Trap geometry in three giant montane pitcher plant species from Borneo is a function of tree shrew body size


Author for correspondence:
Charles Clarke
Tel: +60 3 5514 5809


  • Three Bornean pitcher plant species, Nepenthes lowii, N. rajah and N. macrophylla, produce modified pitchers that ‘capture’ tree shrew faeces for nutritional benefit. Tree shrews (Tupaia montana) feed on exudates produced by glands on the inner surfaces of the pitcher lids and defecate into the pitchers.
  • Here, we tested the hypothesis that pitcher geometry in these species is related to tree shrew body size by comparing the pitcher characteristics with those of five other ‘typical’ (arthropod-trapping) Nepenthes species.
  • We found that only pitchers with large orifices and lids that are concave, elongated and oriented approximately at right angles to the orifice capture faeces. The distance from the tree shrews’ food source (that is, the lid nectar glands) to the front of the pitcher orifice precisely matches the head plus body length of T. montana in the faeces-trapping species, and is a function of orifice size and the angle of lid reflexion.
  • Substantial changes to nutrient acquisition strategies in carnivorous plants may occur through simple modifications to trap geometry. This extraordinary plant–animal interaction adds to a growing body of evidence that Nepenthes represents a candidate model for adaptive radiation with regard to nitrogen sequestration strategies.


Carnivorous plants occur in habitats that are deficient in key nutrients, such as nitrogen and phosphorus (Juniper et al., 1989). They have responded to these deficiencies by producing highly modified leaf organs that serve to attract, trap, retain and digest animals. A diverse array of trapping strategies has evolved, including snap-traps (e.g. Dionaea muscipula), sticky traps (e.g. Drosera spp.), vacuum traps (Utricularia spp.) and pitfall traps (e.g. Nepenthes spp.) (Lloyd, 1942; Juniper et al., 1989). Pitcher plants produce passive, gravity-operated pitfall traps that typically comprise an upper portion which possesses elaborate structures to entice visiting animals into a precarious position. Visitors that lose their footing drop into the fluid-filled lower portion of the trap, which is funnel- or cup-shaped, where they are retained and digested (Clarke, 1997; Phillipps et al., 2009). Nepenthes pitchers possess combinations of epicuticular wax, viscoelastic fluids and slippery pitcher mouths (peristomes) to capture invertebrate prey, most notably Formicidae (Gaume et al., 2002; Bohn & Federle, 2004; Gorb et al., 2004; Gaume & Forterre, 2007; Bauer et al., 2008; DiGiusto et al., 2008; Gaume & DiGiusto, 2009).

Although virtually all prey of Nepenthes examined to date are arthropods (Moran, 1996; Adam, 1997), some species produce very large pitchers and are reputed to trap vertebrates. One such species, Nepenthes rajah, is the world’s largest carnivorous plant, producing pitchers with capacities that may exceed 2 l (Clarke, 1997; Phillipps et al., 2009). Nepenthes rajah is a montane species that occurs only on Mount Kinabalu in northern Borneo. Despite anecdotal accounts of ‘rat catching’ by N. rajah (Clarke, 1997; Phillipps et al., 2009), there is no empirical evidence that vertebrate capture is central to its nutritional strategy. Indeed, Adam (1997) found that N. rajah caught only arthropods, more than 85% of which were Formicidae. We examined 36 N. rajah pitchers at Mesilau on Mount Kinabalu and found predominantly invertebrate prey (79% Formicidae), plus the remains of just one vertebrate, a skink. Since 2001, Sabah Parks personnel have observed only three instances of vertebrate capture by N. rajah at Mesilau. One of us (CC) has observed the contents of several hundred N. rajah pitchers on Mount Kinabalu since 1987, but has never found vertebrate prey. Thus, we conclude that vertebrate capture is rare in N. rajah, and that almost all of its prey could be readily accommodated by much smaller pitchers. Why, then, does it produce such large ones?

Our preliminary observations revealed that 58% (= 36) of N. rajah pitchers at Mesilau contained tree shrew faeces. Nepenthes lowii, another Bornean montane species, demonstrates a remarkable nitrogen sequestration strategy, in which mountain tree shrews (Tupaia montana) defecate into the pitchers whilst feeding on exudates secreted by glands on the inner surface of the pitcher lid. Faeces accounts for 57–100% of foliar nitrogen in this species (Clarke et al., 2009) and N. lowii‘toilet’ pitchers are relatively ineffective arthropod traps. The large orifices and reflexed, concave lids of N. lowii pitchers induce T. montana to sit astride the pitcher whilst feeding, facilitating the ‘capture’ of faeces. These pitcher characteristics are also present in N. rajah, although to a reduced degree: its pitchers retain the ability to trap invertebrates. In addition, we have observed tree shrew faeces in pitchers of N. macrophylla on Mount Trusmadi, c. 60 km south of Mount Kinabalu, on several occasions since 2003. Like N. rajah and N. lowii, this species produces very large pitchers with concave, reflexed lids (Clarke, 1997).

