Better together? Microbial communities associated with enhanced growth of the invasive seagrass Halophila stipulacea when mixed with a native Mediterranean seagrass Cymodocea nodosa

Abstract

Introduction:

Microorganisms are increasingly recognized as important contributors to the invasion success of introduced species into new environments. The globally invasive seagrass Halophila stipulacea flourishes in mixed species assemblages with native seagrasses, where it exhibits enhanced growth, while, in contrast, native seagrasses in mixed environments experience reduced growth. Here, we hypothesize that microbes may support the success of invasive seagrass in mixed Mediterranean environments.

Methods:

We analyzed 16S rRNA genes to characterize phyllosphere microbial diversity associated with the Mediterranean-native seagrass Cymodocea nodosa and the invasive H. stipulacea when grown alone or together, alongside biochemical, morphological, and sediment nutrient measurements.

Results:

Overall, C. nodosa in monoculture harbored a microbiome exhibiting higher ASV richness and a distinct community composition than H. stipulacea. Variation in bacterial diversity associated with hydrogen peroxide (H₂O₂) and internode length suggests that microbial communities of the native seagrass might be shaped by its stress-related physiological condition. Conversely, H. stipulacea’s microbiome was most abundant in mixed environments, with bacteria significantly reduced in monoculture, and bacterial diversity loosely associated with growth, suggesting that bacterial diversity patterns associated with H. stipulacea vary across competition contexts and may be associated with differences in performance in mixed environments.

Discussion/Conclusion:

Our findings suggest that invasive H. stipulacea in the Mediterranean Sea is capable of harboring distinct bacterial assemblages that may be associated with growth patterns observed when co-occurring with the native seagrass C. nodosa. Future work should center on the mechanisms driving H. stipulacea bacterial communities and investigating whether H. stipulacea actively determines its own microbiome, or whether it passively acquires its microbiome by environmental variables.

1. Introduction

Microbial communities are increasingly recognized as key players in plant invasion dynamics and competition (Rudgers et al., 2005; Zhang et al., 2010; Policelli et al., 2019; Elsheikh et al., 2021; Dawkins et al., 2022). They are essential to all forms of life and vital for providing resources to their hosts (Rout and Callaway, 2012; Dawson and Schrama, 2016), as well as driving processes that sustain ecosystems and the planet (Policelli et al., 2019; Zhang et al., 2020; Elsheikh et al., 2021; Hesse et al., 2021; Dawkins et al., 2022). Microbes can influence invasion success and plant competition through several key functions, including nutrient acquisition, stress tolerance, pathogen protection and resource availability (Inderjit and van der Putten, 2010; Dawson and Schrama, 2016; Dawkins et al., 2022). These functions are mediated by mechanisms such as the production of phytotoxins, phytohormones, secondary metabolites, disruption of native symbioses (Cipollini et al., 2012), and shifts in microbial communities that can suppress the growth of their competitors (Fang et al., 2022; Dar et al., 2024).

Studies of plant-microbe interactions in marine environments are much scarcer than those in terrestrial environments, yet suggest that microbes can be important factors in successful species invasions (Funk and Vitousek, 2007). For example, Gribben et al. (2018) found that sediment microbial communities influenced the growth of Caulerpa taxifolia fragments, with intact microbial communities from native seagrass sediments reducing fragment growth, while microbial communities associated with C. taxifolia sediments increased fragment growth, suggesting that sediment microbes can differentially influence invasion outcomes depending on their origin. Additionally, Aires et al. (2013) used microbes to track the invasion source of Caulerpa racemosa, suggesting that associated microbes may assist its acclimation to novel habitats. However, interactions between the host and its associated microbial communities can shift during the invasion process. For example, Saha et al. (2016) demonstrated that the macroalga Gracilaria vermiculophylla becomes vulnerable to epibionts from its native range, but showed increased resistance to epibionts in its invaded range. Growing evidence demonstrates that seagrasses’ microbial communities—particularly those associated with roots, rhizomes, and leaves—play a pivotal role in determining performance, nutrient acquisition, and responses to environmental change (Tarquinio et al., 2019).

Microbes confer beneficial advantages to seagrasses, including enhanced nutrient acquisition, protection against pathogens, biofouling, toxins, and oxidative stress; and promotion of plant growth and developmental processes (Ugarelli et al., 2017). Microbial mediation may help explain patterns of establishment and spread across seagrasses among the three known invasive seagrass species—Zostera japonica, a highly invasive seagrass in the Pacific Northwest whose associated microbes are detoxifying and denitrifying, growth-promoting, nitrogen-fixing, and epibiont-inhibiting (Wojahn, 2016; Crump et al., 2018); Halophila ovalis, whose associated microbes can fix carbon and nitrogen, cycle sulfur, promote growth and stress tolerance, and degrade C₁ products like methanol and methane (Yan et al., 2021); and the globally invasive Halophila stipulacea (Lipkin, 1975; Harrison and Bigley, 1982; Short et al., 2010). Originating from the Red Sea and Indian Ocean, H. stipulacea is a Lessepsian migrant that colonized the Mediterranean Sea 150 years ago after the opening of the Suez Canal in 1869. First observed in the Mediterranean in 1894 (Willette et al., 2014), H. stipulacea is believed to have subsequently been introduced to the Caribbean Sea via boat (Winters et al., 2020), with first sightings in the Caribbean reported in Grenada in 2002 (Willette et al., 2014).

As an opportunistic, phenotypically plastic species, H. stipulacea’s ability to tolerate varying light, depth, temperature, salinity, and hydrodynamic regimes (Rotini et al., 2017) has been associated with its high invasion success, although less work has been done examining how H. stipulacea’s microbial communities might influence this plasticity. To date, most evidence supports the host generalist hypothesis, which posits that H. stipulacea’s ability to adapt to the conditions of its invaded range comes from effectively exploiting the microbial communities of its invaded range (Parker, 2001; Aires et al., 2023), which has been proposed to assist its competition against native seagrasses in mixed communities. In syntopic populations in the Mediterranean Sea, Conte et al. (2023) found that H. stipulacea exhibited unaffected density and largely unaffected bacterial diversity when growing next to the native seagrass, while the native seagrass exhibited decreased density and bacterial diversity. In the Caribbean Sea, Aires et al. (2021) identified bacteria that putatively promoted growth, nutrient acquisition, stress and saline tolerance, detoxification, quorum sensing inhibition, and epiphyte degradation, potentially enabling H. stipulacea to thrive in nonnative territories. Further support comes from Winters et al. (2023), who found that Caribbean H. stipulacea exhibited higher growth rates and biomass than native Red and Mediterranean Seas populations, suggesting a shift in H. stipulacea’s physiology between native and invaded territories, which could potentially be microbe mediated. Put together, positive impacts on the seagrass holobiont, the relationship between microbes and their seagrass hosts, might enable invaders to be more effective competitors than their rivals by enhancing seagrass ecological functions (Ugarelli et al., 2017; Tarquinio et al., 2019; Conte et al., 2021).

H. stipulacea microbial communities may be impacted by its intrinsic characteristics, its environment, or the act of competition itself. It is well understood that plants can mediate their own microbial communities through various genomic, metabolic, and structural strategies (Pascale et al., 2020; Barnes et al., 2025). Cúcio et al. (2016) found that seagrass-associated microbial communities differed among host compartments at local scales, whereas broaderscale patterns were associated with environmental conditions. Similarly, Mejia et al. (2016) observed that H. stipulacea’s epiphytic bacterial communities varied along a depth gradient in the Gulf of Aqaba, and Rotini et al. (2017) found that variation in abiotic factors, including depth, was associated with differences in H. stipulacea microbial community structure.

Microbial communities may also vary in association with plant competition. In terrestrial systems, competition between native and invasive grasses can restructure rhizosphere communities, sometimes amplifying competitive asymmetries (LaForgia et al., 2022; LaForgia et al., 2022), while in crop–weed systems, microbial shifts can mediate allelopathic suppression (Mishra et al., 2013; Kohli et al., 2024). In marine systems, competition between macroalgae (Egan et al., 2017; Brunet et al. 2025) and coral species (Teplitski et al., 2016) has been linked to changes in surface-associated microbial communities. Taken together, these studies suggest microbiome communities can vary in association with competitive interactions. In seagrasses specifically, experimental studies have shown that root and sediment-associated microbial communities can affect plant performance under environmental stress and nutrient limitation (Fuggle et al., 2023; Jongen et al., 2026).

