An ecological perspective on ‘plant carnivory beyond bogs’: nutritional benefits of prey capture for the Mediterranean carnivorous plant Drosophyllum lusitanicum

Abstract

Background and aims

Little is known about the evolutionary and ecological drivers of carnivory in plants, particularly for those terrestrial species that do not occur in typical swamp or bog habitats. The Mediterranean endemic Drosophyllum lusitanicum (Drosophyllaceae) is one of very few terrestrial carnivorous plant species outside of Australia to occur in seasonally dry, fire-prone habitats, and is thus an ecological rarity. Here we assess the nutritional benefits of prey capture for D. lusitanicum under differing levels of soil fertility in situ.

Methods

We measured the total nitrogen and stable nitrogen and carbon isotope ratios of D. lusitanicum leaves, neighbouring non-carnivorous plant leaves, and groups of insect prey in three populations in southern Spain. We calculated trophic enrichment (ε¹⁵N) and estimated the proportion of prey-derived nitrogen (%N_prey) in D. lusitanicum leaves, and related these factors to soil chemistry parameters measured at each site.

Key results

In all three populations studied, D. lusitanicum plants were significantly isotopically enriched compared with neighbouring non-carnivorous plants. We estimated that D. lusitanicum gain ~36 %N_prey at the Puerto de Gáliz site, ~54 %N_prey at the Sierra Carbonera site and ~75 %N_prey at the Montera del Torero site. Enrichment in N isotope (ε¹⁵N) differed considerably among sites; however, it was not found to be significantly related to log₁₀(soil N), log₁₀(soil P) or log₁₀(soil K).

Conclusions

Drosophyllum lusitanicum individuals gain a significant nutritional benefit from captured prey in their natural habitat, exhibiting proportions of prey-derived nitrogen that are similar to those recorded for carnivorous plants occurring in more mesic environments. This study adds to the growing body of literature confirming that carnivory is a highly beneficial nutritional strategy not only in mesic habitats but also in seasonally dry environments, and provides insights to inform conservation strategies for D. lusitanicum in situ.

INTRODUCTION

Carnivorous plants are an ecologically defined group of organisms characterized by their ability to capture and digest prey using specially modified leaves. Amongst the world’s angiosperms, carnivory has independently evolved at least ten times, with >800 species of carnivorous plants currently described worldwide, arising from five orders, 12 families and 19 genera (Fleischmann et al., 2018). This carnivorous phenomenon, which occurs across a range of phylogenetic histories and taxonomic groups, has prompted many researchers to ponder the ecological and evolutionary drivers of successful carnivory. Darwin (1875) provided the first scientific evidence of plant carnivory and suggested that these adaptations must have evolved to supplement limited soil nutrient uptake, particularly given that carnivorous plants often exhibit highly reduced root structures (Adlassnig et al., 2005). A recent synthesis by Givnish et al. (2018) identified several hypotheses regarding carnivorous plant evolution and ecology with respect to the widely accepted cost–benefit model first proposed by Givnish et al. (1984). The first of these hypotheses states that ‘carnivory is most likely to evolve and be favoured ecologically in habitats that are sunny, moist and nutrient-poor’ and a further hypothesis states that ‘optimal investment in carnivory in terrestrial habitats should increase toward the sunniest, moistest, most nutrient-poor sites’ (Givnish et al., 2018). In these sunny, moist, nutrient-poor habitats, e.g. peatlands and bogs (Dionaea muscipula and species of Sarracenia, Pinguicula, Utricularia and Drosera throughout the northern hemisphere) or tropical rainforest (species of Nepenthes throughout East and South-East Asia), it is expected that the net benefits associated with carnivory would significantly outweigh the photosynthetic costs of producing functioning trapping mechanisms (Givnish et al., 1984).

Despite the empirical support for this cost–benefit model (Givnish et al., 2018), there are several carnivorous plant species that partially contradict its primary hypothesis through their occurrence in less mesic, seasonally dry and fire-prone infertile habitats. These notable exceptions include Drosophyllum lusitanicum (Drosophyllaceae) from Western Europe, species of Philcoxia (Plantaginaceae) from Brazil, and species of Drosera (Droseraceae) and Byblis (Byblidaceae) from South-West Australia. These species exhibit a variety of physiological adaptations to seasonal drought stress, such as the well-developed root systems of D. lusitanicum (Carlquist and Wilson, 1995; Cross et al., 2018b) and the tuberous storage organs of many South-West Australian Drosera (Dixon et al., 1980) and at least one species of Philcoxia (Scatigna et al., 2015; Cross et al., 2018b). Furthermore, many species rely on fire for at least one aspect of their reproductive ecology, either requiring fire-related cues for the stimulation of flowering (Dixon and Pate, 1978) or alleviation of seed dormancy and promotion of germination and emergence from a soil seed bank (Cross et al., 2013; Cross et al., 2017, 2018a), or exhibiting significant population regeneration following fires (Paniw et al., 2017b). It has been hypothesized that fire facilitates carnivory by removing competing foliage, increasing light availability and volatizing nitrogen from burnt tissues (Givnish et al., 2018).

