A king and vassals' tale: Molecular signatures of clonal integration in Posidonia oceanica under chronic light shortage

Under unfavourable conditions, clonal plants benefit from physiological integration among ramets, sharing resources and information. Clonal integration can buffer against environmental changes and lets the plant clone work as a ‘macro’ organism. Molecular signals that regulate this phenomenon are completely unknown in marine plants. Here we present a first comprehensive study providing insights into the metabolic role of different types of ramets (i.e. apical vs. vertical) in the foundation species Posidonia oceanica. Plants were exposed to 80% diminishing irradiance level (LL) in a controlled mesocosm system. Subsequent multiscale variations in whole transcriptome expression, global DNA methylation level, photo‐physiology, morphology and fitness‐related traits, were explored at different exposure times. We tested the hypothesis that vertical shoots (the ‘vassals’) can provide vital resources to apical shoots (the ‘kings’) under energy shortage, thus safeguarding the whole clone survival. Whole transcriptome analysis of leaves and shoot‐apical meristems (SAMs) emphasized signatures of molecular integration among ramets, which strongly correlated with higher organization‐level responses. In both shoots types, the exposure to LL resulted in a growth slowdown throughout the experiment, which started from immediate signals in SAMs. In apical shoots, this was linked to an acclimative response, where they were suffering a mild stress condition, while in vertical ones it fell in a more severe stress response. Yet, they suffered from sugar starvation and showed a clear cellular stress response in terms of protein refolding and DNA repair mechanisms. Several epigenetic mechanisms modulated the observed gene‐expression patterns and the cross‐talk between DNA methylation and the cellular energetic status appeared to regulate shoot metabolism under LL. Synthesis. Our results demonstrate a high level of specialization of integrated ramets within seagrass clones and a ‘division of labour’ under adverse conditions. Vertical shoots appear to do ‘most of the job’ especially in terms of resource providing, whereas activated functions in apical shoots were restricted to few important processes, according to an ‘energy‐saving’ strategy. The response of vertical shoots could be seen as a ‘sacrificing response’ allowing the survival of ‘the king’ that is key for ensuring propagation and population maintenance, and for the colonization of new environments.


| INTRODUC TI ON
Plants are modular organisms. When such modules are capable of iterating themselves in an independent manner, thus producing offspring through vegetative propagation, the plant is referred to as clonal (Liu, Liu, & Dong, 2016). Clonally formed offspring are called 'ramets', whereas the whole plant, which can comprise a number of clonal ramets, is referred to as a 'genet' (Harper, 1977). Different ramets belonging to the same genet share the same genotype (Harper, 1977). Within a genet, each ramet has the potential to perform all biological functions, and can be regarded as an independent individual.
Clonal plants dominate diverse terrestrial and marine ecosystems as primary producers, and include many of the most important crops and invasive plants, and some of Earth's largest, tallest and oldest plant species (Douhovnikoff & Dodd, 2015). Clonal integration, that is, the physiological integration taking place among the different ramets for sharing resources and information, is a striking attribute of clonal plants. It plays a crucial role in their ecological and evolutionary success, enabling them to act as a cooperative system (Liu et al., 2016). This is possible since ramets are physically linked to each other through horizontal structures (e.g. rhizomes or stolons) allowing the translocation of various material, including external resources absorbed by plants (e.g. water and nutrients), hormones, photosynthates and secondary metabolites, via interconnected vascular structures (Liu et al., 2016). Clonal integration permits plants to cope with spatio-temporal heterogeneity of the environment.
For example, within a single genet, donor ramets situated in favourable microsites (e.g. with abundant resource supply) can help resource-poor or otherwise adversely placed ramets, to alleviate their shortages (e.g. shading, nutrient depletion and drought) and/ or to tolerate abiotic and biotic stressors (Liu et al., 2016). This has often been observed from parent ramets (older) to offspring ramets (younger/developing), although reciprocal exchange of resources between neighbouring ramets growing in differing quality patches, has also been described (Alpert, 1999). Ultimately, resource sharing through clonal integration results in an increased performance of the recipient part without decreasing that of donor parts (at least in the short term), thus leading to an increased performance of the whole clone (Song et al., 2013). Numerous studies have shown that clonal integration can support ramets to survive in stressful environments, for instance under high salinity (Evans & Whitney, 1992;Pennings & Callaway, 2000) and soil alkalinity stress (Zhang, Yang, Sun, Chen, & Zhang, 2015), or to withstand defoliation by herbivores (Schmid, Puttick, Burgess, & Bazzaz, 1988;Wang et al., 2017).
Seagrasses are clonal rhizomatous plants sharing a similar morphology to that of terrestrial monocotyledons. All seagrass species present a highly organized growth based on the reiteration of ramets, which are composed of a bundle of leaves, a piece of rhizome and a root system. Rhizomes are stems extending either horizontally on (or below) the sediment surface or vertically, raising the leaves towards (or above) the sediment surface (Marbà, Duarte, Alexandre, & Cabaço, 2004). Besides providing mechanical support and nutrient storage, rhizomes are responsible for the extension of the seagrass clone in the space, as well as for connecting adjacent ramets, thus enabling physiological integration (Marbà et al., 2004). Shoots growing vertically and horizontally are called orthotropic (vertical) and plagiotropic (or apical if terminal) shoots respectively. Seagrass beds typically have wide spacing between many vertical shoots with few horizontal apices and are able to spread through those apices (Terrados, Duarte, & Kenworthy, 1997a), which grow horizontally until space has been completely colonized. Plagiotropic shoots can revert into vertical, which leads to the cessation of their horizontal growth, or vertical shoots can branch to produce horizontal ones, when the apical meristem of the original horizontal rhizome dies (Marbà et al., 2004). Rhizome elongation rate, leaf production and turnover are far higher in apical shoots than vertical ones (Marbà & Duarte, 1998). The Mediterranean seagrass Posidonia oceanica has dimorphic rhizomes; hence, it possesses both horizontal (plagiotropic or apical) and vertical (orthotropic) rhizomes, whereas other species such as Zostera spp. have only horizontal rhizomes.
Clonal integration has been demonstrated in seagrasses, for example in the form of nitrogen and carbon translocation among neighbouring ramets (Marbà et al., 2002). Photosynthates and nutrients can be re-allocated within seagrasses mainly towards organs with high metabolic activity, including growing leaves, flowering shoots and remarkably apical shoots, thus resulting in an enhanced clone growth and meadow spreading (Harrison, 1978;Libes & Boudouresque, 1987;Marbà, Hemminga, & Duarte, 2006;Marbà et al., 2002;Schwarzschild & Zieman, 2008a, 2008bTerrados, Duarte, & Kenworthy, 1997b). Clonal integration supports seagrass persistence, ameliorating adverse effects of environmental stressors. For example, Tuya and colleagues Tuya, Viera-Rodríguez, et al., 2013) demonstrated that the preservation of clonal integration in Cymodocea nodosa buffered its physiological performance against shoots could be seen as a 'sacrificing response' allowing the survival of 'the king' that is key for ensuring propagation and population maintenance, and for the colonization of new environments.

