Diverse traits of aquatic plants cannot individually explain their consumption by the generalist gastropod Biomphalairia glabrata

Field sites and sample collections

Thirteen aquatic plant species belonging of different families and that represented a wide range of life-forms (according Cook, 1996) and taxonomic groups were target (Table 1). All plants were collected in aquatic habitats inside the Restinga de Jurubatiba National Park, Rio de Janeiro State, Brazil (SISBio 30299). Immediately after collection, the plants were conditioned in thermal boxes and transported to the laboratory. The species were identified by one of the authors (Tatiana U.P. Konno) and voucher specimens have been deposited at the Herbarium of the Rio de Janeiro Federal University, Brazil. Samples of vegetative parts of each species were separated into three groups: (1) fresh plants for use in susceptibility to consumption assays (within 48 h of collection); (2) lyophilized and ground for the assays using reconstituted plants (as artificial food) and the nutritional analyses; (3) chemical extractions to investigate the defensive capabilities.

Specimens of the generalist herbivorous gastropod B. glabrata (Thomas, Nwanko & Sterry, 1985) (shell diameter = 1.7 ± 0.3 cm) were collected from Cabiúnas lagoon (22°16′S, 41°41′W) by actively searching through clumps of aquatic plant using a sieve. Upon collection, the gastropods were conditioned in thermal aerated boxes containing freshwater and transported to the laboratory. There, they were maintained in plastic containers (10 L) with freshwater changed daily, constant aeration, constant temperature (22 °C) and a 12:12 h light:dark photoperiod. They were fed on lettuce for 48 h until 1 h before the beginning of experiments to avoid starvation that could promote unnaturally feeding behavior during our assays (Cronin & Hay, 1996).

Susceptibility of the fresh whole aquatic plants to B. glabrata

The susceptibility of nine species of aquatic plants to consumption by B. glabrata was assessed according to a previous study (Cronin et al., 2002). Each experimental unit (replicate) consisted of a transparent plastic container (1 L) supplied with non-recirculating freshwater, in which one sample (approximately equal volumes to provide comparable encounter rates) of each of the nine types of aquatic plant was simultaneously offered to a single individual of B. glabrata for 3 days (n = 14) to determine the differential susceptibility of them to this gastropod. Only vegetative parts of each of the nine plant species were used. One sample of each plant species was set up under identical conditions, but without B. glabrata (as a control) to verify possible autogenic changes in biomass (independent of herbivory). Each plant sample was weighed at the beginning and at the end of the experiment, and the biomass consumed by B. glabrata was expressed as a percentage (%) of consumption, calculated according to the equation [H₀ ×(C_f/C₀) −H_f] ×100; where H₀ and H_f represent the biomass of the plant sample exposed to herbivory before and after the assay, respectively, and C₀ and C_f represent the biomass of corresponding control samples of the species from before and after the assay. For this bioassay, 14 replicates of each plant species were used.

Consumption of reconstituted aquatic plants by B. glabrata

To determine if plant structure influenced the consumption by B. glabrata, samples of the 13 collected aquatic plant species were frozen, freeze-dried, ground into a fine powder, and stored at −4 °C for later preparation as artificial foods. Artificial foods (i.e., reconstituted plants) were prepared in agar gel to reproduce the natural dry mass concentration of each species, as described previously (Bolser & Hay, 1998; Cronin et al., 2002). The amount of powdered plant incorporated as artificial foods varied for each species due to the specific relationships between dry mass and wet mass. The artificial food comprising each plant species was prepared by mixing 5.0 mL of distilled water with 0.36 g of agar, heating the mixture in a microwave oven until boiling, and then pouring it into cold water containing desired amounts of freeze-dried powder for each species (i.e., each species was rehydrated with the same amount of cold water to reestablish its wet mass). This mixture was allowed harden onto a plastic screen mesh (∼1 mm² per aperture) and flattened between two layers of wax paper. After cooling and solidification, the screen mesh was cut into smaller pieces of 10  × 10 mm² (total of 100 squares). Accordingly, all artificial foods presented the same morphology and hardness, but the original nutritional value and taste of each species was maintained as nuch as possible. For this bioassay, 30 replicates of each plant species were used.