We tested the hypothesis that pitcher geometry in these ‘faeces-trapping’ species is related to the body size of T. montana, as faecal capture depends on the tree shrew’s hindquarters being positioned correctly over the pitcher orifice whilst the animal feeds on the lid exudates (Clarke et al., 2009). Any such relationship is likely to depend on the parts of the pitchers with which the tree shrews come into direct contact whilst feeding, that is, the peristome surrounding the orifice, and the pitcher lid. We measured pitcher characteristics in eight montane Nepenthes species from northern Borneo, finding that: (a) production of large pitchers is a non-exclusive requirement of the faeces-trapping syndrome (i.e., some non-faeces-trapping species also produce large pitchers); and (b) pitchers of the faeces-trapping species share a unique combination of five physical characteristics, including the distance from the front of the pitcher orifice to the inner surface of the lid, which corresponds exactly to the head+body (hb) length of T. montana. This distance is almost entirely dependent on pitcher orifice size and the angle at which the lid is positioned in relation to the orifice. On this basis, we conclude that the pitcher geometry in N. lowii, N. macrophylla and N. rajah is a function of tree shrew body size.

Our findings demonstrate that extraordinary modifications to nutrient acquisition strategies in carnivorous plants may occur through simple modifications of trap geometry. In N. lowii, N. macrophylla and N. rajah, enlargement of the pitcher orifice and lid angle to match the body size of T. montana, coupled with the production of copious lid exudates, play a key role in the acquisition of supplementary nitrogen.

Materials and Methods

Study sites and measurements

Pitchers of N. rajah (each from a separate plant) were measured and their prey examined in February (20 pitchers) and April (17 pitchers) 2009 at 2050 m asl at the Mesilau landslip, Mount Kinabalu, Sabah, Malaysia (N6.048°, E116.599°). Our sample included almost every suitable pitcher in the population that was accessible without disturbing the surrounding vegetation, which is very fragile. The contents of damaged, very old or very young pitchers are rarely identifiable, so these pitchers were excluded. The following pitcher characteristics were measured: pitcher height, width and depth, lid length and width, orifice depth and width, capacity and volume of contents, lid angle, orifice angle and a parameter that we call ‘fmfs’ (front of the mouth to the food source, Fig. 1). Pitcher height was measured from the bottom of the pitcher to the base of the ‘spur’, a filiform appendage that is the true leaf apex (Lloyd, 1942). Pitcher width and depth were measured at the widest/deepest points below the peristome. ‘Lid angle’ was defined as the angle subtended by the planes of axis of the pitcher lid and orifice, and fmfs was the distance from the front of the orifice to the deepest recess on the inner surface of the pitcher lid. The orifice angle was the departure of the plane of axis (determined as a straight line from the front of the orifice to the point at which the lid is joined to the pitcher) of the orifice from the horizontal. A simple estimate of lid area was calculated using the formula for the area of an ellipse (0.7854AB), where A is the lid length and B is the lid width.

Figure 1.

 Pitcher morphology and characteristics that were measured. (A) Nepenthes rajah pitcher showing (i) Tupaia montana faeces on the inner surface, (ii) fmfs (front of the mouth to the food source), (iii) orifice depth, (iv) lid angle; (B) detail of N. rajah lid showing concave structure, (v) midrib protrusion distance and (vi) lid depth; (C) N. burbidgeae terrestrial pitcher; note the convex lid and lid angle < 90°; (D) N. macrophylla aerial pitcher being measured with a 30 cm ruler by the senior author; note the concave lid, and large orifice depth and lid angle; (E) terrestrial pitcher of N. villosa, showing prominent peristome ribs, elongated ‘neck’ at the rear of the orifice and the convex lid, with lid angle < 90°; (F) aerial pitcher of N. lowii; (G) aerial pitcher of N. reinwardtiana; (H) terrestrial pitcher of N. stenophylla; (I) terrestrial pitcher of N. tentaculata; (J) still image taken from a video recording, showing T. montana sitting astride the orifice of an N. rajah pitcher whilst it feeds on the secretions of the lid glands; note the position of the animal’s hindquarters and tail (inside the pitcher); (K) still image taken from a video recording, showing three T. montana faecal pellets on the inner surface of an N. rajah pitcher. The pellet labelled ‘vii’ was deposited by T. montana during a visit to this pitcher that occurred whilst this recording was being made. Scale bar on all images, 5 cm.