K E Y W O R D S
clonal integration, DNA methylation, energy shortage, marine angiosperms, molecular indicators, RNA-Seq, shoot apical meristem small-scale burial events and nutrient enrichment, similarly to what observed for Thalassia testudinum under localized light limitation (Tomasko & Dawes, 1989). The importance of clonal traits was also revealed in Zostera noltii grown under low-light conditions and organic matter enrichment (Olivé, García-Sánchez, Brun, Vergara, & Pérez-Lloréns, 2009). Specifically, a differential plant response was observed when contrasting levels of organic matter and light were established between the plant apex and distal parts, with harmful effect of organic matter being alleviated when the apex was grown in high light. This demonstrated that apical shoots represent the leading parts of the plant, and are highly sensitive to light deprivation (Olivé et al., 2009).
Light availability is by far the most important factor controlling seagrass growth, survival and depth distribution (Lee, Park, & Kim, 2007;Ralph, Durako, Enríquez, Collier, & Doblin, 2007). This is attributed to the fact that the minimum light requirement for seagrasses is one of the highest among all angiosperms. Underwater irradiance attenuation occurs naturally along several gradients, namely the bathymetric, the canopy and the leaf-epiphytic gradients. In addition, light attenuation may occur indirectly through excess anthropogenic nutrients leading to eutrophication, increased sediment accretion and resuspension, aquaculture and dredging as well as regional weather and oceanic swell patterns (Ralph et al., 2007). Light shortage, due to natural and/or anthropogenically driven processes, can compromise the photosynthetic process and ultimately lead to seagrass loss, as already documented worldwide (Ralph, Tomasko, Moore, Seddon, & Macinnis-Ng, 2006;Short & Wyllie-Echeverria, 1996). Seagrass responses to light limitation at multiple level of organization, from molecular to physiological and morphological levels, and across various spatial scales, from leaf to meadow scale, have been deeply addressed Dattolo et al., 2014;Davey et al., 2018;Kumar et al., 2017;Olesen, Enríquez, Duarte, & Sand-Jensen, 2002;Ralph et al., 2007). Nonetheless, the differential response to light limitation of specific shoot types, and its fundamental implications for the survival of the whole clone, has never been addressed so far, especially in terms of molecular and cellular rearrangements.
The present study aims at disentangling the relationship between apical and vertical shoots in the foundation species P. oceanica undergoing energy deprivation. To this end, we analysed photo-physiological properties, morphology and fitness-related traits (i.e. leaf growth rate and carbohydrate content) of those shoots under chronic low light (LL). Molecular mechanisms underlying such responses were explored through whole transcriptome profiling and global DNA methylation analyses of the different ramets. Our hypothesis is that molecular signals of clonal integration would be seen when the transcriptome profile of the two shoot types is compared under energy shortage, emphasizing unique biological roles for apical shoots ('the kings'), those responsible for colonization and population maintenance through clonal extension, and vertical shoots ('the vassals'), which should provide vital resources for ensuring the whole clone survival to the stress event.

