•Carnivory in plants is presumed to be an adaptation to a low-nutrient environment. Nitrogen (N) from carnivory is expected to become a less important component of the N budget as root N availability increases.
•Here, we investigated the uptake of N via roots versus prey of the carnivorous plant Drosera rotundifolia growing in ombrotrophic bogs along a latitudinal N deposition gradient through Sweden, using a natural abundance stable isotope mass balance technique.
•Drosera rotundifolia plants receiving the lowest level of N deposition obtained a greater proportion of N from prey (57%) than did plants on bogs with higher N deposition (22% at intermediate and 33% at the highest deposition). When adjusted for differences in plant mass, this pattern was also present when considering total prey N uptake (66, 26 and 26 μg prey N per plant at the low, intermediate and high N deposition sites, respectively). The pattern of mass-adjusted root N uptake was opposite to this (47, 75 and 86 μg N per plant).
•Drosera rotundifolia plants in this study switched from reliance on prey N to reliance on root-derived N as a result of increasing N availability from atmospheric N deposition.
Ever since Darwin’s (1875) insightful book, plant biologists have been intrigued by carnivorous plants: how and why they attract, capture and digest animal prey and how they use the nutrients once incorporated into their own tissues (Juniper et al., 1989). Carnivory in plants has evolved independently on at least six occasions (Ellison & Gotelli, 2009), and is widespread, with over 600 carnivorous plant species occurring on every continent, except Antarctica. The convergence on this single trait makes carnivorous plants useful model systems for a number of important evolutionary, ecological and ecophysiological questions (Ellison & Gotelli, 2001; Ellison et al., 2003).
Key to understanding the evolution and ecology of carnivorous plants is the interaction between the uptake of nutrients through their roots and the uptake of prey-derived nutrients (Ellison, 2006). The additional source of nitrogen (N) and phosphorus (P) gained by carnivorous plants from their prey is used for growth, photosynthesis and seed production (Darwin, 1875; Aldenius et al., 1983; Thum, 1988; Thorén & Karlsson, 1998; Adamec, 2002). However, carnivory also carries a cost in reduced efficiency of photosynthesis as a result of the modification of leaves into traps (Givnish et al., 1984; Knight, 1992; Méndez & Karlsson, 1999; Ellison & Farnsworth, 2005). In the cost–benefit model proposed by Givnish et al. (1984), carnivory should be beneficial if the nutrients captured from prey can be used to increase photosynthesis and balance these costs. Givnish et al. (1984) predicted that investment in carnivory should have a net marginal benefit in low-nutrient, high-light and wet sites. As nutrient availability increases and/or light and water availability decreases, the level of investment in carnivory that yields the highest net benefit decreases until noncarnivory is the optimal strategy. This is thought to explain differences in trap complexity and resulting differences in prey nutrient acquisition. For example, the percentage of total plant N that is derived from prey (%Ndfp) varies considerably, from 10% for Sarracenia purpurea to 87% for Drosera pallida (Ellison & Gotelli, 2001). In addition, there can be considerable within-species variation in %Ndfp. For example, Millett et al. (2003) found that %Ndfp for different populations of Drosera rotundifolia was 29–65%. However, no-one has yet identified environmental drivers of within-species differences in %Ndfp.
There is strong evidence that prey capture is less beneficial to carnivorous plants when nutrient availability is increased. In a meta-analysis of 26 experimental ‘prey addition’ studies, Ellison (2006) showed that prey capture results in increased ‘fitness’ only when root N availability is low. This effect is presumably because increased root nutrient availability means that prey N is a smaller proportion of total plant N, and so changes in the amount of this N source have a smaller, or no, impact. In addition, a number of studies have shown that carnivorous species reduce their investment in carnivory when N availability increases. For example, when root N availability is increased, S. purpurea pitcher morphology changes, becoming less pitcher-like and more leaf-like (Ellison & Gotelli, 2002); D. rotundifolia traps become less sticky (Thorén et al., 2003); and Utricularia spp. produce fewer traps (Knight, 1991). It seems likely that this phenotypic plasticity will result in a reduction in the amount of N obtained from prey, although this has never been measured. As such, no-one has yet shown in situ and in experiments in which there has been no manipulation that, when root N availability increases, the amount of N gained from prey decreases and prey N becomes a less important component of the N budget.
