Introduction

In the current context of biodiversity erosion, it is crucial to improve our understanding of the mechanisms underlying plant community dynamics for getting reliable predictions of the fate of plant species. Two decades ago, functional approaches have emerged and have allowed to rebuild community ecology and to produce general principles on assembly mechanisms (Calow, 1987; McGill et al., 2006). By considering the concept of ecological function, where species are no longer described by their taxonomic identities alone but by their biological attributes through the measurements of functional traits (Albert, Thuiller, Yoccoz, Douzet, et al., 2010), functional approaches have been fruitfully developed for addressing a range of key pending ecological questions. Functional traits correspond to any morphological, physiological, and phenological characteristics that impact individual performance, either directly or indirectly, through individual growth, reproduction or survival (Violle et al., 2007). More recent studies evidenced that large trait variations can also occur in the field at inter- but also intraspecific levels (Albert, Thuiller, Yoccoz, Soudant, et al., 2010; Messier et al., 2010; Taudiere & Violle, 2016). These variations would be the response to a range of biological processes operating at different spatial and temporal scales, and ultimately leading to species coexistence.
So far, organismal trait variation has usually been observed through the lens of space, time, or phylogeny independently, and this has resulted in a fragmented view of community assembly (but see Fyllas et al., 2020; Liu & Ng, 2020; Messier et al., 2010 for variance partitioning across spatial and phylogenetic scales). Temporal trait variation is, for example, reported as part of plant seasonal strategies allowing them to permanently adjust trait performance to better deal with external stresses. Adjustment of physiological (photosynthetic capacitye.g. Grassi et al., 2005; Muraoka et al., 2010) or morphological traits (leaf thickness e.g. Fullana-Pericas et al., 2017) is thus frequently reported in plant species. Spatial scales, owing to the diversity of environmental conditions they encompass, also shape overall trait variations, which can arise from species specific responses to various environmental filtering processes. Consistently, both aboveground (Reich et al., 2004) and belowground (Zadworny et al., 2016) traits have been found to spatially vary depending on environmental temperatures (Reich et al., 2004). By subsequently decomposing trait variations through phylogeny (McGill, 2008; Swenson et al., 2006) so that inter- vs. intraspecific trait variations can be considered, it is possible to determine whether species identity is a critical factor in the filtering processes or whether individuals are rather filtered out depending on their trait values. Overall, studies providing information on the scale which carries the most important trait variations are of particular interest, as the results should be promising to understand many community patterns and processes (McGill, 2008; Taudiere & Violle, 2016; Violle et al., 2007, 2012), and thus to assess the scale dependency issue in community ecology (e.g. , Cavender-Bares et al., 2009; Swenson et al., 2006).
Over the last ten years, studies reported the major role of intraspecific trait variability (ITV) in ecological processes, and showed that ITV can account for up to 30% to 40% of trait variance ​​within a plant community (Albert, Thuiller, Yoccoz, Soudant, et al., 2010; Jung et al., 2010). ITV would allow a better resistance to filters (Jung et al., 2010; Violle et al., 2012), in particular by adaptation (genetic variability) and/or phenotypic plasticity (Albert et al., 2011; Byars et al., 2007; Ghalambor et al., 2007). Most morphological traits in plants are key for coping with different environmental filters (seee.g. Lavorel & Garnier, 2002). Responses to both biotic and abiotic filters have indeed been extensively evidenced in traits involved in growth or resource acquisition (in particular Specific Leaf Area SLA, Leaf Dry Matter Content LDMC, height, and root traits) (Bittebiere & Mony, 2015; Butterfield & Callaway, 2013; Colom & Baucom, 2020; Cornelissen et al., 2003; Dostálek et al., 2020; Lavorel & Garnier, 2002), as well as in clonal traits (internode length, specific connection mass) (Bittebiere & Mony, 2015; Huber & Stuefer, 1997; Hutchings et al., 1997; Pilon & Santamaria, 2002; Rusch et al., 2011). It remains however unclear how traits will respond to simultaneously varying biotic and abiotic conditions, as their effects are usually studied separately (see e.g. Dudley & Schmitt, 1996; Greulich & Bornette, 1999; Riis et al., 2012; Suding et al., 2008). In some cases, this knowledge gap may lead to misinterpretation of patterns of traits variations, especially for studies conducted in situ(Cadotte & Tucker, 2017; Kraft et al., 2015). We thus argue that partitioning intraspecific trait variations should help determining the role of traits in species biotic and abiotic resistance to environmental filters (Le Bagousse-Pinguet et al., 2017; R. Liu et al., 2018). Traits would simultaneously respond to both biotic and abiotic conditions, but with varying strengths depending on the trait under consideration.
Variations in traits ultimately influence individual performance (i.e. production of vegetative biomass, and reproduction rates) (Geber & Griffen, 2003) and thus the species fate in the plant community (Ghalambor et al., 2007). However, direction and intensity of the influence of the trait on organismal performance will depend on the measured trait. For instance, in plant individuals, clonal traits mainly drive spatial positioning and resource storage (Klimešová et al., 2021), while aerial (e.g. height, SLA) and root traits (e.g.specific root length) determine efficiency of resources acquisition (Garnier et al., 2004). We thus expect aerial and root traits to have stronger importance on the intensity and direction of individual performance in comparison with clonal traits. Nevertheless, it is also very important to bear in mind that several traits co-vary and/or are subjected to tradeoffs, within plant individuals to coordinate different biological functions (Maire et al., 2013; Valladares et al., 2007; Violle et al., 2007). Yet, literature studies usually neglect these trait relationships that can result in indirect influence on individual performance. The identification of the direct and indirect influence of traits on organismal performance likely imposes hierarchies among plant traits, and this can be revealed through Structural Equation Modelling (Ackerly et al., 2000; Dwyer & Laughlin, 2017; Saiz et al., 2018; Vile et al., 2006).
Recently, a growing number of theories and concepts in ecology have been tested in the sub-Antarctic region, as they represent valuable sentinel ecosystems and open air laboratories (Bergstrom & Chown, 1999). In these regions, climate change, in particular warming, is especially rapid (Walther et al., 2002), with strong impacts on aquatic ecosystems, including ponds (Douce et al., in prep). Moreover, their plant communities (macrophytes) are remarkably poor, resulting in simplified interaction networks. In this study, we used macrophyte communities as system models, and measured different categories of traits (aerial, root, clonal, and of performances) on all pond-occurring species at the French sub-Antarctic Iles Kerguelen. Trait variations were studied across spatial, temporal and phylogeny scales, and covered multiple habitat abiotic and biotic conditions. Consequences for plant individual performance were also explored. The following hypotheses were specifically tested:
  1. Partitioning of trait variance among spatial, temporal, and phylogenetic scales should be quite promising as it would allow to detect the mechanisms involved in plant community assembly.
  2. Traits are expected to simultaneously respond to both biotic and abiotic conditions, but the direction and intensity of responses should be trait-dependent.
  3. Traits related to resource acquisition (aerial, and root traits) would affect individual performance directly, with stronger influence than clonal traits. However, relationships between traits may modulate these influences.