| Experimental design and light treatment
For this study, large P. oceanica fragments bearing several vertical shoots and at least one apical shoot, were collected by SCUBA diving from a shallow-water meadow (8-10 m depth) located around the island of Ischia (Gulf of Naples, Italy; 40°43.849′N, 13°57.089′E) on 16th February 2018 (11:00-12:00 p.m.). Winter P. oceanica plants were chosen to intensify stress responses and shorten response times, as they possess little stored energy in the form of carbohydrates. Plant material was kept in darkened coolers filled with ambient seawater and rapidly transported to the laboratory (within 1-2 hr) to be transplanted in the indoor mesocosm facility of Stazione Zoologica Anton Dohrn (Naples, Italy; Figure 1a; see Ruocco, De Luca, Marín-Guirao, & Procaccini, 2019 for an in-depth description of the system). Twenty-four plant fragments of similar size and shoot number (15-25 connected shoots) were selected to standardize the experiment, and individually attached to the bottom of 12 plastic net cages (40 cm × 30 cm × 10 cm) filled with coarse sediment (two fragments per pot). Two randomly selected cages were then placed in each of six glass aquaria (500 L). Large fragments of P. oceanica were used to ensure unaltered clonal integration and healthy conditions of plants and to resemble the canopy structure of the meadow (Ruocco, De Luca, et al., 2019).
Prior to start the experimental treatment, plants were acclimated for 10 days (t 0 ) to mean prevailing environmental conditions of the sampling site during the study period (temperature: c. 16.5°C; salinity: 37.5; max. noon subsurface irradiance: c. 200 μmol photons m −2 s −1 ; 11 hr:13 hr light:dark photoperiod). Subsequently, irradiance level in half of the tanks (n = 3) was lowered to 40 μmol photons m −2 s −1 for simulating a strong shading event (80% light reduction), while lamps of control tanks (n = 3) were maintained at c. 210 μmol photons m −2 s −1 (Figure 1b). Both values represent max. noon irradiance levels. Temperature and salinity levels were left as in the acclimation phase (T: c. 16.5°C; salinity: 37.3-37.7). Continuous light and temperature measurements were performed by means of LI-COR LI-1400 quantum sensor and HOBO ® Pendant ® UA-002-64 data loggers (Onset Computer Corporation) respectively. Salinity was measured daily in each aquarium using a WTW Cond 3310 portable conductivity meter and kept within the range indicated above by adding freshwater to compensate for evaporation.
The LL exposure lasted 40 days. Molecular, physiological and morphological assessments were carried out at definite time points, according to their expected response timing (Macreadie, Schliep, Rasheed, Chartrand, & Ralph, 2014). Morphological parameters and leaf growth rate were assessed throughout the experiment (i.e. after 15 days-t 1 , 30 days-t 2 , and 40 days of exposure-t 3 ), whereas carbohydrate content was only determined at the end of the experiment (t 3 ). Shoot photosynthetic performance and global DNA methylation were determined at t 1 and t 2 . Genome-wide transcriptome analysis was performed on leaves and shoot-apical meristems (SAMs) of apical and vertical shoots at t 1 , in order to capture the early activation of plant molecular stress signals, likely anticipating physiological and ultimately morphological changes (Ceccherelli et al., 2018;Macreadie et al., 2014). To remove age-related effects, vertical shoots employed for the analyses were selected at least three positions after the apical one/s. Examples of ramets used for this experiment are shown in Figure 1c.

| Shoot morphology and fitness-related traits
A set of vegetative variables (i.e. shoot size, number of leaves per shoot, max. leaf length and width) and fitness-related traits (i.e. leaf growth rate and non-structural carbohydrate content) were determined at the selected time points (see above) on apical and vertical shoots under control and chronic LL. To determine the leaf growth rate, all apical and vertical shoots of one rhizome per tank (at least one apical and three vertical shoots) were marked at the beginning of each experimental phase (t 0 , t 1 and t 2 ) following the Zieman method (Zieman, 1974). Marked fragments were harvested at the end of each experimental phase (t 1 , t 2 and t 3 ) to determine the surface of newly formed tissue below the needle mark (as cm 2 shoot −1 day −1 ).
The total content of non-structural carbohydrates (TNC; soluble sugars and starch) was analysed in leaf and rhizome tissues, to assess the energetic status of the shoots under light shortage. Analyses were conducted following the phenol-sulphuric method modified from DuBois, Gilles, Hamilton, Rebers, and Smith (1956). Briefly, leaf samples (central sections of second-rank leaves) and the first 2 cm of the rhizome apexes were dried and finely ground (c. 50 mg) with a Mixer Mill MM300 (QIAGEN) and tungsten carbide beads (3 mm).
TNC were then solubilized by three sequential extractions with 80% (v/v) ethanol at 80°C for 15 min. After centrifugation (3,000 rpm in an Eppendorf 5810 R centrifuge-rotor A-4-62 for 10 min), the ethanol extract was used for the determination of soluble sugars' content, while the pellet was hydrolysed for starch determination (24 hr at RT) with 3 ml NaOH 0.1 M. For both soluble sugars and starch, 3% aqueous phenol (0.25 ml) and 95%-97% H 2 SO 4 (2.5 ml) were mixed with 1 ml sample in glass tubes and the solution was allowed to rest for 30 min. Absorbance was then read at 490 and 750 nm with an Agilent 8453 UV-Vis spectrophotometer. TNC content was calculated using sucrose calibration curves (standard sucrose 99%, Biorad). Apical and vertical shoots were only analysed for TNC content at t 3 (n = 3).

