Trait-based response-effect frameworks, TBFs, are currently underexplored in seagrass research. The number of studies including the word “trait” increased steadily since the first study from the year 1988, reaching a maximum in the last five years (n=73, Figure 2A ). Only 29 studies were found to use existing TBFs, accounting for only 15% of the total. The “trait-based approach” was first developed in 2002 (Lavorel and Garnier, 2002) for terrestrial plants, and it does not appear in the seagrass literature until 2012. This indicates that the body of knowledge available from terrestrial plant ecology has been under-utilized by seagrass researchers.

Studies focusing on seagrass traits have been mainly developed in the Mediterranean bioregion (27%, Figure 2B ), while the Tropical Atlantic (9%) and Temperate Southern Oceans (11%), showed the lowest number of studies. TBF studies have been homogeneously performed in all bioregions, with the Temperate North Pacific showing the highest number (24%). Differences across bioregions could be attributed to the differential use of the term “trait” across research groups and the seagrass species that are the focus of their study. As a consequence, while the widespread species Zostera marina accounted for 30% of trait studies, Posidonia oceanica, endemic to the Mediterranean Sea, accounted for 19% ( Figure 2C ). This result indicates a certain bias in the use of the trait nomenclature in certain species like P. oceanica, while it simultaneously highlights the problem of research bias and inference from unique species to the others, specifically when trait responses (Viana et al., 2020) can be species-specific.

We found seven studies that linked seagrass dispersal and settlement to traits (3.6% of the total, Figure 3 ), suggesting that seagrass dispersal can be predicted by the traits. Target traits can change according to the life stage of the seagrass plant (seed or vegetative fragment) (Orth et al., 2007; McMahon et al., 2014; Bryan-Brown et al., 2017). In the case of seeds, traits such as flotation capacity and digestibility determine the distance they can disperse and the effectiveness of using animal vectors, respectively (Wu et al., 2016). Seeds also show a high degree of intraspecific variability in size, which determines their settling velocity and dispersal potential (Delefosse et al., 2016). When it comes to settlement, traits like germination rate can be site specific and negatively affected by increasing temperature in Z. marina (Cabaço and Santos, 2010). In addition, the current velocity in the settlement area and the stiffness and flexibility of surrounding shoots limit the settlement capacity of seagrass seeds (Bouma et al., 2009). In the case of vegetative fragments, the plant morphology can partially control their dislodgement resistance, whereas the age and rooting rate determine their capacity for settlement (Lai et al., 2018). Vegetative fragments have the potential for long distance dispersal thanks to long lasting shoot buoyancy and survival, allowing the colonization of new areas (Berković et al., 2014). Additionally, fruits of certain seagrass species allow for long distance dispersal as well, as it is the case of Posidonia australis (10s to 100s of kilometers, Ruiz-Montoya et al., 2015).

The identification of traits controlling dispersal and settlement of seeds and vegetative fragments challenges the perception of dispersal as a stochastic and unpredictable process. In addition, abiotic (temperature, current velocity, wave disturbance and exposure) and biotic (animal vectors, shoot stiffness) factors as well as a combination of these exemplified by the seascape mosaic formed by the plants on an unvegetated substrate affect their dispersal and settlement success. The neutral assembly process hypothesis ( Figure 1 : Hypothesis #1), while not formally tested, seems to be false in the case of seagrasses ( Table 1 ). However, only a small number of studies investigated this question. Further research is needed to test this hypothesis at different scales, with null models as a fundamental tool to assess the relative importance of purely stochastic and niche assembly processes (Mori et al., 2015). This hypothesis has been tested in other marine organisms, including fish communities (Ford and Roberts, 2018), which assemble neutrally at the regional scale but not at the local scale, and woody plants, whose dispersal and settlement are primarily trait-driven (Duarte et al., 2010).