Notably, the carnivorous plant taxa occurring in seasonally dry and fire-prone habitats employ an adhesive trapping mechanism, whereby prey is captured via the secretion of a viscous adhesive mucilage from glands or hairs along the leaf surface. A spectrum of adhesive-trap complexity is apparent amongst these taxa, ranging from the relatively simple glands of Byblis and Philcoxia species to the fully vascularized glands of D. lusitanicum and active tentacles of Drosera species. Drosophyllum lusitanicum and some species of Drosera have also invested in prey-attraction mechanisms, including distinctive ultraviolet patterns visible to insect prey (Joel et al., 1985) and fly-attractive honey-like scents (Bertol et al., 2015; Horner et al., 2018). Clearly, D. lusitanicum and Drosera are highly specialized in the carnivorous habit, and this may be reflected in the long evolutionary histories of each genera, with Drosophyllum and Drosera having phylogenetic ages of 70.4 and 53.4 Mya, respectively (Fleischmann et al., 2018). However, whilst Drosera is one of the most species-rich genera of carnivorous plants (>200 species) with a cosmopolitan distribution including tropical, temperate and Mediterranean climatic zones, Drosophyllum is a monotypic genus (consisting of D. lusitanicum) in the monotypic family Drosophyllaceae, occurring exclusively in Mediterranean heathlands of the south-western Iberian Peninsula and northern Morocco. Given its long evolutionary history, high level of investment in the carnivorous habit and occurrence exclusively on dry (not waterlogged) soils of seasonally (summer) dry habitats, D. lusitanicum is arguably one of the most notable exceptions to the primary hypothesis of the cost–benefit model.

Assessing the costs and benefits of carnivory for D. lusitanicum will provide crucial insights into the evolutionary drivers and ecological significance of the carnivorous syndrome in less mesic, seasonally dry and fire-prone infertile habitats. It may also assist in the design of adequate management strategies for its conservation. Whilst prior research by Paniw et al. (2017a) has shown that prey addition provides a significant growth benefit to D. lusitanicum individuals grown under controlled glasshouse conditions, no studies have yet assessed the ecological benefit of prey capture for D. lusitanicum in its natural habitat. In order to quantify the nutritional benefit from carnivory, we used natural abundance stable isotope techniques based on the theory of trophic isotopic enrichment, whereby organisms tend to be enriched in heavier isotopes when compared with their diet (DeNiro and Epstein, 1976; Boecklen et al., 2011). We measured the natural abundances of stable nitrogen (N) and carbon (C) isotopes as well as total N of D. lusitanicum, neighbouring non-carnivorous plants and groups of insect prey at three sites in southern Spain, and calculated percentages of prey-derived N in D. lusitanicum leaves. We hypothesized that D. lusitanicum would exhibit higher isotopic enrichment compared with neighbouring non-carnivorous plants, and would correspondingly exhibit a proportion of prey-derived N within their total leaf N. In addition to this, we measured soil chemistry parameters at each site, and hypothesized that the trophic enrichment of D. lusitanicum would be greatest in areas with the least fertile soil.

MATERIALS AND METHODS

Sites

Drosophyllum lusitanicum was sampled in November 2015 from three sites representing distinct populations in southern Spain, the centre of the species’ distribution. Sites included Sierra Carbonera (SC: 36°12′35″ N, 5°21′37’’ W), Montera del Torero (MT: 36°13′35″ N, 5°35′08″ W) and Puerto de Gáliz (PG: 36°34′26″ N, 5°32′12″ W). Vegetation at all sites consisted of an overstorey of low shrubs dominated by Erica australis (Ericaceae), Halimium alyssoides (Cistaceae) and Calluna vulgaris (Ericaceae). Erica scoparia (Ericaceae) and Quercus lusitanica (Fagaceae) were also dominant at PG and SC. All sites were located within the Mediterranean climatic zone, characterized by warm, dry summers and cool, wet winters, with mean annual temperatures ranging from 14.6 to 18.4 °C and mean annual rainfall from 700 to 1330 mm (Mejías et al., 2007).

Stable isotope enrichment

In order to quantify the trophic enrichment of D. lusitanicum in comparison with co-occurring non-carnivorous plants and insect prey, we followed the plant and insect sampling methods of Schulze et al. (1991). Replicate leaf samples were collected from the five D. lusitanicum individuals (paired to the soil sampling) along with replicate leaf samples from surrounding non-carnivorous shrub neighbours (Table 1). Any captured insects on D. lusitanicum samples were removed from the adhesive leaves, and fresh insects were collected with an insect net (Table 1). Non-carnivorous plants were identified to the species level and insects were identified to the order level. All leaf and insect samples were stored in paper bags surrounded by silica gel and later dried at 105 °C for 48 h.

Table 1.