| Effective quantum yield
Chlorophyll a fluorescence measurements were performed with a diving-PAM fluorometer (Walz, Germany). The saturation pulse method was used to measure F and F m ′ in apical and vertical shoots of plants after 5 hr of illumination in aquaria. The effective quantum yield of PSII (ΔF/F m ′ = F m ′ − F/F m ′) was calculated as a proxy of plant productivity since it reflects the photosynthetic performance of plants. Chlorophyll a-derived photosynthetic measurements were determined on two apical and two vertical shoots per tank, then values were averaged to be used as individual replicates (i.e. no. of replicates used in statistical tests n = 3; total biological replicates N = 6).

| Global DNA methylation
Leaf material for DNA extraction was obtained from the middle section of second-rank leaves of one apical and one vertical shoot per tank (n = 3). Leaf tissue (about 5-7 cm) was accurately cleaned of epiphytes and dried with silica gel. Genomic DNA was subsequently isolated and quality checked following Ruocco, De F I G U R E 1 (a) View of the experimental system at Stazione Zoologica Anton Dohrn; (b) daily irradiance at the top of the leaf canopy measured with LI-COR LI-1400, in control (light blue) and low-light (dark blue) tanks over a 24-hr cycle; (c) example of vertical and apical shoots of Posidonia oceanica used for this experiment

| Statistical analysis
Two-way GLM repeated measures ANOVAs (RM-ANOVAs) were conducted to detect the effects of shoot type and treatment on plant morphological characteristics, leaf growth rate, effective quantum yield and global DNA methylation, along the course of the experiment. The analysis consisted of two fixed factors: 'Shoot Type' (ST), with two levels (apical and vertical) and 'Light' (L), with two levels (control and low light), and 'Time' as a within-subject factor. This allowed us to avoid the potential non-independence of measurements from the same aquaria. Carbohydrate contents at the end of the experiment (t 3 ) were assessed by two-way ANOVAs with the same fixed factors. Normality of the data was tested using the Shapiro-Wilk test and variance homogeneity was verified using Levene's test. In the case of RM-ANOVAs, the assumption of sphericity was assessed using Mauchly's sphericity test. When parametric assumptions were not met, data were Box-Cox transformed.
Student-Newman-Keuls post hoc test was used whenever significant differences were detected. All ANOVAs were performed using the statistical package STATISTICA (StatSoft, Inc. v. 10).

| RNA extraction, library preparation and sequencing
Leaf sub-samples (c. 5 cm) for RNA extraction were obtained from middle section of mature leaves (third-rank leaf) of vertical and apical shoots (n = 3). In addition, the first most apical 0.5 cm of the rhizome tip, containing the SAM, were also collected from the same shoots (n = 3). Leaf material was gently cleaned from epiphytes and submerged in RNAlater © tissue collection (Ambion, Life Technologies), stored one night at 4°C, and finally stowed at −20°C. Rhizome fragments were cleaned from leaf sheaths and sediment particles and then preserved in LN2 to be definitely stored at −80°C. For both plant organs, total RNA was extracted and checked for purity following Ruocco, De Luca, et al. (2019).
RNA concentration was accurately determined by Qubit ® RNA BR assay kit using the Qubit 2.0 Fluorometer (Thermo Fisher Scientific). RNA quality was assessed by measuring the RNA Integrity Number (RIN) with an Agilent 2100 Bioanalyzer (Agilent Technologies, Inc.); only high-quality (RIN ≥ 7) RNA was used for RNA-Seq analysis. Twenty-four indexed cDNA libraries (2 shoot types × 2 organs × 2 treatments × 3 biological replicates) were constructed with the Illumina TruSeq ® Stranded mRNA Library Prep Kit, and sequenced with an Illumina NextSeq 500 platform (single-ends 1 × 75 cycles) at Genomix4life s.r.l. (Salerno, Italy).

| Functional annotation, differential expression and GO enrichment analysis
Assembled contigs were annotated through sequence similarity search against UniProtKB/Swiss-Prot and National Center for Biotechnology Information (NCBI) non-redundant sequence (Nr) protein databases using BLAST+ tool v2.6.0 (Altschul et al., 1997; e-value cut-off 1e −6 ).
Subsequently, results were loaded on Blast2GO v.5 (Conesa et al., 2005) to retrieve Gene Ontology (GO) terms (e-value cut-off 1e −6 ). Enzyme code (EC) annotation and KEGG maps were also retrieved. Full description of transcriptome assembly and annotation results can be retrieved from Supporting Information (Section 1).
For the differential gene-expression analysis, reads from each biological replicate were individually mapped to the assembled transcriptome using Bowtie v1.1.1 (Langmead, Trapnell, Pop, & Salzberg, 2009), and the expression of each transcript was quantified using the expectation-maximization method (RSEM; Li & Dewey, 2011). Finally, differentially expressed genes (DEGs) for each pairwise comparison were determined using a generalized linear model (GLM) in the edgeR package (Robinson, McCarthy, & Smyth, 2010). The bulk of low-abundance transcripts were removed keeping those having at least a ≥1 cpm (read/count per million) for at least three samples. Transcripts were considered significantly differentially expressed (up-and down-regulated) if FDR-corrected p value < 0.05. Due to the high number of DEGs, a more stringent cut-off of logFC > ±2 and FDR < 0.05 was also applied.
Expression values generated by edgeR were used for examining profiles of expression across different samples through a hierarchical clustering. A heatmap of DEGs was generated using the heatmap3 package in r v3.2.2 (R Core Team, 2015). To assess overall similarity across samples, their relationships were also explored through a PCA on the transposed normalized expression matrix with r v3.2.2.
Venn diagrams to identify shared and unique DEGs between different contrasts were performed with http://bioin forma tics.psb.ugent. be/webto ols/Venn/. GO-term enrichment analysis of DEGs was performed through the Fisher's exact test approach by using the GO enrichment analysis function provided by Blast2GO v.5 with a threshold FDR of 0.05. Due to a large list of enriched GO terms was obtained in SAM comparisons, a further reduction to most specific terms (FDR < 0.01) was carried out. Summarization and visualization of GO terms were performed by using the REVIGO web service (http://revigo.irb.hr/; Supek, Bošnjak, Škunca, & Šmuc, 2011).

