Functional macrophyte trait variation as a response to the source of inorganic carbon acquisition

Study site and field sampling

The study was performed in north-western Poland, in the Pomeranian Lakeland (53°48′51.1N, 17°38′00.9E) in 30 softwater lakes in June and August from 2014 to 2020. All lakes belong to softwater lake types and their environmental conditions represent a wide spectrum of softwater habitat, water acidity (pH 4.1–7.9) and calcium concentration (1.0–18.6 mg/L). The geographic coordinates and morphometric features of these lakes were presented in our previous study (Chmara, Szmeja & Robionek, 2019). To investigate species abundance the aquatic macrophytes were sampled in 30 lakes along depth zones in a transect, perpendicularly to the shoreline. In each of these lakes, one transect was delineated. At each transect, a diver randomly collected macrophyte samples until the maximum macrophyte occurrence depth. Macrophyte abundance was expressed as a cover-plant sample (squares with area = 0.1 m²). For study of several protected macrophyte species, permission of Regional Director for Environmental Protection in Gdańsk, Poland (No. RDOŚ-Gd-WZG.6400.92.2020.AB.2) was obtained. A total of 145 depth zones in 30 transects were designated to determine macrophytes presence and abundance.

List of macrophyte species divided into inorganic carbon acquisition

The information about species inorganic carbon acquisition was collected from the scientific literature: aquatic angiosperms (Maberly & Madsen, 2002; Iversen et al., 2019); bryophytes (Riis & Sand-Jensen, 1997) and charophytes (Baastrup-Spohr et al., 2015). Finally, each species was assigned to one of three inorganic carbon groups. A total of 30 species were selected and included in the statistical analyses.

Measurement of leaf traits

We measured four leaf traits of 30 macrophyte species: leaf area (LA mm²), leaf dry weight (LDW mg), specific leaf area (SLA mm² mg⁻¹) and shape trait circularity [4π(area × perimeter⁻²)]. Leaves were collected in June and August from 2016 to 2020 in 30 softwater lakes. Subsequently, 30 healthy leaves were collected from three to five individuals of each aquatic macrophyte species. Plant species names were checked according to The Plant List (http://www.theplantlist.org/). Leaf traits were determined following standardised methods of Pérez-Harguindeguy et al. (2013). We made measurements of charophytes and bryophytes functional traits. For measurement of charophytes traits we used the branchlets which are equivalents of the leaves of higher plants (Soulié–Märsche, 1999). For measurement of leaf traits, each leaf was photographed while fresh. Photos of bryophyte leaves and branchlets (charophytes) were taken using a Nikon Coolpix MDC Lens camera and Nikon SMZ 1500 stereomicroscope, Tokyo, Japan. Measurements of leaf area were calculated using ImageJ software. Leaf mosses and branchlets of the charophytes were weighed with a precision balance at 0.01 mg resolution. Mosses’ and charophytes’ specific leaf area was calculated as the leaf area (mm²) per unit of leaf dry mass (mg), determined with a precision scale. Leaves of vascular plants were assessed by using a standard flatbed scanner for leaf area; circularity was measured by means of ImageJ ver. 1.46 (http://imagej.nih.gov/ij) open-source software. Circularity was calculated according to Krieger (2014), circularity is mathematically constrained to range from 0 for a line to 1 for a circle.

All leaves were dried at 80 °C for 48 h, and the final dry mass was measured. Specific leaf area was calculated as the leaf area (mm²) per unit of leaf dry mass (mg). Species were classified into (1) free CO₂, (2) atmospheric CO₂ and (3) bicarbonate HCO₃⁻ groups based on the previous studies (Iversen et al., 2019; Maberly & Madsen, 2002).

Detailed information on the qualitative values of aquatic plants traits was archived in the AQUA-PLANT-TRAIT-UGDA DATABASE in the Department of Plant Ecology, University of Gdańsk.

Environmental data

In each lake we collected water samples and measured eight environmental variables during the vegetation seasons (in June and August) from 2016 to 2020. The samples were collected by SCUBA divers. A total of 465 water samples were collected in the depth zones; each sample containing 500 ml. The following environmental factors were determined in the depth zones: depth (m), visibility (m), photosynthetically active radiation (PAR, in % of the light reaching the water surface), pH of water, conductivity (µS/cm), calcium concentration (mgCa²⁺ L⁻¹), total nitrogen (mgN L⁻¹) and total phosphorus (mgP L⁻¹). The measurements were performed according to Eaton et al. (2005). PAR was measured in the depth zones with 0.5 m intervals by means of Licor LI–250 Light Meter.

