INTRODUCTION
Natural microcosms offer opportunities for an array of ecological studies, such as the role of predation in biotic interactions, shaping food webs and altering community structure and ecosystem functions. Phytotelmata (plant-held waters; Kitching Reference KITCHING2001) are popular study models of arthropod–plant mutualism food webs and community structure (Armbruster et al. Reference ARMBRUSTER, HUTCHINSON and COTGREAVE2002, Clarke et al. Reference CLARKE, BAUER, LEE, TUEN, REMBOLD and MORAN2009, Kitching Reference KITCHING2001, Maguire et al. Reference MAGUIRE, BELK and WELLS1968, Moon et al. Reference MOON, ROSSI, DEPAZ, MCKELVEY, ELIAS, WHEELER and MOON2010, Peterson et al. Reference PETERSON, DAY, WOLFE, ELLISON, KOLTER and PRINGLE2008), focusing on the apex predators that reside within the same aquatic habitat as their prey. However, there has been recent interest in the cross-habitat (i.e. terrestrial to aquatic) predatory effects of terrestrial inhabitants (Romero & Srivastava Reference ROMERO and SRIVASTAVA2010).
Phytotelm communities have been the focus of numerous studies of community dynamics for the past three decades (Kitching Reference KITCHING2000, Maguire et al. Reference MAGUIRE, BELK and WELLS1968, Mogi & Chan Reference MOGI and CHAN1997, Mogi & Yong Reference MOGI and YONG1992, Mouquet et al. Reference MOUQUET, DAUFRESNE, GRAY and MILLER2008, Naeem Reference NAEEM1988, Seifert & Seifert Reference SEIFERT and SEIFERT1976). Amongst these, pitchers are unique as these highly modified leaf structures, holding a digestive fluid, lure, trap and kill any organism that falls into these pitfalls. These vessels possess several key traits that facilitate their heterotrophism: chemical and colour cues that attract prey (Bennett & Ellison Reference BENNETT and ELLISON2009, Jurgens et al. Reference JURGENS, EL-SAYED and SUCKLING2009, Schaefer & Ruxton Reference SCHAEFER and RUXTON2008), slippery inner wall surfaces (Gorb et al. Reference GORB, KASTNER, PERESSADKO, ARZT, GAUME, ROWE and GORB2004, Scholz et al. Reference SCHOLZ, BÜCKINS, DOLGE, ERLINGHAGEN, WETH, HISCHEN, MAYER, HOFFMANN, RIEDERER, RIEDEL and BAUMGARTNER2010), and a highly viscous (Di Giusto et al. Reference DI GIUSTO, GROSBOIS, FARGEAS, MARSHALL and GAUME2008), acidic and hypoxic fluid (due to decomposing insects). Collectively, these factors constitute a hostile environment and pose a challenge for any organisms. Rather, numerous organisms have adapted to living, many exclusively, within pitchers as nepenthebionts (i.e. obligate Nepenthes pitcher dwellers), holding positions as apex predators, mesopredators and scavengers that regulate top-down and bottom-up forces (Kneitel & Miller Reference KNEITEL and MILLER2002). In comparison, little is known about the ecological importance of terrestrial phytotelm organisms (Greeney Reference GREENEY2001).
Although many terrestrial phytotelm inhabitants are known (for a review see Greeney Reference GREENEY2001), those that forage across terrestrial–aquatic environments are seldom recorded. Clarke & Kitching (Reference CLARKE and KITCHING1995) probably provided the first empirical evidence of cross-habitat predation in the golden ant Camponotus schmitzi and its host plant Nepenthes bicalcarata, where the ant significantly influenced the abundance of filter-feeding mosquito larvae. The nepenthebiont crab spiders (Araneae: Thomisidae) are the only other group of organisms documented to forage into Nepenthes pitcher fluid (Barthlott et al. Reference BARTHLOTT, POREMBSKI, SEINE and THEISEN2007, Clarke Reference CLARKE1997, Reference CLARKE2001; Phillipps et al. Reference PHILLIPPS, LAMB and LEE2008, Pollard Reference POLLARD2005; Figure 1) for live dipteran larvae (Clarke Reference CLARKE1998, Moran Reference MORAN1993). These larvae are Nepenthes obligates and assume key roles within the food web (Figure 1); however, nothing is known about the aquatic foraging ability of the terrestrial crab spiders and their potential to alter dipteran larval abundance.
Figure 1. Food web of a Nepenthes gracilis pitcher (modified from Clarke & Kitching Reference CLARKE and KITCHING1993, Phillipps et al. Reference PHILLIPPS, LAMB and LEE2008, Tan Reference TAN1997) organized in trophic levels. Arrows in bold indicate predator–prey interactions involved in this study.