Number (n) of Drosophyllum lusitanicum, neighbouring non-carnivorous plants and insect samples collected for stable isotope analysis at three sites in southern Spain

	Montera del Torero	n	Puerto de Gáliz	n	Sierra Carbonera	n
Carnivorous plants	Drosophyllum lusitanicum	5	Drosophyllum lusitanicum	5	Drosophyllum lusitanicum	5
Non-carnivorous plants	Erica australis	5	Erica australis	5	Erica australis	5
	Halimium alyssoides	5	Halimium alyssoides	5	Halimium alyssoides	5
	Calluna vulgaris	5	Calluna vulgaris	5	Calluna vulgaris	5
			Erica scoparia	2	Erica scoparia	3
			Quercus lusitanica	2	Quercus lusitanica	3
Insects	Diptera (flies)	5	Diptera (flies)	4	Diptera (flies)	1
	Lepidoptera (moths)	6	Lepidoptera (moths)	2	Neuroptera (lacewings)	5

Stable isotope analyses of all plant and insect samples were conducted at the BayCEER Laboratory of Isotope Biogeochemistry at the University of Bayreuth (Germany) in 2016. Plant samples were first prepared by carefully cleaning away any remaining soil or insect residue using fine forceps. The dried plant and insect samples were individually ground into a fine powder (ball mill MM 200, Retsch, Haan, Germany) and weighed into tin capsules (~1 mg for insects and ~2.8–3.5 mg for plants) with the aid of an electrical microbalance (CPA2P, Sartorius, Göttingen, Germany). These samples were analysed for ¹⁵N and ¹³C abundances and total N and C concentrations using an elemental analysis–isotope ratio mass spectrometry system (EA-IRMS; elemental analyser: NA 1108, CE Instruments, Milan, Italy; interface: ConFlo III, Finnigan MAT, Bremen, Germany; IRMS: Delta S, Finnigan MAT) with the software ISODAT (V 2.0, Thermo Fisher Scientific). Atmospheric N₂ or Vienna PeeDee Belemnite (V-PDB) were used as standards for N and C isotope abundances and calibrated by the secondary standard substances N1 and N2 or ANU sucrose and NBS 19, respectively, provided by the International Atomic Energy Agency (Vienna, Austria); acetanilide [C₆H₅NH(COCH₃)] (C, 71.09 %; H, 6.71 %; N, 10.36 %; O, 11.84 %; Merck, Darmstadt, Germany) was used as the standard for the total N and C concentration calibration and for reproducibility control. The isotope abundance measurements had a resolution of ±0.2‰.

Measured isotope abundances are denoted as δ values, which were calculated as δ¹⁵N or δ¹³C = (R_sample/R_standard − 1) × 1000 (‰), where R_sample and R_standard are the ratios of heavy to light isotope of the samples and the respective standard. To enable comparisons of ¹⁵N abundances between the three sites, we normalized the data using an isotope enrichment factor approach described by Preiss and Gebauer (2008). Normalized enrichment factors (ε) were calculated from measured δ values as ε  =  δX − δR, where δX is a single δ¹⁵N or δ¹³C value of any sample, and δR is the mean δ¹⁵N or δ¹³C value of all non-carnivorous reference plant samples by site.

Percentage nitrogen from prey

In order to calculate the proportion of prey-derived N within the total leaf N of D. lusitanicum, we initially followed the methods of Schulze et al. (1991), wherein δ¹⁵N values of a single D. lusitanicum individual (δS), along with all prey (δP) and non-carnivorous plants (δR) at that D. lusitanicum individuals’ site are utilized in a two-source linear mixing model (eqn 1). Given that Diptera were the only prey order collected at all three sites, and based on observations by Bertol et al. (2015) that Diptera are the predominant prey caught by D. lusitanicum across its entire geographical range, we also tested a mixing model utilizing only Diptera δ¹⁵N values as a prey source (eqn 2). Following from this, we tested a mixing model utilizing the calculated site-independent Diptera ε¹⁵N values averaged across all three sites (eqn 3). Given that D. lusitanicum are generalists in their prey capture and should have access to a similar composition of prey across all sites regardless of what was randomly sampled (M. Paniw, pers. comm.), we also tested a model utilizing the mean (eqn 4) and median (eqn 5) site-independent ε¹⁵N values of all prey items averaged across all three sites.

$${\displaystyle \begin{array}{l}{\displaystyle \mathrm{\%}{\mathrm{N}}_{\mathrm{p}\mathrm{r}\mathrm{e}\mathrm{y}}=\text{©}(\delta \mathrm{S}-\delta \mathrm{R})/(\delta \mathrm{P}-\delta \mathrm{R})\text{©}\times \text{©}100}\end{array}}$$ (1)

$${\displaystyle \begin{array}{l}{\displaystyle \mathrm{\%}{\mathrm{N}}_{\mathrm{D}\mathrm{i}\mathrm{p}\mathrm{t}\mathrm{e}\mathrm{r}\mathrm{a}}=\text{©}(\delta \mathrm{S}-\delta \mathrm{R})/(\delta {\mathrm{P}}_{\mathrm{D}\mathrm{i}\mathrm{p}\mathrm{t}\mathrm{e}\mathrm{r}\mathrm{a}}-\delta \mathrm{R})\text{©}\times \text{©}100}\end{array}}$$ (2)

$${\displaystyle \begin{array}{l}{\displaystyle \mathrm{\%}{\mathrm{N}}_{\mathrm{D}\mathrm{i}\mathrm{p}\mathrm{t}\mathrm{e}\mathrm{r}\mathrm{a}}=\text{©}(\epsilon \mathrm{S}-\epsilon \mathrm{R})/(\epsilon {\mathrm{P}}_{\mathrm{D}\mathrm{i}\mathrm{p}\mathrm{t}\mathrm{e}\mathrm{r}\mathrm{a}}-\epsilon \mathrm{R})\text{©}\times \text{©}100}\end{array}}$$ (3)

