Materials and methods

Study model

The study was carried out in the field, at the French sub-Antarctic Iles Kerguelen (South Indian Ocean) (48°30–50°S,68°27–70°35E). The climate of this archipelago is characterized by cold temperatures (annual mean of 4.6 °C with few variations - Frenot et al., 2006; Lebouvier et al. 2011), and abundant precipitations (range from 500 mm to 3200 mm along an East-West gradient - Frenot et al., 1998 and Meteo France). Many plant communities from polar regions are paucispecific, resulting in simplified plant-plant interactions, and subjected to highly reduced anthropogenically-induced environmental pressure. As such, they constitute a relevant model to infer mechanisms of community assembly, especially within constrained abiotic environments (Bergstrom & Chown, 1999).
The study was conducted in ponds from three sites located along the shore of the main island of the archipelago (Supporting information 1): Cap Molloy, Isthme Bas, and Cap Ratmanoff. These three sites are less than 40 km apart, and display similar climates. Conversely, overall abiotic conditions of the ponds are more variable locally as compared with variation levels among sites (Douce et al., submitted). For each site, 15 ponds of varying area were selected to overcome bias related to system size, which influences water temperatures (Bornette & Puijalon, 2011). All investigated ponds but one (180800 m²) have relatively small size (average area = 181.57 ± 382.39 m²) (Douce et al., in prep). They are shallow freshwater systems, enriched by nutrient inputs from marine fauna like seabirds and seals owing to their littoral proximity (Smith, 2008). Mean water depth was relatively similar among ponds (average of the 45 ponds: 15.52 ± 8.72 cm, Douce et al., in prep).
Macrophyte communities are mainly composed by six native species:Limosella australis R.Br. (Scrophulariaceae), Callitriche antarctica Engelm (Plantaginaceae), Juncus scheuchzerioïdesGaudich. (Juncaceae), together with three Ranunculaceae species:Ranunculus biternatus Smith, R. pseudotrullifoliusSkottsb., and R. moseleyi Hook.f. All of these species are perennials displaying a clonal network structure, i.e. plants are composed of related ramets (shoots with leaves and roots) that are connected through plagiotropic stems. C. antarctica, L. australis, R. pseudotrullifolius , and R. moseleyi are hydrophytes, whileR. biternatus and J. scheuchzerioides are helophytes. OnlyC. antarctica harbors a floating canopy, while the five other species display a rosette architecture and are rooted in pond sediments.

Field sampling

The study was conducted in 2017, 2018, 2019 and 2020, and the sampling of plant ramets was conducted each year, in the three sites, in November. In each pond, we randomly positioned three to five 1m × 1m quadrats depending on the pond area. In each quadrat, we collected one ramet (with its leaves and roots, and one connection internode) of each of the six above-mentioned species when present, to measure plant traits. In addition, we recorded species abundances within each quadrat.
Pond abiotic conditions were monitored every three months over a year, starting from November 2019, and ending in November 2020. For each sampling date and pond, water samples were collected manually for nutrients and phytoplankton concentration analyses. Sediment samples were also collected for subsequent nutrients analyses. In each pond, the three samples were collected at three distinct points of the pond, and were further mixed together to consider nutrient spatial heterogeneity. One composite water sample (30 mL) and one composite sediment sample (50 mL) were obtained per pond. As abiotic variables are likely to vary within ponds, we randomly positioned three (pond area < 5 m²) to five (pond area > 5 m²) quadrats (1 m²) and measured: water mean depth (based on three measurements), pH, electric conductivity (EC) and mean dissolved oxygen (DO – average of three depths in the water column) in each quadrat. These measurements were performed using a multiparameter HQ40D HACH sensor (C ± 0.1 mS.cm-1, DO ± 0.01 mg.L-1). Water temperature of the whole pond was also recorded every half hour using loggers (Hobo MX2202) positioned in the middle of the water column (From September 2020 onwards). Water temperatures were then simulated from November 2019 to August 2020 using their relationship with air temperatures recorded by Meteo France (2020 records, Port-aux-Français) [see Douce et al., in prep]. Each pond area (length × width) was measured in November 2019.

Trait measurements

Seven functional traits were measured on ramets sampled from the ponds; for each specimen, leaves, roots, and one connection internode was collected The selected traits characterize ramet growth strategies (Table 1), and are known to be responsive to a range of biotic and abiotic variables (Table 1). Three main categories of traits were investigated: aerial, clonal and root traits (see Table 1 for details).
Specific Leaf Area (SLA) and Leaf Dry Matter Content (LDMC) were measured following Cornelissen et al. (2003) on one healthy and mature leaf randomly per ramet. Internode length was measured on one of the internodes attached to the ramet. Internode mass was calculated as the ratio between internode dry mass and length (g.cm−1). Root mass was calculated as the ratio between root dry mass and maximum root length (g.cm−1). We also estimated performance of ramets resulting from their growth strategy, by measuring their total biomass (i.e. dry masses of leaves, stems, roots and one connection internode). Measurements of dry masses were performed after drying plant parts for 48 hours in an oven at 65 °C. All plants were cleaned in water before measurements.
Table 1. The plant functional traits measured and their associated responses to abiotic and biotic conditions.