Picky carnivorous plants? Investigating preferences for preys’ trophic levels – a stable isotope natural abundance approach with two terrestrial and two aquatic Lentibulariaceae tested in Central Europe

Abstract

Background and Aims

Stable isotope two-source linear mixing models are frequently used to calculate the nutrient-uptake efficiency of carnivorous plants from pooled prey. This study aimed to separate prey into three trophic levels as pooled prey limits statements about the contribution of a specific trophic level to the nutrition of carnivorous plants. Phytoplankton were used as an autotrophic reference for aquatic plants as the lack of suitable reference plants impedes calculation of their efficiency.

Methods

Terrestrial (Pinguicula) and aquatic (Utricularia) carnivorous plants alongside autotrophic reference plants and potential prey from six sites in Germany and Austria were analysed for their stable isotope natural abundances (δ¹⁵N, δ¹³C). A two-source linear mixing model was applied to calculate the nutrient-uptake efficiency of carnivorous plants from pooled prey. Prey preferences were determined using a Bayesian inference isotope mixing model.

Key Results

Phytophagous prey represented the main contribution to the nutrition of Pinguicula (approx. 55 %), while higher trophic levels contributed a smaller amount (diverse approx. 27 %, zoophagous approx. 17 %). As well as around 48 % nitrogen, a small proportion of carbon (approx. 9 %) from prey was recovered in the tissue of plants. Aquatic Utricularia australis received 29 % and U. minor 21 % nitrogen from zooplankton when applying phytoplankton as the autotrophic reference.

Conclusions

The separation of prey animals into trophic levels revealed a major nutritional contribution of lower trophic level prey (phytophagous) for temperate Pinguicula species. Naturally, prey of higher trophic levels (diverse, zoophagous) are rarer, resulting in a smaller chance of being captured. Phytoplankton represents an adequate autotrophic reference for aquatic systems to estimate the contribution of zooplankton-derived nitrogen to the tissue of carnivorous plants. The autonomous firing of Utricularia bladders results in the additional capture of phytoplankton, calling for new aquatic references to determine the nutritional importance of phytoplankton for aquatic carnivorous plants.

Lentibulariaceae, Pinguicula alpina, Pinguicula vulgaris, Utricularia minor, Utricularia australis, stable isotopes, nitrogen, carbon, mixing model, trophic level, plankton

INTRODUCTION

Contrary to popular belief, not all green plants gain all their nutrients essential for living solely from photosynthesis and root uptake (e.g. Těšitel et al., 2010; Givnish, 2015; Gebauer et al., 2016). Carnivorous plants are one group of plants that tap into additional nutrient sources. As well as from photosynthesis and some root uptake, they can obtain mineral nutrients (e.g. nitrogen, potassium and phosphorus) by trapping and digesting animal prey (Ellison and Gotelli, 2009; Givnish, 2015; Pavlovič and Saganová, 2015; Ellison and Adamec, 2018). To contrive this capture, they evolved complex attraction and trapping structures (Givnish, 2015; Ellison and Adamec, 2018; Fleischmann et al., 2018) that have fascinated humans probably even before their first description by Charles Darwin (1875).

Carnivory has established itself in 12 plant families (Pavlovič and Saganová, 2015; Givnish, 2015; Fleischmann et al., 2018), which are spread over almost every climatic zone (Brewer and Schlauer, 2018). Therefore, the reasons for the establishment of plant carnivory are not primarily restricted to climatic zones but rather to habitat properties. Most often, habitats of carnivorous plants have in common that they are moist, rich in sunlight and offer a soil poor in nutrients (Givnish et al., 1984; Givnish, 2015; Ellison and Gotelli, 2001; Brewer and Schlauer, 2018). By inhabiting such niches, carnivorous plants can circumvent competition with other ‘non-adapted’ plants which often cannot thrive in such harsh conditions and become successful competitors themselves under these conditions. Nevertheless, adaption to carnivory results in a trade-off between benefits from prey capture and costs resulting from construction and preservation of traps, e.g. loss of photosynthetic leaf area (Ellison and Gotelli, 2001; Ellison, 2006; Pavlovič and Saganová, 2015). Due to the costly investments in trap structures (high carbon demand and secondary metabolites processed) (Pavlovič and Saganová, 2015) and increased carbon loss (Ellison and Adamec, 2011), the plant is able to adapt and reduce trap structures under increasing root-derived nutrient supply (Pavlovič et al., 2010; Millett et al., 2012).

Studies about nutrient fluxes from prey to carnivorous plants and the costs and benefits of preserving this life strategy can help improve our understanding of the role of carnivorous plants in ecosystem functioning and the importance of these plants in their unique environments. A method to analyse these questions in a representative and minimally invasive way is the use of stable isotopes to illustrate nutrient fluxes in carnivorous plant–environment interactions (Schulze et al., 1991). Following the idea of the cost–benefit model, an enrichment of the heavy stable nitrogen isotope (¹⁵N) can elucidate the nutritional gain from prey-derived nitrogen to plant tissue (e.g. Schulze et al., 1991, 1997; Moran et al., 2001; Pavlovič et al., 2011). The heavy stable carbon isotope (¹³C) demonstrates the costs induced by the construction of trapping organs, reduced photosynthesis, stomatal behaviour and transpiration (Moran et al., 2001; Ellison and Adamec, 2011; Pavlovič et al., 2011; Pavlovič and Saganová, 2015). By applying the two-source linear mixing model for carnivorous plants established by Schulze et al. (1991), the efficiency of translocating prey-derived nitrogen to the tissues of carnivorous plants can be calculated. This model is based on the different ¹⁵N enrichment in each trophic level of organisms. The lower endpoint is given by photo-autotrophic, non-carnivorous reference plants, which are depleted in ¹⁵N. Since the enrichment of the heavy nitrogen isotope increases by about 3 ‰ per trophic level (DeNiro and Epstein, 1981; Post, 2002), prey animals are often highly ¹⁵N-enriched and set the upper endpoint of the mixing model. A similar enrichment per trophic level (approx. 1.6 ‰) was described for carbon (DeNiro and Epstein, 1978; Parnell et al., 2010). The efficiency of a carnivorous plant in translocating nutrients to its tissue does not depend only on the type of prey captured, but also on the trap structure and geometry, trap size and the habitat (Schulze et al., 1991, 2001; Chin et al., 2010, 2014). Therefore, the efficiency of a carnivorous plant in taking up prey-derived nutrients can decrease during unfavourable conditions (e.g. less prey availability, drought or anthropogenic nitrogen deposition) (Givnish et al., 1984, 2018; Ellison and Gotelli, 2001; Ellison, 2006; Pavlovič et al., 2011; Millett et al., 2012; Givnish, 2015).