Each of these measurements was also made on seven other montane Nepenthes species (20 pitchers each, all from separate plants), including N. burbidgeae, N. reinwardtiana, N. stenophylla, N. tentaculata (= 10) and N. villosa on Mount Kinabalu (1800–2600 m asl) and N. lowii, N. macrophylla and N. tentaculata (= 10) on Mount Trusmadi (N5.556°, E116.508°; 2000–2600 m asl). Pitchers were selected on the basis of the criterion that they were representative in size and geometry of those produced by mature plants, even if they were terrestrial types. The switch in pitcher morphology that occurs as the plants mature is well documented (Clarke, 1997; Gaume & Di Giusto, 2009), but is not uniform across different species, particularly the montane ones (Clarke, 1997). For instance, young plants of N. lowii, N. macrophylla and N. rajah produce small pitchers that trap arthropods in the normal manner for the genus and are incapable of trapping T. montana faeces because of obvious size constraints. When N. rajah plants reach maturity, they do not switch to the production of aerial pitchers and only a few of the latter have ever been seen (Kurata, 1976; Clarke, 1997). By contrast, plants of N. lowii and N. macrophylla produce few terrestrial pitchers and switch to the production of aerial ones at a relatively young age. With the exception of one terrestrial pitcher from an immature plant of N. macrophylla, plus three of N. rajah, a scarcity of terrestrial pitchers (or immature plants, in the case of N. rajah) of the faeces-trapping species at both sites prevented us from examining them quantitatively in this study.

The remaining species are ‘typical’ in that their pitchers trap arthropods only. Equal numbers of aerial and terrestrial pitchers were sampled for N. reinwardtiana, N. stenophylla, N. tentaculata and N. villosa. Nepenthes burbidgeae is a rare species that produces strongly dimorphic pitchers (Clarke, 1997), but there were very few aerial pitchers of this species available to us during this study, and so we sampled only terrestrial ones. One N. rajah pitcher that was examined in February died before we returned to the site in April. As most of our measurements of pitcher characteristics were made during the April visit, this pitcher was excluded from our analyses.

We measured the degree to which the inner surfaces of the pitcher lids were concave or convex. For instance, N. rajah lids are distinctly concave, whereas those of N. reinwardtiana are virtually flat (Fig. 1a,g). Concavity was measured by placing a ruler against the margins of the pitcher lid at the widest point on the abaxial side and positioning a second ruler perpendicular to the first to measure the distance from the inner edge of the first ruler to: (a) the deepest recess on the inner rear surface of the pitcher lid (‘lid depth’) and (b) the midrib (‘midrib depth’). For lids that were convex, the ruler was placed against the margins on the adaxial surface. Positive values refer to concavity, negative ones to convexity, with a value of zero denoting a flat lid. The extent to which the midrib protrudes from the inner rear surface of the lid (‘midrib protrusion’) was determined for each pitcher by subtracting the midrib depth from the lid depth.

Observations of visitors to pitchers

All pitchers were examined for evidence of faecal capture. We simply scored pitchers as ‘+’ or ‘−’ for faecal capture, because faecal pellets are usually washed into the pitcher fluid soon after deposition (sometimes immediately, but generally within 1 wk), where they dissolve and are uncountable. However, those pitchers that received faeces generally did so frequently. We counted the numbers of pellets deposited above the fluid in 21 pitchers that were photographed in the field (11 pitchers of N. lowii and 10 of N. rajah). To confirm that T. montana is the vertebrate responsible for defecating into the pitchers, video recordings were made on three pitchers of N. rajah (each from a separate plant) and two pitchers of N. macrophylla (both on the same plant) using a tripod-mounted digital video camera (Panasonic SDR-S7, Panasonic Corporation, Osaka, Japan). Recordings were made during daylight hours only between 10.00 and 17.00 h in May 2009. In most recordings (including Video S1, see Supporting Information), the camera was rotated 90° clockwise to ensure that the pitcher occupied as much of the field of view as possible.

Statistical analyses

Decisions about hypotheses were made against a critical value of  0.01. Measures of location are stated as means ± 1 standard or angular deviation (depending on whether the data were on a linear or circular scale) throughout. Where necessary, data were transformed to satisfy the assumptions of parametric methods. For the comparison of lid and mouth angles, data were pooled into two groups in order to distinguish between the faeces-trapping species (N. rajah, N. lowii and N. macrophylla) and the ‘typical’ ones (N. burbidgeae, N. reinwardtiana, N. stenophylla, N. tentaculata and N. villosa). Angular measurements were transformed to radians for all statistical analyses. Principal component analysis and multiple regression were used for exploratory analyses, and through the latter method we detected a strong linear relationship (r= 0.94) between fmfs, orifice depth and lid angle, where inline image. This regression provided an effective visual means of describing the relationship between these variables in the context of T. montana body size (Fig. 3). However, as a result of interspecific variation in the data, the residuals of this regression were not normally distributed. We therefore combined the two predictor variables into a single one, where inline image. We then normalized the residuals using log10 transformations of both the response (y) and predictor (x) variables. Simple linear regression using OLS and ANOVA of regression were then performed on the transformed data. All analyses of linear data were performed using Minitab v. 12.23; analyses of angular data were performed manually following Zar (1999).