Despite evidence from the controlled competition experiment demonstrating enhanced performance of invasive H. stipulacea when growing with native seagrass (Chiquillo et al., 2022), the role of microbes in shaping these interactions remains unresolved. Across both the Mediterranean and Caribbean Seas, H. stipulacea tends to dominate mixed assemblages, exhibiting enhanced growth when co-occurring with native seagrasses compared to when it grows alone—often to the detriment of native species (Chiquillo et al., 2022). In contrast, native seagrasses Syringodium filiforme and Cymodocea nodosa lost shoots when grown with H. stipulacea at full density and only showed positive growth in monoculture. Invasive species capable of driving their own invasions are typically superior competitors (Fleming and Dibble, 2015). However, given the importance of microbes to plants (Barrow et al., 2008), microbial interactions may also contribute to the competitive performance of H. stipulacea. In this study, we examine how microbial communities differ when H. stipulacea is grown alone or together with native seagrasses, previously associated with enhanced H. stipulacea performance, and discuss whether the observed patterns are consistent with (1) context-dependent variation in H. stipulacea-associated microbial communities, (2) stochastic processes shaped by the environment, or (3) differences in microbial communities associated with H. stipulacea across growth conditions that may be associated with physiological responses.

The objective of this study is to understand whether microbial communities differ across competition contexts that are already known to enhance Halophila stipulacea performance in mixed Mediterranean seagrass communities. Using a controlled competition experiment, we compared microbial diversity, community composition, and biochemical indicators in both the invasive H. stipulacea and the native Cymodocea nodosa under monoculture and mixed conditions. We focus on phyllosphere-associated microbial communities, given that shoot dynamics responded strongly when the seagrasses grew mixed and alone. We hypothesized that (1) H. stipulacea would exhibit differences in microbial community composition between mixed species communities and monoculture, and (2) microbial assemblages would correlate with physiological traits such as oxidative stress and shoot growth. We therefore assessed microbial alpha and beta diversity, differential taxon abundance, and associations with environmental descriptors, interpreting these patterns as correlative with plant performance and physiological state. Our findings offer new insights into marine plant–microbe interactions and generate testable hypotheses regarding potential microbial-mediated mechanisms that may contribute to biological invasions in seagrass ecosystems.

2. Materials and methods

2.1. Experimental design of species assemblages

To examine associations between microbial communities and seagrass species interactions across treatments, we expanded upon a previously published experiment (Chiquillo et al., 2022), which manipulated seagrass community composition (monocultures and mixed communities) and density in controlled mesocosms to assess interactions between the native Cymodocea nodosa and invasive Halophila stipulacea. The experiment included four treatments with ten mesocosm replicates each: (1) native and invasive grown together at natural densities, (2) native and invasive together at reduced densities, (3) invasive monoculture, and (4) native monoculture. All mesocosms were initiated from naturally mixed seagrass communities containing both H. stipulacea and C. nodosa collected from naturally co-occurring donor beds. On SCUBA, intact seagrass cores were extracted with clear polycarbonate coring tubes (30 cm height × 7.2 cm diameter), each containing whole plants and associated epiphytes. Two intact cores were placed into each 24-L flow-through cylindrical mesocosm. Additional sterilized sand was added to standardize sediment depth (~15 cm) and allow continued rhizome expansion within each mesocosm.

Experimental treatments were then established by selectively modifying species composition from these naturally mixed communities. In the natural density treatment, sediments were disturbed to rhizome depth to mimic the disturbance associated with treatment establishment, but no seagrass tissue was removed. Invasive monocultures were created by removing C. nodosa, native monocultures by removing H. stipulacea, and reduced-density mixed treatments were established by removing approximately half of the shoots of both species while maintaining their co-occurrence. At the start of the experiment H. stipulacea averaged 5.9 ± 0.1 shoots, while the native seagrass began on average with 3.0 ± 0.2 shoots.

Experimental units were maintained in flow-through mesocosms for six weeks under natural conditions, from October 16 to December 3, 2018 at the Marine Aquaculture Center within the Fisheries Aquatic Research Center, Larnaka, Cyprus.

In this study, we investigate the microbial diversity and composition of the phyllosphere and seagrass tissue from each of these experimental treatments to test whether seagrass community structure and density impact associated microbial communities.

2.2. Plant descriptors

To evaluate how plant physiological responses shift in response to seagrass competition, as described in Chiquillo et al. (2022), samples were collected from both H. stipulacea and C. nodosa at the end of the competition experiment to record the biochemical, morphological (e.g., number of shoots), and soil variables. We measured photosynthetic leaf pigments, indirectly linked to photosynthetic activity in each treatment, by extracting photosynthetic pigments (chlorophyll a and chlorophyll b, total carotenoids) from 50 mg of blade tissue per treatment and species, prepared as composite samples from at least 10 randomly selected leaves following modified methods (Supplementary Table S1). Specifically, we determined chlorophyll a and b concentrations (mg/g of fresh weight, FW) from lyophilized samples after DMSO extractions, recording average absorbances (A) at 470, 652, 665, and 750 nm from duplicate readings on a Spectroquant Pharo 100 spectrophotometer. Carotenoids and chlorophyll concentrations were estimated using equations proposed by Sims and Gamon 2002.

To investigate if competition leads to oxidative stress, we measured lipid peroxidation as an indicator of cellular damage. We tested for both hydrogen peroxide (H₂O₂) and malondialdehyde (MDA) using tissue from five haphazardly selected blades from each mesocosm replicate per treatment and species; these blades were pooled to create three composite samples per treatment (~ 4.1-4.9 mg). While pooling reduces resolution at the individual blade level, it also minimizes variability among blades within each mesocosm replicate. Lipid peroxidation was determined from the measurement of MDA content resulting from the thiobarbituric acid (TBA) reaction (Minotti and Aust, 1987) using an extinction coefficient of 155 mM-1 cm-1. Hydrogen peroxide was quantified using the KI method, as described by Velikova et al. (2000).

Next, to determine whether growing conditions impacted seagrass growth, we measured growth by calculating the difference in the number of shoots (final (F) minus initial (I)) over the 6-week experiment from Chiquillo et al. (2022), which we used as a standardized proxy for vegetative growth, as well as above and belowground biomass and internode length taken from cores at the end of the experiment.

To determine nutrient levels in the sediment, we cored 50 mL falcon tubes in random locations within each replicate of each mesocosm and combined them to create a composite sediment sample for each treatment before and after the experiment. Additionally, to measure dissolved nutrients in the water, we collected five water samples from each mesocosm before and after the experiment. We filtered water using a vacuum pump and using 0.2uM pore size. We used Nitrite test kit (#1.14776.002), and Nitrate test in seawater (kit #1.14942.001). Extracts were read twice using Spectroquant Pharo 100 spectrophotometer.

2.3. Bacterial communities: DNA extraction and 16S rRNA amplicon sequencing of seagrass-associated bacterial community

To investigate how microbial communities might change based on its response to intraspecific seagrass competition conducted by Chiquillo et al. (2022), we sampled seagrass phyllospheres at two time points: the onset and conclusion of the experiment. At the start of the experiment, we grouped by treatment and mesocosm origin, collected one blade per species from each of the 10 experimental units, and then pooled blades into four composite samples per treatment. In the full mixed-density treatment (where both native and invasive grew together in natural density); in this treatment only sediment was disturbed to mimic the disturbance of the other treatments, but blades were not removed (n=16). At the end of the experiment, we again collected one blade of each experimental unit for each species from each treatment. Blades were grouped together and then randomly redistributed into four composite samples, tissue originating from multiple independent, experimental mesocosm units (n=24). This compositing strategy preserves treatment-level inference while reducing within-replicate technical variation among mesocosms; however, it does not resolve mesocosm-level heterogeneity.

To characterize bacterial communities, each composite sample was scraped from all the epibionts (including small pieces of tissue blade) used for DNA extraction. Tissue blades were gently scraped into sterile Petri dishes using forceps and sterile razors. Each sample was first washed with 10 ml of washing solution (200 mM Tris–HCl pH 8, 10 mM EDTA, and 0.24% Triton X-100; Kadivar and Stapleton 2003). During the scraping process, some tissue fragments were also released into the wash. The resulting supernatant was vortexed (3× 30 s) to detach and collect the bacteria. The plant parts were then removed with sterile tweezers, and the remaining solution was centrifuged at 5000 rpm for 20 min to obtain the recovered pellet, which may have contained residual plant material, for subsequent DNA extraction using the Qiagen Power Soil^® DNA isolation kit (Qiagen, Carlsbad, CA, USA), following the manufacturer’s instructions. A single blank DNA extraction using sterile water was conducted alongside sample extractions as a contamination control and carried through downstream processing.