Despite being the largest carnivorous plant family (Fleischmann et al., 2018) and offering fascinating trapping structures (e.g. bladders or corkscrew), members of the Lentibulariaceae Rich. have been insufficiently investigated concerning their efficiency to obtain and incorporate prey-derived nutrients. There exists some knowledge about the influences of fertilization for the genus Pinguicula L. or the functioning of bladders of the genus Utricularia L. (Knight and Frost, 1991; Aldenius et al., 1983; Reifenrath et al., 2006), but a knowledge gap is present regarding the ecology and efficiency in nutrient uptake and translocation from prey in the Lentibulariaceae plant family. Additionally, comparable studies are often performed as laboratory feeding or tracer experiments and cannot be transferred to field conditions. Here we aim to complement knowledge for temperate species of the genera Pinguicula and Utricularia (Lentibulariaceae) by determining their nutrient gain and feeding preferences by applying a stable isotope natural abundance approach. Terrestrial Pinguicula has evolved sticky leaves to trap animal prey, consisting of small flying and dwelling insects (Aldenius et al., 1983; Fleischmann and Roccia, 2018) of different trophic levels, which affects the stable isotope signature of the carnivorous plant (Schulze et al., 2001, but for Dionaea Sol. ex J.Ellis). Temperate aquatic Utricularia establishes suction bladders to trap prey (Meyers and Strickler, 1979; Knight and Frost, 1991; Reifenrath et al., 2006; Peroutka et al., 2008; Jobson et al., 2018), which is described to be composed mostly of crustaceans and mosquito larvae (Jobson and Morris, 2001; Richards, 2001), but also algae and pollen due to an autonomous firing of the bladders (Peroutka et al., 2008; Adamec, 2011, 2018; Koller-Peroutka et al., 2015; Ellwood et al., 2019). Utricularia plants often form a mutualism with microfauna to attract microcrustaceans, thereby increasing trapping success and taking advantage of by-products of this microfauna (Albert et al., 2010).

Previous studies have identified the prey species caught by carnivorous species but used a prey pool for nutrient transfer analyses (e.g. Schulze et al., 1991; Millett et al., 2012), resulting in a mixed isotopic signal and wider variances due to the combination of different trophic levels in the prey. However, a study by Schulze et al. (2001) discovered distinct differences in ¹⁵N and ¹³C values of prey animals dependent on their nutritional behaviour (e.g. feeding on C₃ or C₄ plants, feeding on other animals). Therefore, in this study we aim to separate prey organisms into three main trophic level (phytophagous, diverse and zoophagous) and calculate their contribution to the nutrition of carnivorous plants, which can improve efficiency estimates and demonstrate the importance of a specific prey group. Additionally, the aquatic habitat of Utricularia requires an adaptation of autotrophic references. A study by Koller-Peroutka et al. (2015) applied some terrestrial autotrophic plants as references due to a lack of suitable aquatic species, which resulted in unrealistically high efficiencies calculated for the carnivorous plants. We therefore aim to apply phytoplankton as an autotrophic reference, given that these organisms are free floating and non-rooting similar to Utricularia and reflect the nitrogen signal of the surrounding water (Friday and Quarmby, 1994; Xin et al., 2010).

We performed a field condition analysis of temperate Pinguicula and Utricularia species to determine their efficiency to translocate prey-derived nutrients, and we demonstrate that the carnivorous species are enriched in ¹⁵N compared with autotrophic references, while prey is most enriched depending on its trophic level. The strength of the enrichment in ¹⁵N will depend on the preferred trophic level of prey, with moderate enrichment for a preference for phytophagous prey and higher enrichment for a preference for zoophagous prey. We will show costs of carnivory with ¹³C either being enriched due to stomatal closure or depleted due to higher transpiration compared with non-carnivorous references. Phytoplankton depleted in ¹⁵N will be applied as the autotrophic reference for ¹⁵N-enriched aquatic Utricularia while zooplankton represent the most ¹⁵N-enriched trophic stage.

MATERIALS AND METHODS

Study sites and sample preparation

Terrestrial Pinguicula vulgaris L. and P. alpina L. and aquatic Utricularia australis R.Br. and U. minor L. (Fig. 1) were collected at six sites in Germany (North-Eastern Bavaria) and Austria (Vorarlberg) from June to August 2015 and 2016. Sampling was conducted in five 1 m² plots according to the methodology established by Gebauer and Meyer (2003). Pinguicula species were sampled twice, in 2015 close to the location Garfülla and in 2016 close to both Damüls and Bingarten. At the Garfülla site (creek plot), P. alpina grew in a moist meadow in a montane climate on limestone soil near the Faludriga stream (47°11′23″ N, 9°55′23″S, 1326 ± 12 m a.s.l.), and P. vulgaris (47°11′41″N, 9°53′20″E, 1041 ± 6 m a.s.l.) was sampled on limestone near a forest inclination (precipice plot). The Damüls site (marshland plot) comprised an open, montane limestone soil meadow for P. alpina (47°16′48″N, 9°53′1″E, 1487 ± 11 m a.s.l.), while the Bingarten site (shaded plot) of P. vulgaris was located at a spruce forest glade in a temperate climate on phyllite–quarzite–tuffit soil (49°51′47″N, 11°59′32″E, 590 ± 4 m a.s.l.). Three fully expanded, clean leaves were taken from each Pinguicula species together with leaves from three autotrophic reference plants each (Supplementary Data Table S1). We excluded Tofieldia calyculata as a reference plant for our analysis as this plant revealed isotopic nitrogen patterns which were significantly different from the other references. Very recently, other members of the plant family Tofieldiaceae were identified as carnivorous or protocarnivorous. Capture of prey by Triantha japonica (Sean Graham, pers. comm.) was identified, and loss of plastid NADH dehydrogenase for T. occidentalis and ndh loss for T. glutinosa, indicating heterotrophy (Ross et al., 2016).

Fig. 1.