| Morphological and photo-physiological responses of apical and vertical shoots to chronic LL
On average, total shoot size was higher in vertical than apical shoots (Table 1). Apical shoots generally contained a significantly higher number of leaves per shoot and a lower maximum leaf length (Table 1). Both variables decreased in apical and vertical shoots under chronic LL (Table 2), although such variations were not significant at any sampling time points. LL had mild effect on max. leaf width, with a significant reduction observed at t 1 (~5%) and t 3 (~6%-7%; Tables 1 and 2), and no significant differences between shoot types. LL exposure caused a global reduction in shoot size that was especially evident after 30 and 40 days of exposure, for both apical (~31%) and vertical (~26%-27%) shoots (Tables 1 and 2).
At photo-physiological level, effective photochemical efficiency (ΔF/F m ′) was significantly reduced by 30% and 39% in LL plants at t 1 and t 2 , respectively, without any differences between apical and vertical shoots (Figure 2; Table 2).

| Effects of chronic LL exposure on fitness traits of apical and vertical shoots
LL greatly slowed down leaf growth rate, as it was significantly reduced at all sampling time points, in shaded with respect to control plants ( Figure 3; Table 2). Specifically, already after 15 days of exposure to 40 μmol photons m −2 s −1 , leaf growth was reduced by 50% and 41% in apical and vertical shoots respectively ( Figure 3).
After 1-month exposure, a further decline of up to 62% and 55% in apical and vertical shoots, was observed (Figure 3). At the end of the experiment (t 3 , 40 days of exposure) leaf growth rate of apical shoot was 78% lower than controls, whereas for vertical shoots Shoot size (cm 2 /shoot)

Max leaf width (cm)
15 days (t 1 ) TA B L E 1 Plant morphological characteristics in apical and vertical shoots at t 1 , t 2 and the end of the exposure (t 3 ). Values are means (SE) for n = 3 TA B L E 2 Two-way RM-ANOVAs to assess the effect of shoot type (ST) and low-light treatment (L) on plant morphological and photophysiological characteristics along the course of the experiment. p < 0.05 are in bold, p < 0.

Apical
Changes in effective quantum yield (ΔF/F m ′) of apical and vertical shoots at t 1 and t 2 . Data are mean ± SE (n = 3). Black and grey bars represent controls and low-light (LL) plants respectively. Asterisks indicate significant differences between control and LL conditions. Full results of two-way RM-ANOVAs are reported in Table 2. *p < 0.05 F I G U R E 3 Changes in leaf growth rate of apical and vertical shoots at t 1 , t 2 and t 3 . Data are mean ± SE (n = 3). Black and grey bars represent controls and low-light (LL) plants respectively. Asterisks indicate significant differences between control and LL conditions. Different letters indicate significant differences between shoot types, for each light treatment. Full results of twoway RM-ANOVAs are reported in Table 2. **p < 0.01, ***p < 0.001 the decrease was still around 50%, similarly to what observed at t 2 ( Figure 3). Overall, the reduction in leaf growth rate was always greater in apical than vertical shoots ( Figure 3).
The factor 'shoot type' had a significant effect on total non-structural carbohydrate (TNC) accumulation in leaf tissues, yet there was a significant interaction with the factor 'light' (ST × L), as depicted by two-way ANOVAs (Table 3). Specifically, at t 3 while TNC content in leaf tissue of vertical shoots exposed to LL fell down dramatically (42%, p < 0.05 SNK), in apical shoots it showed values significantly higher than vertical shoots (50%, p < 0.05 SNK) and even slightly higher than those displayed by their own control (ns; Figure 4; Table 4). The same pattern was not visible in rhizomes (Figure 4).

| Effects of chronic LL on global DNA methylation levels of apical and vertical shoots
Overall, the two-way RM-ANOVA highlighted a significant

| General description of differential geneexpression patterns in apical and vertical shoots
The profile of expression across different samples at gene level was firstly explored through a hierarchical clustering ( Figure S3a) Figure S3b).