Here we investigate the nepenthebiont crab spiders’ foraging ability to alter the aquatic larval abundance of the tropical pitcher plant Nepenthes gracilis. We hypothesize that the crab spiders’ presence significantly affects the abundance of nepenthebiont dipteran larvae species in N. gracilis. As pitchers provide natural microcosms amenable to both laboratory and field experiments (Srivastava et al. Reference SRIVASTAVA, KOLASA, BENGTSSON, GONZALEZ, LAWLER, MILLER, MUNGUIA, ROMANUK, SCHNEIDER and TRZCINSKI2004), we conducted laboratory-based studies to investigate the aquatic foraging behaviour of the crab spiders by introducing one individual into one pitcher containing any of the three dipteran larvae species; a reduction in larval abundance, coupled with weight increase in the spider will verify this ability. Field experiments then determined the spiders’ influence on the abundance of these aquatic dipteran larvae. A significant decrease in larval abundance in the presence of the crab spiders will suggest the ecological importance of these predators, with probable implications for the Nepenthes food web, and highlight the importance of the ecological roles of terrestrial animals with aquatic predatory traits within phytotelmata.
METHODS
Study species
Nepenthes gracilis Korth. (Figure 2a) is widespread in Singapore, its pitchers home to a diverse macrofauna dominated by insects (Kitching Reference KITCHING2001). Larvae of several dipteran species (mosquitoes Tripteriodes spp. and Culex spp., scuttle fly Endonepenthia schuitemakeri (Schmitz, 1932) and gall midge Lestodiplosis spp. (Figure 2d–g)) occupy the various trophic zones within these pitchers (Figure 1). Residing within pitchers and above the fluid are two species of thomsids; the red crab spider Misumenops nepenthicola (Pocock, 1898) (Figure 2b) and the yellow crab spider Thomisus nepenthiphilus (Fage, 1930) (Figure 2c). Only the foraging of M. nepenthicola, but not Thomisus nepenthiphilus, has been recorded within N. gracilis (Kitching Reference KITCHING2000).
Figure 2. Organisms involved in this study: a freshly opened Nepenthes gracilis pitcher (a), red crab spider Misumenops nepenthicola (b), yellow crab spider Thomisus nepenthiphilus (c), mosquito larvae Tripteriodes spp. (d) and Culex spp. (e), scuttle fly larva Endonepenthia schuitemakeri (f), and gall midge larva Lestodiplosis spp. (g). Scale bar represents 1 cm (a, b, c) and 1 mm (d, e, f, g).
Laboratory experiments
We investigated the foraging behaviour of Misumenops nepenthicola and Thomisus nepenthiphilus on dipteran larvae that dwell within Nepenthes pitchers using 10 adult females (M. nepenthicola; body length (mean ± SD): 5.4 ± 0.5 mm, T. nepenthiphilus; body length: 7.3 ± 0.5 mm) and 200 mosquito larvae (Tripteriodes spp.; body length: approximately 4 mm), all collected from Kent Ridge Park, a secondary forest in Singapore. Pitcher fluid (collected from 20 pitchers) was filtered to remove detritus and live organisms. We also purchased N. gracilis from a local nursery; these were maintained in clear plastic tanks (39 × 24.5 × 30 cm). Twenty fresh pitchers (mean ± SD: height: 7.30 ± 1.65 cm; width: 1.50 ± 0.28 cm) were selected, each rinsed thoroughly (using distilled water from a squirt bottle) prior to the experiment. Pitcher contents were discarded, and the fluid replaced by those collected in the field (2 ml of fluid per pitcher) prior to experiment. Two circular plastic containers (4.3 cm diameter × 11.2 cm height) ensured containment of an individual spider in each pitcher. Experiment was limited to 5 d because many N. gracilis had withered and most mosquito larvae had moulted into pupae and emerged as adults in earlier trials that lasted 1 wk. We accounted for the number of larvae (10 Tripteroides spp.) on days 1 and 3 to check for cannibalism. The mass of each crab spider (to the nearest 0.00001 g) was recorded after collection (i.e. day 1) from Kent Ridge Park, prior to their introduction into the pitchers and maintained on sugar solution ad libitum via dental roll soaked in diluted sugar solution till day 3. On day 5, we took note of larvae carcases (i.e. dead larvae not eaten by the spider) to ensure that all larvae were accounted for.
We repeated the above procedure but used two dipteran larvae species in separate trials: the carrion-feeding scuttle fly larva E. schuitemakeri (length: 0.4 cm) and the predatory gall midge larva Lestodiplosis spp. (length: 0.2 cm). Only five individuals of each species were used due to their lower abundance observed in the field (pers. obs.). We excluded T. nepenthiphilus from this and further experiments because it did not forage on aquatic mosquito larvae. All experimental animals, plants and units were maintained in a laboratory under controlled environmental conditions (relative humidity 80–85%; temperature 25 °C ± 1 °C; light regime 12:12 h; lights on at 0800 h).