Much research has been performed on the responses of seagrass traits to environmental drivers, making a total of 72% of the studies found in the literature review. There was, however, a large imbalance between the study of seagrass traits under the abiotic (89%) vs the biotic (11%) filter ( Figure 3 ). This indicates that the knowledge of seagrasses is focused on the study of its fundamental niche, i.e. the major physico-chemical constraints in the system. For example, the vertical zonation of tropical seagrasses was explained by physiological traits controlling their ability to tolerate high irradiances and nutrient inputs (Björk et al., 1999). Other examples focus on the assessment of the fundamental niche of individual species. Halophila decipiens occupies a wide range of irradiances and temperatures, due to its phenotypic plasticity (Gorman et al., 2016). Z. marina has a low niche specialization in the Baltic Sea, allowing this species to exist under variable environmental conditions in comparison to other macrophytes (Herkül et al., 2018). The three most frequent abiotic drivers studied were temperature (28.3%), nutrients (24.4%) and light (17.3%), whereas the least studied include the effect of freshwater input (1.5%) or metal pollution (2.4%).

Traits have been therefore used as indicators of environmental change, and their response is both driver and species-specific. This can be illustrated using the example of temperature. An increase in temperature within the optimal range fosters leaf growth (Thalassia hemprichii: Viana et al., 2020; Enhalus acoroides: Artika et al., 2020) and leaf size (Z. marina: Young Kim and Seob Choi, 2004; Ondiviela Eizaguirre et al., 2018; DuBois et al., 2019; Halodule wrightii: Sordo et al., 2011; Zostera noltei: Ondiviela Eizaguirre et al., 2018; T. hemprichii: Viana et al., 2020; Cymodocea serrulata: Viana et al., 2020; E. acoroides: Artika et al., 2020; Artika et al., 2021; Zostera capensis: Beltrand et al., 2022). However, when the optimal temperature for a species is surpassed, heat stress reduces these two traits (Posidonia oceanica: Traboni et al., 2018; Halophila ovalis: Ontoria et al., 2020, Halophila stipulacea: Viana et al., 2020). Co-inhabiting species can have different thermal optima (Collier et al., 2011), and their trait responses can give fundamental information on how future warming will affect seagrass communities. Traits can therefore inform about the responses of seagrass to environmental change, these responses being species-specific or general among seagrass species. The diversification of research to different species and bioregions is therefore fundamental to predict how seagrasses will deal with future global change scenarios, as even co-inhabiting species may respond in different ways (Agawin et al., 2001).

The prevalence of the use of morphological traits (55.9%) among all other trait categories is worth mentioning (biochemical 31.5%; physiological 29.1%; growth 28.3%; mechanical 15.7%; reproductive 10.2%). This is likely explained by their relatively easy and inexpensive measurements compared to physiological measurements of e.g., photosynthetic efficiency (Hernán et al., 2016; Llagostera et al., 2016) or enzymatic activity (Alexandre et al., 2004; Alexandre et al., 2010), which require specialized equipment, technical staff and laboratories. There are, therefore, extrinsic economical and technical reasons that constrain scientific questions in seagrass research. This trait type imbalance may impede a deeper understanding of responses of seagrasses to abiotic drivers, as physiological and biochemical indicators are recommended over morphological ones for early stress detection in seagrasses (Roca et al., 2016).

The study of morphological (52%), biochemical (50%) and growth traits (34.8%) under the biotic filter was more balanced compared to the abiotic filter. Most of the traits were studied in response to eutrophication (36.9%) and/or herbivory (23.9%). Eutrophication is considered one of the main threats to seagrasses, as it can lead to a phase shift in primary producers from seagrass to macroalgal dominance (Duarte, 1995; Orth et al., 2006; Burkholder et al., 2007; Waycott et al., 2009). Eutrophication affects different compartments of the ecosystem (seagrass, micro- and macroalgae, epiphytic organisms), changing their relative abundances and causing changes in the light penetration or the redox potential in the sediment (Burkholder et al., 2007). Seagrass species respond to eutrophication through their traits, again highlighting their value as environmental change indicators. Under eutrophication, Cymodocea nodosa increases the nutrient content in its leaves, while reducing its fiber content and biomechanical properties (Jiménez-Ramos et al., 2018a). Similarly, P. oceanica increases the nutritional quality of its leaves under fertilization, while reducing plant growth (Ravaglioli et al., 2018). In the case of herbivory, seagrass plants respond to this driver using a comprehensive array of traits, including growth compensation, changes in their nitrogen content or mobilization of carbohydrates (Sanmartí et al., 2014) and their morphology and growth form can predict grazing impacts on a global scale (Poore et al., 2012).