To characterize the diversity and relative abundance of prokaryotes in the different treatments, we sequenced 16S rRNA V4 using primers 515F (GTGYCAGCMGCCGCGGTAA) and 806 R (GGACTACNVGGGTWTCTAAT) (Caporaso et al., 2011) as per the Earth Microbiome Project protocol https://earthmicrobiome.org/protocols-and-standards/16s/. Each sample was amplified in triplicate 25 uL reactions, using the Qiagen Multiplex PCR kit. Thermocycling parameters began with a denaturation step of 94 °C for 3 min, followed by 35 cycles of 94 °C for 45 sec, annealing at 50 °C for 60 sec, and extending at 72 °C for 90 sec, with a final extension of 72 °C for 10 min. Once validated on an agarose gel, triplicate reactions were pooled and cleaned using Agencourt AMPure magnetic beads. Further, we evaluated taxonomic composition with morphological, biochemical traits, and soil nutrient (nitrate/nitrite) assays, allowing us to identify microbial indicators associated with competitive outcomes.

2.4. Sequencing and bioinformatics

Pooled PCR samples were dual-indexed using the Illumina Nextera XT Indexing kit and then cleaned a second time with AMPure magnetic beads, including a PCR negative control (blank) to detect and remove any incidental DNA contamination introduced during the process. Indexed samples were then quantified using a Qubit dsDNA BR kit from Invitrogen to ensure that samples were pooled into equimolar ratios. The pooled library was subsequently sent for sequencing on an Illumina MiSeq v3 with a 20% PhiX at the TCGB core at UCLA (300 bp x 2).

To conduct quality control, amplicon sequence variant (ASV) taxonomy assignment, and community diversity analyses, we processed sequencing data using the QIIME 2 (v. 2024.5.0) microbiome data science platform (Bolyen et al., 2019). Briefly, we demultiplexed and denoised raw sequence data using the DADA2 pipeline (Callahan et al., 2016) and merged the output into a feature table for analysis. Sequences were then trimmed at 10 bp and truncated at 270 and 215 bp for the forward and reverse directions, respectively. Next, ASVs were assigned taxonomy using a naive Bayes taxonomy classifier trained on the SILVA v138.1 SSU NR99 database trimmed to the 515F/806R region (Quast et al., 2013), employing clustering at 99% similarity, and ASVs represented by fewer than 10 total reads were removed prior to downstream analyses. Rarefaction curves were used to determine whether we measure sufficient sequencing depth given the diversity within the sample. We removed reads that matched archaea, chloroplasts, eukaryotes, and mitochondria to ensure we analyzed only reads belonging to bacteria. Subsequent analyses were conducted in R (v.4.4.1) (R Core Team, 2024), using the `qiime2R` package to translate QIIME 2 artifacts (Bisanz, 2025).

16S rRNA microbial community composition in the phyllosphere was analyzed for both native and invasive seagrass species. Using observed ASVs, we calculated alpha diversity using Shannon diversity and Pielou evenness metrics with the `otuSummary` package (Yang, 2023). Differences in alpha-diversity metrics among groups were assessed using Kruskal–Wallis to evaluate treatment effects on these metrics. For alpha- and beta-diversity analyses, samples were rarefied to a sequencing depth of 1,500 reads per sample.

Using the same rarefied data set, beta diversity was assessed using Bray–Curtis dissimilarities. Community compositional patterns were visualized with non-metric multidimensional scaling (NMDS) using the vegan package in R (Oksanen et al., 2025). Overall differences in community composition among groups were tested using PERMANOVA with 999 permutations, and ANOSIM was used as a complementary multivariate test. Pairwise PERMANOVA comparisons were conducted using pairwiseAdonis (Martinez Arbizu, 2020) with 999 permutations. Where applicable, P values from pairwise comparisons were adjusted for multiple testing using the Benjamini–Hochberg procedure.

Because PERMANOVA can be sensitive to differences in within-group dispersion, homogeneity of multivariate dispersion was assessed separately using betadisper and permutation testing. To further evaluate whether observed differences reflected unequal multivariate dispersion among groups, we conducted a permutational test of homogeneity of multivariate dispersions ‘PERMDISP’ with 999 permutations (Anderson and Walsh, 2013). This allowed us to distinguish differences in community composition among groups from differences in within-group heterogeneity.

We created a principal coordinate analysis (PCoA) plot to explore how environmental factors, such as our plant descriptors, influenced bacterial community composition in our samples using the full n=18 samples. The plot and covariate statistics were produced using ‘vegan’, ‘ecodist’, and 999 permutations (Oksanen et al., 2025; Goslee and Urban, 2007), after which we ran PERMANOVA and an Analysis of similarity matrix (ANOSIM) tests to detect overall group differences and a pairwise PERMANOVA test to detect pairwise differences with ‘pairwiseAdonis’ (Martinez Arbizu, 2020) with 999 permutations. Core microbiomes were determined in R as taxa appearing within at least 80% of samples, as done in Aires et al. (2021). Venn diagrams comparing the core microbiomes of our samples were generated using the ‘venn’ package (Dusa, 2022). Heatmaps were then produced using ‘pheatmap’ (Kolde 2015) to visualize sample clustering and identify which taxa within the core microbiomes were most abundant. To visualize bacterial compositions of our samples, we created bar plots of relative taxonomic abundance using ‘ggplot2’ (Wickham et al., 2025).

To evaluate the robustness of diversity patterns to the normalization approach, we applied Scaling with Ranked Subsampling (SRS) using the q2-srs plugin in QIIME 2 (Heidrich et al., 2021). SRS normalization was performed on the feature table using default parameters, and rarefaction-comparison curves were generated to assess consistency with rarefaction-based approaches. Alpha diversity (Shannon diversity and Pielou evenness) and beta diversity (Bray–Curtis distances) were recalculated from the SRS-normalized table. Statistical analyses, including Kruskal–Wallis tests, PERMANOVA, ANOSIM, and betadisper, were conducted as described above to allow direct comparison with rarefied results.

To identify associations between microbial taxa and experimental variables, we used Microbiome Multivariable Associations with Linear Models (MaAsLin 2; Mallick et al., 2021), an R package designed to assess multivariable associations by integrating microbial data with a broad spectrum of experimental factors correcting for multiple comparisons. Taxon relative abundances were modeled as a function of treatment and associated metadata variables. Covariates included biochemical (e.g., H₂O₂, MDA, chlorophylls) parameters, morphological (e.g., internode length, number of shoots, biomass), and environmental factors. No additional filtering was performed. No additional prevalence or abundance filtering was applied before the MaAsLin2 analysis, and it was conducted on all 18 samples. Relative abundance data were used as input, and zero values were handled using MaAsLin2 default procedures (log transformation with the addition of a pseudo-count). Multiple testing correction was performed using the false discovery rate (FDR) (Benjamini-Hochberg), and association was considered significant using FDR q-values of 0.05. Microbial relative abundance was modeled as a function of treatment and associated metadata variable. Prior to model fitting, covariates included biochemical (e.g., H₂O₂), morphological (e.g., shoot number, internode length), and experimental treatment variables, depending on species-specific analyses were considered. All models were run using MaAsLin2 with default settings unless otherwise specified. To assess robustness given the small effective sample size, we conducted an additional sensitivity analysis using a reduced predictor set. For native samples, models included H₂O₂ and internode length as predictors, while for invasive samples, models included shoot number. Results from reduced models were qualitatively consistent with the full models, supporting the robustness of identified associations.

Taxonomic identities were used to infer potential ecological functions based on previously reported associations in the literature; however, 16S rRNA amplicon sequencing does not directly assess functional genes or microbial metabolic activity.

3. Results

3.1. Plant descriptors

Native C. nodosa had higher photosynthetic pigment levels, including chlorophyll a, chlorophyll b and carotenoids, when grown in either natural or reduced mixed communities than when grown alone (Table 1). This pattern was opposite for the invasive H. stipulacea, resulting in significant treatment x species interactions for chlorophyll a (p=0.0042), chlorophyll b (p=0.0175), and carotenoids (p=0.003); Supplementary Table S2). Hydrogen peroxide (H₂O₂), a marker of cellular stress, showed a significant species effect, not between species and treatment (LME; df=12; p < 0.0001; Supplementary Table S2). Overall, H₂O₂ concentrations were more than sevenfold higher in the native species (mean = 4.18 ± 0.42 SE μmol/g) than in the invasive species (mean = 0.62 ± 0.09 SE μmol/g), indicating substantially greater oxidative stress in the native seagrass. In contrast, the native seagrass C. nodosa had higher MDA levels than the invasive seagrass, resulting in a significant treatment x species interaction (LME; df=12, p = 0.049, Supplementary Table S2). Average MDA levels in native C. nodosa were 9.18 ± 0.84μmol/g, and for invasive H. stipulacea was 7.83 ± 0.34 μmol/g.

Table 1.

Averages of biochemical, morphological, and soil parameters measured across native and invasive seagrass species under six treatments: native mixed, native reduced mixed, native alone, invasive mixed, invasive reduced mixed, and invasive alone, and the standard error.