Tupaia montana visits and defecates into pitchers of N. lowii, N. macrophylla and N. rajah

Tupaia montana faeces were observed in pitchers of N. rajah (21 of 36 pitchers), N. macrophylla (17 of 20) and N. lowii (18 of 20), but not in any pitchers of N. burbidgeae, N. reinwardtiana, N. stenophylla, N. tentaculata or N. villosa. The mean number of faecal pellets present above the fluid on the inner surfaces of N. lowii and N. rajah pitchers was 1.95 ± 0.23 (range, 1–4; = 21). Assuming that this represents the mean number of faecal pellets deposited per week, during the course of the operative life of a pitcher (at least 3 months), we estimate that the total number of faecal pellets deposited in any given pitcher would range from approximately 10 to 30. Although T. montana faecal pellets contain significantly less nitrogen by weight than do ants (4.9 ± 0.4% vs 9.8 ± 0.1%; < 0.0001; = 11.36; df = 18), a steady input of faeces has nonetheless been shown to contribute between 57 and 100% of foliar nitrogen in N. lowii (Clarke et al., 2009).

Video recordings of three N. rajah pitchers (4 h on one pitcher, 2 h on another and 1 h on a third) detected seven visits by T. montana (Fig. 1j), one of which resulted in the deposition of faeces into the pitcher (Fig. 1k). One hour of recording of two N. macrophylla pitchers (both on the same plant) detected a single visit (to both pitchers) by T. montana. The mean duration of a pitcher visit was 24 ± 7 s (range, 6–61 s; = 9; Video S1, see Supporting Information). No other vertebrates were seen to visit pitchers of these species (either in our video recordings or unquantified visual observations), but observations were not made at night and anecdotal records of vertebrate capture by N. rajah demonstrate that reptiles, amphibians and small mammals do visit its pitchers. Despite this, we have never observed evidence of faecal inputs by vertebrates other than T. montana. On this basis, and from the observations of Clarke (1997) and Clarke et al. (2009), we concluded that, although a variety of vertebrates may visit pitchers of the faeces-trapping species, only T. montana defecates into them habitually, and that this interaction is the principal source of vertebrate-derived N for N. lowii, N. rajah and N. macrophylla. Clearly, other vertebrates [such as mountain blackeyes (Chlorocharis emiliae), which are known to feed on the lid gland exudates of N. lowii (Clarke et al., 2009)] may occasionally defecate into pitchers, and such inputs could be of short-term benefit to an individual plant; however, from our observations of hundreds of pitchers of N. lowii, N. macrophylla and N. rajah, we believe that such inputs are rare and unlikely to be of significant long-term benefit to populations of these Nepenthes species.

The glands on the pitcher lids of N. lowii, N. macrophylla and N. rajah produce copious exudates. Those of N. lowii were described by Clarke et al. (2009) as ‘white and buttery’ in texture. The secretions of the lid glands of N. macrophylla are less copious and of thinner consistency and, when not consumed by T. montana (and perhaps other vertebrates), they accumulate as a thin, white crust at the base of the lid. In N. rajah, the lid gland secretions are clear and watery in texture and do not seem to accumulate on the pitcher lids. We do not know whether this is a result of rapid consumption by T. montana or run-off as a consequence of their low viscosity. The composition and nutritional value of these exudates to T. montana have yet to be investigated. In addition to an insectivorous diet, T. montana is highly frugivorous, specializing in crushing fruits to extract the sugary juices. It may therefore be considered potentially preadapted to feeding on the sugar-rich nectars of Nepenthes pitchers (Emmons, 1991).

Composition of biomass in pitchers of the eight Nepenthes species examined

Pitchers of N. burbidgeae, N. reinwardtiana, N. stenophylla, N. tentaculata and N. villosa were found to contain only arthropod prey and, very occasionally, small fragments of leaf litter. No evidence of vertebrate faecal capture was detected in any of the pitchers of these species (this finding is also supported by informal observations of the contents of hundreds of pitchers of these species by us since 1987). Aerial pitchers of N. lowii and terrestrial pitchers of N. rajah trap considerable amounts of plant matter, including leaf litter, senescent orchid flowers and seeds of sedges. Although these inputs may contribute to the plants’ nitrogen budget, their value to the plants has yet to be investigated. In N. macrophylla pitchers, the lid overhangs the orifice, preventing the entry of leaf litter, and very little plant matter was observed within them. We also observed abundant invertebrate prey in pitchers of both N. rajah and N. macrophylla; therefore, unlike the aerial pitchers of N. lowii, those of these species have clearly retained the ability to function as arthropod traps.

Interspecific comparisons of pitcher characteristics and geometry

Analyses of interspecific variation in pitcher characteristics (Table 1) revealed that, as a group, the faeces-trapping species have few unique features, differing significantly from their typical congeners on the basis of fmfs, lid angle, orifice depth and lid length, area and concavity. Principal component analysis yielded five components that had a cumulative contribution rate of 97.3% (Table 2). Subsequent analysis was conducted on the first two components, whose combined contribution was 81.8%. The first component appeared to describe variation in pitcher size, whereas the second component contrasted fmfs, lid angle and mouth angle with the other pitcher characteristics (Fig. 2), with pitchers on mature plants of the faeces-trapping species being distinguished from those of the others. By contrast, small pitchers of N. rajah and N. macrophylla that were produced by immature plants were grouped with the ‘typical’, arthropod-trapping species in Fig. 2. We feel that this provides tentative support for the hypothesis that small pitchers produced by immature plants of N. lowii, N. macrophylla and N. rajah cannot trap faeces and function as arthropod traps in the typical manner for the genus (Clarke et al., 2009).

