Inquiline predator increases nutrient-cycling efficiency of Nepenthes rafflesiana pitchers

Abstract

The modified-leaf pitchers of Nepenthes rafflesiana pitcher plants are aquatic, allochthonous ecosystems that are inhabited by specialist inquilines and sustained by the input of invertebrate prey. Detritivorous inquilines are known to increase the nutrient-cycling efficiency (NCE) of pitchers but it is unclear whether predatory inquilines that prey on these detritivores decrease the NCE of pitchers by reducing detritivore populations or increase the NCE of pitchers by processing nutrients that may otherwise be locked up in detritivore biomass. Nepenthosyrphus is a small and poorly studied genus of hoverflies and the larvae of one such species is a facultatively detritivorous predator that inhabits the pitchers of N. rafflesiana. We fitted a consumer–resource model to experimental data collected from this system. Simulations showed that systems containing the predator at equilibrium almost always had higher NCEs than those containing only prey (detritivore) species. We showed using a combination of simulated predator/prey exclusions that the processing of the resource through multiple pathways and trophic levels in this system is more efficient than that accomplished through fewer pathways and trophic levels. Our results thus support the vertical diversity hypothesis, which predicts that greater diversity across trophic levels results in greater ecosystem functioning.

1. Introduction

Some terrestrial plants, such as pitcher plants, bromeliads, heliconias and trees hold small volumes of freshwater within their leaves, leaf axils, flowers or stems [1–3]. These water bodies can become habitats for specialized aquatic organisms [1]. Such habitats are known as phytotelmata (sing. phytotelma), and the organisms that inhabit them, as inquilines [1,2]. Phytotelmata are useful ecological models because of their small size, manipulatability and the relative simplicity of the communities that inhabit them [4–6].

Most phytotelmata are allochthonous [7]—that is, they are sustained by energy/nutrients that originate outside of the system, instead of by energy/nutrients that are fixed by autotrophs within it (autochthonous). In the case of carnivorous pitcher plants, invertebrate prey trapped by pitchers are the primary energy/nutrient source. Detritivorous inquilines are known to accelerate [8–11] the breakdown of these detrital resources into simpler compounds such as ammonia, which can be absorbed by the host plant. Because carnivorous pitcher plant individuals are often nutrient limited in their natural environments [12], this process represents a mechanism by which detritivorous inquilines benefit their hosts in a relationship that may be termed as a digestive/nutritional mutualism [13]. Furthermore, if the phytotelma is a miniature ecosystem, then principles concerning the nutrient-cycling efficiency (NCE) of phytotelmata may be extrapolated to larger, more complex ecosystems.

Inquiline predators often exert top-down control on phytotelma communities through trophic cascades [1,7,14–16], but how predators affect the NCE of ecosystems is not well understood [17]. Studies from diverse systems including phytotelmata (reviewed in [17]) have found that the presence of predators is often correlated with higher NCEs. However, mechanistic explanations for this observation are largely descriptive and suggest that the effect of predators on the NCE of systems depends on the number of trophic levels beneath them [17]. Mouquet et al. [18] showed using a theoretical model that was parametrized with field data of the trumpet pitcher plant (Sarracenia purpurea) phytotelma that the top predator Wyeomia smithii (Culicidae) increases NCE by suppressing mesopredators (protozoa and rotifers) that consume detritivorous bacteria. However, these findings seemed to suggest that (top) predators alter NCE only indirectly, by releasing lower trophic-level detritivores from (meso-) predation pressure. Other studies also argue that microbe populations are the primary nutrient processors in the S. purpurea system [19]. It thus remains unclear (i) if predation and nutrient transfer through the food chain can directly increase the NCE of an ecosystem and (ii) how direct consumption of detrital resources by predators may alter previous findings.