Gradients of anthropogenic N deposition provide a good opportunity to test fundamental questions on responses to N availability. Ombrotrophic bogs and the plants that grow on them are particularly sensitive to changes in atmospheric N deposition because, by definition, their entire N budget is derived from wet and dry deposition. As a result, plants growing on ombrotrophic bogs show clear responses to increasing levels of N deposition in terms of their N uptake and use (Heijmans et al., 2002; Bragazza et al., 2004). This ultimately results in species loss and community change (e.g., Gunnarsson et al., 2002; Bubier et al., 2007). The carnivorous plant D. rotundifolia grows predominantly on ombrotrophic bogs. It is a relatively common plant with a widespread, circumboreal distribution. Furthermore, D. rotundifolia is rooted directly among the living Sphagnum shoots, with an intricate competitive balance: increased N availability may have negative direct effects on Drosera (Redbo-Torstensson, 1994), but also increase its ability to compete with Sphagnum (Svensson, 1995).
In this study, we used natural abundance stable isotope ratio measurements (δ15N) to quantify the amount of N contained in individual D. rotundifolia plants that originated from prey sources and from root uptake. We made these measurements at three ombrotrophic bogs in Sweden, each of which received different amounts of background N deposition. We primarily aimed to test the following hypotheses: (1) that increasing N deposition (and therefore root availability) results in (a) increased root N uptake, (b) increased growth and therefore plant size and (c) plants that are more N replete; (2) that increasing N deposition results in N from prey capture becoming a less important component of the total N budget of the plant; and (3) that plants become increasingly N replete as the amount of prey N incorporated into their tissues increases.
Materials and Methods
Study sites and sampling
Three ombrotrophic mires were identified in Sweden along a latitudinal N deposition gradient (Table 1). Sampling was made in undisturbed, completely open, ombrotrophic parts with a hollow hummock microtopography and typical bog vegetation. The central and northern sites were dominated by Sphagnum fuscum and S. balticum, whereas the southern site was dominated by S. fuscum, S. magellanicum and S. cuspidatum. At each site, 10 S. fuscum hummocks were chosen as study plots. Sampling was restricted to S. fuscum hummocks to avoid differences in hydrology and in Sphagnum growth – the effect of N deposition on Sphagnum growth differs between hummock and lawn communities (Limpens et al., 2011). Sampling was carried out between 8th and 20th August 2010. At each plot, the following samples were removed: 15 D. rotundifolia L. plants, a sample of S. fuscum capitula and a sample of potential prey. All previously captured insects were removed from each D. rotundifolia plant and the old growth was removed, leaving the current year’s root, stem, leaves and flowers. These 15 plants were pooled to form a single sample per plot for analysis (n = 10 per site). The S. fuscum sample consisted of a total of approximately 10 cm2 of capitula (the top 1 cm of the plant) cut from at least three separate places on the hummock. These were pooled to form a single sample. Any other Sphagnum species were carefully removed from the sample under a dissecting microscope. The sample of potential prey was collected by placing a 24 × 10-cm2 yellow sticky insect trap on each plot. This was left for a minimum of 24 h, after which a sample of insects < 2 mm in length (representing the typical prey size for D. rotundifolia) was carefully removed using fine tweezers.
Table 1. Locations, precipitation, temperature, growing season length and nitrogen (N) deposition of the three sites used in the study
Mean annual precipitation (mm yr−1)
Mean temperature January/July (°C)
Mean growing season length (d)a
Growing season average temperature (ºC)a,b
N deposition (g N m−2 yr−1)c
aBased on data from IVL Svenska Miljöinstitutet (http://luftweb.smhi.se (accessed 15 October 2011)), mean values for 2006–2009 inclusive.
bGrowing season is the number of days with mean temperature ≥ 5°C.
cModelled N deposition data from EMEP model (http://www.emep.int (accessed 30 September 2011)). Data are mean values for 2004–2009 inclusive.