Carnivorous plant trap structures. (A) Sticky leaf trap of Pinguicula. The scale bar = 1 cm. (B) Bladder trap of Utricularia. The scale bar = 0.5 cm. Photographs were taken with a Sony α6000 camera.

Potential prey animals (n_P.alpina = 32, n_P.vulgaris = 24) were either collected from leaves if still alive or sampled with the aid of an exhauster (11 × 1 cm, Bioform, Nuremberg, Germany) and shock-frozen in glass vessels. Before oven drying, they were identified and graded per trophic level with identification literature (Müller et al., 2015). Prey consisted of cicadas (Auchenorrhyncha, Hemiptera), moths (Coleophora), spiders (Arachnida), dance flies (Empididae, Diptera), caddisflies (Trichoptera), mosquitoes (Nematocera), springtails (Collembola), drain flies (Psychodidae, Nematocera), flies (Brachycera, Diptera), bugs (Heteroptera), beetles (Coleoptera) and ants (Formicidae, Hymenoptera). After drying (105 °C, 48 h), prey animals were stored in desiccators filled with silica gel.

Utricularia species were hand harvested in water bodies in five plots together with autotrophic (phytoplankton) and heterotrophic (zooplankton) reference material. Utricularia australis grew in a pond close to North-Köglitz (49°51′32″N, 11°56′2″E, 485 m a.s.l.) with a mean water conductivity of 183 μS cm^–1 (22∙1 °C) (conductivity meter WTW 82362, Weilheim, Germany) and a pH of 5∙4 (pH meter HI 991301, ATP Messtechnik & Waagen, Ettenheim, Germany), while U. minor was sampled in a forest lake close to Parkstein-Hütten (49°41′32″N, 12°1′19″E, 411 m a.s.l.) rich in humic acids with a mean water conductivity of 136 μS cm^–1 (20∙0 °C) and a pH of 5∙2. Plant material of Utricularia species was stored in sealed bags on ice as described by Wang et al. (2015) before transport to the laboratory and cleaning with deionized water. The collected shoot segments were between 2 and 3 months old. Bladders were separated from shoot material with tweezers, opened, and rinsed with deionized water to ensure that only bladders free of any visible prey were analysed and to avoid an influence of the microbial community inside the bladders on the isotopic values. The cleaned bladders and the shoots were dried and analysed separately for nutrient translocation from trapping bladder to plant tissue. Additionally, three species of autotrophic reference plants from the shore were collected to test for their suitability as references for aquatic Utricularia. These terrestrial references were prepared as described for the autotrophic references of Pinguicula species. Plant material of carnivorous species and references were oven-dried to constant weight at 105 °C for 48 h before being ground to a powder using a ball mill (MM2, RETSCH, Haan, Germany). Ground plant material was stored in desiccators filled with silica gel until measurement.

Plankton samples of Utricularia habitats were retrieved with a plankton net (mesh size 250 μm, Hydro Bios, Kiel, Germany) at each sampling location. The pond where U. australis was found was dominated by zooplankton, while the forest lake where U. minor grew was characterized by phytoplankton. Zooplankton consisted of Daphnia and Bosmina (cladocerans), mosquito larvae, polychaetes, copepods (Cyclops), nematodes (Adenophorea, Secernentea) and diving beetles (Dytiscidae, Coleoptera). Regarding the phytoplankton, species of the Euglenophyta (Euglena and Colacium), Streptophyta (Micrasterias, Cosmarium and Closterium), Bacillariophyta (Pinnularia, Cymbella and Nitzschia), Dinophyta (Peridinium), Chlorophyta (Chlamydomonas and Gloeocystis), Xanthophyta (Tribonema) and Cyanobacteria (Anabaena) were identified. Plankton was sub-sampled (pure and alcohol fixed: 70 % ethanol) in PE bottles. Unfixed plankton samples were filtered onto glass fibre filters (Whatman GF/F, diameter 47 mm, 0.7 μm thick; Whatman plc, GE Healthcare, Maidstone, UK) and dried in an oven at 105 °C for 24 h as described by Syväranta et al. (2006). Success of filtration was tested microscopically. Filters were stored in Petri dishes until small parts of the filters were punched out and weighed into 5 × 12 mm tin capsules.

For both carnivorous plants, pollen grains (e.g. from Pinus or Picea) and debris are often found as potential prey (Karlsson et al., 1994; Koller-Peroutka et al., 2015). During sampling in the field, no significant amounts of pollen were identified at the sampling sites. In addition, microscopic observation of the water samples revealed only a few pollen grains, which were excluded from the measurement by selecting the filter parts mostly free of pollen. To focus on the nutrient transfer from living prey, the impact of pollen grains or debris was not taken into account in this study.

In total, 20 samples of two Pinguicula species, ten phylloclade samples of two Utricularia species, 90 samples from 11 accompanying autotrophic reference plant species and 56 prey animals were collected, together with ten water samples each for zooplankton and phytoplankton accompanying Utricularia species (Supplementary Data Table S1).

Determination of stable isotope natural abundance and calculations

Measurements of nitrogen (N) and carbon (C) stable isotopic ratios (¹⁵N/¹⁴N, ¹³C/¹²C) were conducted with an EA-IRMS (Elemental analyser: NA 1108, CE Instruments, Milan, Italy; Interface: ConFlo III, Finnigan MAT, Bremen, Germany; IRMS: Delta S, Finnigan MAT, Bremen, Germany) at the BayCEER Laboratory of Isotope Biogeochemistry at the University of Bayreuth (Germany). This method allows the simultaneous measurement of the nitrogen and carbon concentrations of the samples (mmol g d. wt^-1) (Supplementary Data Table S1). Ground sample material was weighed into tin capsules (4 × 6 mm; animals up to 1 mg, plants 2.8–3.5 mg) with the aid of an electric micro balance (CPA2P, Sartorius, Göttingen, Germany). Air was used as the standard for nitrogen, and Vienna Pee Dee Belemite (V-PDB) as the standard for carbon. Acetanilide [C₆H₅NH(COCH₃), MERCK KGaA, Darmstadt, Germany] was used to calibrate the C/N concentrations and analysed with different sample weight each set of ten samples. The calibration of the nitrogen isotope compositions was done with the IAEA standards N1 and N2, and calibration for carbon isotopes with the IAEA standards ANU sucrose and NBS 19. For the C and N evaluation of the samples, the software ISODAT (V: 2.0, Thermo Fisher Scientific) was applied. The resolution of the measurement is ± 0.2 ‰. Isotopic values are annotated as the delta value (δ), defined as: δ¹⁵N or δ¹³C = (R_sample/R_standard – 1) × 1000 (‰), with R being the ratio of the heavy to the light isotope of the sample or the respective standard.