Parameter	Native mixed	Native reduced mixed	Native alone	Invasive mixed	Invasive reduced mixed	Invasive alone
Chl a* (mg/FW)	0.006 ± 0.002	0.007 ± 0.000	0.005 ± 0.001	0.005 ± 0.001	0.003 ± 0.001	0.004 ± 0.001
Chl b * (mg/FW)	0.003 ± 0.001	0.004 ± 0.002	0.003 ± 0.001	0.003 ± 0.000	0.002 ± 0.001	0.002 ± 0.001
Chl a / Chl b	2.11 ± 0.84	1.75 ± 0.58	1.75 ± 0.43	1.56 ± 0.19	1.56 ± 0.77	1.89 ± 0.54
Car *	0.002 ± 0.001	0.002 ± 0.000	0.001 ± 0.001	0.002 ± 0.001	0.001 ± 0.000	0.001 ± 0.001
MDA* (μmol/g)	9.33 ± 1.76	7.67 ± 0.49	10.56 ± 1.79	7.27 ± 0.35	8.45 ± 1.22	7.78 ± 0.18
H2O2* (μmol/g)	3.73 ± 1.05	3.80 ± 0.42	5.01 ± 0.90	0.49 ± 0.26	0.59 ± 0.20	0.78 ± 0.06
Change in number of shoots	−1 ± 0.4	0.2 ± 0.1	1.7 ± .1	3.7 ± 1.3	4.3 ± 1.5	0.5 ± 1.00
Aboveground biomass (g)	0.18 ± 0.03	0.22 ± .04	0.55 ± 0.06	0.46 ± 0.08	0.37 ± 0.08	0.38 ± 0.06
Belowground biomass (g)	1.92 ± 0.35	1.71 ± 0.36	3.15 ± 0.74	2.22 ± 0.39	1.79 ± 0.26	1.32 ± 0.23
Internode length* (cm)	3.17 ± 0.46	2.78 ± 0.20	2.08 ± 0.33	0.99 ± 0.04	1.05 ± 0.17	1.10 ± 0.00
Nitrates (mg/L)	0.93 ± 0.23	0.97 ± 0.12	1.00 ± 0.17	0.93 ± 0.23	0.97 ± 0.12	1.10 ± 0.00
Nitrites (mg/L)	0.06 ± 0.00	0.02 ± 0.00	0.05 ± 0.00	0.06 ± 0.00	0.02 ± 0.00	0.06 ± 0.02

Biochemical parameters include chlorophyll a (Chl a), chlorophyll b (Chl b), total chlorophyll (Total Chl), chlorophyll a/b ratio (Chl a/Chl b), carotenoids (Car), Malondialdehyde (MDA), and hydrogen peroxide (H₂O₂). Morphological traits include the number of shoots and internode length. Soil parameters include nitrate and nitrite concentrations. Asterisks indicate significant effects from the linear mixed effects model:

^*p < 0.001for the interaction Treatment × Species. Bolded parameters are values analyzed and shown in Supplementary Table 2.

Average nitrate concentrations in native mixed sediment were 0.93 ± 0.23 SE mg/L, while nitrite levels averaged at 0.06 ± 0.04 SE mg/L (Table 1). Similar values were observed in reduced mixed densities, with nitrates at 0.97± 0.12 SE mg/L and nitrites at 0.02 ± 0.03 SE mg/L. Native alone treatments had an average nitrate of 1.00 ± 0.17 SE mg/L and nitrite concentration of 0.053 ± 0.004 SE mg/L respectively, while the invasive showed 1.10 ± 0 SE mg/L for nitrates and 0.06 ± 0.02 SE mg/L for nitrites. Overall, nitrate and nitrite concentrations did not significantly differ between treatments and species.

Morphological values for the number of aboveground and belowground shoots, presented in Table 1 are published in Chiquillo et al. (2022), and no additional statistical analyses were conducted on these data. Briefly, competition experiments in the Caribbean and Mediterranean showed that H. stipulacea performed better with native species than alone, while the native species was directly affected by H. stipulacea; where it lost shoots and biomass in the presence of the invasive. In the Mediterranean, H. stipulacea increased by 3.7 ± 1.3 shoots (+59.2%) when mixed but showed no change alone, whereas the native Cymodocea nodosa lost −1 ± 0.4 shoots (−29.2%) when mixed. Aboveground biomass patterns were similar: natives produced significantly more biomass when alone (0.55 ± 0.06 g) than mixed (0.18 ± 0.03 g; p < 0.01), while the invasive species maintained or increased biomass in mixed treatments. In the reduced-density mixed treatment, H.stipulacea grew by 4.3 ± 1.5 SE shoots, an increase from an initial of +150.0% ± 50.7 SE (initial shoot density 2.9 ± 0.1). Surprisingly, when density was reduced in the mixed community, the native increased in shoots by 0.2 ± 0.1 SE, which is a +15.0% ± 10.7 SE increase. When H.stipulacea grew in reduced mixed treatments, final aboveground biomass was 0.37 ± 0.08 g (estimated initial 0.24 ± 0.03 g), indicating considerable regrowth during the experiment. When the native seagrass grew in the reduced-density mixed treatment, its aboveground biomass of 0.22 ± 0.04 g was not different from when grown in the native full mixed treatment, despite the experimental reduction. However, internode length was not included in Chiquillo et al., 2022, and we found a species differences in the length of the rhizomal internodes between species; the average internode length of the native species was 3-fold longer than the invasive (Table 1; Supplementary Table S2).

3.2. Native and invasive seagrass phyllosphere

A total of 301,907 reads were sequenced from 41 sequencing libraries (40 biological samples and 1 negative extraction control). After quality control, joining of paired-end reads, and removal of chimeras, 169,068 high-quality reads remained, representing approximately ~56% of the original dataset. The extraction blank yielded 2,769 raw reads; however, no reads passed quality filtering, denoising, merging, or chimera removal, resulting in zero retained sequences after processing.

Rarefaction analysis was conducted on all 40 samples originally sequenced, demonstrating adequate sequencing depth with a maximum of 46,777 reads, indicating sufficient sampling effort to capture the bacterial diversity present (Supplementary Figure S1A). The remaining treatments—mixed at full density and mixed at reduced density—for each species exhibited intermediate ASV richness, while sequencing depth remained sufficient across all treatments with no consistent differences among treatment groups.

A total of 18 samples were retained for microbial analyses after completing morphometric, biochemical, and soil nutrient (nitrate/nitrite) assays. To standardize sampling effort for diversity analyses, we rarefied our n=18 samples to a sequencing depth of 1,500, resulting in the loss of one sample. Initial samples were excluded due to the absence of morphometric and biochemical metadata, and because only the natural-density treatment included sediment disturbance, which prevented a direct comparison across treatments. Samples were rarefied to a depth of 1,500 reads based on inspection of rarefaction curves (Supplementary Figure S1A), which indicated that most samples approached asymptotic richness at relatively low sequencing depths. This threshold was selected to balance sequencing depth with sample retention. This was an important consideration because the final timepoint samples were paired with morphometric, biochemical, and sediment nutrient (nitrate/nitrite) datasets; thus, maximizing overlap across datasets was necessary to maintain statistical power and enable integrative analyses. The exclusion of baseline samples from downstream analyses limits our ability to infer temporal changes across the experiment, and analyses therefore focused on treatment-driven differences at the final timepoint. Rarefaction curves for 17 samples, which were used exclusively for alpha and beta diversity analyses (Figure 1 Part II), reached saturation at a maximum sequencing depth of 32,142 reads, confirming that sequencing coverage was sufficient to represent the microbial diversity in the dataset (Supplementary Figure S1B). The native-alone and native-reduced treatments exhibited the steepest initial slopes and highest plateau values, consistent with their elevated ASV richness. In contrast, samples from the invasive-mixed and invasive reduced mixed treatments plateaued at lower read depths, corresponding to lower bacterial diversity in those communities.

FIGURE 1.

Flowchart of the sampling design and microbial sequencing workflow. Part I – Overall sampling (n = 40). Four mesocosm treatments were established following data from Chiquillo et al. (2022) (A) Mixed (native and invasive seagrass species co-occurring at natural densities), (B) Reduced mixed (native and invasive species co-occurring at reduced densities to assess density-dependent effects), (C) Native alone (monoculture of Cymodocea nodosa), and (D) Invasive alone (monoculture of Halophila stipulacea). Mesocosm symbol required attribution from: [images/graphics from] Integration and Application Network (ian.umces.edu/media-library). DNA extractions were conducted on four composite samples per treatment × species at the initial timepoint (n = 16) and final (n = 24) time points for 40 biological samples and 1 negative. The full mixed treatment lacked initial samples (see Supplementary Table 1). Raw sequence reads underwent quality filtering, denoising, chimera removal, and ASV inference using the SILVA database. Sequencing depth adequacy and sample coverage were evaluated using rarefaction curves, maintaining only bacterial taxa. Part II – Sampling Competitive Outcomes (n = 18). Downstream analyses assessed alpha diversity, beta diversity, and differential abundance for taxa under competitive interactions.