$${\displaystyle \begin{array}{l}{\displaystyle \mathrm{\%}{\mathrm{N}}_{\mathrm{p}\mathrm{r}\mathrm{e}\mathrm{y}}=\text{©}(\epsilon \mathrm{S}-\epsilon \mathrm{R})/(\epsilon {\mathrm{P}}_{\mathrm{m}\mathrm{e}\mathrm{a}\mathrm{n}}-\epsilon \mathrm{R})\text{©}\times \text{©}100}\end{array}}$$ (4)

$${\displaystyle \begin{array}{l}{\displaystyle \mathrm{\%}{\mathrm{N}}_{\mathrm{p}\mathrm{r}\mathrm{e}\mathrm{y}}=\text{©}(\epsilon \mathrm{S}-\epsilon \mathrm{R})/(\epsilon {\mathrm{P}}_{\mathrm{m}\mathrm{e}\mathrm{d}\mathrm{i}\mathrm{a}\mathrm{n}}-\epsilon \mathrm{R})\text{©}\times \text{©}100}\end{array}}$$ (5)

Soil chemistry

In order to characterize the soil chemistry at each site, ~500 g of soil was collected from the top 5 cm of soil immediately adjacent to five D. lusitanicum individuals at each site. All soil samples were dried at 105 °C for 48 h to remove any moisture, then analysed at the BayCEER Laboratory of Isotope Biogeochemistry (total N and C, pH and electrical conductivity) or BayCEER Analytical Chemistry (macro- and micronutrients and heavy metals) at the University of Bayreuth, Germany. Total N and total carbon (C) were measured using an elemental analyser (NA 1108, CE Instruments, Milan, Italy), and the concentrations of various macronutrients (Ca, K, Mg, P, S), micronutrients (Al, B, Co, Cu, Fe, Mn, Mo, Na, Ni, Se, Zn) and heavy metals (As, Cd, Pb) were measured by inductively coupled plasma-optical emission spectrometry (Vista Pro Radial, Varian, Palo Alto, CA, USA) after a Mehlich 3 extraction (Mehlich, 1984). Electrical conductivity (EC) and acidity of the soil were measured by placing EC and pH electrodes in a 1:5 slurry of soil:water at 25 and 22.5 °C respectively.

Statistical analyses

All statistical analyses were performed using R (version 3.4.1; R Core Team, 2014), with the significance level set to α = 0.05.

A non-parametric Kruskal–Wallis rank sum test was used to test differences in δ¹⁵N between D. lusitanicum and neighbouring non-carnivorous plants at each site. This test was repeated to test differences in δ¹⁵N between D. lusitanicum and insect prey, differences in δ¹⁵N between the three orders of insects, and differences in δ¹⁵N between the five species of non-carnivorous plants. These tests were repeated for δ¹³C, ε¹⁵N, ε¹³C and total N concentrations.

Where individual soil samples at a site had nutrient concentrations below detection limits, they were assigned the detection limit concentration for the following analyses. Statistical differences in individual soil parameters between each site were measured by a Kruskal–Wallis rank sum test, followed by a pairwise Wilcoxon rank sum test with a Benjamini and Hochberg (1995)P-value adjustment method. These differences were visualized using principal component analysis, with principal components calculated from log₁₀-transformed soil chemistry data and plotted using the devtools (Wickham et al., 2018) and ggbiplot (Vu, 2011) packages in R.

The relationship between soil chemistry and D. lusitanicum nutrition was analysed by fitting a linear mixed-effects model with log₁₀(soil N) used as the explanatory variable and trophic enrichment of D. lusitanicum (ε¹⁵N) as the response variable. We included ‘site’ as a random effect on model means. We performed a likelihood ratio test (Vuong, 1989) using the R package lme4 (Bates et al., 2014) to assess the significance of log₁₀(soil N) in explaining the variation in ε¹⁵N by comparing the model including log₁₀(soil N) as predictor with a null model excluding the effect of log₁₀(soil N). These analyses were repeated using log₁₀(soil P) and log₁₀(soil K) as the explanatory variables, and using total N concentration of D. lusitanicum leaves as the response variable.

RESULTS

Isotopic enrichment of D. lusitanicum

Drosophyllum lusitanicum individuals had significantly greater δ¹⁵N compared with non-carnivorous plants at all sites (P ≤ 0.001 in all cases). The δ¹⁵N of non-carnivorous plant species did not vary significantly among species at each site (P > 0.05 in all cases). Overall, the trophic enrichment (ε¹⁵N) of D. lusitanicum was greatest at MT (~6.4 ± 0.12, n = 5), ~1.5 times greater than at SC (~4.6 ± 0.27, n = 5) or PG (~3.1 ± 0.06, n = 5) (Fig. 1).

Fig. 1.

Trophic enrichment factors of carbon (ε¹³C) and nitrogen (ε¹⁵N) for Drosophyllum lusitanicum at each of the three sites, along with non-carnivorous plants and insect prey across the three sites. Data are mean ± s.e.