One of the main questions posed in the conceptual TBF is the relative importance of the abiotic vs the biotic filters ( Figure 1 : Hypothesis 2) to better understand the fundamental and realized niches of seagrass species. The available trait-based studies have focused separately on the fundamental and realized niches of seagrass species. We found that only 17.1% of the studies included both biotic and abiotic factors simultaneously, which indicates the existence of a knowledge gap in this topic. Trait-based approaches suggest the study of the convergence and divergence of trait distributions to unravel the relative effects of the abiotic and biotic filters (see review by Grime, 2006). Herbaceous plant communities tend to diverge at the species level but converge at the trait level (Fukami et al., 2005), indicating that environmental forces select for functional groups but not for species identities, which are historically contingent. This finding exemplifies the two schools of thought in ecological community assembly. On one hand, it has been suggested that different species coexist, occupying different niches (Diamond, 1975). On the other hand, members of the same plant community tend to exhibit similarity in plant traits, therefore showing overlapping niches (Clements, 1916).

The drivers that shape communities, namely environmental drivers, competition, and disturbance regime, can act at different spatial scales (Díaz et al., 1998; Pierce et al., 2007). To disentangle these effects, it is necessary to calculate the functional diversity (FD) and, specifically, functional trait dissimilarity among species within and among communities (Petchey and Gaston, 2002; De Bello et al., 2009). If the functional dissimilarity is lower than a set of random species (null model, see Götzenberger et al., 2016), this indicates trait convergence due to environmental filtering, whereas the opposite indicates trait divergence, and therefore a dominance of competition and/or a disturbance regime allowing for differential life-history strategies (Mason et al., 2007; Mouillot et al., 2007; Petchey et al., 2007). Despite the knowledge that seagrass traits change under environmental drivers (Roca et al., 2016) and that these traits affect interspecific competition (Moreira-Saporiti et al., 2021b), there are only a handful of examples in which trait convergence and divergence ( Figure 1 : Hypothesis #2) have been tested, and uniquely in Z. marina at the intraspecific level. Z. marina communities have a higher trait diversity with higher genetic relatedness among genotypes, indicating that trait divergence may be selected among competing genotypes (Stachowicz et al., 2013). Similarly, niche differentiation through trait divergence is suggested as an explanation for the positive correlation between its genotypic and trait diversity (Abbott et al., 2017). However, transplant experiments have shown that the species morphology changes under local environmental conditions, resembling the morphology of local populations (Ruesink, 2018). The results of these studies indicate that divergence occurs at the local level among competing plants, while convergence seems to occur at a larger scale.

The large body of literature on response traits under abiotic and biotic factors shows that there is a wealth of data that can be reassessed to answer questions in the context of the TBF presented here. In addition, the study of response traits uses a variety of nomenclature that could not be included in this review (e.g. indicators, responses…), further increasing the volume of data available. Despite the presentation of neutral and niche assembly theories separately, both are not incompatible. While some species can be stochastically eliminated from the local community (not reaching a suitable habitat or a random event leading to local extinction, see Sp6 in Figure 1 ), the subset of species that successfully colonized a community undergo a process of niche assembly. The disentanglement of fundamental and realized niches under a TBF is currently unexplored in seagrasses, providing an opportunity to answer fundamental research questions under global change that includes both abiotic and biotic drivers.

The study of the phylogenetic and genotypic control of traits is quite prevalent in the seagrass literature (23.3% of studies, Figure 3 ). There is ample evidence that genotypic richness covaries with phenotypic variation in functionally relevant traits, such as leaf morphology and shoot productivity within P. australis (Evans et al., 2016). In contrast, genetic diversity is a poor proxy for trait differentiation in Z. marina (Abbott et al., 2018). In P. oceanica a reproductive trait like flower abundance was negatively correlated to genotypic diversity and positively correlated to heterozygosity (Jahnke et al., 2015a), while there was a correlation of genetic indices and their response to environmental conditions (Jahnke et al., 2015b). Ecosystem functions like the accumulation of biomass and susceptibility to herbivory are also genotypically controlled in Z. marina (Tomas et al., 2011), with genotypes differing in key traits related to these processes. Similarly, nutrient uptake rates differed among genotypes of Z. marina (Hughes et al., 2009).

These studies confirm that there is genetic control of seagrass traits ( Figure 1 : Hypothesis #3) and, consequently, of ecosystem functions and services (Díaz et al., 2013). However, this control is species-specific. Knowledge gaps in this area lay in the lack of information from most of the seagrass species, as the genus Zostera and Posidonia accumulate 82.2% of the studies.