All alpha and beta diversity calculations were done with rarefied data and using a max of 17 individuals per sample. All remaining analyses, including relative abundance, core microbiome identification, and MaAsLin2 modeling, were performed on the final dataset (n=18), comprising of 826 ASVs, and the average number of ASVs per sample was 220.

3.3. Alpha and beta diversity identified of the phyllosphere

The rarefaction allowed us to detect 826 bacterial ASVs identified across all taxonomic levels. A mentioned in methods, it led to exclusion of one sample, and rarefaction curves demonstrated that sequencing coverage was sufficient to capture the majority of bacterial diversity across samples (Supplementary Figure S1). Native and invasive seagrasses portrayed analogous Shannon alpha bacterial diversity measurements (Figure 2A), with no significant relationships between treatment and species (p = 0.171; Supplementary Table S3). Pielou evenness also showed no significant relationships between groups (p = 0.763; Supplementary Table S4).

FIGURE 2.

Alpha bacterial diversity of the seagrass samples split by treatment and species. Points are color-coded according to treatment (in fully natural mixed (pink), reduced mixed (blue) and alone (green). Stippled box plots represent the native Cymodocea nodosa, while solid box plots represent the invasive Halophila stipulacea. (A) Shannon (p= 0.171) and Pielou alpha diversity (p = 0.763) measures were analyzed on the species level. (B) Beta bacterial diversity of native versus invasive seagrasses grown in fully natural mixed (pink), reduced mixed (blue) and alone (green) in a non-metric multidimensional scaling (NMDS) plot with Bray-Curtis distances analyzed on the species level. Circles represent the native Cymodocea nodosa, while triangles represent the invasive Halophila stipulacea.

Beta diversity analysis using based on Bray-Curtis dissimilarities showed significant overall differences in microbial community composition among groups (PERMANOVA: F_5,11 = 3.857, R² = 0.637, p = 0.001), with group identity explaining 63.7% of the variation. ANOSIM supported this overall separation (R = 0.713, p = 0.001). Native and invasive seagrass samples clustered separately in non-metric multidimensional scaling (NMDS) ordination space (Figure 2B). Pairwise PERMANOVA comparisons were not significant after multiple testing correction (Supplementary Table S5). Tests for homogeneity of multivariate dispersion indicated no significant differences among groups (betadisper permutation test, p = 0.505, PERMDISP, p = 0.528), indicating that differences in dispersion did not account for the observed overall PERMANOVA result.

Results from the SRS-normalized dataset were consistent with the rarefied analyses. Alpha diversity patterns were consistent with the rarefied dataset, with marginally non-significant differences among groups (Kruskal–Wallis p = 0.0666). Beta diversity analyses also supported consistent results, with significant overall differences (PERMANOVA p = 0.001; ANOSIM R = 0.737, p = 0.001), non-significant pairwise PERMANOVA comparisons, and non-significant dispersion effects (betadisper permutation test, p = 0.607) (Supplementary Figure S2; Supplementary Table S6).

The top phyla associated with the phyllosphere of native and invasive seagrass showed Proteobacteria (Figure 3A), which accounted for 72.1% of the analyzed sequence reads across all samples. The second and third most abundant phyla across all samples were Bacteroidota (7.7%) and Planctomycetota (3.7%). Additional common phyla found among all groups were Actinobacteriota (2.7%), Verrucomicrobiota (2.5%), Desulfobacterota (2.4%), and Myxococcota (1.6%).

FIGURE 3.

Bacterial relative abundance plots of the seagrass samples split by species and treatment. (A) Relative abundances of the top 25 most common phyla in native and invasive seagrass samples. (B) Relative abundances of the top 25 most abundant bacterial taxa observed in native and invasive seagrass samples. Taxa are reported at the highest taxonomic level that could be confidently assigned (e.g., genus, family, or class).

The top 25 most abundant taxa marked strong differences in bacterial proportions between the two species of seagrass (Figure 3B). The top genus detected was Rhodobacteraceae, which accounted for 26.1% of the analyzed sequence reads across all samples. Surprisingly, the native seagrass C. nodosa showed relatively elevated levels of Bacteroidota (10.5-11.8%) compared with the invasive seagrass (2.3-5%). We also found higher proportions of uncultured genera of the family Oceanicella (5-14.2%), Microtrichaceae (3.1-5.8%), and other genera such as Hyphomonas (3-4.3%), Methylotenera (2.5-4.9%), and Lentimonas (2.1-3.9%) in the native seagrass compared to the invasive seagrass, where it was absent or low in recognizable reads. Compared to the native seagrass, which were absent or low in recognizable reads, there was an increased abundance of the genus Cobetia in the invasive H. stipulacea (2.1-4.6%).

3.4. Bacteria significantly associated with Cymodocea nodosa grown in mixed densities versus alone

Of the top 25 taxa, a family within the Rhodobacteraceae in the native seagrass C. nodosa possessed a higher relative abundance of the genus Oceanicella when grown in mixed density (10.6%) and reduced mixed densities (14.2%) compared to when it grew alone (5%). Additionally, we found higher relative abundance of an uncultured taxon of the family Methyloligellaceae in mixed (10.6%) and reduced mixed (8.1%) treatments compared to when it grew alone (4.1%). When the native seagrass grew alone, the native seagrass had elevated levels of the genus Eionea (24.9%) versus when it grew with the invasive in mixed (0.6%) and reduced mixed habitats (0.1%). Lastly, when native C. nodosa grew alone, we identified elevated levels of the phyla Camplyobacterota (3.6%) when taxa abundance was very low or unrecognizable in all other treatments.

Results from MaAslin2 analysis showed that when the native C. nodosa grew alone there were three significantly positive associations with the genera Desulfobulbus (B = 1.107, q- value= 0.00214), Bermanella (B = 1.152, q- value= 0.002), and the PS1 clade of the order Parvibaculales (B = 3.916, q- value= 0.003) compared to when it grew mixed or in reduced mixed densities (Figure 4A). We found an increased association with the families Arcobacteraceae (B = 5.677, q-value = 0.048), Desulfococcaceae (B = 1.445, q-value = 0.083), the genera Sediminispirochaeta (B = 1.482, q-value = 0.048) and Aminicenantales (B = 4.232, q-value = 0.071). In contrast, the genus Vibrio was negatively associated with the native C. nodosa when it grew alone compared to when it grew in mixed and reduced density communities (B = −4.530, q-value = 0.019). Lastly, to assess robustness effects, we ran a simplified model where parameterization included only the primary predictors, and our analyses confirmed the main patterns observed.

FIGURE 4.

Microbiome multivariable associations with linear models (MaAsLin 2) heatmaps based on 27 covariates. A red color (+) indicates positive correlation, while a blue color (−) indicates negative correlation. FDR q-value cutoff value was 0.05. Statistical significance is based on FDR-adjusted q-values (q ≤ 0.05). Effect sizes (coefficients) are shown for taxa significantly associated with metadata variables. (A) Heatmap of taxa correlated with significant changes in native groups only compared to the native mixed group with a low number of shoots. (B) Heatmap of taxa correlated with significant changes in invasive groups only compared to the invasive mixed group with a high number of shoots.

3.5. Bacteria significantly associated with the invasive Halophila stipulacea grown mixed versus alone

The top 25 taxa within the invasive H. stipulacea samples were generally dominated by the family Pseduoaltermonadaceae and Vibrionaceae. Within the family Pseduoaltermonadaceae, the genus Pseudoalteromonas was notably pronounced across the invasive seagrass treatments. However, when H. stipulacea grew in mixed densities, results showed a higher relative abundance of the genus Pseudoalteromonas (32.7%) compared to when it grew in a reduced mixed community (14%) and when it grew alone (6.5%). Specifically, H. stipulacea showed a pronounced presence of the genus Vibrio and Photobacterium in mixed density (16.7% and 9.1%, respectively) and reduced mixed density treatments (30% and 4.7%, respectively) versus when it grew alone (18.6 and 0.1%, respectively). Additionally, we found H. stipulacea tended to have a higher relative abundance of unassigned genera of the family Cellvibrionaceae in reduced mixed densities (7%) compared to when it grew in mixed densities (1.1%) and when it grew alone (3.6%).