Across all three sites, the δ¹³C values of D. lusitanicum ranged from −26.23 to −29.20 (n = 15), and δ¹³C values of all non-carnivorous plants ranged from −25.28 to −29.93 (n = 15). There were no significant differences between the C isotope ratios of D. lusitanicum and neighbouring non-carnivorous plants at MT and at PG (P > 0.05 in both cases); however, the ε¹³C of D. lusitanicum at SC was greater than that of neighbouring non-carnivorous plants (P = 0.034) (Fig. 1).

Contribution of prey to total leaf N

Overall, the average isotopic enrichment (ε¹⁵N) of insect prey sampled across the three sites was ~9.41 ± 3.05 (n = 23) and the median was 8.53. The ε¹⁵N values for Diptera (~12.61 ± 3.61, n = 10) and Lepidoptera (~9.24 ± 1.80, n = 8) were significantly greater than those for D. lusitanicum (P = 0.0047 and 0.015, respectively); however, ε¹⁵N for Neuroptera (~3.30 ± 1.34, n = 5) was not significantly different from that for D. lusitanicum at the same site (P = 0.60).

The %N_prey estimates differed among the four calculation methods (Table 2). The greatest difference was observed at site SC, where D. lusitanicum was estimated to gain nearly twice the %N_prey using eqn (1) (site-dependent δ¹⁵N values, including only the Neuroptera and one Diptera sampled at SC) compared with eqns (4) and (5) (site-independent ε¹⁵N values, including trophic enrichment of Neuroptera, Diptera and Lepidoptera sampled across all sites). Estimates of %N_Diptera (eqns 2 and 3) followed the same pattern across the three sites as the latter estimates of %N_prey (eqns 4 and 5). (i.e. greatest percentage in MT, lowest in PG).

Table 2.

Estimates of percentage of prey-derived nitrogen for Drosophyllum lusitanicum at three sites in southern Spain using five calculation methods

Calculation method (equation number)	Montera del Torero	Sierra Carbonera	Puerto de Gáliz
(1) %Nprey = (δS – δR)/(δP – δR) × 100	50 ± 0.9	87 ± 5.0	42 ± 0.9
(2) %NDiptera = (δS – δR)/(δPDiptera – δR) × 100	42 ± 0.8	35 ± 0.7	30 ± 1.7
(3) %NDiptera = (εS – εR)/(εPDiptera – εR) × 100	51 ± 1.0	37 ± 2.1	24 ± 0.5
(4) %Nprey = (εS – εR)/(εPmean – εR) × 100	68 ± 1.3	49 ± 2.8	33 ± 0.7
(5) %Nprey = (εS – εR)/(εPmedian – εR) × 100	75 ± 1.4	54 ± 3.1	36 ± 0.8

Data are mean ± s.e.

δ refers to δ¹⁵N values, ε refers to ε¹⁵N values, S refers to a Drosophyllum lusitanicum individual, P refers to prey species, and R refers to non-carnivorous plants.

In order to take all potential prey items into account, eqn (4) was selected for all subsequent analyses. Overall, a large proportion (~33–68 %, n = 15) of the leaf total N of D. lusitanicum was derived from captured prey (Fig. 2). Compared with neighbouring non-carnivorous plants, D. lusitanicum exhibited between 2 and 3.5 times higher leaf total N across the three sites (P ≤ 0.001 in all cases; Fig. 2). Leaf total N varied markedly between sites for D. lusitanicum, with D. lusitanicum at SC having 1.5 times greater leaf total N compared with D. lusitanicum at MT and PG.

Fig. 2.

Total leaf nitrogen of non-carnivorous plants and Drosophyllum lusitanicum at three sites in southern Spain, with estimates of the proportion of total leaf nitrogen derived from prey (dark grey bars) versus total leaf nitrogen derived from soil (light grey bars). Data are mean ± s.e.

Relationship between soil chemistry and D. lusitanicum nutrition

Each of the three D. lusitanicum populations exhibited a distinctive soil chemistry profile, with >82 % of the observed variation explained by the first two principal components (Fig. 3). Compared with soils at PG and SC, soils at MT exhibited the lowest concentrations of several macronutrients, particularly N, P and K (Table 3). Drosophyllum lusitanicum from MT exhibited the highest N isotope enrichment and highest %N_prey, while SC exhibited the highest total N concentrations (Fig. 4). N isotope enrichment (ε¹⁵N) and leaf total N concentrations differed considerably among sites (94 % of the variance in the linear mixed models for ε¹⁵N explained by site, and 66 % of the variance in the linear mixed models for total N concentrations explained by site). However, neither ε¹⁵N nor leaf total N concentration was significantly related to log₁₀(soil N), log₁₀(soil P) or log₁₀(soil K).

Fig. 3.

Principal component analysis showing the ordination (projection of axes 1 and 2) of Drosophyllum lusitanicum microhabitats at three sites in southern Spain, as constrained by log₁₀-transformed soil chemistry variables, including total carbon (C), macronutrients (N, P, K, Ca, Mg, S), micronutrients (Al, Cu, Fe, Mn, Na, Ni, Zn), heavy metals (Pb), pH and electrical conductivity (EC).