The 13 artificial foods were simultaneously offered to one specimen of B. glabrata in 1.0 L plastic containers supplied with non-recirculating freshwater (n = 30) and consumption of them was calculated by scoring the number of squares of artificial food that were consumed. A fixed consumption stopping rule was applied (Lockwood III, 1998), where all assays were termined once >50% of all artificial food had been consumed.

Defensive capacity of the plant extracts against B. glabrata

The defensive capabilities of the extracts from ten aquatic plants against B. glabrata were evaluated by assaying B. glabrata’s consumption of an artificial food containing the extract (treatment) and a respective control (without extract). A lipophilic extract of each plant species was incorporated at a naturally occuring concentration (extract weight equivalent to the plant dry weight) to artificial food (treatment and control) prepared as described previously (Bolser & Hay, 1998; Cronin et al., 2002). For treatment foods, lipophilic aquatic plant extracts were first dissolved in CH₂Cl₂, added to a freeze-dried lettuce:broccoli mixture (1:1) that lacks defensive compounds, and then the solvent was removed by rotary evaporation. This procedure resulted in a uniform coating of the lipophilic extracts on the lettuvce:broccoli plant particles (Hay, Kappel & Fenical, 1994). For the control food, only the solvent was incoprporated into the lettuce:broccoli mixture. Artificial foods (treatment or control) were then prepared by adding 0.36 g of agar to 5.0 mL of distilled water, heating the mixture in a microwave oven until boiling, and then pouring it into 2 mL of cold water containing 0.5 g of freeze-dried powder of the lettuce:broccoli (1:1) mixture with (treatment) or without (control) extract. Treatment and control artificial foods were hardened separately onto a plastic screen mesh (∼1 mm² per aperure) and flattened between two layers of wax paper. After the artificial foods had cooled and solidified, the screen mesh was cut into strips comprising 100 squares (10 mm²) of artificial food pieces of 10  × 10 mm square (control and treatment). The feeding assays involved simultaneously offering a treatment (one particular plant extract) and a control strip of artificial food to one individual B. glabrata as a two-way choice. The assays were carried out in small plastic aquaria (500 mL) containing 250 mL of the water collected from Cabiúnas Lagoon. The number of replicates for each extract (species) varied between 18 and 26, and gastropod specimens were not used for more than one replicate.

Food consumption was estimated as described above under the section Consumption of reconstituted freshwater plants by B. glabrata, and the fixed-consumption stopping rule was again applied. Percentage inhibited consumption was was calculated according to the differences between consumption of control and treatment artificial foods, quantified as % Inhibition = [(C –T)/C] ×100 for each replicate, where C and T correspond to control and treatment squares consumed, respectively.

Extracts preparation

Extracts were obtained from a total of 10 aquatic plant species. Their vegetative parts were separated and air-dried under shade and in a climatized room at ambient temperature (mean 21 °C) for 48 h. In order to minimize the intra-individual variability among organisms, each plant was gathered and subjected to extraction together. Dried plants (X. brevifolia, 47.7 g of dry weight, d.w.; E. interstincta 27.3 g d.w.; T. domingensis 26.3 g d.w.; P. illinoensis 19.6 g d.w.; N. humboldtiana 95.0 g d.w.; E. azurea 54.7 g d.w.; S. auriculata 31.3 g d.w.; P. stratiotes 42.0 g d.w.; O. salzmannii 29.8 g d.w.; U. foliosa 193.0 g d.w.) were ground down and extracted with a mixture of dichloromethane:methanol, 1:1 (HPLC grade, Tedia), using 20 mL of solvent for each 1 g of plant dry mass for one day at room temperature. Then the solvent was filtered out and the biological material was re-extracted using the same solvent/plant mass ratio. This procedure was repeated twice, and then the solvents removed from these extractions were combined and dried by rotary evaporation (Buchi R210, V850, B491, Switzerland) to yield a total of 10 different plant extracts that were used in the feeding assays and to quantify phenolic contents. The extract from Potamogeton illinoensis was further subjected to chemical profile analysis by Proton Nuclear Magnetic Resonance (¹H NMR).