Results from MaAsLin2 analyses showed that when invasive H. stipulacea seagrass grew alone and in reduced densities, there was a negative association the genera Zeaxanthinibacter (B = −1.175, q-value= 0.015) and Tenacibaculum (B = −1.348, q-value= 0.027) compared to when it grows in mixed densities (Figure 4B). Interestingly, the genus Photobacterium (B = −6.877, q-value= 0.036) was negatively associated with the invasive H. stipulacea when it grew alone compared to when it grew in mixed and reduced density communities. The family Rhodobacteraceae was positively associated with the invasive seagrass when it grew in monoculture compared to when it grew in mixed environments (B = 2.478, q-value = 0.139). When the invasive H. stipulacea was grown in reduced density there was a negative association with the genus Rhodospirillaceae (B = −1.699, q-value= 0.152) and a positive association with the genus Perspicuibacter (B = 3.008, q-value = 0.176) compared to when it grew mixed or in monoculture. To assess potential overparameterization, we conducted a robustness check using reduced predictor sets (native: discrete and continuous internode length and H₂O₂; invasive: discrete and continuous shoot number), and these reduced models produced results consistent with the full analysis, supporting the robustness of the observed associations (Supplementary Figure 3).

3.6. Environmental covariates associated with native and invasive bacterial diversity

The relationship between beta diversity and environmental covariates revealed a clear clustering between native and invasive seagrass samples forming distinct groups (Figure 5). Invasive seagrass displayed a broader ordination space, consistent with higher heterogeneity, suggesting more variable microbial assemblages relative to the native samples. Among the environmental covariates, hydrogen peroxide (H₂O₂)—a proxy for cellular stress—was significantly associated with microbial community structure of native C. nodosa (p = 0.003, R² = 0.610; Table 2). Additionally, covariate internode length, indicative of space-foraging or habitat-seeking behavior, was also significantly correlated with bacterial composition of native C. nodosa (p = 0.008, R² = 0.492). However, the number of shoots, often linked to overall growth and productivity, showed a weaker association with the invasive H. stipulacea (p = 0.060, R² = 0.329).

FIGURE 5.

Principal coordinate analysis (PCoA) plot with Bray-Curtis distances on the species level. Native C. nodosa samples are mapped as circles, while invasive H. stipulacea samples are mapped as triangles. Points are color-coded according to treatment. Arrows signify the strength and direction of correlations between environmental covariates and the ordination plot. Only the top three statistically significant environmental covariates are displayed.

Table 2.

Environmental covariate statistics from PCoA.

Parameter	Dim1	Dim2	R2	Pr(>r)
Chl a	−0.512	−0.859	0.124	0.343
Chl b	0.206	−0.979	0.091	0.493
Total Chl	−0.202	−0.979	0.202	0.189
Chl a / Chl b	−0.999	−0.036	0.067	0.587
Car	−0.293	−0.956	0.024	0.807
MDA	−0.314	−0.950	0.157	0.289
H2O2	−0.505	−0.863	0.610	0.003
Nitrite	−0.307	0.952	0.054	0.647
Nitrate	0.414	−0.910	0.029	0.767
Number of shoots	0.225	0.974	0.329	0.060
Aboveground biomass	−0.385	0.923	0.110	0.450
Belowground biomass	−0.981	−0.192	0.062	0.639
Internode length	−0.533	−0.846	0.492	0.008

Significant values are bolded. Chlorophyll a, Chlorophyll b, Carotenoids, MDA, and H₂O₂. Numerator (numDF) and denominator (denDF).

3.7. Core microbiome of native and invasive seagrass

We identified a total of 53 genera occurring in at least 80% of the samples that were shared among both native and invasive seagrass, representing their core microbiome (Figure 6A; Supplementary Table S7). Two genera were present in all groups, except when the invasive was growing alone, and were identified as an uncultured genus of the class Alphaproteobacteria and the genus Aquibacter. The native Cymodocea nodosa exhibited a core microbiome comprising 87 genera (Figure 6B; Supplementary Table S8), with no taxa unique to only one or two groups. In contrast, the invasive H. stipulacea harbored a smaller core of 24 genera (Figure 6C; Supplementary Table S9), and similarly, no taxa were exclusive to only one or two groups.

FIGURE 6.

Venn diagrams of unique and concordant genera observed in the core microbiome of the seagrass samples. Empty cells indicate zero values. (A) Unique and concordant genera observed in 80% of the combined native and invasive seagrass samples. (B) Unique and concordant genera observed in 80% of the native C. nodosa samples only. (C) Unique and concordant genera observed in 80% of the invasive H. stipulacea samples only.

Interestingly, we detected more differentially abundant bacterial genera of the core microbiome between native and invasive seagrass species when growing in mixed communities or alone (Figure 7A). Bacterial communities appeared more similar when the native seagrass grew in reduced mixed treatments and when it grew alone, than when grown with the invasive. This trend is the opposite for the invasive H. stipulacea, where bacterial communities looked similar to each other when grown in natural and in reduced mixed treatments, compared to when grown alone.

FIGURE 7.

Heatmaps of genera observed in the core microbiome of native and invasive seagrass samples using Bray-Curtis distances and the Ward clustering algorithm on the genus level. A red color indicates a higher frequency of taxa (similarity), while a blue color indicates a lower frequency of taxa (dissimilarity). Hues indicate intensity. (A) Genera observed in 80% of the native C. nodosa samples only. (B) Genera observed in 80% of the invasive H. stipulacea samples only.

In the core microbiome of the native seagrass Cymodocea nodosa, microbial community composition resembled more closely when the native seagrass growing with the invasive H. stipulacea in reduced mixed densities and when it grew alone, compared to taxa when it grew in natural conditions with the invasive H. stipulacea (Figure 7A). Over 35 taxa– such as Bdellovibrio, members of Saprospiraceae, Cyclobacteriaceae, and Flavobacteriaceae– were more abundant when the native C. nodosa grew alone compared to when it was grown in mixed assemblages with the invasive species at full or reduced densities.

In contrast, only a limited set of taxa—including Pseudorhodobacter and members of Rhodobacteraceae—showed higher abundance when growing at native fully mixed densities with the invasive seagrass, consistent with the possibility that interspecific interactions influence the composition of specific microbial associates. In the core microbiome of the invasive seagrass, many taxa that were highly abundant in mixed environments were absent when the invasive seagrass grew alone (Figure 7B). For example, Bdellovibrio and other members of the class Alphaproteobacteria showed increased activity in mixed conditions but were negatively associated when the invasive seagrass grew alone. Conversely, when the invasive species grew alone, we observed a positive association with Rhodobacteraceae and the genus Eionea, which were lower in abundance when the invasive seagrass grew in mixed and reduced mixed densities with the native seagrass.

4. Discussion

This is the first study to show that microbial community composition differs between mixed C. nodosa and H. stipulacea assemblages and monocultures in the Mediterranean, consistent with interactions between native and invasive seagrasses. Our results suggest that the microbial community composition of H. stipulacea varies with the outcome of species interactions in mixed communities. Consistent with the increased shoot production of H. stipulacea and reduced shoot production of C. nodosa when grown in mixed assemblages, H. stipulacea exhibited increased ASV richness in the presence of the native seagrass. In contrast, the native seagrass had a more diverse core microbiome and increased oxidative stress when grown alone, as well as lower ASV richness in the presence of the invasive seagrass. Importantly, combining these data with biochemical and morphological data from Chiquillo et al. (2022) suggests that phyllosphere microbial patterns are associated with the growth of invasive H. stipulacea, and we hypothesize that it may relate to traits linked to invasion success; future work is needed to determine the functional roles of these microbial taxa.

Our results reveal an asymmetric pattern: H. stipulacea exhibited greater ASV richness in the presence of C. nodosa, while C. nodosa’s microbial diversity decreased in the presence of H. stipulacea. Although much of the observed variation was primarily driven by species identity rather than specific treatment contrasts, these patterns suggest that competitive interactions may be associated with shifts in the composition of seagrass-associated microbial communities, with potential implications for invasion dynamics. While we note that quality control and rarefaction reduced sequencing depth across samples, resulting in relatively low read counts for some analyses, rarefaction curves indicated sufficient coverage for diversity estimation, and treatment-level comparisons were therefore interpreted within this constrained sequencing depth. Given that H. stipulacea thrives in both disturbed and open habitats and successfully invades intact native communities, our results are consistent with a negative association between H. stipulacea co-occurrence and C. nodosa density and bacterial diversity, which may be relevant to understanding the microbial dynamics associated with the invasion ecology of H. stipulacea.

4.1. Native species, though diverse, may fail to resist invasion, at their own cost

We found that the native C. nodosa exhibited the highest microbial diversity in monoculture, along with a more diverse core microbiome. When grown with H. stipulacea, C. nodosa’s diversity declined, and plant density was reduced (Sghaier et al. 2014; Mannino et al. 2023). Similar to Conte et al. (2023), we observed that C. nodosa’s highest diversity occurred when it grew alone, whereas in mixed assemblages, it exhibited reduced density, lower diversity, and higher phenol content, which is a biochemical indicator of stress (Mejia et al., 2016).