North (N): 64°10′N, 19°35′E
Central (C): 60°01′N, 17°22′E
South (S): 56°51′N, 13°28′E
Schulze et al. (1991) pioneered the use of stable isotopes for the measurement of prey versus root N uptake by carnivorous plants. The use of natural abundance measurements of stable isotopes allows the estimation of the relative contribution of two distinct N sources to a single pool, and can be employed in any system in which there are only two N sources (or, less accurately, where they comprise the vast majority of the total N) and where the δ15N value of the two sources is different. The δ15N value of the pool is then a result of the δ15N value of each source and the relative contribution of each source to the pool. This method is used extensively for animals in trophic ecology (Boecklen et al., 2011), and was initially used in plants to determine the relative contribution of N2 from atmospheric fixation and N from root uptake to the N budget of N2-fixing plants (Shearer & Kohl, 1986). As a result of discrimination during metabolism, organisms at higher trophic levels tend to become relatively enriched in 15N, generally by 3–4‰ (Post, 2002), although the entire range of δ15N enrichment factors is far greater (for a review, see Vanderklift & Ponsard, 2003). As a result, the insect prey of carnivorous plants is expected to be enriched relative to N taken up through the roots. The δ15N of a D. rotundifolia plant is therefore defined by the relative contribution of these two N sources to the total N budget. The technique has been used in a number of studies to measure prey contribution to the total N content of Drosera spp. (Schulze et al., 1991), D. rotundifolia (Millett et al., 2003, 2012) and Nepenthes spp. (Moran et al., 2001).
All samples were kept cold before being dried for 72 h at 70°C; this was performed within 5 d of sample collection. Drosera rotundifolia plants were weighed, and all plant samples were milled to a fine powder with a ball mill (Retsch MM200). Insect samples were ground to a fine powder in a small pestle and mortar. The δ15N of all tissues was analysed using a Costech ECS 4010 elemental analyser linked to a Thermo Delta XP Plus isotope ratio mass spectrometer. Results are given using the δ notation expressed in units of per mil (‰), where δ15N = (Rsample/Rreference) − 1 × 1000, and R = 15N/14N. Data are reported with respect to the primary international reference AIR (i.e. = 0‰), and the measurements are calibrated using the internal standards gelatin, alanine and glycine, which are calibrated against secondary international reference materials. %N and %C are determined from the mass spectrometric output using tryptophan as a standard.
The contribution of insect-derived N to the total N content of D. rotundifolia was calculated using a simple one-isotope, two-source, end-member mixing model as follows:
(%Ndfp, percentage of N derived from insect prey; δ15NDROSERA, δ15N of the pooled sample of D. rotundifolia plants; δ15NSPHAGNUM, δ15N of the capitula of S. fuscum in which D. rotundifolia is growing; δ15NINSECT, δ15N of the sample of the insects available as prey). δ15NDROSERA and δ15NSPHAGNUM are the values for each plot, and δ15NINSECT is the mean value for each site. The mean value was used because it seems likely that the insects form a single population on each site because they are mobile.
Drosera rotundifolia dry mass, %N and %Ndfp were used to calculate the following variables: total N per plant, total prey-derived N per plant (Ndfp) and total root-derived N per plant (Ndfr). To test the hypotheses relating to ecophysiological changes in response to N deposition (hypotheses 1 and 2), we used one-way ANOVA and ANCOVA in PASW Statistics (IBM) to test for differences between the three sites. Differences in Ndfp and Ndfr between sites were first analysed with a one-way ANOVA. As a result of the strong relationship between plant size and Ndfp and Ndfr, we then analysed Ndfp and Ndfr using plant mass as a covariate. Thus, Ndfp and Ndfr were compared between sites at a common mass (the mean of the independent variable). We present values on an unadjusted basis and adjusted for differences in mass (mass adjusted) where appropriate. Where the data did not confirm to the assumptions of homoscedasticity, they were log10 transformed. We used linear regression to test our third hypothesis by testing the relationship between the average amount of prey N (Ndfp and %Ndfp) in plants on each hummock and the average tissue N concentration (%N). In order to account for differences in plant mass in this analysis, we used ‘size-standardized’ measures of prey- and root-derived N content, RNdfp and RNdfr (RNdfp = Ndfp/plant mass; RNdfr = Ndfr/plant mass).