Drivers for the fractionation of ¹³C in aquatic habitats strongly depend on habitat conditions (e.g. pH, temperature and boundary layer effect) and the diffusion coefficient for gases is 10⁴ times lower in water than in air. Therefore, here the ¹³C values were not considered for the analysis of Utricularia. Single δ values and nitrogen concentrations of the Pinguicula and Utricularia species, their respective reference plants, prey animals, phytoplankton and zooplankton analysed in this study are listed in Supplementary Data Table S1.

Calculations and statistical analyses

To enable a comparison of the isotopic values of carnivorous species from different locations, a transformation of the δ values to epsilon values (ε) according to Preiss and Gebauer (2008) was performed: ε_TP = δ_TP – δ_meanRP, with TP being the target plant/target prey animal and RP representing the reference species of the respective location.

The contribution of animal N to the nitrogen content of the carnivorous plant (% N_CP, nitrogen-uptake efficiency) was calculated following the methodology of Schulze et al. (1991) adapted from Shearer and Kohl (1989). Therefore, the δ¹⁵N value of the carnivorous plant (δ¹⁵N_CP), the δ¹⁵N of the non-carnivorous reference plants (δ¹⁵N_RP) and the δ¹⁵N value of the prey animals (δ¹⁵N_PA) were applied:

$${\displaystyle \begin{array}{l}{\displaystyle \mathrm{\%}{\mathrm{N}}_{\mathrm{C}\mathrm{P}}=\frac{{\delta }^{15}{\mathrm{N}}_{\mathrm{C}\mathrm{P}}-{\delta }^{15}{\mathrm{N}}_{\mathrm{R}\mathrm{P}}}{{\delta }^{15}{\mathrm{N}}_{\mathrm{P}\mathrm{A}}-{\delta }^{15}{\mathrm{N}}_{\mathrm{R}\mathrm{P}}}}\end{array}}$$

The carbon gain efficiency (%C_CP) of carnivorous plants from prey animals was estimated independently from carbon stable isotopes to prevent influences of soil respiration and plant physiological behaviour towards environmental conditions (e.g. stomatal regulation). The calculation is based on the proportion of the prey-derived carbon content (cC_PD) that contributes to the total carbon content of the plant (cC_CP), given in concentrations (c, mmol g d. wt^–1).

We used the nitrogen-uptake efficiency of the carnivorous plant (%N_CP), calculated with the use of the end member mixing model, to determine the amount of carbon (cC_PD) that was captured concurrent with nitrogen. The estimated prey-derived carbon dry weight (cC_PD) was set into relation to the total carbon concentration of the carnivorous plants (cC_CP), resulting in the carbon gain efficiency of the carnivorous plants (%C_CP).

First, the total prey-derived nitrogen concentration of the carnivorous plant (cN_PD) was calculated over the total nitrogen concentration of the carnivorous plant (from soil and prey, cN_CP) related to the nitrogen-uptake efficiency (%N_CP).

$${\displaystyle \begin{array}{l}{\displaystyle \left[c{\mathrm{N}}_{\mathrm{P}\mathrm{D}}\right]=\left[c{\mathrm{N}}_{\mathrm{C}\mathrm{P}}\right]\times \left[\mathrm{\%}{\mathrm{N}}_{\mathrm{C}\mathrm{P}}\right]/100\mathrm{\%}}\end{array}}$$

Secondly, the total prey-derived carbon concentration of the carnivorous plant (cC_PD) was determined by relating the total prey-derived nitrogen concentration of the carnivorous plants to the C/N ratio of the respective prey animals (C_P/N_P).

$${\displaystyle \begin{array}{l}{\displaystyle \left[c{\mathrm{C}}_{\mathrm{P}\mathrm{D}}\right]=\left[c{\mathrm{N}}_{\mathrm{P}\mathrm{D}}\right]\times \left[{\mathrm{C}}_{\mathrm{P}}/{\mathrm{N}}_{\mathrm{P}}\right]}\end{array}}$$

To calculate the prey-derived carbon-uptake efficiency, the total concentration of prey-derived carbon of the carnivorous plant (cC_CP) was related to the concentration of prey-derived carbon within the carnivorous plant (cC_PD).

$${\displaystyle \begin{array}{l}{\displaystyle \mathrm{\%}{\mathrm{C}}_{\mathrm{C}\mathrm{P}}=\left[c{\mathrm{C}}_{\mathrm{P}\mathrm{D}}\right]\times 100\mathrm{\%}/\left[c{\mathrm{C}}_{\mathrm{C}\mathrm{P}}\right]}\end{array}}$$

Statistical tests were generated using the software RStudio [version V: 0.99.467 (R Development Core Team, 2016)] and SigmaPlot (V: 11.0 Systat Software). The data were tested for normality (Shapiro and Wilk, 1965) and homogeneity [Levene test; package ‘car’, (Levene, 1960; Fox and Weisberg, 2011)]. Depending on the number of groups, either Student’s t-test and one-way analysis of variance (ANOVA) for parametric data, or Wilcoxon–Mann–Whitney U-test and Kruskal–Wallis H-test for non-parametric data were applied (Zar, 2010). Within the Kruskal–Wallis H-test, Dunn’s method was applied for the pairwise comparison procedure when an unequal sample size was apparent (Q; Dunn, 1964), while the Tukey method was used for equal sized samples (q; Tukey, 1949). The level of significance was set to α < 0∙05.

For source partitioning of prey-derived nutrients, a Bayesian inference isotopic mixing model was applied following the R package ‘SIAR’ version 4.2 (Parnell et al., 2010; Parnell and Jackson, 2013). It contained the stable isotope signatures of the consumer plants (P. vulgaris and P. alpina) and stable isotope signatures and concentrations of nitrogen and carbon of the source animals separated into phytophagous, diverse and zoophagous prey.