Table 3.

Chemistry parameters for soil collected immediately adjacent to sampled Drosophyllum lusitanicum individuals at three sites in southern Spain, including total carbon (C), macronutrients (N, P, K, Ca, Mg, S), micronutrients (Al, Cu, Fe, Mn, Na, Ni, Zn), heavy metals (Pb), pH and electrical conductivity

Variable	Montera del Torero	Sierra Carbonera	Puerto de Gáliz	P †
N (mg kg−1)	448 ± 57.0a	1978 ± 97.4b	2627 ± 648.6b	*
P (mg kg−1)	0.9 ± 0.12a‡	6 ± 0.3b	4 ± 1.0b	**
K (mg kg−1)	31 ± 2.6a	42 ± 3.0b	54 ± 2.9c	**
Mg (mg kg−1)	53 ± 6.5a	53 ± 5.7a	93 ± 9.3b	**
C (mg kg−1)	6672 ± 1194.1a	37215 ± 2244.8b	55620 ± 15347.0ab	*
Ca (mg kg−1)	129 ± 22.3a	274 ± 51.3a	480 ± 116.6a	*
S (mg kg−1)	10 ± 2.0a	10 ± 0.9a	14 ± 2.0 a	n.s.
Al (mg kg−1)	433 ± 11.7a	239 ± 20.6b	434 ± 28.4a	**
Cu (mg kg−1)	0.3 ± 0.03a	0.2 ± 0.01a‡	0.4 ± 0.05a	*
Fe (mg kg−1)	17 ± 0.6a	77 ± 5.6b	95 ± 27.4b	*
Mn (mg kg−1)	0.8 ± 0.20a	1 ± 0.2a	9 ± 3.2a	*
Na (mg kg−1)	6 ± 0.6a	5 ± 0.6a	5 ± 0.4a	n.s.
Ni (mg kg−1)	<0.2‡	0.3 ± 0.03a	0.3 ± 0.03a‡	n.s.
Zn (mg kg−1)	0.2 ± 0.20a	0.6 ± 0.24a	1 ± 0.4a	n.s.
Pb (mg kg−1)	0.6 ± 0.24a	3 ± 0.4b	2 ± 0.6ab	*
pH	4 ± 0.04a	5 ± 0.1b	5 ± 0.1b	**
Electrical conductivity (µS cm−1)	97 ± 9.4a	181 ± 16.2b	270 ± 52.8b	*

Data are mean ± s.e.

Superscript lower-case letters represent the results of pairwise comparisons using the Wilcoxon rank sum test contrasting among sites; values followed by the same letters are not significantly different among sites for each characteristic, at P = 0.05.

^†Main effect of site on each chemistry parameter according to a Kruskal–Wallis rank sum test (*P < 0.05, **P < 0.01, ***P < 0.001, n.s., not significant).

^‡One or more samples reported below detection limit. These samples were assigned their detection limit for statistical analyses.

Fig. 4.

Scatterplots (x,y) of nitrogen trophic enrichment factors (ε¹⁵N) of Drosophyllum lusitanicum against log₁₀-transformed (A) soil total nitrogen, (C) soil total phosphorus and (E) soil total potassium, and leaf total nitrogen of Drosophyllum lusitanicum against log₁₀-transformed (B) soil total nitrogen, (D) soil total phosphorus and (F) soil total potassium, at three sites in southern Spain. DW, dry weight. Data are mean ± 1 s.e.

DISCUSSION

Nutritional benefit of carnivory for D. lusitanicum

The present study provides an ecological perspective on the nutrition of D. lusitanicum, building on prior research (Darwin, 1875; Bertol et al., 2015; Paniw et al., 2017a) to confirm the status of the Mediterranean endemic D. lusitanicum as a highly successful carnivorous plant. As predicted, D. lusitanicum had significant N isotope enrichment compared with neighbouring non-carnivorous plants in all three native populations (Fig. 1). Based on the theory of trophic isotopic enrichment (DeNiro and Epstein, 1976; Boecklen et al., 2011), these results indicate that mature D. lusitanicum individuals gain a significant proportion of their leaf total N from captured insect prey. Indeed, every D. lusitanicum individual sampled was estimated to gain at least a quarter of its leaf total N from insect prey (Table 2), with no D. lusitanicum individuals relying solely on soil for their N nutrition (Fig. 2). Furthermore, D. lusitanicum had a significantly greater leaf total N compared with neighbouring non-carnivorous plants (Fig. 2). This is contrary to the findings of a global meta-analysis by Ellison (2006), which postulated that carnivorous plants tend to have lower leaf nutrient content compared with non-carnivorous plants. Regardless of the differences between the various %N_prey calculation methods (see Discussion below), insect prey is clearly a highly significant source of N nutrition for D. lusitanicum in its natural habitat.