The concentration of N in S. fuscum capitula increased from north to south, that is, with increasing N deposition (Table 2), being significantly lower at the low and mid deposition sites than at the high deposition site.
Table 2. Measured and derived variables for Drosera rotundifolia, Sphagnum fuscum and a sample of D. rotundifolia prey at three ombrotrophic bogs in Sweden
Dros. mass (mg)1
Dros. N (μg per plant)1
Dros. Ndfr (μg per plant)1
Dros. Ndfp (μg per plant)1
Sph., S. fuscum; Dros., D. rotundifolia; Ndfp, total N from prey (unadjusted for plant size); Ndfr, total N from root uptake (unadjusted for plant size).
1Log10 transformed before analysis.
2Comparing differences between sites. Within a variable, sites with different letters are significantly different from each other (Fisher’s least significant difference, P < 0.05).
There were large statistically significant differences in the size of D. rotundifolia plants between sites (Table 2). The dry mass of plants growing at the south site was nearly three times that of those at the north site; the central site was intermediate. In addition, the total amount of N and the total amount of root-derived N (Ndfr) contained in each plant followed this pattern, with an approximately threefold increase in N content from north to south (Table 2). However, although there were statistically significant differences between sites, the total amount of prey-derived N (Ndfp) did not follow this pattern (Table 2). Plants growing at the south site contained the most prey-derived N and those at the central site the least. There was also no clear pattern in the %N concentration of D. rotundifolia plants along the latitudinal gradient. Plants growing at the central site had lower tissue %N concentrations than those at the north and south sites.
Sphagnum fuscum had negative δ15N values (range across sites − 4.10‰ to − 3.70‰); prey had positive values (1.37–4.97‰). As a result, there were large and significant (P < 0.001) differences in δ15N between S. fuscum and the sample of potential insect prey. This difference was 5.07‰ at the north site, 9.07‰ at the central site and 7.86‰ at the south site. When the δ15N values of S. fuscum, insect prey and D. rotundifolia were used to estimate the amounts of prey- and root-derived N in D. rotundifolia, there were clear and statistically significant differences in the source of N for the plants along the sampled gradient. When adjusted for differences in plant mass, Ndfp was considerably higher in plants from the low deposition site than from the other sites (Fig. 1a; P < 0.001). They had also incorporated less root-derived N (Ndfr) into their tissue (Fig. 1b; P < 0.001) and, as a result of these patterns of prey and root N uptake, D. rotundifolia plants growing at low N deposition obtained, on average, over one-half of their N from prey relative to c. 20–30% at the other sites (Fig. 1c, P < 0.001).
When considering differences between plots (both within and between sites), there were some clear relationships between the measured variables. Larger D. rotundifolia plants contained more N than smaller plants. This pattern was consistent for root-derived, prey-derived and total N (Fig. 2, P < 0.001). When adjusted for differences in mass, D. rotundifolia plants with higher prey N content (RNdfp) tended to have higher tissue %N content (Fig. 3a, r2= 0.19, P = 0.006). This relationship was similar, but weaker, when %Ndfp rather than the amount of prey N was considered (Fig. 3b, r2= 0.149, P = 0.035). However, when adjusted for differences in plant mass, there were no significant relationships between root N content (RNdfr) and %N (r2= 0.001, P = 0.855). There were weak, but significant, correlations between the %N of S. fuscum capitula on a hummock and the nutrition of D. rotundifolia growing on the hummock. Drosera rotundifolia plants growing on hummocks in which S. fuscum contained relatively high tissue %N tended to be larger (r2= 0.15, P = 0.033), had higher tissue %N concentrations (r2= 0.15, P = 0.033), contained more N in their tissues in total (r2 = 0.18, P = 0.020) and contained more root-derived N (RNdfr: r2 = 0.18, P = 0.020). However, there was no relationship with mass-adjusted RNdfp (r2= 0.036, P = 0.313) or %Ndfp (r2= 0.204, P = 0.197).