The model was run with 500 000 iterations and a burn-in with an initial discard of 50 000 which is suitable for most data sets. The output results in 30 000 posterior draws. Following the recommendation of Phillips et al. (2014), the model includes fractionation values per trophic level as a correction (¹⁵N, 3∙54 ± 0∙74; ¹³C, 1∙63 ± 0∙63). Those are in fact influenced by several factors, for instance the environment and animals’ metabolism, which can cause some variance (Phillips et al., 2014). Significances were tested by comparing the interquartile range applying Student’s t-test after validating for normal distribution using the Shapiro–Wilk test.

RESULTS

Significant differences in δ¹⁵N and δ¹³C values between Pinguicula species (δ¹⁵N, P = 0∙015; δ¹³C, P = 0∙001), reference species (δ¹⁵N, P = 0∙024; δ¹³C, P < 0∙001) and some prey groups graded by trophic level (phytophagous δ¹⁵N, P = 0∙868; δ¹³C, P = 0∙452; diverse δ¹⁵N, P = 0∙117; δ¹³C, P = 0∙757; zoophagous δ¹⁵N, P = 0∙016; δ¹³C, P = 0∙158) were detected. The same holds true for aquatic Utricularia species (δ¹⁵N, P = 0∙008; δ¹³C, P = 0∙315) and plankton groups (δ¹⁵N, P < 0∙001; δ¹³C, P < 0∙001). These significant differences indicate a location dependency. Therefore, δ values (Table 1) were transformed into location-independent ε values (Figs 2 and 3), which will be referred to below.

Table 1.

Mean δ¹³C and δ¹⁵N values ± s.d. of the carnivorous target species and their respective reference plants separated by sampling locations

Species	Location	δ13C	δ15N
P. vulgaris*	Precipice plot	–32.9 ± 0.6	–0.6 ± 0.4
Reference plants		–30.6 ± 0.7	–2.4 ± 1.1
P. vulgaris	Shaded plot	–28.5 ± 0.3	–0.7 ± 4.0
Reference plants		–30.3 ± 0.8	0.2 ± 4.7
P. alpina*	Creek plot	–33.6 ± 1.0	–3.4 ± 0.6
Reference plants		–31.9 ± 0.6	–5.3 ± 0.9
P. alpina	Marshland plot	–28.8 ± 0.2	2.3 ± 1.6
Reference plants		–30.2 ± 2.2	–1.1 ± 3.1
U. australis	North-Eastern Bavaria	–29.9 ± 0.4	8.9 ± 0.9
U. minor		–32.5 ± 5.4	3.6 ± 0.6
Reference phytoplankton		–27.0 ± 0.1	2.6 ± 0.3

*Tofieldia calyculata was excluded as a reference plants (see the Material and Methods).

Fig. 2.

Mean enrichment factors ε¹³C and ε¹⁵N and standard deviation of four terrestrial, carnivorous Pinguicula species (n = 20) from six sites in Germany and Austria compared with fully photoautotrophic reference plants (n = 50, green box) and potential prey animals (n = 56). While physiological behaviour (e.g. transpiration and stomatal closure) drives ¹³C depletion during high photosynthetic and transpirational activity (precipice plot and creek plot), an enrichment results from stomatal closure and prey-derived carbon uptake (shaded plot, marshland plot). The Pinguicula species mainly locate intermediate between reference plants and prey animals in ¹⁵N due to prey capture (except for P. vulgaris at the shaded site), whereby phytophagous animals show the highest contribution to the total nitrogen supply.

Fig. 3.

Mean enrichment factors ε¹⁵N and standard deviation of aquatic carnivorous Utricularia species (n = 10) from Germany, compared with photoautotrophic phytoplankton as reference and zooplankton as potential prey. Carnivorous Utricularia species are expected to locate intermediate between phytoplankton and zooplankton in ¹⁵N due to prey capture (20–30 % zooplankton-derived nitrogen). Statistical significant differences (ANOVA, Tukey post-hoc) between groups are highlighted by different letters.

Natural abundances of stable isotopes for Pinguicula

For the carbon stable isotope (ε¹³C), the Kruskal–Wallis test calculated significant differences between the groups of reference plants, Pinguicula species and prey animals [H(2) = 66∙134, P = 0∙001]. Dunn’s post-hoc test estimated significances between prey animals and Pinguicula species [Q(2) = 5∙669, P < 0∙05] as well as prey and reference plants [Q(2) = 7∙468, P < 0∙05], while Pinguicula species and reference plants did not vary significantly.

Within the Pinguicula species, differences in ε¹³C [H(3) = 15∙251, P < 0.01] were discovered (Fig. 2). The creek and precipice plot Pinguicula species were more depleted, while marshland and shaded Pinguicula species were enriched in ε¹³C (Fig. 2). Pinguicula species were significantly distinguishable from their phytophagous [t(33) = –4∙117, P < 0∙001], diverse [U(28) = 18, P < 0∙001] and zoophagous [U(48) = 23, P < 0∙001] prey animals in ε¹³C. According to Kruskal–Wallis test, the trophic levels of prey animals were not different in ε¹³C [H(2) = 5∙001, P = 0.08].

Nitrogen stable isotope (ε¹⁵N) showed a stepwise enrichment from depleted reference plants (0 ± 1∙93 ‰) to moderately enriched Pinguicula species (1∙52 ± 2∙06 ‰) to the most enriched prey animals (4∙97 ± 4∙00 ‰) [H(2) = 51∙988, P < 0∙001] (Fig. 2; Supplementary Data Fig. S1). Dunn’s post-hoc test confirms significant differences between all three mentioned groups. The Pinguicula species of the four locations show significant differences [H(3) = 14∙246, P = 0∙003]. After applying Dunn’s post-hoc test, marshland Pinguicula and shaded Pinguicula were significantly distinguishable (P < 0∙05) from each other and from all other Pinguicula species in ¹⁵N, while P. alpina (creek plot) and P. vulgaris (precipice plot) were not significantly different (P > 0.05) in ¹⁵N (Supplementary Data Fig. S1). The nitrogen concentration was significantly lower for shaded P. vulgaris and the respective reference plants than for the other Pinguicula species and their reference plants [F(3) = 39∙882, P < 0∙001].

Separated into trophic levels, prey animals differed in ε¹⁵N when comparing zoophagous with phytophagous and zoophagous with diverse prey, respectively [Q(2) = 4∙979, P < 0∙05; Q(2) = 2∙838, P < 0∙05] (Supplementary Data Fig. S1). Pinguicula species were similar to phytophagous prey [t(34) = 0∙341, P > 0∙05], but differed significantly from diverse [U(28) = 32, P = 0∙003] and zoophagous prey animals [U(48) = 42, P < 0∙001] (Fig. 2).

