedbacks through elongating stolons, because they may also use the seeds to distribute to better conditions. We conclude that biotic plant-soil feedbacks can negatively in uence the growth of the daughter ramets, rather than the mother ramet, consequently reduce the growth of the whole clone. However, we did not nd the evidence that physiological integration can facilitate the growth of the daughter ramets suffered from negative feedbacks, and that the daughter ramets cannot escape from such negative feedbacks through elongation of internode length either. These results were largely attributed to the defenses of the daughter and mother ramets against their local species-speci c soil pathogens. However, we should notice that these results were only true for H. vulgaris, but not for G. longituba, and thus a general conclusion requires investigations with a large number of clonal plants. Moreover, we constrained the growth of the daughter and mother ramets in separate containers where the soil could be treated as homogeneous by the daughter and mother ramets. However, soil abiotic and biotic conditions are often spatially variable in natural circumstance, challenging plants with a heterogeneous environment consisting of more and less favorable patches of soil (Xue et al. 2018b Numbers are F-values of two-way ANOVAs. Superscripts give P: *** P < 0.001, ** P < 0.01 and * P < 0.05. Total mass, root mass, leaf mass and internode length of the whole clone were log-transformed. Table 2 Effects of the inoculum treatment (inoculated with the unsterilized vs. sterilized conspeci cconditioned soil) of the mother and daughter ramets on the growth and morphology of (A) the daughter ramets, (B) the mother ramets and (C) the whole clone (mother plus daughter ramets) of Glechoma longituba Numbers are F-values of two-way ANOVAs. Superscripts give P: ** P < 0.01 and * P < 0.05. Total mass of the whole clone was log-transformed. Figure 1 Schematic representation of the experimental design. The experiment consisted of two phases. In the conditioning phase, the soil in each pot was grown with ve ramets of Hydrocotyle vulgaris or Glechoma longituba for 12 weeks. In the feedback phase, for each species, a mother ramet was grown in a pot (circle) lled with a bulk soil inoculated with either a unsterilized or a sterilized conspeci c-conditioned soil collected from the conditioning phase, and its daughter ramets connected by a primary stolon were grown in a tray (rectangle) lled with the bulk soil inoculated with either a unsterilized or sterilized conspeci c-conditioned soil collected from the conditioning phase. Treatment codes: LL -both mother and daughter ramets were grown in the bulk soil inoculated with the unsterilized conspeci c-conditioned soil, LS -the mother and daughter ramets were grown in the bulk soil inoculated with the unsterilized and sterilized conspeci c-conditioned soil, respectively, SL -the mother and daughter ramets were grown in the bulk soil inoculated with the sterilized and unsterilized conspeci c-conditioned soil, respectively, and SS -both mother and daughter ramets were grown in the bulk soil inoculated with the sterilized conspeci c-conditioned soil. Conspeci c-conditioned soil inoculation treatments: LL -"mother" and "daughter" ramets grown in soil inoculated with unsterilized conspeci c soil, LS -mother with unsterilized and daughter with sterilised inoculum, SL -mother with sterilized and daughter with unsterilized inoculum, and SS -mother and daughter with sterilized inoculum. Conspeci c-conditioned soil inoculation treatments: LL -"mother" and "daughter" ramets grown in soil inoculated with unsterilized conspeci c soil, LS -mother with unsterilized and daughter with sterilised inoculum, SL -mother with sterilized and daughter with unsterilized inoculum, and SS -mother and daughter with sterilized inoculum.