Table 1.   Comparisons of pitcher dimensions and characteristics
Pitcher characteristicNepenthes species
Tupaia montana faeces-trapping species‘Typical’ speciesANOVA
  1. For all analyses of variance, < 0.001. Superscripts (a,b etc.) denote groups of species for which there is no significant difference for the relevant pitcher characteristic, based on Tukey’s pairwise comparisons of means. Means that lack superscripts are significantly different from all other means for that dimension/characteristic.

  2. fmfs, front of the mouth to the food source.

Orifice depth (mm)143.3 ± 11.94ab129.5 ± 17.9b154.1 ± 33.0a61.3 ± 13.1c50.8 ± 6.6cd51.4 ± 8.5cd35.5 ± 5.6d50.1 ± 7.2cd180.31
Capacity (ml)235.6 ± 61.7b551.9 ± 168.0a722.8 ± 414.9a197.7 ± 107.4bc95.2 ± 22.4bc129.4 ± 50.4bc27.1 ± 12.0c134.6 ± 49.1bc39.50
fmfs (mm)176.9 ± 12.4a196.3 ± 23.0a177.3 ± 40.8a75.4 ± 18.1b57.8 ± 8.3bc64.8 ± 11.2b42.5 ± 9.4c74.6 ± 17.3b171.82
Lid angle (deg)107.1 ± 10.9103.5 ± 13.783.3 ± 14.456.6 ± 13.956.6 ± 13.757.9 ± 15.178.7 ± 16.250.7 ± 14.8
Pooled lid angle (deg)95.19 ± 17.1960.15 ± 17.76
Mouth angle (deg)2.9 ± 7.737.2 ± 6.86.8 ± 13.138.2 ± 17.931.6 ± 6.857.9 ± 15.151.9 ± 10.347.1 ± 8.9
Pooled mouth angle (deg)16.3 ± 17.245.3 ± 15.5
Lid concavity (mm)24.8 ± 3.6a15.1 ± 7.926.7 ± 13.7a-0.7 ± 4.9b0.2 ± 2.8b3.0 ± 3.0b0.2 ± 0.9b-0.6 ± 2.7b65.62
Lid midrib protrusion (mm)8.2 ± 2.5a,b,c5.7 ± 2.6b,c10.6 ± 4.6a5.0 ± 4.3b,c2.5 ± 2.0b,d0.8 ± 3.1d1.5 ± 1.2d6.0 ± 5.3c23.05
Lid length (mm)103.8 ± 20.6a121.3 ± 14.9a163.5 ± 41.968.7 ± 13.7b49.0 ± 6.0bc58.1 ± 8.8b33.6 ± 4.8c57.4 ± 8.4b108.09
Lid width (mm)57.4 ± 5.8bc112.2 ± 12.8a131.0 ± 46.5a66.7 ± 12.6b43.6 ± 5.6c62.5 ± 9.7bc20.4 ± 3.560.3 ± 7.9bc65.46
Lid area (cm2)70.0 ± 24.1a72.8 ± 15.9a129.5 ± 51.436.7 ± 18.2b14.2 ± 7.8b,c20.9 ± 7.7b,c7.0 ± 2.2c28.4 ± 7.8b,c69.28
Height (mm)163.4 ± 17.0b270.6 ± 27.1a246.9 ± 53.6a155.8 ± 26.3b203.6 ± 23.4c215.1 ± 28.8c103.6 ± 18.7143.0 ± 17.8b68.73
Width (mm)84.4 ± 17.4a75.7 ± 10.4a97.5 ± 19.165.9 ± 14.0a36.4 ± 16.6bc44.7 ± 11.4b25.9 ± 5.3c62.0 ± 8.8a72.76
Depth (mm)69.8 ± 12.5a71.5 ± 10.6a97.7 ± 26.560.3 ± 13.1ab40.0 ± 4.8de47.3 ± 9.2cd27.7 ± 4.9e55.5 ± 8.2bc57.41
Table 2.   Eigenvectors and contributions of the principal components and scores for variables included in the principal component analysis (PCA)
  PC1 PC2 PC3 PC4 PC5
  1. fmfs, front of the mouth to the food source.

Proportional contribution0.7150.1020.0690.0650.022
Cumulative contribution0.7150.8180.8870.9520.973
Mouth length−0.4080.0740.035−0.0790.018
Lid angle−0.2560.828−0.0670.265−0.181
Mouth angle0.3090.150−0.874−0.0380.279
Lid area−0.390−0.215−0.167−0.193−0.204
Lid concavity−0.3840.0350.100−0.1750.862
Midrib protrusion−0.290−0.383−0.2000.8470.066
Figure 2.