Data analysis

Macrophyte species were divided into three carbon acquisition groups: (1) free CO₂, (2) atmospheric CO₂; (3) bicarbonate HCO₃⁻. We assessed differences in the functional traits (LA, LDW, SLA, circularity) into carbon acquisition groups using basic statistics and we applied coefficients of variations formula (CV = traits (SD)/traits (mean) × 100%, where SD - standard deviation). To test traits variations into three carbon acquisition groups, non-metric multidimensional scaling (nMDS) was performed. The nMDS algorithm was then used as Bray–Curtis distances between samples. The nMDS analysis was run in PAST ver. 4.05. To compare the values of leaf traits grouped into carbon acquisition, we used the non-parametric Kruskal–Wallis test, followed by Dunn’s multiple comparisons post hoc test. All trait values were log₁₀-transformed.

The RLQ analysis (R-mode; Q-mode; and L-link between R and Q) was applied to investigate the relationships between species functional traits (Q) and environmental variables (R) constrained by species abundance (L) (Dolédec et al., 1996; Stefanidis & Papastergiadou, 2019). This method, since its development, is widely applied in functional trait studies that combine separate analyses on multiple datasets to identify the relationships between traits and environmental variables, weighed by the abundances of species (Dolédec et al., 1996; Stefanidis & Papastergiadou, 2019; Zervas et al., 2019). Similarly to the method procedure described by Stefanidis & Papastergiadou (2019), the first step for RLQ analysis implementation is to create the ordinations analysis on each table, R, L and Q separately. Table R with the environmental variables is limited only by quantitative data; thus, the Principal Component Analysis (PCA) was applied as was pointed by Stefanidis & Papastergiadou (2019). As Zervas et al. (2019) explained that the Correspondence Analysis (CA) was performed on species data in table L. Next, the data analysis procedure for the functional trait table, Q Hill and Smith analysis (Hill & Smith, 1976) were used. The fundamental assumptions of this RLQ method are ordination positioning based on results of CA analysis depended on scores of sites and species data from table L, next the row weights obtained from PCA and in the end the results values of Hill and Smith analysis based on data from table Q (Zervas et al., 2019). The maximum covariance between data of the functional traits and related to them, the environmental variables are shown on the obtained graphs and reports corresponding to those graphs in the R software environment (Dray et al., 2014). Following Stefanidis & Papastergiadou (2019) data analysis procedure, this relationship’s overall significance was tested using a global Monte-Carlo test depending on the rows from table R and those of table Q. As Stefanidis & Papastergiadou (2019) explained, the contribution of each trait and environmental parameter to total inertia was used and presented as a measure of relative importance and helped us to identify the most important traits and environmental factors. All analyses related to the RLQ method were performed by using the ade4 package library (Dray & Dufour, 2007) in R environment version 4.0.2 (R Core Team, 2020).

Differences in macrophyte leaf traits between carbon acquisition groups

In total, functional trait data of 30 macrophyte species was collected, species were grouped into three carbon acquisition groups: free CO₂ (12 species), atmospheric CO₂ (3 species) and bicarbonate HCO₃⁻ (15 species; Table 1). Within the free CO₂ group we observed mainly mosses (including Sphagnum mosses) and isoetids, but in the atmospheric CO₂ group only floating-leaved species. Within the bicarbonate acquisition group we noted charophytes and vascular plants belonging to different growth-forms and leaf types. Interspecific traits variations ranged broadly, the means of LA, LDW, SLA and circularity were 0.85–4,095.7 mm², 0.003–36,120.5 mg, 10.9–342.4 mm² mg⁻¹, 0.005–0.920, respectively. LA varied among carbon acquisition groups from 81.0 mm² in the free CO₂ group to 7,473.4 mm² in the atmospheric CO₂ group, LDW ranged from 2.5 mg in free CO₂ group to 751.5 mg in the atmospheric CO₂ group and SLA ranged from 16.9 mm² mg⁻¹ in the atmospheric CO₂ group to 172.1 mm² mg⁻¹ in free CO₂ group (Table 2). Table 2 shows high interspecific variability among macrophyte functional traits, with coefficients (CV) of variation ranging from 19.96% to 264.45%. For circularity and SLA, the CV was lower than that of LA and LDW.