Physical, nutritional and chemical properties of the aquatic plants

Several tissue properties known to influence herbivory were measured for the 13 collected aquatic plant species. Powdered aquatic plants were analyzed for total nitrogen using the Kjeldahl method, as described previously (Conklin-Brittain et al., 1999), and the protein content was estimated by multiplying the value for total nitrogen by a factor of 6.25 (see Barbarino & Lourenço, 2005). Ash content was accurately determined via combustion of powdered biomass in a muffle oven at 550 °C (Jones, 2001). Lipids were extracted with diethyl ether using a Soxhlet system (see details in Ramluckana, Moodley & Bux, 2014). Results for ash and lipids contents are expressed as a percentage (%) by dry mass, total nitrogen as g% nitrogen (N) and protein content as g% protein (P). Phenol content was quantified from crude extracts of the plants using a modified Folin-Ciocalteu colorimetric method (Zhang et al., 2006) and the results are expressed as milligrams of GAE (gallic acid equivalent)/g of the extract. All analyses were performed in triplicate.

Chemical profile of Potamogeton illinoensis by ¹H NMR

The chemical profile P. illinoensis, the most active extract against B. glabrata, was analyzed by ¹H NMR in order to identify its major lipophilic compounds. NMR analysis was performed on the Varian VNMRS apparatus with 11.75 T (500 MHz) magnet. Deuterated chloroform was used to solubilize the sample.

Data analysis

To evaluate the various relationships between the consumption by B. glabrata and the different plant traits, data on consumption and the various physical, nutritional and chemical properties of the plant species s were compared according to species and also plant life-forms.

Non-parametric tests were used to analyze the herbivory data, as data was not show normally distributed. To compare among species, data on percentage consumption (%) whole plants and artificial foods (reconstituted plants) by B. glabrata were analyzed using the non-parametric Friedman test due data dependency (all species were offered simultaneously to the herbivore). Therefore, the comparison between the different life-form groups the % of consumption data of the whole and reconstituted aquatic plants were analyzed by Kruskal-Wallis non-parametric analysis of variance due to inequality in the sample size of the groups. For pair-wise comparisons of the non-parametric tests, the post-hoc Durbin-Conover and Dwass-Steel-Critchlow-Fligner test were performed to the Friedman and Kruskal-Wallis analysis, respectively. One Way ANOVA, followed by a post-hoc Tukey test was applied for comparisons of physical, nutritional and chemical properties of the aquatic plants. Pearson correlation coefficients were used to evaluate relationships among percentage consumption of whole plants, artificial reconstituted plants, extract-mediated inhibition of consumption, and various physical, nutritional and chemical traits of the plant species. For all statistical analyses, results were considered significant when p < 0.05. All statistical analyses of variance and correlation were performed in Jamovi version 1.6 freeware (R Core Team. 2020; Jamovi The Project, 2021).

A multivariate approach was conducted using the exploratory descriptive Principal Components Analysis (PCA) to summarize all data and allow access and visualization of a possible relationship between plant traits and herbivory inhibition. The consumption data, and physical, nutritional and chemical traits were summarized in a matrix that was submitted to a PCA. The physical, nutritional and chemical traits were used as supplementary quantitative variables. The PCA was performed in R version 4.0.3 with the function “PCAshiny” in the “Factoshiny” package (Vaissie, Monge & Husson, 2021). The scores results were represented by scatter plots of the principal components (PC) for visualization of the discriminant pattern of the samples (species and plant life-form) according to their distribution along the first two PC axes (PC1 =Dim 1 and PC2 =Dim 2). The loading results (consumption data, physical, nutritional and chemical traits) used to discrimination of the samples were represented by scatter plots.

Susceptibility of intact plants to B. glabrata herbivory

The percentage of fresh biomass of the nine aquatic species consumed by B. glabrata ranged from 2.08% (E. interstincta) to 0.45% (P. illinoensis) (median values, Fig. 1A), revealing three groups, two of which differed significantly from each other (χ² = 34.15, df = 8, p < 0.001, Tukey test). All plant species were consumed to some degree, but with the following hierarchy in terms of diminishing preference (Fig. 1A): E. interstincta > N. pulchella > E. azurea > H. nymphoides (group 1) > O. salzmannii > T. domingensis > N. humboldtiana (intermediate group, not significantly different from groups 1 and 2, p > 0.05) > S. auriculata > P. illinoensis (group 2).

There was no significant pattern in terms of plant biomass consumed according to plant life-forms (H = 349.94, df = 4, p = 0.055) (Fig. 2A), though B. glabrata had consumed more emergent plants (2.08%) than emergent floating plants (0.90%).