Nepenthosyrphus is a poorly studied genus of hoverflies (Diptera: Syrphidae) whose predatory larvae are found only in the fluid-filled pitfall traps of tropical pitcher plants (Nepenthes spp.) in Southeast Asia [20–22]. In this study, we examined how a Nepenthosyrphus sp. inquiline predator (figure 1a,b; a taxonomic description of the Nepenthosyrphus sp. used in our study is currently underway) in Nepenthes rafflesiana (figure 1e), alters the NCE of N. rafflesiana pitchers. This was done by formulating a consumer–resource type model and fitting it to data collected from in vitro experiments conducted on Nepenthosyrphus sp., Endonepenthia schuitemakeri (Diptera: Phoridae; figure 1c,d; a detritivorous prey species) and ant-carcass detrital resources. Predators alter communities through trophic (changes in abundances of prey species) and non-trophic (changes in behaviours of prey in response to predator(s) [23]) pathways [17]. Both trophic and non-trophic effects can increase NCEs, but non-trophic effects are often subtle and tend to be more system specific, while trophic ones are often stronger and more generalizable across systems [17]. We thus consider only trophic effects in this study.

Figure 1.

Nepenthosyrphus sp. female adult (a) and larvae in a lower pitcher of Nepenthes rafflesiana (b); Endonepenthia schuitemakeri female adult (c); and fourth instar larva (d); N. rafflesiana lower pitchers in situ (e). Scale bars: 1 mm (a,c,d); 5 cm (b,e). (Online version in colour.)

2. Material and methods

(a). Sample collection

All pitcher fluids and inquilines were collected by pipetting out the contents of N. rafflesiana lower pitchers found in the Nee Soon Swamp Forest (1°22′29.93″ N, 103°48′40.63″ E) and Kent Ridge Park (1°17′10.0″ N 103°47′20.1″ E) in Singapore. The inquiline species E. schuitemakeri (Phoridae) and Nepenthosyrphus sp. (Syrphidae) were isolated from the samples, kept in filtered pitcher fluids and starved for 24 h prior to use in experiments. Pitcher fluids were filtered through Whatman^® grade 4 filter paper (GE Healthcare Life Sciences, Buckinghamshire, UK) to remove prey debris and other metazoan inquilines and then stored in sterile, 50 ml, centrifuge tubes at 4°C for not more than two weeks prior to use in experiments. Leaf nests of Oecophylla smaragdina ants were harvested from the National University of Singapore (NUS) Kent Ridge Campus (1°17′46.3″ N 103°46′33.0″ E) and the colony was killed by freezing at −20°C for 24 h. Carcasses of the major workers were picked out by hand and used as the detrital resource in experiments.

(b). Nutrient cycling model

The trophic interactions between a single resource, prey and predator species may be modelled using the set of differential equations:

$$\left.\begin{array}{rl}\frac{\mathrm{d}R}{\mathrm{d}t}=& \hspace{0.17em}c(S-R)-f_{10}{N}_{1}R-f_{20}{N}_{2}R-kR\\ \frac{\mathrm{d}{N}_1}{\mathrm{d}t}=\hspace{0.17em}& N_1(\hspace{0.17em}{f}_{10}R{\alpha }_{10}-f_{21}{N}_2-d_1)\\ \mathrm{and}\frac{\mathrm{d}{N}_2}{\mathrm{d}t}=\hspace{0.17em}& N_2(\hspace{0.17em}{f}_{21}{N}_1{\alpha }_{21}+f_{20}R{\alpha }_{20}-d_2)\end{array}\right\}$$ 2.1

In this model, the predator also consumes the resource directly (i.e. an omnivorous or facultatively detritivorous predator). This was done because earlier experiments had established that Nepenthosyrphus sp. feeds on both E. schuitemakeri detritivores and ant prey carcasses, albeit with a preference for the former (electronic supplementary material, Appendix S1). These equations follow the linear forms of the well-characterized consumer–resource family of models that are used extensively in community ecology [24,25] and that have also been used to model predation [26]. They express resource ($R$), prey ($N_1$) and predator ($N_2$) densities as functions of the consumption rates ($f_{ij}$), conversion efficiencies (${\alpha }_{ij}$) and mortality rates ($d_i$) of each species and the resource(s) it consumes (table 1).