| Transcriptomic response of apical and vertical leaves to LL
Even though a reduced total number of DEGs was identified in vertical leaves in respect to apical ones under LL (see above), a substantial higher number of enriched GO terms (as biological processes, GO-BPs) was associated with the former contrast. With a total of 36 GO-BP terms, the transcriptome reprogramming observed in leaves of vertical shoots under LL appeared to be much more complex and multifaceted than that of apical shoots, which was restricted to a total of 13 GO-BP enriched terms (Table S4). Yet, few enriched biological functions were in common between the two contrasts, while a large part of them was specifically associated to the response of apical or vertical shoots (Table S4).

| Shared response of apical and vertical leaves
Among shared DEGs identified between apical and vertical contrasts, it is worth noticing the presence of transcripts involved in key plant metabolic processes, such as photosynthesis, chlorophyll biosynthesis and glycolysis/gluconeogenesis, as significantly downregulated under LL (Table S3). Similarly, some transcripts encoding for amino acid, oligopeptide and nitrate transporters were among top downregulated ones (Table S3). Shared upregulated genes included some transcripts for stress-related proteins and interestingly GALACTINOL-SUCROSE GALACTOSYLTRANSFERASE 6, which is known to be induced by dark (Table S3).

| Specific response of apical leaves
Exclusive GO enriched BPs in apical leaves included plant hormonerelated signalling pathways and response to plant hormones (negative regulation of cytokinin-activated signalling pathway and response to abscisic acid), and secondary metabolite-related metabolic processes (pigment biosynthetic process, regulation of phenylpropanoid metabolism, flavonoid metabolic process and geranylgeranyl diphosphate biosynthetic process; Figure 6a; Table S4). Most transcripts specifically associated with hormone signalling pathways were transcriptional factors, and were generally upregulated in LL. In addition, hormone receptors like ABSCISIC ACID RECEPTOR PYL 8 and some genes related to the auxin-activated signalling pathway (e.g. AUXIN EFFLUX CARRIER 8) were also overexpressed. The same was observed for proteins involved in secondary metabolite biosynthesis, such as terpenes and anthocyanins (Table S3).
Among downregulated genes, it is worth remarking the presence of two transcripts involved in phototropism and photoperiodism (

| Specific response of vertical leaves
As commented above, a higher number of GO enriched BPs was recognized in vertical leaves under LL. Among these, the most significant ones (FDR < 0.01) were 'cellular protein modification processes, regulation of transcription, oligopeptide transport, cellular response to stress and negative regulation of cytokinin-activated signalling pathway' (Figure 6b; Table S4). Several other GO-BPs were enriched at FDR < 0.05 for example, amino-acid import, phloem nitrate loading, developmental growth involved in morphogenesis, organic hydroxy compound metabolic process, positive regulation of proteolysis, secondary metabolic process and regulation of response to external stimulus ( Figure 6b; Table S4).
Overall, transcripts involved in phloem nitrate loading, oligopeptide/amino acid transport, as well as carbohydrate transport, were downregulated in LL (Table S3). Upregulated transcripts were mostly included in the GO categories regulation of transcription, cellular protein modification, cellular stress response and response to ethylene.
Several transcripts with a role in protein repair were found overexpressed in LL, including many chaperones and chaperone regulators, members of the universal stress and LEA protein families, as well as some proteins involved in DNA damage response (Table S3). Lastly, the enzyme SUCROSE SYNTHASE 4, a fundamental sucrose-cleaving enzyme, was found among overexpressed transcript in LL.

| Transcriptomic response of apical and vertical SAMs to LL
Similarly to what observed for leaves, a higher number of DEGs was recognized in the SAM of apical rather than vertical shoots were shared between the two contrasts, where the remaining part was specifically associated with the response of apical or vertical shoots (Tables S5 and S6).

| Shared response of apical and vertical SAMs
Surprisingly, many structural and functional components involved in the photosynthetic process, chlorophyll biosynthesis and carbonassimilation pathways, were identified as differentially expressed in the transcriptome of SAMs. The vast majority of these DEGs were strongly downregulated under LL in both shoot types. Among downregulated transcripts were photosystem subunits, electron transportrelated proteins and proteins assisting photosystem assembly and repair (

| Specific response of apical SAMs
In the meristem of apical shoots, fundamental responsive functions were enriched, as for example, those related to plant development  Table S5). Among DEGs included in these categories, many of them were downregulated, including transcripts encoding for cyclins and transcriptional factors (Table S3).

| Specific response of vertical SAMs
Under LL, a significant higher number of BPs was enriched in the meristem of vertical shoots (see Figure 7b; Table S6). Top enriched functions included those related to chloroplast assembly and arrangement of constituent parts (chloroplast organization, chloroplast RNA modification and chloroplast RNA processing) that were not identified in apical shoots. A vast majority of transcripts involved in these processes were down-expressed in LL, for example the PALE CRESS protein, which is required for chloroplast differentiation, RNA POLYMERASE SIGMA FACTOR SIGE or TOC75-3, which is an essential protein required for the import of protein precursors into chloroplasts. Other unique GO-BPs were those related to sugar responses