The Kruskal-Wallis tests showed significant differences among the four functional traits in relation to carbon acquisition groups (LA, SLA, LDW, Fig. 1, p < 0.001). LA, LDW and circularity in the free atmospheric CO₂ group were significantly higher compared to the other groups (values of Kruskal–Wallis test: χ² = 192.0, df = 2, p < 0.001; χ² = 199.2, df = 2, p < 0.001) LA and LDW of bicarbonate acquisition group varied not that much as in the free CO₂ and atmospheric CO₂ groups. However, SLA in the free CO₂ group showed relatively lower log-values than the other groups. Circularity in the free CO₂ and bicarbonate groups did not differ significantly (p = 0.48, Fig. 1). Additionally, the nonmetric multidimensional scaling (nMDS) analysis showed functional trait differences among carbon acquisition groups (Fig. 2), ANOSIM statistics of the assessed groups: R = 0.42, p = 0.002. Generally, nMDS diagram illustrates that the atmospheric CO₂ species were the least overlapping in the diagram, while the other two groups showed more similarities.

Environmental effects on leaf traits

The RLQ analysis explained very well the cross-covariance between species functional traits and environmental variables. First two RLQ axes explained 94.40% of the total inertia (1st axis = 76.79%; 2nd axis = 17.61%; Fig.1D; Table 3, Table S1). The first axis with environmental variables differentiated sites with higher conductivity, calcium ion concentration and pH of water (Fig. 3A). Among the environmental variables, Ca²⁺ concentration, conductivity and visibility have a higher share in the total inertia (Table 4). In the species functional traits, the first axis was positively correlated with the measured leaf dry weight (LDW), leaf area (LA). The second axis was correlated with the SLA and circularity which were negatively correlated to each other (Fig. 3B). Regarding the macrophyte functional traits, LA and LDW (Table 4) have the highest share in the total inertia. Species were also discriminated against each other according to these two axes (Fig. 3C). All investigated species were placed along the second axis where Sphagnum mosses and Warnstorfia exannuata were placed at the bottom, the species from elodeids and isoetids group dominated in the centre. At the top, the species with floating leaves were positioned. An exception to the mentioned rule was noted for Nuphar lutea whose position was more related to the first axis, and it was placed in the right-upper corner (Fig. 3C).

Response of leaf traits to carbon acquisition

The pH of water of the sampled lakes was between 4.1 and 7.9, which represents the full spectrum of softwater habitat. Under these conditions macrophytes take up three inorganic carbon forms (free carbon dioxide, atmospheric carbon dioxide and bicarbonate). Our results showed significant differences in macrophyte traits as a response to the source of inorganic carbon acquisition (Table 1, Figs. 1 and 2). Recent studies reported trait differences in growth-forms (Pierce et al., 2012), native/alien aquatic plants species (Lukács et al., 2017) and leaf types (Liu, Liu & Xing, 2021). We found no reports of functional traits of macrophytes with different inorganic carbon acquisition groups. We found an extreme range of leaf economic spectrum (leaf area, leaf dry weight and specific leaf area) and wide range of shape trait, circularity (Table 2). Among these groups we found high LA and LDW interspecific variations expressed as coefficients of variations. The highest values of specific leaf area were found in free CO₂ acquisition groups, especially Sphagnum mosses and Warnstorfia exannulata. Leaves of these species are small and extremely thin with typical one-cell thickness. The one-cell thick leaves permit the light and free CO₂ to reach photosynthetic cells directly (Glime, 2014). Furthermore, the consequence of high SLA is rapid and economic acquisition of CO₂ as a typical trade-off between rapid acquisition and conservation of resources (Wright et al., 2004).