Our results are consistent with the hypothesis that C. nodosa’s bacterial diversity is correlated with cellular stress. In mixed environments, C. nodosa also exhibited increased internode length, although functional significance of the morphological response remains unclear. One possible explanation is that microbial communities may vary with plant competition, with potential consequences affecting physiology, performance, and fitness of native species. Conte et al. (2023) also observed that C. nodosa boasted the highest bacterial diversity in monoculture, while C. nodosa reduced its density, depicted lower bacterial diversity, and had higher phenol content in syntopy with H. stipulacea. High phenol content is linked with stress in seagrasses and is equipped as a defense mechanism against threats such as predation, low light regimes, pollution, ocean acidification, and competition by marine macrophytes (Mejia et al., 2016). Conte et al. (2023), alongside Chiquillo et al. (2022), observed that C. nodosa exhibited the greatest increase in shoot number when grown alone, suggest that C. nodosa’s may exhibit different microbial associations when it cooccurs with H. stipulacea. These shifts seem to mirror patterns seen in other systems with theoretical and empircal work show soil microbes can shape plant competitive outcomes (Ke and Wan 2020). More broadly, plant associated microbiomes are known to shift under environmental and physiological stress (Perreault and Laforest-Lapointe, 2022; Zandi and Schnug, 2022), consistent with the elevated hydrogen peroxide we observed in native species.

While the majority of studies focus on how microbes mediate competitive outcomes between plants, our approach examined how interspecific competition itself hosts differential microbial communities. Next steps would be to test for direct microbial transfer between species. Contrary to expectations, the native seagrass exhibited significantly higher H₂O₂ concentrations, consistent with greater oxidative stress levels compared to H. stipulacea. The capacity to scavenge reactive oxygen species such as H₂O₂ is a key determinant of a plant’s tolerance to environmental stress. Higher H₂O₂ has been observed in competitive superiors where elevated H₂O₂ concentrations were seen in monocultures compared to mixed assemblages (Ouyang et al., 2023; Navas et al., 1999); for example a noxious marine endophyte, P. Crispus, which is classified as a competitive superior, had higher H₂O₂ concentrations in monoculture than mixed (Chalanika De Silva and Asaeda, 2017). However, our results suggest that despite being competitively inferior to H. stipulacea, C. nodosa’s disproportionately high H₂O₂ levels may indicate that oxidative stress in this case reflects a competitive disadvantage rather than superiority.

4.2. Associations between H. stipulacea and distinct microbial taxa

Competitive interactions between H. stipulacea and three other native seagrasses, H. decipiens, S. filiforme, and T. testudinum, have shown displacement of natives from the Caribbean (Steiner & Willette 2010; Muthukrishnan et al., 2020; Willette and Ambrose, 2012), and its recent expansion of H. stipulacea into South Florida has been reported (Campbell et al., 2025). Not surprisingly, even when alone, H. stipulacea maintained a distinct microbial community, indicating that its microbial associations are not restircted to mixed assemblages and may support its own invasion success, possibly by influencing its microbial community or exploiting its environment to its advantage.

In a field experiment examining seagrass interactions in the Mediterranean, Conte et al. (2023) found evidence that H. stipulacea’s microbial communities are also plastic when growing with the native C. nodosa, displaying a high bacterial variability, supporting the host generalist hypothesis. In addition to negatively impacting the microbial community of the native seagrass, our results indicate that H. stipulacea was associated with bacterial taxa previously reported to potentially possess beneficial functions in mixed environments (Supplementary Table S10), without the need for a disturbance to remove native seagrass species. We know that H. stipulacea is a strong invader even in monoculture, where the species can colonize open habitat (Willette et al., 2014; Winters et al., 2020; Smulders et al., 2017). Like Conte et al. (2023), we found that H. stipulacea harbored a less diverse microbiome than C. nodosa in the Mediterranean Sea and that between the two species, C. nodosa hosted the richest and most abundant microbiome in monoculture, while H. stipulacea bacterial diversity was highest in mixed environments.

A limited set of taxa- including Rhodobacteraceae sp. might explain the ability of H. stipulacea when grown in a mixed environment, as they have been previously associated with assisting with carbon and sulfur cycling as well as photosynthesis in marine anaerobic environments (Pujalte et al., 2014; Crump et al., 2018), producing antibacterial compounds (Hurtado-McCormick et al., 2019), and are potentially involved in biofilm formation (Vogel et al., 2020; Conte et al., 2023). They have been frequently identified in seagrasses, including H. stipulacea (Mejia et al., 2016; Rotini et al., 2017; Stuij, 2018; Hurtado-McCormick et al., 2019; Garcias-Bonet et al., 2021; Ling et al., 2021; Conte et al., 2023) and have been associated with H. stipulacea leaves in high light and hydrodynamic regimes (Mejia et al., 2016).

Results from the competition experiment in the Mediterranean Sea add to the existing knowledge of H. stipulacea microbial diversity. We observed lower microbial community diversity compared to previous observations in both mixed and monoculture environments across the Red Sea (Wahbeh and Mahasneh, 1984; Pereg et al., 1994; Weidner et al., 1996, 2000; Mejia et al., 2016; Rotini et al., 2017; Garcias-Bonet et al., 2021; Abdel-Wahab et al., 2021), Mediterranean (Conte et al., 2021, 2023), and Caribbean Seas (Stuij, 2018; Aires et al., 2021). This contrast may be due to the experimental context of our study, as samples were collected six weeks after transplantation, rather than directly from natural donor sites. Although we did not detect the same taxa, our findings align with Conte et al. (2023), who identified beneficial H. stipulacea bacteria in the Mediterranean Sea such as the nitrogen-fixing Microcystaceae, Nisaeaceae, Phormidiaceae; sulfur-fixing Desulfocapsacaceae; and leaf colonizing Pirellulaceae. Our results overlap with taxa reported by Aires et al. (2021), who found several halotolerant genera in H. stipulacea with elevated abundance in plant growth-promoting Halomonas, quorum-inhibiting Salinimonas and Salinicola, heavy metal-tolerant Lysinibacillus, and halotolerant genera such as Rhodococcus and Azospirillum, which support seed germination under high salinity.

Due to constraints, our study examined only phyllosphere bacteria associated with the native and invasive seagrass, although it is widely accepted that the rhizosphere exhibits a richer microbial community (Rotini et al., 2017; Li et al., 2025). Our focus on phyllosphere-associated microbes was justified because the available microbial samples and observed treatment responses were derived primarily from above-ground tissues and shoot dynamics. However, below-ground microbial communities remain an important limitation of this study and a critical direction for future research. It is likely that rhizosphere bacteria also play important roles in H. stipulacea invasion success, particularly since seagrass above- and belowground compartments possess distinct microbes (Stuij, 2018; Winters et al., 2020; Aires et al., 2021) and sediment microbes can contribute to nutrient acquisition and other critical plant processes. Given their recognized role in nutrient cycling and acquisition in seagrasses (Zhang et al., 2020; Zhang et al., 2024), future studies integrating rhizosphere microbiome analyses with sediment chemistry would provide important mechanistic insight into the observed nitrate dynamics. Especially, since experimental evidence has linked that both root- and sediment-associated microbiota to seagrass performance under nutrient variability and stress, particularly through root-linked microbial functions that buffer environmental change (Fuggle et al., 2023; Jongen et al., 2026). Broader syntheses further emphasize that these below-ground microbial processes are foundational to seagrass health and should be explicitly integrated into restoration and predictive frameworks (Jongen et al., 2024). Similarly, experimental work on the invasive macroalga Caulerpa taxifolia has shown that sediment microbial communities can influence invasion outcomes by either promoting invader growth or providing biotic resistance from native seagrass sediments, highlighting the broader importance of below-ground microbial processes in marine plant invasions (Gribben et al., 2018).