F I G U R E 7 Graphical depiction of enriched GO-BPs in shoot-apical meristems of apical (a) and vertical (b) shoots under low-light
(FDR < 0.01). GO-BP terms are coloured by semantic similarity to other GO terms and bubble size reflects the abs_log10_pvalue of the GO-term in the Fisher test. The two-dimensional semantic space was generated by the REVIGO web service and signalling (e.g. cellular response to sucrose starvation, sugar-mediated signalling pathway and glucose metabolic process) and amino acid metabolism (regulation of cellular amino acid metabolic process and branched-chain amino acid catabolic process). Enzymes with a role in sucrose starvation were generally overexpressed in LL, as were genes involved in sugar-mediated signalling pathway (Table S3).

| D ISCUSS I ON
Here we explored for the first time at the molecular level, the differential behaviour of apical and vertical shoots in the foundation seagrass species P. oceanica under chronic light shortage. Our hypothesis was that under an energetic crisis, metabolic rearrangements occurring in vertical shoots ('the vassals') would be devoted to provide resources for the apical one ('the king'), representing the leading part of the clone, to withstand the unfavourable event (Liu et al., 2016).
Following this view, the response of vertical shoots could be seen as a 'sacrificing response' allowing the survival of 'the king' that is key for ensuring propagation and population maintenance, and for the colonization of new more favourable environments (i.e. 'escape' strategy). Our multiscale analysis of physiological, morphological and molecular plasticity under LL exposure, allowed finding signatures of clonal integration between the two shoot types, in support of our initial hypothesis.

| Physiological and morphological evidences supporting the 'king and vassals' hypothesis
Under chronic LL, apical and vertical shoots showed a similar response in terms of photosynthetic performance, with a large reduction (about 30%-40%) in the effective quantum yield of PSII (ΔF/F m ′) that was especially evident after 1-month exposure. This is a typical response of P. oceanica to diminish irradiance level, in agreement with previous observations (Dattolo et al., 2014Procaccini et al., 2017). However, the response of apical and vertical ramets starts to diverge when analysing shoot morphology and fitness-related traits under light shortage. Although both shoot types greatly slowed down their leaf growth rate throughout the experiment, for apical shoots this was in a much greater extent with respect to vertical ones, reaching around 80% reduction after 40 days of exposure.
Ultimately, this resulted in a larger reduction in their maximum leaf length and width and overall shoot size at the end of the experiment.
Structural changes as the reduction of plant size are considered the main adaptive mechanisms to offset light reductions in large-sized seagrass species, aimed at maximizing light exposure of photosynthetic tissues and minimizing respiratory demands (Collier, Lavery, Ralph, & Masini, 2008;Dattolo et al., 2017;Olesen et al., 2002;Ralph et al., 2007). Here LL-induced growth arrest and decrease in leaf biomass would allow apical shoots to save fundamental resources (e.g. sugars) needed to withstand the temporary stress event (Kosová, Vítámvás, Prášil, & Renaut, 2011;. This is coherent with our assessment of the energetic status of the two shoot types by measuring the total content of non-structural carbohydrates. After 40 days of exposure to extreme LL, TNC content in leaf tissue of apical shoots was even slightly higher than the control, while in vertical shoots it dropped down dramatically, being around half of that measured in apical ones. Although this pattern was not confirmed in rhizomes, probably because the duration of the experiment was too brief to see an effect in these organs, this suggests the possibility of a translocation of photosynthates between ramets, as previously demonstrated in seagrasses (Harrison, 1978;Marbà et al., 2002Marbà et al., , 2006Terrados et al., 1997b) and terrestrial clonal plants (e.g. Duchoslavová & Jansa, 2018;Qian, Li, Han, & Sun, 2010). As apical shoots did not increment their photosynthetic performance under LL, the 'conservative' strategy they put in place in terms of growth arrest and reduction of leaf biomass, together with a possible translocation of photosynthates from neighbouring vertical ramets, would have allowed maintaining their sugar reserves constant all along the experiment.

| Molecular signatures of clonal integration under energy shortage
Our comparative transcriptome analysis revealed different BPs enriched in apical and vertical shoots exposed to chronic LL, and only a small portion of shared BPs. Surprisingly, although a higher number of DEGs was generally found in both leaves and SAMs of apical shoots, those were associated to a reduced number of GO-BPs, with respect to vertical ones. This clearly demonstrates a high level of specialization of integrated ramets within seagrass clones and a 'division of labour' of different shoot types under adverse conditions, as observed in terrestrial plants (Stuefer, During, & de Kroon, 1994). Overall, under light shortage, the response of apical shoots appeared to be less complex and restricted to few important functions, whereas that of vertical shoots was more heterogeneous and involved a wide variety of processes. Below, we discuss the main gene categories/cellular functions associated with physiological integration mechanisms among ramets.

| Hormone response/signalling
In the leaf transcriptome of apical shoots, there were only a few enriched BPs under LL. Top GO terms were those associated with the regulation of phytohormone signalling pathways and response to hormones (e.g. negative regulation of cytokinin-activated signalling pathway) and many transcripts within these categories were upregulated. In addition, several transcripts involved in the auxinactivated signalling pathway and transport, were also among top overexpressed genes. Cytokinin-related signalling pathway was also enriched in vertical leaves under LL. In terrestrial systems, there is considerable evidence that hormones can cause differences in biomass of plant parts in response to different resource availability, and can also regulate translocation between branches (Voesenek & Blom, 1996). In particular, two major types of plant hormones, auxins and cytokinins, can direct the transport of carbohydrates and nutrients between plant parts (Alpert, Holzapfel, & Benson, 2002 (Bielach, Hrtyan, & Tognetti, 2017;Kuhn, Geisler, Bigler, & Ringli, 2011;Peer & Murphy, 2007). By controlling the processes of phytohormone transport and distribution, flavonoids could indirectly modulate patterns of resource accumulation in P. oceanica ramets under LL (Bielach et al., 2017;Peer & Murphy, 2007).