Nutrient gain estimation

Nitrogen gain was calculated successfully for precipice plot P. vulgaris, creek plot P. alpina and marshland P. alpina in ranges of 50∙7 ± 5∙9, 37∙0 ± 9∙1 and 54∙8 ± 15∙8 %, respectively. For the shaded P. vulgaris, insignificance of isotopic values between the carnivorous plant, reference plants and some prey groups resulted in unreliable nutrient gain efficiencies. The mean values of the calculated carbon-derived nutrients (methodology described in the Materials and Methods) from prey ranged between 7∙2 ± 1∙0 and 7∙8 ± 1∙6 % for precipice P. vulgaris and creek plot P. alpina and 12∙6 ± 3∙2 % for marshland P. alpina.

Pinguicula prey preferences

The Bayesian inference isotopic mixing model calculated the highest nutrient gain from phytophagous prey of about 55 ± 4∙7 % for investigated Pinguicula species, with an interquartile range of 36–74 %. Diverse prey contributed about 27 ± 1∙5 %, while the lowest contribution to Pinguicula’s nutrient supply from prey was found for zoophagous prey (17 ± 3∙2 %), with an interquartile range between 13 and 42 % and between 4 and 30 %, respectively (Supplementary Data Fig. S2). These results were significant for all combinations of Pinguicula species (P < 0∙01). The marshland P. alpina consumed the highest amount of phytophagous prey, which was significantly more (65 %, P < 0∙05) than this species consumed at the precipice and creek plots, where 50 % of the total nutrient supply from prey was gained from phytophagous prey. In contrast, 20 % of the total nutrient supply from prey was received from diverse prey for all investigated Pinguicula species across all plots but tended to be lower for marshland P. alpina. The contribution of zoophagous prey was equally low for all Pinguicula species, but lowest for marshland P. alpina (P < 0∙05) (Supplementary Data Fig. S2). The Bayesian inference mixing model suggests a negative regression for phytophagous prey compared with both diverse (–0∙75) and zoophagous (–0∙52) prey when considering all investigated species. This emphasizes that phytophagous prey represent the trophic level comprising the main contribution to total nutrient supply from prey. Pinguicula vulgaris at the shaded plot were excluded from the Bayesian inference isotopic mixing model calculations, as absence of nitrogen stable isotope enrichment indicated no significant nitrogen transfer from prey to plant.

Abundance of natural stable isotopes for Utricularia

One-way ANOVA confirmed a statistically significant enrichment of Utricularia species (1∙2 ± 0∙7 ‰) in ε¹⁵N, compared with depleted phytoplankton (0 ± 0∙3 ‰) and most enriched zooplankton (4∙8 ± 0∙8 ‰) [F(3) = 95∙438, P < 0∙001]. Tukey post-hoc procedures did not demonstrate significant differences between U. australis and U. minor [q(3) = 1∙426, P > 0∙05] (Fig. 3). For U. minor and U. australis, 20∙8 ± 11∙2 % and 29∙3 ± 18∙4 % nitrogen gain, respectively, was calculated

Most of the terrestrial autotrophic shore plants collected for their suitability as references showed an ¹⁵N enrichment (data not shown) caused by their rooting in the ¹⁵N-enriched sediment of the water bodies. They were replaced by phytoplankton in this analysis, as the aquatic, free floating phytoplankton resemble the growing conditions of carnivorous Utricularia better than terrestrial plant species.

DISCUSSION

Applying a stable isotope natural abundance approach, our results show that terrestrial and aquatic carnivorous Lentibulariaceae successfully receive prey-derived nitrogen. For terrestrial carnivorous Pinguicula, a Bayesian inference mixing model visualized a preference for phytophagous prey. Carbon isotopic values illustrate a physiological response to habitat conditions and marginal carbon supply from prey.

Pinguicula: nitrogen translocation and prey preferences

The investigated Pinguicula species, except for the shaded P. vulgaris, effectively translocated prey-derived nitrogen to their tissue. This beneficial nutrient supply was invested into growth and reproduction, as indicated by flowering and fruiting individuals (Ellison and Gotelli, 2001; Givnish, 2015; Givnish et al., 2018). Under similar habitat conditions, a sticky Pinguicula leaf is as effective as a rosette leaf of Drosera species comprising the same trapping mechanism (Schulze and Schulze, 1990; Schulze et al., 1991; Ellison and Gotelli, 2001; Millett et al., 2012), indicating the success of flypaper traps in capturing flying and dwelling insects (Fleischmann and Roccia, 2018).

Of the pool of flying and dwelling prey animals, Pinguicula reveal the main capture of phytophagous prey, potentially caused by their naturally higher frequency compared with higher trophic levels of prey. Thus, diverse prey are captured before zoophagous prey; however, this needs to be interpreted with caution. Prey graded as diverse comprised a wide range of feeding mechanisms, which often represent a mixture of different sources, resulting in a high variation of isotopic values. Following the enrichment per trophic level in heavy nitrogen and carbon isotopes described by DeNiro and Epstein (1978, 1981), Post (2002) and Phillips et al. (2014), diverse prey feeding on mycelia and plant sap will show values closer to phytophagous prey while those feeding on blood diverge closer to zoophagous prey. To lessen this issue, the determination of a prey’s sex (sex-dependent nutrient source) and consideration of insect metamorphosis is recommended. Insects surviving without nutrition as imagoes could reveal the isotopic signature of their larval stage or be even more enriched due to excreting ¹⁵N-depleted faeces, resulting in a mixture of the nutritional signatures within the isotopic signals of the investigated imago (Tibbets et al., 2008; but see Ziegler, 1995). This nutritional effect holds true not only for different insect development stages but also for different insect diets (Ziegler, 1995; Schulze et al., 2001). In their study on the efficiency of the Venus flytrap (Dionaea muscipula), Schulze et al. (2001) identified large differences in both carbon and nitrogen stable isotopes between zoophagous ants and herbivorous grasshoppers. Even within the group of grasshoppers, a large variation in ¹³C was discovered depending on whether the grasshopper was feeding on a C₃ or a C₄ plant (Schulze et al., 2001). These diet-dependent distinct isotopic signals were mirrored in the isotopic values of the carnivorous Dionaea leaves and can be found in the Pinguicula leaves in this study. For Dionaea, this effect is accompanied by a leaf size dependency of prey type capture (Schulze et al., 2001), which demands further investigation for Pinguicula. The third group, zoophagous animals, contributed marginally to the nutrition of Pinguicula. This finding is consistent with the naturally lower occurrence of zoophagous animals compared with phytophagous animals. Additionally, ants were observed crossing the sticky leaves without being captured. Zamora (1990) observed kleptoparasitism by ants stealing prey from the sticky leaves of Pinguicula. However, it is not likely to influence the isotopic signature of the carnivorous plant as the prey is removed from the leaf before it can be digested. This observation hints at the possibility that some of the zoophagous animals are attracted by captured prey of lower trophic levels and get stuck on the leaves accidentally when feeding. This assumption is similar to the pollination mechanism of many carnivorous plants which create long flower stalks or have temporal separation to prevent their pollinators from being caught accidentally (Givnish, 1989; Ellison and Gotelli, 2001). Apart from animal prey, plant material such as pollen grains can contribute to the nutrition of Pinguicula species (Harder and Zemlin, 1967; Karlsson et al., 1994). This plant-based nutrient source could be analysed for its potential as an additional tropic level impacting the nutrition of Pinguicula species in further studies.