4.3. Alternative hypotheses: microbes as potential drivers or passengers of competitive interactions

One possible explanation for reduced microbial diversity in C. nodosa when grown with H. stipulacea is allelopathy. An untested possiblility is that H. stipulacea harbors microbial associates that produce compounds negatively affecting the native species, a mechanism by which microbes are known to mediate plant allelopathy (Cipollini et al., 2012; Greer et al., 2014). For example, we found a strong positive association between H. stipulacea in intraspecific competition and the genus Perspicuibacter, which has been previously associated with sulfur cycling, in contrast to mixed assemblages. This pattern differs from studies reporting strong negative effects of interspecific competition among seagrasses. We also observed a positive association between H. stipulacea grown alone and members of Rhodobacteraceae. While allelopathy is well documented in terrestrial plants (Bais et al., 2003; Orr et al., 2005; Chengxu et al., 2011; Greer et al., 2016; Chen et al., 2017; Kalisz et al., 2021) and in some seagrass–macroalgae interactions (Dumay et al., 2004; Pergent et al., 2008; Raniello et al., 2007), to our knowledge, seagrass–seagrass allelopathy has not been reported. Vibrio, significantly more associated with C. nodosa in mixed assemblages, is notable for its algicidal (Sakata et al., 2011; Setiabundi et al., 2026), antimicrobial (Millan, 2022; Wietz et al., 2010; Wietz et al., 2013; Romero-González et al., 2023; Alattas et al., 2024), and pathogenic properties (Sampaio et al., 2022; Alattas et al., 2024), as well as its production of siderophores—iron-chelating compounds with potential allelopathic effects (Wietz et al., 2013; Alattas et al., 2024; Dar et al., 2024). Results from Saunier et al. (2025) show that Halophila stipulacea has a distinct volatilome compared to native seagrasses, including compounds such as apocarotenoids and isothiocyanates. These compounds have been previously identified as allelopathic (Moreno et al., 2021), supporting a possible chemical mechanism underlying the patterns we observe, that could be investigated in future studies. In addition, we found Pseudomonas in the core taxa shared across H. stipulacea treatments (Supplementary Table S9). Members of this genus are known to produce bioactive compounds, including antimicrobials and siderophores, which can alter microbial community structure and inhibit competitors. Together, these observations raise the possibility that both host chemistry and associated microbes mediate allelopathic interactions and potentially facilitate invasion success, similar to other terrestrial plant communities (Cavieres, 2021).

We hypothesize that another possible mechanism is that H. stipulacea may alter the phyllosphere community of C. nodosa in mixed assemblages. Desulfobulbus, the most significantly associated genus with C. nodosa grown alone (B = 1.107, q = 0.00214), is a sulfate-reducer common in seagrass holobionts and sediments (Küsel et al., 1999; Ling et al., 2021; Zhou et al., 2021). It has been linked to nutrient cycling, hydrocarbon degradation, and nitrogen fixation (Zhou et al., 2021; Wang et al., 2025). Other taxa significantly associated with C. nodosa alone include Bermanella, known for hydrocarbon degradation (Hu et al., 2017; Ribicic et al., 2018; Peeb et al., 2022; Howe et al., 2024), and the PS1 clade, related to nitrogen-fixing Rhizobiales adapted to low-nutrient conditions (Jimenez-Infante et al., 2014; Stuij, 2018; Zhai et al., 2022). Changes in the abundance of such taxa may be associated with microbial functions in mixed assemblages and could alter microbial functions associated with native seagrass performance when grown alone.

For H. stipulacea, several taxa were less abundant when grown alone than in mixed assemblages, suggesting a capacity to retain microbial taxa previously associated with beneficial functions (Cardini et al. 2018). These included Zeaxanthinibacter (photoprotective pigment production; Demmig-Adams et al., 2022; Demmig-Adams et al., 2020), Tenacibaculum (organic matter degradation), Photobacterium (nitrogen fixation, nutrient provisioning), Rhodospirillales (anaerobic photosynthesis), and Micavibrionales (potential oil-degrader interactions). While 16S rRNA sequencing does not confirm functional activity, these taxonomic affiliations provide a hypothesis that these traits could enhance H. stipulacea growth and resilience. We also did not account for archaea, protists, viruses, or fungal communities, which are known to mediate nutrient acquisition, host–microbe interactions, and organismal expansion (Van der Putten et al., 2007). As climate change is expected to promote the expansion of H. stipulacea in the Mediterranean (Winters et al., 2020; Beca-Carretero et al., 2024; Nguyen et al., 2020), understanding whether these microbial shifts are a cause or a consequence of competitive success remains a critical next step.

4.4. Concluding statement

This study illuminates that phyllosphere microbial communities associated with Halophila stipulacea vary with growth conditions and co-occurrence with the native seagrass, Cymodocea nodosa, in the Mediterranean Sea. Specifically, we show how microbial communities alter in relation to competition. When competition coincides with enhanced invasive performance, microbial communities also differ (Brooker, 2006). Ultimately, our results identify microbial community patterns associated with H. stipulacea phyllosphere and its enhanced performance in mixed Mediterranean environments; however, further research is necessary into the mechanisms behind this association. Specifically, future investigations should focus on whether H. stipulacea selectively associates with microbial taxa previously reported to possess beneficial functions, whether putative microbe-mediated allelopathic interactions occur, or whether its microbiome is passively shaped by community assembly processes. We acknowledge that the compositing may reduce estimates of within treatment variance, and should therefore be interpreted as a conservative estimate of microbial community variability. Overall, we present valuable new insights into plant-microbe interactions in marine ecosystems and the possible benefits of seagrass-microbe interactions in H. stipulacea invasions.

Supplementary Material

Table 1

SUPPLEMENTARY TABLE 1

Number of samples across treatments for biochemical, morphological, sediment, and microbial analyses in native, Cymodocea nodosa, and invasive, Halophila stipulacea. Values represent the number of composite samples (n). Treatments include Naturally Mixed, Reduced Mixed, Native Alone, and Invasive Alone. Photosynthetic pigments include chlorophyll a, chlorophyll b, carotenoids and oxidative stress markers include H₂O₂ and MDA. Morphological traits include growth, biomass, and internode length measured per treatment. Sediment nutrients include nitrate and nitrite concentrations. Microbial datasets include composite samples collected at the initial time point and final time point, with a subset retained for downstream analyses. PCR negative controls were included to account for contamination and sequencing artifacts.

SUPPLEMENTARY TABLE 2

Linear mixed effects model of chlorophyll a, chlorophyll b, carotenoids, MDA and H₂O₂. Numerator (numDF) and denominator (denDF), F-value and p-value are summarized.

SUPPLEMENTARY TABLE 3

Kruskal-Wallis statistics for Shannon alpha diversity. Significant values are bolded.

SUPPLEMENTARY TABLE 4

Kruskal-Wallis statistics for Pielou evenness (alpha diversity).

SUPPLEMENTARY TABLE 5

PERMANOVA, pairwise PERMANOVA, ANOSIM, and betadisper statistics for beta diversity. Significant values are bolded.

SUPPLEMENTARY TABLE 6

PERMANOVA, ANOSIM, and betadisper statistics for Scaling with Ranked Subsampling (SRS) beta diversity for all 18 samples. Significant values are bolded. Pairwise PERMANOVA: Not Significant.

SUPPLEMENTARY TABLE 7

Table of core taxa shared among all six groups and all groups except invasive alone (see Figure 6A).

SUPPLEMENTARY TABLE 8

Table of core taxa shared among C. nodosa groups (see Figure 6B).

SUPPLEMENTARY TABLE 9

Table of core taxa shared among H. stipulacea treatments (see Figure 6C).

SUPPLEMENTARY TABLE 10

Core microbiome taxa identified in mixed and reduced mixed treatments in the H. stipulacea heatmap and their putative function.

Suppl materials

SUPPLEMENTARY FIGURE 1

(A) Rarefaction curve of total bacterial diversity for n=40 initial and final samples. (B) Rarefaction curve of total bacterial diversity for n=18 final samples.

SUPPLEMENTARY FIGURE 2

Alpha bacterial diversity of the seagrass samples split by treatment and species, using Scaling with Ranked Subsampling (SRS). (A) Native, Cymodocea nodosa, fully natural mixed (red) box plot, dark blue represents native and invasive grown in reduced mixed treatments and blue represents the native when it grows alone, while the yellow box plot represents the invasive Halophila stipulacea, grown in fully natural mixed and pink is when the invasive grows alone. SRS measures were analyzed on the species level with marginal significance (p = 0.06657). (B) Beta bacterial diversity of native versus invasive seagrasses grown in fully natural mixed (pink), reduced mixed (blue), and alone (green) in a non-metric multidimensional scaling (NMDS) plot with Bray-Curtis distances analyzed on the species level. Circles represent the native Cymodocea nodosa, while triangles represent the invasive Halophila stipulacea. Points are color-coded according by species and treatment.

SUPPLEMENTARY FIGURE 3

Microbiome multivariable associations with linear models (MaAsLin 2) heatmaps based on the primary predictors of interest for each species. A red color (+) indicates positive correlation, while a blue color (−) indicates negative correlation. FDR q-value cutoff value was 0.05. Statistical significance is based on FDR-adjusted q-values (q ≤ 0.05). Effect sizes (coefficients) are shown for taxa significantly associated with metadata variables. (A) Heatmap of taxa correlated with significant changes in native groups only compared to the native mixed group with a low number of shoots. (B) Heatmap of taxa correlated with significant changes in invasive groups only compared to the invasive mixed group with a high number of shoots.

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fmars.2026.1871379/full#supplementary-material

Data availability statement

The 16S rRNA raw sequence data have been deposited in the SRA of NCBI with BioProject Number PRJNA1291277 and assigned accession numbers SRR34539037-SRR34539076.