In seagrass ecosystems, characterized by low plant species richness, intraspecific variation is likely to play a more important role than in terrestrial ecosystems. In comparison to terrestrial plant lineages, the taxonomic diversity of seagrass is low with all species belonging to four Alismatales families. Indeed, many temperate meadows are monospecific, and most tropical meadows consist of only a handful of co-occurring species (Short et al., 2007).

Species and populations can differ for the level of plasticity (i.e., amplitude of the genotypes’ reaction norm), which is a fundamental trait affecting genotype persistence in changing environments (Pazzaglia et al., 2021). The plasticity of populations and genotypes is given by different levels of genetic variability, encompassing clonal somatic mutations and epigenetic changes. Several studies have indicated that the intraspecific trait variability of seagrass species is key for their survival. For example, the plant size of Z. marina (the predominant species in the northern hemisphere) spans more than two orders of magnitude across its distribution range (Ruesink, 2018), and different genotypes show large differences in nutrient uptake capacity and key photosynthetic parameters when grown in a “common garden” (Hughes et al., 2009). Even putatively less plastic species such as P. oceanica display a large variation in the acclimation to environmental factors (e.g. heat, Marín-Guirao et al., 2018). This feature, potentially supported by high intra-specific and intra-clonal (epi-)genetic diversity, enables seagrasses to cope with major environmental changes (Maxwell et al., 2014) and has most likely contributed to their successful colonization of shallow coastal zones along five continents, despite their low taxonomic diversity. Intraspecific variability in traits does not only occur at the species level, but also at the shoot, rhizome and clone levels. Epigenetic differences are even present within the same rhizome, which foster clonal persistence both within the same shoot (Ruocco et al., 2021) and within the same leaf (Ruocco et al., 2019a; Ruocco et al., 2019b). Recent evidence even points out that within single clones, somatic mutations lead to differentiation of ramets (= clone mates), with the potential to result in phenotypic differences within clones (Yu et al., 2020).

This body of literature highlights the importance of intraspecific trait variability in the response of seagrasses to disturbances, their resilience and capacity for ecosystem functions provision. However, we found only one example of the simultaneous study of intra- and interspecific variability of structural and nutritional traits, which drive palatability and herbivory in seagrasses (Jiménez-Ramos et al., 2018b). Future research assessing the relative importance of inter- vs intraspecific variability in both response and effect traits ( Figure 1 : Hypothesis #4) will be necessary to understand the relative role of intra- and interspecific diversity in seagrass ecosystem functions.

The links between effect traits and ecosystem functions were tested in 26.9% of the studies ( Figure 3 ). Herbivory (38.4%) and primary production (23%) were the most studied functions. Morphological (53.8%) and biochemical (46.1%) traits were the most used in the assessment of functions ( Figure 3 ). Examples include wave attenuation, which is explained by a combination of morphological and mechanical traits including blade stiffness, shoot density and leaf length (Bouma et al., 2005; Paul et al., 2012) or herbivory of Z. noltei, which is mediated by both structural and nutritional leaf traits (Martínez-Crego et al., 2016). Examples of more nuanced, indirect interactions between traits and functions include the reduction of the canopy height in P. oceanica by grazing, thereby increasing the predation risk on associated sea urchins (Pagès et al., 2012).

There are, therefore, clear mechanistic links between seagrass effect traits and ecosystem functions. However, at the community level, there is the question of whether effect traits control ecosystem functions through dominance (CWM) or complementarity (FD) ( Figure 1 : Hypothesis #5). In addition, the link between traits and ecosystem functions can be environmentally constrained ( Figure 1 : Hypothesis #6). A great number of studies on ecosystem functions included environmental metrics (71.1%).