 Scores for all pitchers for principal components (PCs) 1 and 2. Key to symbols: open symbols: triangle, apex up, Nepenthes burbidgeae; squares, N. lowii; diamonds, N. macrophylla; large circles, N. rajah; +, N. reinwardtiana; ×, N. stenophylla; small circles, N. tentaculata; triangles, apex down, N. villosa; closed symbols for N. lowii, N. rajah and N. macrophylla denote pitchers that trapped faeces. Arrows denote pitchers of N. rajah and N. macrophylla sampled from immature plants.

For the remaining pitcher characteristics, the principal source of overlap between the two groups of species involves N. lowii and N. burbidgeae. The pitchers of these species are equivalent in size and lid width, showing that, although the production of pitchers large enough to accommodate an animal the size of T. montana is an obvious requirement of the faeces-trapping syndrome, other modifications to trap geometry are also necessary: despite their large capacities, N. burbidgeae pitchers do not receive faecal inputs because of their small fmfs, orifice size and lid angle (Table 1). This is apparent in Fig. 2, where N. burbidgeae is grouped with the arthropod-trapping species, despite the lack of a significant difference in pitcher capacity between this species and N. lowii (Table 1).

The relationship between pitcher geometry and tree shrew body size

The mean lid angle for N. lowii, N. macrophylla and N. rajah combined is significantly greater than the combined mean lid angle of the other five species (Table 1; Watson–Williams test; F1,174 = 406.53; < 0.001). The large lid angle and pitcher orifice of N. lowii pitchers play an important role in the ‘capture’ of T. montana faeces by inducing the animals to sit astride the pitcher whilst feeding on the lid gland exudates (Clarke et al., 2009). In N. lowii, N. macrophylla and N. rajah, only pitchers with orifice depths of more than 100 mm received faecal inputs, suggesting that pitchers must exceed a size threshold before T. montana is able to defecate into them. The hb length of T. montana is 156–227 mm (Payne et al., 1985), which frequently exceeds the orifice depth of pitchers that receive faeces, indicating that some pitchers are too small to accommodate tree shrews. However, the lid gland exudates are positioned behind the orifice in the faeces-trapping species because of their large lid angles. The distance from the food source to the front of the orifice (‘fmfs’, Fig. 1) is 156–241 mm in all pitchers that receive faecal inputs (except one, see below), corresponding precisely to the hb length of T. montana. The regression of fmfs against mouth length and lid angle [= 1.18− 0.212, where = log10 fmfs and = log10(mouth length + 2 × lid angle)] is significant for the pooled data for all eight Nepenthes species (F1,174 = 2073.31; < 0.001; r2 = 0.923; Fig. 3). Only pitchers with orifice depth/lid angle combinations that exceed the minimum hb length of T. montana receive faeces and, among the species we examined, these are produced only by N. lowii, N. macrophylla and N. rajah.

Figure 3.

 Regression of fmfs (front of the mouth to the food source) vs orifice depth and lid angle for the eight Nepenthes species studied: = 1.18x − 0.212; where = Log10fmfs and = Log10(mouth length + 2 × lid angle). Key to symbols: triangles, apex up, N. burbidgeae; squares, N. lowii; diamonds, N. macrophylla; large circles, N. rajah; +, N. reinwardtiana; ×, N. stenophylla; small circles, N. tentaculata; triangles, apex down, N. villosa. Closed symbols for N. lowii, N. rajah and N. macrophylla denote pitchers that trapped faeces. Arrow denotes N. rajah pitcher with fmfs = 143 mm, which was visited by Tupaia montana, but faeces was deposited on the outside of the pitcher and on the ground beside it.

We found that T. montana feeds on the lid gland secretions of small N. rajah pitchers on immature plants, but, apart from one exception, no pitchers with fmfs < hb length of T. montana received faecal inputs. The exception was an N. rajah pitcher in which fmfs = 143 mm, but, in this case, virtually all of the faeces were deposited on the ground adjacent to the pitcher, suggesting that the orifice depth/lid angle combination was insufficient for T. montana to climb onto the pitcher and defecate into it. These observations support our assertion that only pitchers with fmfs large enough to accommodate T. montana receive faecal inputs, whereas the deposition of faeces around this pitcher supports the finding of Kawamichi & Kawamichi (1979) that tree shrews mark the location of valuable food resources with faeces. Despite the obvious strength of the relationship between fmfs and T. montana body size, we do not assert that only pitchers with fmfs > 156 mm can or will receive faecal inputs. Exceptions to this pattern are likely to be observed when more detailed studies of the T. montana–Nepenthes association are performed. However, we expect such exceptions to be uncommon, given that the combination of pitcher size and geometry appears to be fundamental to the positioning of the tree shrew’s hindquarters over the pitcher orifice whilst it feeds.