Consumption of reconstituted aquatic plants by B. glabrata

When we eliminated interspecific differences arising from physical/structural characteristics, but retained the taste, nutritional and chemical properties of the thirteen aquatic plants by offering reconstituted palnts as artificial food to B. glabrata, we obtained different results from those observed for fresh plants (Fig. 1A). Again, the gastropod consumed all 13 plants (Fig. 1B), but in significantly different amounts (χ² = 144.36, df = 12, p < 0.001, Tukey test). Moreover, distinct groupings were not observed and the hierarchy of diminishing preference was quite different: T. domingensis > X. brevifolia > N. pulchella >  E. interstincta > O. salzmannii > H. nymphoides > C. thalictroides > P. stratiotes > N. humboldtiana > S. auriculata > E. azurea > U. foliosa > P. illinoensis. For instance, Eichornia azurea that was highly consumed as a fresh plant was one of the least preferred as a reconstituted food. However, P. illinoensis remained unpalatable the least palatable (3.00% consumption, median value), suggesting it harbor a component that exerts an inhibitory effect on consumption by B. glabrata. In contrast, T. domingensis that was one of the least susceptible to consumption as a fresh plant, was the most preferred (% consumption, median value) as reconstituted food, indicating that it possesses structural components that deter consumption by this gastropod.

Considering the life-forms of the plants reconstituted as artificial food (Fig. 2B), we noted a significant decreasing preference by B. glabrata (H = 197.10, df = 6, p < 0.001, Tukey test): fixed emergent > helophyte > submerged to helophyte > emergent > emergent free floating > emergent floating > submerged. Accordingly, B. glabrata appears to exhibit a preference for consuming aquatic plants that occupy the marginal areas of the lagoons from which they were collected.

Defensive capabilities of the plant extracts

The nine extracts of aquatic plants we assayed exhibited significantly different defensive effects (H = 101.10, df = 8, p < 0.001, Tukey test) against consumption by B. glabrata (Fig. 1C), with inhibition (%) ranging from 100% (P. illinoensis) to 13.3% (S. auriculata), and with a hierarchy of diminished inhibition of: P. illinoensis > X. brevifolia > T. domingensis > E. azurea > N. humboldtiana > U. foliosa > E. interstincta > P. stratiotes > S. auriculata. In contrast, the extract from O. salzmannii significantly stimulated consumption by the gastropod. When we focused in plant life-forms, we found significant differences between species in terms of extract-mediated herbivory inhibition (H = 93.00, df = 6, p < 0.001, Tukey test), with the following hierarchy (Fig. 2C): emergent floating = helophyte > fixed emergent > submerged > emergent > emergent free floating. Chemicals from submerged and helophyte plants stimulated the consumption. For the emergent floating plant P. illinoensis, this result reiterates that obtained in previous assays about susceptibility to consumption by B. glabrata (Figs. 1A and 2A), as well as their consumption as reconstituted food (Figs. 1B and 2B), since in both they were the least consumed food items.

Physical, nutritional and chemical properties of the aquatic plants

The physical, nutritional and general chemical defense properties we assessed varied significantly among the plant species and also when analysed to life-forms (Table 2) Dry mass, ash, lipid, total nitrogen, protein, and phenol levels varied from five-fold (g% nitrogen) to almost thirty-fold (% lipid). Salvinia auriculata displayed the lowest percentage dry mass, whereas X. brevifolia scored highest in this respect, reflecting their life-form as submerged and helophyte plants, respectivley. Ash (%) was lowest in E. interstincta (7.10 ± 0.18%) and highest in P. stratiotes (39.42 ± 0.47%), and more generally was lower among emergent free-floating plants. Lipid contents dysplaied the widest variation, ranging from 1.16 ± 0.07% in E. interstincta to 20.52 ± 1.03% in submerged to helophyte O. salzmannii. This later species also presented the lowest levels of nitrogen and protein (0.46 ± 0.02 g% N and 2.85 ± 0.13 g% P, respectively) in marked contrast to N. pulchella that harbored the highest amounts (2.73 ± 0.03 g% N and 17.08 ± 0.16 g% P). Phenol levels also showed considerable amount variation, ranging from 3.84 ± 0.02 mg GAE/g extract in X. brevifolia to 77.98 ± 0.09 mg GAE/g extract in T. domingensis, representing helophyte and fixed emergent plants, respectively.