In contrast, macrophytes using atmospheric CO₂ differ in leaf functional traits compared to previous groups. Leaves of aquatic plants that float, have stomata at upper surface (Rudall & Knowles, 2013). Moreover, they tend to decrease SLA and increase LDW and LA, and are more oval. Low CV of circularity indicated small shape differences. Leaf area trait of emergent macrophytes correlated with nutrient concentration (Garnier et al., 2001; Wright, Reich & Westoby, 2001). The RLQ analysis showed that Nuphar lutea leaf traits related to the first axis correlated with conductivity, calcium ion concentration and pH of water (Table 4, Fig. 3C). In our study area, Nuphar lutea occurred most often in acidic softwater lakes and softwater-lobelia lakes with acidophytic mosses, where it forms heterophyllous leaves, floating leaves with long petioles and submersed leaves with short petioles. The functional traits of these leaves are different; they take up free carbon dioxide and atmospheric carbon dioxide.

In our study, the number of 15 species (50%) in the bicarbonate acquisition group is close to 44% of the total 131 investigated submerged aquatic plants with the capability of using HCO₃⁻ investigated by Iversen et al. (2019). It should be noted that those plant species use bicarbonate as a carbon source but in the conditions where this source is limited they might also use CO₂, which is sometimes not strictly pointed out in the available literature (Maberly & Madsen, 2002; Iversen et al., 2019). Our study was performed within a huge range of the pH of water (from 4.1 to 7.9); thus, the species we investigated had the suitable conditions to use both above-mentioned carbon forms for the process of photosynthesis. Our study showed that the LA and LWD functional traits of bicarbonate acquisition group varied not that much as in the free CO₂ and atmospheric CO₂ groups (Table 2, Fig. 1), which might be related to the adaptation to permanently submerged conditions and ability to considerably take up carbon and other nutrients mainly from water (Maberly & Gontero, 2018). Our study confirmed that low specific leaf area in aquatic macrophytes might reflect the dominance of bicarbonate users (Lukács et al., 2019). Moreover, the CV of circularity was the highest in this group (Table 2), which is related to the greater variability of different types of macrophytes species (charophytes and vascular plants belonging to different growth forms and leaf types).

Differences in ecological strategies between carbon acquisition groups

Our study found high traits variations in the carbon acquisition groups. These findings showed the rapid carbon acquisition strategy of macrophyte species in softwater lakes. We agree with the previous study showing that aquatic plants exhibit numerous strategies to increase carbon uptake (Maberly & Gontero, 2018). This diversity explains well the three carbon acquisition strategies: avoidance, exploitation and amelioration (sensu Klavsen, Madsen & Maberly, 2011). We investigated that macrophytes follow these carbon acquisition strategies in the softwater lakes. Firstly, mosses with small leaves, extreme thin and high SLA live and grow in microhabitats with locally high free carbon dioxide and employ the avoidance strategies. Secondly, isoetids follow the exploitation strategies which involve morpho-anatomical features (lacunae in leaves and roots, thick cuticles) to higher concentrations of CO₂. Thirdly, these strategies also include floating-leaved species (e.g. Nuphar lutea, P. amphibia and P. natans) with access to CO₂ in the atmosphere. Furthermore, the amelioration strategies with energy-requiring processes utilise bicarbonate as a source of carbon. For water with air-equilibrium carbon dioxide concentration, the energy cost of photorespiration with diffusive carbon dioxide entry can exceed that of a carbon dioxide concentrating mechanism (often involving bicarbonate entry), which can largely suppress Rubisco oxygenase and hence photorespiration (Raven, Beardall & Giordano, 2014). In our study, amelioration strategies include numerous growth-forms: potamids, myriophyllids and charophytes (Table 1).

Knowledge gaps

The available scientific data on aquatic plant shape traits is incomplete, non-representative, mainly descriptive and has never been evaluated. A recent study based on the two leaf-shape types of macrophytes (flat-leaf type, needle-leaf type) showed different adaptive strategies to lake eutrophication and water depth-leaf shape relationships (Liu, Liu & Xing, 2021). Other studies emphasised that leaf shape, as an important phenotypic trait, can reflect the adaptation of macrophytes to environmental constraints (Maberly & Gontero, 2018; Pierce et al., 2012).

Our study indicated significant differences in leaf shapes as a response to carbon acquisition and higher contribution to total inertia of the first axis in RLQ analysis (Table 4, Fig. 3B). Furthermore, these findings are based on the trait metric expressed as circularity showing for the first time that aquatic macrophytes represent almost full spectrum of circularity metric (0.018–0.92). We propose circularity as a leaf-shape trait quick and easy to measure. It would also be interesting to assess relationships between leaf circularity and other traits into carbon acquisition groups.