Table 1.

State variables (a) and parameters (b), and their definitions, units and estimated values (of parameters only).

model terms		definition	units	specific parameter	estimated values (± s.e.)
(a) state variables	$C$	soluble nitrogen in pitcher fluids (nutrient cycling)	mmol	—	—
	$R$	dry mass of detrital resource	mg	—	—
	$N_1$	fresh mass of dipteran larvae	mg	—	—
	$N_2$	fresh mass of predator	mg	—	—
(b) parameters	$k$	rate of enzymatic (and microbial) digestion of detrital resource	d−1	$k$	0.236 (±0.068)
	$c$	turnover rate of detrital resource	d−1	$c$	—
	$S$	supply of detrital resource	Mg	$S$	—
	${\alpha }_{ij}$	feeding efficiency (gain in biomass of species i for every unit of biomass of species j or the resource consumed)	—	α10	3.047 (±0.428)
				α20	1.724 (±1.097)
				α21	0.142 (±0.017)
	${\beta }_{ij}$	excretion coefficient (amount of nitrogen produced per unit mass of species j or the resource consumed by species i)	mmol mg−1	β0	0.432 (±0.158)
				β10	0.108 (±0.020)
				β20	0.066 (±0.048)
				β21	0.040 (±0.001)
	$f_{ij}$	maximal feeding rate (rate of change in mass of species j per unit mass of species i and j present)	mg−1 d−1	$f_{10}$	0.082 (±0.018)
				$f_{20}$	0.016 (±0.008)
				$f_{21}$	0.042 (±0.004)
	$d_i$	mortality rate of species i	d−1	$d_1$	—
				$d_2$	—

If nitrogen is produced proportionally with each of the energy conversion processes present in the model, then the total concentration of nitrogen in the system can be modelled by the following equation, where ${\beta }_{ij}$ represents the nitrogen release constants of each corresponding energy conversion process:

$$\begin{array}{c}\frac{\mathrm{d}C}{\mathrm{d}t}={\beta }_{0}kR+{\beta }_{10}{f}_{10}{N}_{1}R+{\beta }_{20}{f}_{20}{N}_{2}R+{\beta }_{21}{f}_{21}{N}_2{N}_1\hspace{0.17em}.\end{array}$$ 2.2

If or when the system of differential equations in (2.1) arrives at an equilibrium, dC/dt becomes a constant.

(c). Parameter estimation

Four experimental set-ups were established, with each representing only a subset of the variables and processes described in the full model, and model parameters were sequentially estimated from these. These experimental set-ups were established in 50 ml, plastic, Cellstar^® centrifuge tubes filled with 10 ml of filtered N. rafflesiana lower pitcher fluids and were each conducted for 8 days. Each experimental set-up was replicated thrice and distributed evenly over three experimental batches. The first set-up contained only ant carcasses (hereafter referred to as detritus) and was used to estimate the parameters k and ${\beta }_0$, which were associated with the enzyme-mediated resource breakdown. The second set-up contained only the detritus and eight E. schuitemakeri 1st to 2nd instar larva individuals (hereafter referred to as prey). It was used to estimate the parameters $f_{10}$, ${\alpha }_{10}$ and ${\beta }_{10}$ associated with the consumption of detritus by prey. The third set-up contained only detritus and one individual of Nepenthosyrphus sp. (hereafter referred to as the predator), and was used to estimate the parameters $f_{20}$, ${\alpha }_{20}$ and ${\beta }_{20}$, which were associated with the consumption of detritus by the predator. The final parameter, $f_{21}$, the maximal feeding rate of the predator on prey, was estimated from a separate set of in vitro feeding experiments (electronic supplementary material, Appendix S1). Parameter estimation is detailed in electronic supplementary material, Appendix S2.