In this first in situ study without manipulations of root N availability or prey capture rates, we show that, when N deposition (and therefore availability) increases, D. rotundifolia plants incorporate less prey N and more root-derived N into their tissues, and therefore obtain a lower percentage of their N from prey. This overall result supports the cost–benefit model proposed by Givnish et al. (1984) in an ecological context.
A somewhat surprising result was that the shift from predominantly prey to root reliance occurred at very low N deposition (< 0.4 g N m−2 yr−1). This can be related to the literature on critical loads of N on other peatland processes. A central ecosystem service of peatlands is carbon balance, and a detailed literature review (Limpens et al., 2011) has indicated that the key process, Sphagnum growth, starts to decline at a tissue concentration of c. 10 mg g−1, and this is reached at N depositions (natural or experimental) of c. 1 g N m−2 yr−1. For Sphagnum, the natural low N levels in bogs makes growth N limited (Aerts et al., 1992), a limitation that can be avoided by D. rotundifolia by carnivory (Ellison, 2006).
Prey N uptake was presumably lower when N deposition was higher because of reduced investment in prey capture. Previous studies have shown that carnivorous plants reduce investment in trapping prey when N availability to the roots is increased (Thorén et al., 2003; Ellison & Gotelli, 2002). There is a presumption that this reduces prey capture and the uptake of prey-derived nutrients. We provide good evidence that these changes might result in reduced uptake of prey N, at least for D. rotundifolia. The plants at the site receiving the highest level of N deposition took up considerably less prey N than those at the lower N deposition sites, after prey N uptake was adjusted for differences in plant mass. One mechanism might be a reduction in trap stickiness. This response has been shown in a glasshouse study (Thorén et al., 2003) and would presumably reduce prey capture by allowing more prey to escape once trapped. However, this may also be caused by other mechanisms. For example, the red colour of D. rotundifolia leaves is thought to be a prey attraction mechanism, and could conceivably respond to N availability. Root N uptake was presumably higher because of increased pore water N concentration, resulting in increased uptake, as shown by Heijmans et al. (2002).
Our results support the key assumption that, when N availability increases, the N nutrition of carnivorous plants shifts away from reliance on prey-derived N towards reliance on root-derived N. These results also support the suggestion by Thorén et al. (2003) that carnivorous plants might switch from investment in nutrient uptake through prey capture to nutrient uptake through their roots when N availability increases. This impact of increased N availability explains the observed lack of response to prey addition when root N availability is high in previous studies (reviewed by Ellison, 2006), because additional N inputs from prey seem likely to have little benefit if the plant’s N requirements are met by root uptake.
We also provide some evidence of the nutritional benefit of prey capture. Plants that gained more N from prey (either mass-adjusted Ndfp or %Ndfp) had an enhanced nutritional status (higher tissue %N). The relationship with %Ndfp was shown by Millett et al. (2003), but the relationship with Ndfp has not been demonstrated previously. This pattern was absent for root-derived N. Conversely, growth (i.e. mass) was closely correlated to total plant N content. It is not possible to determine causality, but previous studies have shown variable impacts of prey addition and root N addition on the growth and N content of carnivorous plants (Ellison, 2006). Prey N uptake and root N uptake are also linked in some carnivorous plant species (Adamec, 2002). We also show that larger plants contain more Ndfp. Thus, Schulze & Schulze’s (1990) finding that increased leaf area (and presumably plant mass) resulted in higher rates of prey capture in D. rotundifolia translates into an increase in prey N uptake for the plants. We did not measure P or potassium (K) content or uptake in the plants. However, these nutrients have both been shown to be important components of carnivorous plant responses to changes in root nutrient availability and prey availability, because carnivorous plant growth often appears to be co-limited by these nutrients (Ellison, 2006).