There is no significant difference between the combined mean orifice angles of the faeces-trapping species and the typical ones (Table 1; Watson–Williams test; F1,174 =0.057; > 0.5). Although a horizontal orifice may have some role in trap function in N. rajah and N. lowii, this feature is absent in N. macrophylla (Table 1), and so faecal capture is not contingent on the orientation of the pitcher orifice.

The other characteristics that are unique to the faeces-trapping species are the concave structure and elongation of the pitcher lids (Table 1). Concavity results in the nectar glands being deeply recessed within the structure, making access difficult for T. montana unless it is directly in front of the inner surface of the lid (i.e. sitting astride the pitcher). Elongation of the lid accentuates the concavity effect by inducing T. montana to move forward towards the rear of the orifice whilst feeding at the nectar glands near the lid apex. Clarke et al. (2009) have demonstrated that, once T. montana finishes feeding at an N. lowii pitcher, it has effectively ‘climbed’ part-way up the pitcher lid, moving its hindquarters over the centre or rear of the orifice in the process. We observed that tree shrews initially approach N. rajah pitchers from the side, in which case they must reach around the incurved outer margin of the lid to access the closest nectar glands. However, in this position, they can only reach the glands on the half of the lid that is closest to them. The only way in which the animals can reach all parts of the lid during a single visit is to position themselves directly in front of the lid, so that they can easily reach the glands on either side of the midrib (Video S1, see Supporting Information).


The faecal capture syndrome in Nepenthes

Our findings demonstrate that pitcher size and geometry in N. lowii, N. macrophylla and N. rajah are a function of T. montana body size. The unique combination of fmfs > 156 mm (itself a function of orifice depth and lid angle) and lid concavity/elongation possessed by these species facilitates supplementary nitrogen acquisition through the capture of T. montana faeces. fmfs is the distinguishing characteristic of the faeces-trapping syndrome, as the distance from the food source to the front of the pitcher orifice matches the hb length of T. montana.

The association with T. montana represents an extraordinary example of co-evolution and specialization comparable with many plant–pollinator systems (e.g. Lunau, 2004), but the modifications to trap geometry that facilitate the syndrome are relatively subtle and involve few, if any, unique pitcher characteristics. For example, several other species produce copious amounts of nectar (e.g. N. bicalcarata and N. jacquelineae), N. northiana, N. burbidgeae and N. truncata produce large pitchers, and N. dubia and N. ampullaria have lid angles in excess of 100° (Clarke, 2001). The fact that the switch in primary pitcher function from carnivory to faecal capture (with or without the retention of carnivorous traits) requires minimal morphological or physiological modification strongly suggests evolutionary preadaptation, analogous to the role of preadaptation in pollinator shift in entomophilous flowers (e.g. Schiestl & Cozzolino, 2008).

Although the faecal capture syndrome appears to comprise the pitcher characteristics described above (and fmfs in particular), there is noticeable divergence among the pitcher characteristics of N. lowii, N. macrophylla and N. rajah, which appears to reflect differing degrees of specialization towards (or reliance on) faecal capture. Nepenthes lowii appears to be highly specialized, with aerial pitchers that have mostly lost the ability to trap arthropods (Clarke et al., 2009). Nepenthes rajah and N. macrophylla pitchers have retained the ability to trap arthropods, but those of N. rajah are enormous, have horizontal orifices, very broad peristomes with short teeth and rest on the ground, whereas those of N. macrophylla have steep mouth angles, narrower peristomes with pronounced teeth and are produced on climbing stems in the upper montane forest canopy. The arthropod prey spectra of these species have yet to be examined in detail, and it is apparent that we still know little about the precise functioning of their pitchers with regard to prey capture.

It appears that anecdotal records of rodent capture in N. rajah relate to rare, chance events and are not a fundamental component of this species’ nitrogen acquisition strategy. Nevertheless, we observed that T. montana had some difficulty straddling the pitcher orifice on the largest pitchers of this species (Video S1, see Supporting Information). We doubt that tree shrews are commonly (if ever) trapped by N. rajah pitchers, but it is possible that a partially digested tree shrew inside a pitcher might resemble a rodent to a casual observer, and as no mammalian prey of N. rajah has been identified to genus, it could be that some of the ‘rats’ observed in N. rajah pitchers were in fact clumsy tree shrews.

In addition to N. lowii, N. macrophylla and N. rajah, one other Bornean montane species, Nepenthes ephippiata, produces pitchers that possess the characteristics associated with faecal capture. Like N. lowii, N. ephippiata pitchers have large orifices and lid angles, reduced peristomes, and elongated, concave pitcher lids whose glands produce copious amounts of white, buttery exudates. These species are thought to be closely related (Clarke, 1997; Jebb & Cheek, 1997) and, although detailed, field-based examinations of N. ephippiata have yet to be made, we predict that it will be found to share a similar association with T. montana.