It must be noted that the model was parameterized using physiological constants determined at the level of the individual or sub-population but ecological inferences were drawn from an extrapolation of the model to the level of the population. This approach prevented us from estimating mortality rates ($d_1$ and $d_2$; natural death rates owing to disease or age), which are a population attribute. Instead, we ran numerical simulations until equilibrium conditions were reached using the estimated parameters and all possible combinations of 101 $d_1$ and 101 $d_2$ values between 0 and 1 d⁻¹. We then identified the range of mortality rate values that could give rise to coexistence of predator and prey, since the coexistence of the two in nature is understood. All analyses were performed in R v. 3.4.0 [27].

3. Results and discussion

Nepenthosyrphus sp. predators were observed, and filmed for the first time (electronic supplementary material, Appendix S6–S8), consuming large numbers of E. schuitemakeri prey. Each specimen consumed all eight individuals within the 8-day duration of each experiment. Eleven of the 15 model parameters were estimated from the experiments (electronic supplementary material, table S2.1), leaving only four unknown parameters: environmental resource supply ($S$; $c\times S$ is the pitcher prey capture rate in the model, but since this rate may be assumed to be proportionate to available prey in the environment, we simply refer to S hereafter as the environmental resource supply) and turnover (c) rates and mortality rates of prey (d₁) and predator (d₂) species. Although the true values of these four parameters cannot be known (indeed, they are likely to vary with environmental contexts), they must combine to lie within the parameter space that permits coexistence.

We explored the d₁ and d₂ parameter space by fixing $c=1$ and $S=10$ (which are biologically plausible values), while setting the other parameters to the values estimated from our data. Simulations run using these parameter values all resulted in stable equilibrium conditions (electronic supplementary material, figure S3.0) and suggested that low predator mortality rates (d₂) result in the extinction of prey while high values of predator (d₂) and prey (d₁) mortality rates result in the extinction of the predator at equilibrium (figure 2a). This overall trend was not altered by resource supply (S) or turnover (c) rates (electronic supplementary material, figures S3.1a–5a). Environmental resource supply rates (S) also determined equilibrium outcomes, with very low supplies leading to the extinction of both predator and prey, low supplies leading to the extinction of predator, intermediate supplies leading to coexistence and high supplies leading to the extinction of prey, at equilibria (figure 2b). Resource turnover rates (c; which represents both a rate of resupply and loss) also moderated equilibrium outcomes (figure 2b), but more significantly, a threshold value of this parameter determined whether the effect of detritivores on pitcher NCE was positive (as suggested by the literature) or negative (i.e. detritivores are parasitic) (figure 2c; also electronic supplementary material, figures S3.1b–5b). We were able to determine analytically that this threshold was $c=k({\beta }_0-{\beta }_1)/{\beta }_1$ (electronic supplementary material, Appendix S5).

Figure 2.

Equilibrium outcomes of simulations along (a) gradients of prey (d₁) and predator (d₂) mortality rates and (b) gradients of resource turnover (c) supply (S) rates, and comparison of NCEs between the various equilibria (c–f) across c–S parameter space. Each cell represents a simulation, and its colour, the outcome described in the legends (panels (c–f) share the same legend). The cross symbol in panel (a) represents the values of d₁ and d₂ in panels (b–f) (which share the same axes), while that in panel (b) represents the values of c and S in panel (a). Equilibrium conditions denote the persistence or extinction of predator or prey species when stable states are reached in simulations: trivial equilibrium = extinction of both predator and prey species; predator-dominant (pred-dom) equilibrium = extinction of prey species; prey-dominant (prey-dom) equilibrium = extinction of predator species; coexistence (coex) equilibrium = persistence of both predator and prey species.