At all sites, D. lusitanicum individuals exhibited levels of nutritional benefit from prey that are similar to those recorded for carnivorous plants occurring across tropical, temperate and Mediterranean habitats. For example, the %N_prey for D. lusitanicum is similar to that for other carnivorous plants with adhesive trapping mechanisms, including Drosera species from the Mediterranean Southwest Australian Floristic Region (Drosera with climbing or erect morphological forms have been estimated to gain 50 %N_prey on average; Schulze et al., 1991), Drosera species from temperate western Europe (estimated to gain up to ~50– 57 %N_prey depending on site conditions; Millet et al., 2003, 2012; Cook et al., 2017), Pinguicula alpina from temperate central Europe (estimated to gain between ~37 and 55 %N_prey; Klink et al., 2019) and Roridula species from the South African Cape Floristic Region (estimated to indirectly gain between ~40 and 70 %N_prey via a digestive–mutualism relationship with resident hemipterans; Anderson and Midgley, 2002, 2003). The %N_prey for D. lusitanicum is also similar to that for carnivorous plants using alternative trapping mechanisms, including Dionaea muscipula with a snap-trap mechanism (estimated to gain ~75 %N_prey soon after fire; Schulze et al., 2001), Utricularia species with suction-trap mechanisms (U. australis and U. minor estimated to gain ~29 and 21 %N_prey, respectively; Klink et al., 2019) and various Nepenthes species with pitfall-trap mechanisms (N. rafflesiana and N. albomarginata estimated to gain ~53 and 68 %N_prey, respectively; Moran et al., 2001). Clearly, despite the expected high costs of carnivory for species occurring in less mesic habitats (Givnish et al., 1984, 2018), some carnivorous plant species are still capable of thriving in these non-waterlogged, drier conditions, particularly those species with mechanisms to avoid drought stress (e.g. tubers, extensive roots).

As has been shown for species of Drosera and Pinguicula and for Aldrovanda vesiculosa (Fabian-Galan and Salageanu, 1968; Dixon et al., 1980; Klink et al., 2019), it is possible that D. lusitanicum may be capable of gaining some C from prey in addition to N benefit. The results of the present study suggest some enrichment of the natural ¹³C isotope abundances of D. lusitanicum at one study site (SC) compared with neighbouring non-carnivorous plants, though this enrichment is not evident at the other two sites. However, prey-derived C alone may not explain these results sufficiently, as the major source of C and thus variability in δ¹³C is photosynthesis (Farquhar et al., 1989). The range of δ¹³C measured for D. lusitanicum in this study indicates that this species has a C3 photosynthetic pathway (Bender, 1971). The variability in δ¹³C could indicate that D. lusitanicum plants at SC have a lower intercellular concentration of CO₂ compared with neighbouring non-carnivorous plants, which may be caused by higher stomatal closure and/or higher rates of CO₂ fixation. Given that D. lusitanicum plants at SC have 3.5 times greater leaf total N than neighbouring non-carnivorous plants, higher CO₂ fixation via photosynthesis could be a likely explanation for the variation in δ¹³C. However, in order to resolve this, future studies ought to compare photosynthetic gas exchange measurements for D. lusitanicum with non-carnivorous neighbours, and further investigate the possible role of carnivory in C nutrition (as per Fabian-Galan and Salageanu, 1968; Dixon et al., 1980; Klink et al., 2019).

Calculation methods for %N_prey

In the seminal paper by Schulze et al. (1991) on the use of natural abundance stable isotope methods to determine the nutritional benefit of plant carnivory, the mixing model calculation methods involved site-dependent δ¹⁵N values of the target carnivorous plant, neighbouring non-carnivorous plants and insect prey (eqn 1). In this study, we tested several alternative versions of these calculation methods to estimate the %N_prey utilized by D. lusitanicum, and found notable differences between each method. Using the same calculation methods as Schulze et al. (1991), D. lusitanicum from SC were estimated to gain ~87 %N_prey, approximately twice the %N_prey for D. lusitanicum from PG and MT. However, we suggest these relatively high %N_prey estimates at SC may be skewed due to the relatively low δ¹⁵N values of Neuroptera, which were the dominant prey item sampled at SC and which were not sampled at MT or PG. Given that Diptera were sampled at all three sites, and based on the observation that Diptera are the dominant prey item throughout the geographical range of D. lusitanicum (Bertol et al., 2015), we tested two %N_Diptera calculation methods using site-dependent δ¹⁵N values of Diptera only at each site (eqn 2) and site-independent ε¹⁵N values of Diptera only averaged across all three sites (eqn 3). These methods resulted in more consistent estimates across the three sites, with both exhibiting a pattern whereby D. lusitanicum at MT exhibited the highest %N_Diptera and D. lusitanicum at PG exhibited the lowest %N_Diptera. The same overall pattern emerges when using site-independent ε¹⁵N values for all prey across the three sites (eqns 4 and 5) to account for differences in the types of prey randomly sampled at each site and given that D. lusitanicum has been observed to access similar prey at each site (M. Paniw, pers. comm.). In order to further refine these calculation methods, future research ought to investigate these observations with long-term monitoring of prey capture across the geographical range of D. lusitanicum, according to the methods of Cook et al. (2017) for Drosera rotundifolia. However, in cases where intensive prey sampling is limited or not possible, we recommend that the calculation methods using site-independent ε¹⁵N values (e.g. eqn 4) could be used.