The role of faecal capture in carnivorous plants

These Nepenthes species are not the only carnivorous plants that derive nutritional benefit from animal faeces: two species of Roridula from southern Africa obtain supplementary nitrogen from the faeces of hemipteran bugs which feed on insects that are trapped by sticky secretions produced by stalked glands on these species’ leaves (Midgley & Stock, 1998; Anderson, 2005; Plachno et al., 2009). Faeces are a valuable source of nitrogen, and it is perhaps surprising that so few carnivorous plants appear to benefit from ‘trapping’ them. Furthermore, N. lowii, N. macrophylla and N. rajah may receive more than faecal inputs from tree shrews. Clarke et al. (2009) noted that T. montana scent mark N. lowii pitchers and, although we have not yet observed tree shrews urinating into pitchers, the likelihood of this happening is high, and the uric acid and creatinine present in urine would be immediately available to the plant. Although faecal capture by Nepenthes appears to be confined to a few montane species from Borneo, the potential for a wider range of species to benefit from vertebrate urine is high. Almost any Nepenthes species that produces large pitchers (e.g. N. attenboroughii, N. northiana, N. rafflesiana, N. truncata) could receive such inputs from a variety of mammals. For example, bats have been observed roosting during the daytime in pitchers of N. rafflesiana and N. bicalcarata (J.A. Moran & C. Clarke, pers. obs.); the potential for urine inputs into the pitchers under such circumstances is also high.

Nepenthes as a candidate model for adaptive radiation

The Bornean Nepenthes flora comprises c. 36 species (Phillipps et al., 2009), 20 of which are montane. Given that three of the montane species have evolved to capture T. montana faeces as a source of supplementary nitrogen (perhaps four, if we include N. ephippiata), it is apparent that alternative nitrogen sequestration strategies (apart from unspecialized arthropod capture) are relatively common, occurring in up to 20% of Bornean montane species. In fact, Nepenthes demonstrates remarkable variety in pitcher morphology, and the relatively small amount of ecological research to date indicates that this broad array of morphologies is mirrored by the variety of strategies employed to sequester nitrogen. For example, in addition to producing scent, N. rafflesiana pitchers exhibit colour contrast signals that correspond to the visual sensitivity maxima of many of its anthophilous insect prey taxa, allowing the exploitation of prey populations inaccessible to sympatric congeners (Moran, 1996; Moran et al., 1999), N. albomarginata pitchers produce a lichen-mimicking tissue to target termites (Moran et al., 2001; Merbach et al., 2002) and N. ampullaria appears to be evolving away from a strictly carnivorous mode of nutrition, with c. 35% of its nitrogen derived from leaf litter (Cresswell, 1998; Moran et al., 2003).

The current study therefore adds to a growing body of evidence suggesting that the genus Nepenthes may be a candidate model for adaptive radiation with regard to nitrogen sequestration strategies. Adaptive radiation may involve reproductive strategies (Aquilegia spp.; Whittall & Hodges, 2007), adaptations to varying degrees of water deficit (the Hawaiian silversword alliance; Baldwin, 1997) or the use of a range of habitat types (Caribbean Anolis spp.; Butler et al., 2007). Criteria for an adaptive radiation include rapid speciation, recent common ancestry, correlation between phenotype and environment, and trait utility (Schluter, 2000). Although the origin of the monotypic Nepenthaceae is currently unresolved, DNA evidence suggests recent and rapid speciation on the Sunda Shelf (Meimberg & Heubl, 2006). The Greater Sunda Islands west of Wallace’s Line were linked by land bridges during the Pleistocene. Despite a short period of isolation since the end of the last glacial period, these islands are notable for the high endemism of Nepenthes species (e.g. 77% and 76% for Borneo and Sumatra, respectively), and only a handful of species have a geographical distribution that encompasses more than one island. It is arguable that phenotype is correlated with environment: ‘typical’ lowland species commonly possess relatively unadorned pitchers, whereas montane species produce pitchers with a range of extreme morphologies (current study), at least partly in response to the pressure of low invertebrate availability at altitude (Collins, 1980; Clarke et al., 2009). Finally, with regard to trait utility, the manipulation of pitcher characteristics has demonstrated the value of features including colour, scent and fluid properties (Moran, 1996; Moran et al., 1999; Gaume et al., 2002; Bohn & Federle, 2004; Gaume & Forterre, 2007; Bauer et al., 2008; DiGiusto et al., 2008), as well as the nutritional advantages of pitcher production itself (Schulze et al., 1997; Moran & Moran, 1998; Pavlovičet al., 2009).


We thank Dr Maklarin Lakim and Sabah Parks for assistance and permission to conduct the research, Ansou Gunsalam, Sukaibin Sumail and Sonja Raub for assistance in the field and three anonymous referees for their helpful comments on the manuscript. This research was conducted in accordance with Sabah Parks research permit #TS/PTD/5/4 Jld. 34(55) and was funded by the School of Science, Monash University Sunway Campus, Malaysia.