Since the presence of a predator species results in reductions in prey populations, it may be supposed that the predator only increases NCE when the prey decreases NCE. However, our simulations suggested that predator–prey systems (simulations that included both predator and prey) almost always had higher NCEs than prey-only ones (simulations that excluded the predator) (figure 2d), regardless of whether these detritivorous prey had independently positive or negative effects on NCE (see also electronic supplementary material, figures S3.1c–5c). Similarly, predator–prey systems almost always had higher NCEs than enzyme-only ones (simulations that excluded both predator and prey) (figure 2e; electronic supplementary material, figure S3.1d–5d).

It may also be argued that the direct consumption of detritus by the predator made it function as a detritivore, and that these results thus only further confirm that detritivores increase the NCE of phytotelmata. However, our simulations showed that, across large regions of parameter space, predator–prey systems could achieve higher NCEs than predator-only ones (figure 2f). Furthermore, predator-only systems did not always have higher NCEs than enzyme-only ones (electronic supplementary material, figures S3.1e–5e), suggesting that the detritivorous habit of the predator could reduce NCEs under some parameter conditions. Yet even under these circumstances, predator–prey systems still achieved higher NCEs than both prey-only (electronic supplementary material, figures S3.1c–5c) and enzyme-only (electronic supplementary material, figures S3.1d–5d) systems. This meant that the direct consumption of detritus by the predator in itself was not efficient enough to increase NCEs very greatly. Thus, the positive effects of the predator on NCE are not attributable, alone, to its ability to directly consume detrital resources.

Collectively, these findings suggest that the Nepenthosyrphus sp. inquiline predator has a positive effect on the NCE of N. rafflesiana lower pitcher phytotelmata under most natural scenarios. This effect is not merely the consequence of Nepenthosyrphus sp.'s consumption of ant-carcass resources, and neither is it the mere result of a trophic cascade that results in the suppression of (potentially parasitic) detritivore populations. Instead, it is likely that the processing of a resource through multiple pathways (in this case, enzymatic, detritivorous and predatory pathways) and trophic levels is more efficient than that accomplished through fewer pathways and trophic levels. This is consistent with the vertical diversity hypothesis, which predicts that greater diversity across trophic levels results in greater ecosystem functioning [28,29].

(a). Limitations

A key weakness of the study was the method of estimation of α parameters. In consumer–resource models, α represents the efficiencies with which consumed resource biomass is transformed into consumer population biomass. This conversion efficiency is determined jointly by both individual growth and reproduction, but we have used the simplifying assumption that they are based wholly upon individual growth.

Plastic tubes may have been inexact replicas of live N. rafflesiana pitchers, which continuously replenish oxygen and remove dissolved nutrients (e.g. ammonium) from their fluids [10]. From our experience, Nepenthosyrphus sp. and E. schuitemakeri inquilines are unlikely to have been affected by the ranges of oxygen or ammonium in the experiments (W. N. Lam 2019, personal observation), but it is unclear if the pitcher fluid microbe communities were. Future studies should use in situ observations/manipulations to validate the findings of these in vitro experiments.

Ethics

Our work complies with the laws of the Republic of Singapore, which does not regulate the use of invertebrate animals in research. Fieldwork was carried out under the National Parks Board Permit no. NP/RP13-008-6c.

Data accessibility

Data available from the Dryad Digital Repository: https://doi.org/10.5061/dryad.1c59zw3rk [30].

Authors' contributions

W.N.L. and H.T.W.T. composed the study design. Y.Y.C. and W.N.L. collected the specimens. Y.Y.C. conducted the experiments and analysed the data. W.N.L. and F.W.S.L. developed the model and fitted the data to it. W.N.L. and F.W.S.L wrote the manuscript together and all other authors contributed to revisions. All authors approved of the final version of the manuscript and agreed to be held accountable for all aspects of the work presented here.

Competing interests

We declare we have no competing interests.

Funding

This project was funded by the Fourth Ah Meng Memorial Conservation Fund.