Relationships between plant traits and its consumption by B. glabrata

We did not uncover significant correlations between the susceptibility to consumption of fresh plants consumption of reconstituted plants, and the defensive capacity of the corresponding plant extracts (Pearson correlation coefficients, Fig. 3). Likewise, no significant relationships were observed between these three parameters and most of the various analyzed plant traits (ash, lipids, protein, total nitrogen and phenol) (Fig. 4), apart from a negative correlation between % extract inhibition and % lipids (R_Pearson = −0,75; p = 0.02). When we observe the PCA result, the first two axes of our PCA analysis summarized 76.78 % of the data variation (PC1 = 44.19 % and PC2 = 32.59 %, Fig. 5A, Fig. 5B and Fig. 5C). The ordination of the samples (Fig. 5A and Fig. 5B) and variables (Fig. 5C) formed by these axes support our results described in previous sections. Overall, the PCA divided the samples into two groups: group 1 in the PC1 negative and PC2 positive quadrant; and group 2 in the PC1 positive and PC2 positive quadrant. In terms of species, the ordination of the of species sample (Fig. 5A), P. illinoensis, E. azurea and U. foliosa and formed group 1, and E. interstincta, X. brevifolia, T. domingensis, N. pulchella and H. nymphoides formed group 2. Ordination according to plant life-form (Fig. 5B) discriminated two groups, i.e., emergent floating and submerged plants (group 1) and helophyte, fixed emergent and emergent plants (group 2). The ordination of the variables (loading, Fig. 5C) revealed that the group 1 encompassing plants whose extracts exhibited higher defensive activity against B. glabrata, exemplified by P. illionensis. The group 2 (Fig. 5C) represents the plants most susceptible to or preferred by B. glabrata as fresh or reconstituted food, in addition to revealing a strong correlation between dry mass and both consumption of fresh and reconstituted plants.

PCA analysis showed that the first two axes summarized 76.78% of the data variation (PC1 = 44.19% and PC2 = 32.59%, Figs. 5A, 5B, 5C). The ordination of the samples (Figs. 5A and 5B) and the variables in the space (Fig. 5C) formed by these axes reiterated the previous analysis of the data of the percentage of consumption of whole plants, reconstituted plants, inhibition of the extracts, and the several analyzed plant physical, nutritional and chemical traits and their correlations. Overall, the PCA divided the samples into two main groups: group 1 in the PC1 negative and PC2 in positive quadrant; and group 2 in the PC1 positive and PC2 positive quadrant. As main results, in the ordination of the species sample (Fig. 5A), it is possible to observe a group formed by P. illinoensis, E. azurea, U. foliosa and N. humboldtiana (group 1); and another group formed by E. interstincta, X. brevifolia, T. domingensis, N. pulchella and H. nymphoides (group 2). The ordination by life-form (Fig. 5B) discriminated two large plant groups: emergent floating and submerged plants (group 1) and helophyte, fixed emergent and emergent plants (group 2). The ordination of the variables (Fig. 5C) revealed that the group I includes those plants whose extracts exhibited higher defensive action against B. glabrata, mainly P. illinoensis. In contrast, group 2 (Fig. 5C) represents those plants most susceptible to or preferred by B. glabrata when offered to this gastropod either as a whole or as a reconstituted plant.

Chemical profile of P. illinoensis extract

The extract from P. illinoensis proved the most defensively capable against B. glabrata among the tested plants (Fig. 1C), so we further investigated its chemical profile by ¹H NMR. Characteristic signals of fatty acids were observed in the aliphatic region, i.e., between δ 0.86 parts per million (ppm) and δ 1.70 ppm. An α-CH₂ acyl group was observed at δ 2.32 (m), and strong overlapping resonance signals of methylene groups were also observed at δ 1.25 (sl). Additionally, we detected signals of exocyclic methylene (δ 4.0–5.0 ppm), methyl groups (δ 0.5–1.5 ppm) and hydrogen bonded to the sp² carbon of furane or lactone moieties (δ 6.09–7.11 ppm), which are characteristics of labdane diterpenes commonly reported in species of the Potamogeton genus (Waridel et al., 2003) were also detected.