Influence of soil fertility

Soil fertility has long been considered to be a driver of the evolution of carnivorous plants and an important component of their ecology (Darwin, 1875; Cook et al., 2017; Givnish et al., 2018), and thus we tested whether the nutritional benefits of carnivory for D. lusitanicum were more pronounced in areas with less fertile soils. We found that the plants with the highest ε¹⁵N occurred at the site with the lowest soil N, P and K (Fig. 4A, C, E); however, the relationships of ε¹⁵N with log₁₀(soil N), log₁₀(soil P) and log₁₀(soil K) were not statistically significant. Whilst these results suggest that there could be an influence of soil chemistry on a broader scale, additional research is clearly required to fully understand the variation in ε¹⁵N observed. Future studies should investigate the influence of soil chemistry alongside other factors, such as plant age, investment in carnivory (e.g. leaf stickiness), prey diversity and availability, local climate conditions and levels of habitat disturbance.

Implications for the evolution and ecology of carnivorous plants

The results of the present study add to the growing body of literature confirming that carnivory is a successful nutrient acquisition strategy not only in ‘typical’ sunny and waterlogged environments (Givnish et al., 2018), but in less mesic habitats too (Schulze et al., 1991; Paniw et al., 2017a). Given its high level of investment in the carnivorous habit (with highly viscous mucilage, vascularized gland structure, a strong honey-like scent and production of digestive enzymes), it is perhaps unsurprising that D. lusitanicum would receive a significant nutritional benefit from prey. Future studies should further investigate the links between investment in the carnivorous syndrome and success of prey capture and nutrient assimilation, particularly for carnivorous plant species occurring in less mesic environments. In particular, we highlight the gap in our current scientific knowledge for Byblis gigantea and Byblis lamellata occurring in seasonally xeric environments in the Southwest Australian Floristic Region, especially given the high degree of morphological convergence these two species have with D. lusitanicum. All three are perennial sub-shrubs, with an upright rosette growth form, elongated glandular leaves and well-developed root systems, but the trapping hairs of Byblis species are much less complex in structure than the trapping glands of D. lusitanicum (Darwin, 1875). Comparative studies of convergent carnivorous plant taxa such as these could provide novel insights into the relationship between investment and reward for carnivorous plants, particularly for those species occurring in seasonally xeric environments.

Implications for conservation

The high degree of reliance on prey-derived N exhibited by D. lusitanicum individuals in their natural habitat may have significant implications for the ongoing conservation and management of this species. While D. lusitanicum has not yet been formally assessed by the International Union for the Conservation of Nature (Clarke et al., 2018), it is widely considered to be a conservation-dependent species (Correia and Freitas, 2002; Garrido et al., 2003; Jennings and Rohr, 2011; Paniw et al., 2015; Cross et al., 2017). Critically, D. lusitanicum tends to occur in isolated populations, often with small individual numbers, and is thus threatened by natural processes such as genetic drift and demographic stochasticity (Muller and Deil, 2001), as well as broad-scale anthropogenic disturbances such as urban and industrial development and heavy livestock pressure (Correia and Freitas, 2002; Garrido et al., 2003). In order to successfully manage and conserve D. lusitanicum populations, the ecology and reproductive biology of the species must also be considered. Recent studies by Paniw et al. (2017b) and Cross et al. (2017) have highlighted the importance of fire management for the regeneration of D. lusitanicum populations, whilst Salces-Castellano et al. (2016) reveal the importance of insect pollinator populations for their role in improving the seed set of D. lusitanicum individuals. The present study adds to this body of ecological knowledge, revealing the significant contribution of insect prey to the nutrition of mature D. lusitanicum individuals in their natural habitat, particularly at sites with poorer soil fertility. We recommend that management plans should strive to conserve the diversity and abundance of insect populations throughout the geographical range of D. lusitanicum (as per Cardoso et al., 2011). Given the possible influences of soil chemistry, future research could also consider the effects of N deposition and eutrophication on the growth and nutrition of wild D. lusitanicum populations (as per Millett et al., 2012), particularly in regions affected by land-use change. It is clear that a holistic management approach addressing ecology and threatening processes at both landscape and local scales is integral for the conservation of this highly specialized and rare carnivorous plant.

Conclusions

The present study provides an ecological perspective on the nutrition of D. lusitanicum by quantifying the nutritional benefit of carnivory in situ. This work builds on prior research (Darwin, 1875; Bertol et al., 2015; Paniw et al., 2017a) to confirm the status of the Mediterranean endemic D. lusitanicum as a highly successful carnivorous plant. Understanding the ecological significance of carnivory in less mesic, seasonally dry and fire-prone infertile habitats provides crucial insights into the evolution of the carnivorous syndrome, and can better inform conservation practices for threatened populations of D. lusitanicum. Future research needs to investigate the role of prey and photosynthesis in the C nutrition of D. lusitanicum, evaluate the nutritional benefits of carnivory for convergent taxa such as Byblis, and further assess the relationship between investments in carnivory and reward from prey.

FUNDING

This work was financially supported by Spanish MINECO-FEDER funding (project HERRIZA; grant CGL2015-64007-P) (FEDER ID: 501100008530), and by a grant from the International Carnivorous Plants Society (no funder ID available).