The hypothesis of the control of ecosystem functions by functional complementarity (FD) versus dominance (CWM) have been barely tested in seagrass ecosystems, with only a handful of examples found in the literature review ( Table 2 ). Regarding the dominance hypothesis, CWM has been found as a reliable predictor of primary production in marine and brackish plant communities, including Z. marina (Gustafsson and Norkko, 2019). Particularly, plant height had positive effects on primary production, while the effects of other traits were environmentally constrained ( Table 2 ). In the case of carbon storage, geophysical attributes seem to constrain any effect of seagrass traits (Belshe et al., 2018). Complementarity alone was tested in one study (Abbott et al., 2017, Table 2 ), showing that the Rao quadratic entropy index of trait diversity can predict invertebrate abundance. We found only two studies assessing simultaneously the effect of dominance and complementarity on ecosystem functions. In the case of habitat provision for fishes, trait complementarity had no effect, while the dominance of structurally more complex plants positively affected fish abundance (Jones et al., 2021). In the case of primary production, dominance of taller plants with bigger leaves positively affected production (Angove et al., 2020), while complementarity was discarded as a significant driver.

There were, however, several studies on Z. marina assessing both the dominance and complementarity effects of genotypic diversity on ecosystem functions. Primary production is influenced by genotypic diversity of Z. marina at the plot level (Abbott et al., 2017). There is also evidence of intraspecific niche complementarity in the partitioned nutrient uptake of genotypes of Z. marina (Hughes et al., 2009). Dominance and complementarity hypotheses have been tested simultaneously in one study in Z. marina (Hughes and Stachowicz, 2011). Biomass production was higher in polycultures (i.e., higher complementarity) at high disturbance levels, whereas under no disturbance, monocultures (i.e., dominance) outperformed polycultures. Additionally, polycultures outperformed monocultures in shoot and biomass production under a macroalgal bloom. It is worth mentioning that, despite not being included in the literature review due to not being focused on the study of traits, there have been studies in communities including Z. marina linking taxonomic richness to resistance to shading (Gustafsson and Boström, 2013) and complementarity to increased biomass production (Salo et al., 2009).

The scarcity of trait complementarity vs dominance data on seagrasses highlights the complexity of assessing their relative importance, particularly under a changing environment. To test the dominance and complementarity hypotheses it is fundamental to find effect traits with proven mechanistic relationships with ecosystem functions. These relationships may be environmentally controlled and therefore it is necessary to include relevant environmental metrics in the study of ecosystem functions (van der Plas et al., 2020). This has been barely tested in seagrass communities and only in the case of three ecosystem functions (primary production, habitat provision for invertebrates and fishes and carbon storage, see Table 2 ). There is therefore a big knowledge gap in our understanding of how the functional traits of seagrass communities are linked to ecosystem functions, and how this provision will be altered under global change.

The worldwide rate of seagrass loss and the numerous threats to seagrass ecosystems (Orth et al., 2006; Waycott et al., 2009) call for the assessment of the vulnerability of the ecosystem functions provided by seagrasses. Therefore, it is necessary to study the correlation between response and effect in seagrass ecosystems ( Figure 1 : Hypothesis #7).

As stated in previous sections, the study of the response of traits to environmental drivers is common, particularly in the case of temperature or light (Tanaka and Nakaoka, 2006; Mota et al., 2018). Traits are sensitive indicators of plant stress under environmental change (Roca et al., 2016). When it comes to ecosystem functions, their vulnerability is generally discussed in terms of seagrass loss, i.e., the loss of the seagrass meadows would mean the end of certain ecosystem function provision (Trevathan-Tackett et al., 2018). However, one important missing link is the identification of response traits that drive function effects. The rationale is that, despite the presence of seagrasses, a change in their trait values or the replacement by another species or taxa with different traits may alter ecosystem function provision. This hypothesis has not been explicitly tested in seagrass ecosystems ( Figure 1 : Hypothesis #7).

The concept of ecosystem service has gained increasing importance in the last two decades, as a tool to couple science with environmental policymaking and management (Costanza et al., 1997; Costanza et al., 2017). The identification of effect traits responsible for the provision of ecosystem services is of fundamental importance to develop a taxon-independent metric that could be incorporated into policymaking and guide coastal management strategies. Ruiz-Frau et al. (2017) classified ecosystem functions performed by seagrass in ecosystem services based on the TEEB - The Economics of Ecosystems and Biodiversity categorization created in TEEB (2009). For example, fisheries are classified as food provision, while carbon burial and storage are classified as gas and climate regulation. Knowing which functions underlie each ecosystem service, and how to relate simple trait metrics to ecosystem function and service vulnerability, is fundamental to achieve a holistic view of seagrass response, function provision and service provision under a changing environment.