| Stress response
In agreement with our hypothesis, GO-terms related to the cellular stress response (CSR; sensu Kültz, 2005)  suggesting that molecular chaperoning, DNA repair and apoptosis inhibition processes are the instant stress signals exerted by the species (Traboni et al., 2018).

| Growth arrest and sucrose starvation
The negative regulation of functions related to cell growth and proliferation is one of the key responses of plants to non-lethal abiotic/ biotic stressors and it has a high adaptive value, as the resultant plant is more likely to survive (Kitsios & Doonan, 2011). Our study revealed that LL slowed down shoot size and leaf growth rate in both shoot types, although with a different extent. Notably, LL-induced growth arrest started at the level of meristems, as clearly highlighted from the underlying gene-expression responses. In general, the decrease of plant size can be attributed both to a reduction in cell number as well as cell growth (Kitsios & Doonan, 2011). In apical shoots, both ways can be postulated, as many GO-terms related to these processes were enriched (e.g. regulation of G2/M transition of mitotic cell cycle, mitotic metaphase plate congression, cytokinesis by cell plate formation, cell wall modification involved in multidimensional cell growth).
In particular, the reduction in cell number can be attributed to the observed suppression of cell cycle-related transcripts (e.g. cyclins), resulting in cell-cycle arrest at the G1/S and G2/M checkpoints, prolonged S-phase progression and/or delayed entry into mitosis (De Veylder, Beeckman, & Inzé, 2007).
In vertical shoots, we found a similar pattern, although this did not seem to be related to an acclimative response. The inactivation of genes required for cell-cycle progression can arise from the activation of DNA stress checkpoints, which also induces DNA repair-related genes. This coordinated action ensures that cells repair their damaged genome before they proceed into mitosis (De Veylder et al., 2007). The analysis of vertical SAMs revealed the presence of many GO terms associated with DNA damage, and many genes related to these processes were found overexpressed in LL (see above).
The inhibition of cell proliferation and cell-cycle arrest observed in these shoots could actually result from the interplay between CSR mechanisms and sucrose starvation (Riou-Khamlichi, Menges, Healy, & Murray, 2000;Yu, 1999).
Sucrose is the major organic carbon form exported from source to sink organs (Rosa et al., 2009) and a key modulator of cell division rates, as its availability to proliferating meristematic cells reflects the overall photosynthetic capacity and prevailing environmental conditions (Koch, 1996). In addition, sucrose acts as a signalling molecule modulating gene expression (Koch, 1996;Yu, 1999). In SAMs of vertical shoots, these functions were particularly represented (e.g. cellular response to sucrose starvation and sugar-mediated signalling pathway), and this is consistent with our assessment of TNC However, what appears to be very clear is the effect of sucrose starvation on nitrate and amino acid metabolism. In sugar-starved cells, a decrease in enzymatic activities related to nitrate reduction/ assimilation and protein synthesis is generally observed to protect cells against nutrient stress, together with an overexpression of genes related to amino acid catabolism, as an alternative way to sustain respiration and metabolic processes (Yu, 1999

| Response of the SAM to LL: A new early warning indicator in seagrass research?
Under LL, the transcriptomic response of apical and vertical SAMs was always much greater than that identified in leaves.
Surprisingly, fundamental processes such as photosynthesis, car- revealing an early acquisition of photosynthetic capacity (Dalal et al., 2018). Another important consideration is that LL exposure seems to significantly impair the SAM transcriptional machinery responsible for chloroplast biogenesis and later for the establishment of photosynthetic competence, and these gene-expression changes likely anticipated leaf-related responses.
Other enriched functions in SAMs included those related to meristem growth and maintenance, development of plant organs/tissues (e.g. phloem/xylem histogenesis), as well as DNA damage/repair. These functions were only marginally represented in leaves. Ultimately, this experiment revealed that the stress response in P. oceanica exposed to chronic LL starts primarily at the level of meristems, which appeared to be the most sensitive plant parts, with the lowest tolerance threshold.
SAMs are fundamental structures ensuring organogenesis over the whole plant's life. Accordingly, plants evolved special mechanisms to safeguard the genome integrity of these cell niches through cell-cycle arrest, DNA repair and at last selective programmed cell death programmes that are different from those of differentiated cells (Fulcher & Sablowski, 2009;Hefner, Huefner, & Britt, 2006).
From an ecological perspective, this opens a new view where the SAM-related response could be considered a fundamental indicator of seagrass stress status. If further studies will demonstrate that the molecular response of SAM to other abiotic/biotic stressors occurs not only in a greater extent, but also in much earlier than leaves, the role of the latter should be reconsidered. Specific protocols should be developed to exploit this key plant organ and to identify suitable target genes to be used as early-warning molecular monitoring tools (Macreadie et al., 2014).

PEER R E V I E W
The peer review history for this article is available at https://publo ns.