Overall, the nutrient gain of Pinguicula from Austria is quite similar when comparing different species, P. vulgaris and P. alpina. The occupation of a similar habitat as well as prey availability could be an explanation for this finding. Nevertheless, this result suggests a spatial component, because the quantity of specific prey differs without influencing the overall preference for phytophagous animals.

In contrast to these successful nutrient gains, shaded P. vulgaris suffers from ineffectiveness in receiving prey-derived nitrogen, which may result from habitat-specific stress factors limiting plant performance. Apart from a low nitrogen supply, this P. vulgaris population may suffer from insufficient availability of other nutrients such as phosphorus, potassium and micronutrients. This nutritional disadvantage limits a plant’s growth and trapping success as the plant changes to a cost-saving strategy (Aldenius et al., 1983; Karlsson and Carlsson, 1984; Givnish, 1989) and is enhanced by the shady conditions at the forest site. Accompanied by reduced photosynthetic rates due to shading, the already normally low photosynthetic rates of P. vulgaris may reach detrimental levels (Ellison, 2006). When carnivory loses its advantage as light becomes a limiting factor, the plant will minimize the costs of maintaining carnivorous structures (Givnish, 1989; Zamora et al., 1998; Ellison and Gotelli, 2001; Ellison, 2006), resulting in a reduction of mucilage production for flypaper trap Pinguicula. Additionally, ribulose-1,5-bisphosphate carboxylase-oxygenase (Rubisco) produced from nitrogen increases CO₂ fixation rates, but its levels are low during scarce nitrogen supply (Ellison, 2006). Nevertheless, all these disadvantages do not imply that the carnivorous plant is not capturing any prey (some plots showed success in prey capture), but suggests that the success may be limited. Here, based on the linear mixing model approach, prey capture could not be visualized as the model components were not sufficiently distinguishable from each other. Nonetheless, inhabiting a shaded, but moist and potentially prey-rich habitat could serve as an adaptational compromise for not having to compete in a sunny but dry site, as found for Mexican and European Pinguicula species (Fleischmann and Roccia, 2018) and shade-loving Drosera species in Queensland rain forests (Givnish et al. 2018).

Plant performance and carbon gain

Carbon stable isotope abundances of plants may be affected by (1) the plant’s photosynthetic pathway and physiological performance in its environment (e.g. transpiration) (Dawson et al., 2002); (2) soil respiration (Dawson et al., 2002) (especially for leaf rosettes growing just above the soil surface); and (3) carbon input from sources other than photosynthesis (Těšitel, 2010; Merckx, 2013). The carbon isotopic values of all but the shaded Pinguicula can be explained by their physiological behaviour accompanied by soil respiration effects. The Pinguicula in the creek and precipice plots grew in a sunny, moist habitat allowing for efficient photosynthesis, outweighed by a higher transpiration rate due to an increased leaf surface covered by sticky hairs. As a result, these Pinguicula species reveal a depletion in ¹³C compared with the autotrophic reference plant species. For shaded Pinguicula, the lack of light potentially leads to a reduction in carbon assimilation, as found for Dionaea muscipula by Schulze et al. (2001), and reduced transpiration, leading to more enriched ¹³C values compared with reference plants. This finding aligns with the disadvantages of a shaded habitat as described by the cost–benefit model of carnivory established by Givnish et al. (1984). Additionally, soil respiration may also shift the isotopic carbon value in a more enriched direction (Ekblad and Högberg, 2001).

The marshland plot was bright, but less sunny than the creek and precipice plots. The more enriched ¹³C signatures of marshland Pinguiculas may be explained by reduced transpiration due to higher humidity and influences of soil respiration on the rosette leaves. Nevertheless, these effects were not discovered for rosette-leaved reference plants such as Potentilla and Plantago, which should be affected in a similar way. Therefore, the assumption is that the ¹³C values of marshland Pinguicula species are, in addition, indicating some carbon gain from prey. The ¹³C values of animals follow a trophic enrichment shift as described for nitrogen (DeNiro and Epstein, 1978; Post, 2002), such that some prey-derived carbon could contribute to an isotopic enrichment of the carnivorous plants. Even though the contribution of prey-derived carbon is small (approx. 10 %), it represents a considerable factor when taking into account the ¹³C signatures of carnivorous plants. A carbon gain from prey seems possible for the other Pinguicula species but appears counterbalanced by the above-mentioned factors inducing their depletion in ¹³C.