Tissue %N content in D. rotundifolia remained remarkably stable at all sites, despite large differences in N deposition and, presumably, availability. This might indicate that this parameter is well regulated by the plants. This regulation might be caused by the ‘dilution effect’ of increased growth rates and by the reduction in prey N uptake when root N uptake increases. However, this regulation might also be explained by changes in leaf turnover, which can be rapid in D. rotundifolia (Schulze & Schulze, 1990). If higher N availability increases the rate of leaf turnover, N losses from the plant will increase. Furthermore, Butler & Ellison (2007) showed that S. purpurea relies, to a large extent, on stored N for use in subsequent years, but no studies have compared the use of prey-derived N with that of root-derived N. Our results provide a tentative suggestion that root- and prey-derived N might be used differently, with both N sources used for growth, but prey N also being allocated to storage processes/mechanisms (resulting in increased %N).
The natural abundance stable isotope method has been used in a number of studies to estimate the contribution of prey N to the N budget of carnivorous plants (for a summary, see Brearley, 2011). However, the method makes a number of assumptions (detailed by Boddey et al. (2000) for N2 fixation and outlined by Millett et al. (2012) for carnivorous plants). For example, it is assumed that root-derived N discrimination between 15N and 14N is the same for S. fuscum and D. rotundifolia. This seems a reasonable assumption. Millett et al. (2003) found that δ15N of Sphagnum was almost identical to the lowest value of δ15N for individual D. rotundifolia plants. Glandless mutants of D. erythrorhiza were used by Schulze et al. (1991), and these would be an ‘ideal’ reference plant, but no such mutants are known for D. rotundifolia. In addition, it is assumed that there is no discrimination between 15N and 14N during the uptake of prey N. Any deviation from these assumptions will affect the precision of the point estimate for %Ndfp. Boecklen et al. (2011) found that the 95% confidence interval of the point estimate for the percentage contribution of the two end-points spanned, on average, 33%. This uncertainty is caused by the variability in the δ15N values of the two end-points. We believe that our estimates of %Ndfp can therefore probably only provide good qualitative or semi-quantitative estimates. However, the differences in Ndfp between our lowest N deposition site and the higher N deposition sites are relatively large, as well as being statistically significant. We therefore consider these patterns to be biologically significant.
Our three sites were located on a latitudinal gradient. The purpose of this was to enable the large differences in N deposition to be exploited. There are other differences between the sites that will also clearly affect the plants, but this deposition gradient has been used successfully in studies of nitrogen effects on other peatland processes (e.g. Breeuwer et al., 2008; Granath et al., 2009). That we confined sampling to the S. fuscum hummock microhabitat within truly ombrotrophic peatlands should minimize the latitudinal effects. There is significantly higher precipitation at the southern, highest N deposition site. Higher rainfall might reduce prey capture by washing prey from the traps, but the largest shift from prey reliance was between the central and northern sites with similar precipitation. Differences in plant size, as a result of increased N deposition, are likely to be enhanced by the differences in temperature and, in particular, growing season length. However, we cannot conceive a mechanism by which changes in these temperature variables might create the observed differences in prey N uptake. Indeed, increased temperature and growing season at the central and southern sites would presumably increase prey N uptake because of increased rates of digestion. Furthermore, the large differences in D. rotundifolia plant size and the strong correlation between size and root and prey N content confirm our use of mass-based calculations of the tissue content of these two N sources.
In conclusion, understanding the responses of ecosystems to anthropogenically elevated N deposition is essential if we are to mitigate the impacts. Ombrotrophic bogs are particularly sensitive to N deposition. We provide good evidence that a key process for the carnivorous plant D. rotundifolia (i.e. prey N use) is significantly impacted, even by low rates of N deposition. We conclude that anthropogenic N deposition results in reduced reliance on carnivory by the D. rotundifolia plants in this study. Carnivory might therefore be a Hobson’s choice in an ecological sense, as well as in an evolutionary sense (as suggested by Ellison, 2006). As such, even for carnivorous plants, high reliance on carnivory might be the strategy of ‘last resort’ when N availability is very low. Nonetheless, the residual costs of carnivory probably remain, resulting in reduced competitive ability in these higher N deposition sites. Manipulative experiments are necessary to elucidate the main interactions between foliar and root nutrient uptake in carnivorous plants, and the investment in carnivory.
Stable isotope analyses were undertaken at the Life Sciences Mass Spectrometry Facility, funded by the UK Natural Environment Research Council (EK113-08/07). Many thanks are due to two anonymous reviewers for their helpful comments on the manuscript.