Nitrogen distribution and adapted reference systems for aquatic Utricularia

The application of phytoplankton as an autotrophic reference is suitable for calculating the contribution of nitrogen from zooplankton to the nutrition of Utricularia. Utricularia were similarly successful to other aquatic and terrestrial carnivorous species (e.g. Polypompholyx multifida, 21 %, Drosera rotundifolia, 26.5 %, Cephalotus follicularis, 26.1 %) (Schulze et al., 1991, 1997). For U. vulgaris, Friday and Quarmby (1994) demonstrated approx. 30 % prey-derived total plant nitrogen transferred to immature tissues. Another proportion of N is gained from capturing pollen, phytoplankton and other plant material (Koller-Peroutka et al., 2015; Ellwood et al., 2019) and even their own mutualistic microfauna (Richards, 2001). The remaining nitrogen supply is then received from the surrounding water medium (Friday and Quarmby, 1994). Despite this knowledge, more investigation of shoot uptake affinity for mineral and organic nutrients from surrounding water and the influence of prey capture on stimulating water-derived nutrient uptake is highly recommended (Adamec, 2018). This complicated network of different nutrient sources demands further investigation to decrypt the nutrient gain performance of Utricularia. One possibility could be the application of a Baysian inference model as used for Pinguicula to shed light on the contribution of the different nutrient sources. An interesting focus in addition to the living prey organisms such as phytoplankton and zooplankton could be the analysis of pollen grains and debris, which are often found inside the bladders at the respective habitats (Koller-Peroutka et al., 2015; Adamec, 2018). Further analyses on the nutritional contribution of this prey would help to decrypt this nutritional network further. Furthermore, as only a part of the bladders of Utricularia capture large macroscopic prey during their life (Koller-Peroutka et al., 2015; Adamec, 2018), a correlation of the amount of bladders with macroscopic prey with the δ¹⁵N values of the carnivorous plant represents an interesting future research question.

Regarding the nitrogen isotopic signatures of bladders and bulk shoot tissue, the equal ¹⁵N enrichment indicates that prey inside of the bladders were already digested when bladders were collected. Both Utricularia species seem to translocate the gained nitrogen from the traps to the shoot tissue and to distribute it equally over the plants’ tissue. Friday and Quarmby (1994) discovered that prey-derived ¹⁵N was quickly taken up by the plant and translocated to immature tissues. Variation of the total nitrogen content depended on the plant’s physiological age (Friday and Quarmby, 1994).

As described by Koller-Peroutka et al. (2015), the calculation of phytoplankton-derived nutrients requires adequate autotrophic references that are aquatic, submersed and non-rooting (or just by shoots) such as Utricularia. The stable isotope signatures of terrestrial shore plants were affected by their rooting in the lake sediment. Lake sediments often reveal anaerobic conditions where denitrification by bacteria occurs (Reinhardt et al., 2006). The bacteria discriminate against the ¹⁵N isotope such that the N₂ produced becomes depleted in ¹⁵N, while the NO₃^– remaining in the sediment is ¹⁵N enriched (Reinhardt et al., 2006). We assume the enriched nitrogen is taken up by these plants and translocated to their tissue, resulting in enriched δ¹⁵N values for these autotrophic reference plants. A bladderless Utricularia species would represent an optimal reference, allowing for the determination of phytoplankton-derived nutrients in the nutrient supply of aquatic Utricularia, which are frequently identified inside the bladders (Koller-Peroutka et al., 2015; Ellwood et al., 2019). At least in temperate central Europe, the main issue is the existence of such submersed non-rooting plants. Plants with absent bladders would represent the most suitable reference for efficiency analyses, as shown by the utilization of a glandless Drosera mutant by Schulze et al. (1991). For temperate water bodies, Ceratophyllum demersum, aquatic rootless Lemna trisulca or the liverwort Ricciocarpos natans could represent a suitable reference; these plants are commonly found in Utricularia habitats.

Conclusion

Summarizing, temperate terrestrial and aquatic Lentibulariaceae were analysed for their efficiency to receive prey-derived nutrients by applying a stable isotope natural abundance approach. Sticky leaf Pinguicula species were as effective as other carnivorous sticky leaf plants, while phytoplankton as well as zooplankton were found to be suitable reference systems for the aquatic bladderwort Utricularia. Conducting a Bayesian inference model, this study enables a better understanding of carnivorous plant nutrition, while further studies are recommended to improve the clarity of the trophic-level separation. The tested Pinguicula individuals mostly fed on phytophagous prey, receiving nitrogen and possibly some carbon. Nitrogen translocation from zooplankton to aquatic Utricularia was successfully calculated applying phytoplankton as the autotrophic reference. Isotope signatures additionally suggest the capture of phytoplankton and plant material by autonomous bladder firing. Quantification of phytoplankton-derived nutrients still demands an autotrophic reference system revealing an equal life form resembling that of Utricularia. If such a reference is applied in the nutrient gain calculation, the importance and ecological influence of specific trophic levels of prey to the behaviour and performance of carnivorous plants will be guided to a new level.

The main new findings of this study are as follows. (1) Sticky leaf Pinguicula species receive most of their heterotrophically gained N supply from phytophagous prey; higher trophic levels play a minor role. A small amount of carbon is received from prey by Pinguicula species. (2) Phytoplankton can be applied as the autotrophic reference for the nutrient gain from zooplankton by aquatic Utricularia. (3) Recommendations for future work are further modifications of the reference system, allowing determination of the nutritional contributions of all food web members (e.g. bladder microfauna, pollen, detritus, algae and zooplankton) of Utricularia. A possible correlation of the number of bladders capturing macroscopic prey and the ¹⁵N values of the Utricularia plant should also be considered for future research.

SUPPLEMENTARY DATA

Supplementary data are available online at https://academic.oup.com/aob and consist of the following. Figure S1: boxplots of mean ε¹⁵N values demonstrating differences between Pinguicula species, autotrophic reference plant and prey animals. Figure S2: Bayesian inference model converted to boxplots (whiskers in one interquartile range, extreme values excluded), conducted on ¹³C and ¹⁵N stable isotope values and C and N concentrations of carnivorous plants as consumers together with phytophagous, diverse and zoophagous prey as the nutrient sources of consumers. Table S1: single nitrogen concentrations, single δ¹³C and δ¹⁵N values, and trophic grading of all plant and animal samples utilized in this study.

ACKNOWLEDGEMENTS

We thank the technicians Christine Tiroch, Petra Eckert and Isolde Baumann (BayCEER Laboratory of Isotope Biogeochemisty) for technical assistance with stable isotope natural abundance measurements. We thank Dr Pedro Gerstberger for information about the habitats of the carnivorous species, Professor Dr Christian Laforsch and Dr Max Rabus for their plankton net and the camera-equipped binocular, Professor Dr Konrad Dettner for revising the animal identification, and Laura Skates and Adrienne Keller for native speakers’ language improvement. We also thank the Regierung der Oberpfalz (Bavaria) for the permission for sampling the carnivorous plant species investigated in this study.