Field experiments
We investigated the relationship between M. nepenthicola and the abundance of phytotelm dipteran larvae in natural occurring populations of N. gracilis at Kent Ridge Park, Singapore. As M. nepenthicola abandoned shorter pitchers (pers. obs.), we only used pitchers more than 6 cm high and unopened at time of selection. From two separate experimental periods (22 October 2009–3 December 2009; 24 December 2009–4 February 2010), we tagged a total of 65 unopened N. gracilis. These were surveyed twice a week from the time they opened to the time they withered or until the end of the field experiment, whichever came first. We also introduced a 2-wk colonization period to allow establishment of secondary consumers and scavengers, as freshly opened pitchers harboured neither aquatic dipteran larvae nor crab spiders (pers. obs.).
Field experiments commenced 2 wk after the pitchers had opened (Figure 3); a small number of pitchers that harboured spiders were excluded from our data. Contents of qualified pitchers were emptied into individual collection vials and pitchers rinsed with water via a squirt bottle into a second vial to remove residual contents. Both vials were then transported to a laboratory and the live aquatic dipteran larvae identified (based on morphospecies) and counted under a stereomicroscope. We returned all dipteran larvae and contents to their respective pitchers on the same day, and introduced one female adult M. nepenthicola (0.50 ± 0.10 cm) into each pitcher for 1 wk. We repeated the above procedure (i.e. counting of larvae and returning contents to the respective pitcher) two more times; with the resident crab spider first removed and maintained in the laboratory (sugar solution provided ad libitum) and finally reintroduced to the same pitcher (Figure 3). We used 7 d per treatment (i.e. spider absent/present) because laboratory experiments revealed that many M. nepenthicola had consumed most of their prey within 2 d. Introduction, removal and subsequent re-introduction of M. nepenthicola into pitchers over 4 wk enabled the realistic testing of this spider as a predator along with other concurrent activities (e.g. egg-laying by dipteran adults, newly hatched or moulted aquatic dipteran larvae, newly emerged dipteran adults from pupae, other predation and parasitic activities, and changes to pitcher detritus) that can affect larval abundance.
Figure 3. Summary of field-based experiment depicting the periods of colonization, presence and absence of crab spiders, and sequence of data collection and pitcher manipulation (i.e. introduction and removal of spider).
We inspected all pitchers for spiders every 3–4 d to ensure its status (i.e. spider present or absent). If we found a spider in a pitcher designated as ‘spider absent’, the pitcher fluid and its contents were first collected before the pitcher was filled with distilled water to the brim so that removal, only if the spider surfaced at the mouth of the pitcher upon depletion of its air supply, was easy. This approach is necessary as, upon disturbance, M. nepenthicola never fails to drop into the fluid and stays at the bottom of the pitcher until its air supply (i.e. air bubble entrapped in a small pit on the abdomen ventral side) is depleted after several minutes. Pitcher contents were returned after the spider was removed. We also reintroduced M. nepenthicola into pitchers that were supposed to hold a spider but were otherwise absent; a spider usually climbed on and into a pitcher within a few minutes. We assumed that these newly introduced adult female crab spiders, collected from Kent Ridge Park on the same day, have similar satiation levels to other conspecifics in experimental pitchers. We attached a pair of Velcro® fasteners (lightly smeared with Singer® Oil twice a week) on each pitcher's leaf blade to dissuade experimental spiders from leaving their designated pitchers and non-experimental crab spiders from entering experimental pitchers. We also excluded, from data analyses, a small number of pitchers with withered lids and/or contained egg sacs.
Statistical analysis
We compared the larval abundance, in the absence and presence of spiders, of the same pitcher using a related sampling approach, the Friedman test (PASW Statistics, version 18; significance level at 0.05) and a non-parametric pairwise comparison (Siegel & Castellan Reference SIEGEL and CASTELLAN1988) for multiple group comparisons of related samples. We only considered pairwise comparisons when the corresponding Friedman test was significant (i.e. absolute difference value exceeds the corresponding critical difference, denoting a significant difference for that respective pair; Siegel & Castellan Reference SIEGEL and CASTELLAN1988). We only report relevant pairwise comparisons of interest to our hypotheses.
We adopted a related-sampling approach owing to the limited abundance and occurrences of M. nepenthicola within pitchers from only one site in Singapore (i.e. Kent Ridge Park). Each pitcher was used as its own control to minimize any potential confounding variables. We sought to minimize the possibility of temporal effects by (1) repeating the procedure of spider introduction one more time for each pitcher, and (2) carrying out the entire experimental procedure on two separate occasions.
RESULTS
Laboratory experiments
The weight of M. nepenthicola increased significantly on day 5 (χ2 = 15.8, df = 2, P < 0.001; Figure 4a) after removal from experimental pitchers. The mean larval abundance of Tripteroides spp. reduced significantly after M. nepenthicola was introduced on the third day (χ2 = 20.0, df = 2, P < 0.001; Figure 4b). There was no change in larval abundance of Tripteroides spp. when Thomisus nepenthiphilus was introduced (Figure 4b) and a significant decrease in predator weight was observed (χ2 = 20.0, df = 2, P < 0.001; Figure 4a). Hence, T. nepenthiphilus was excluded from further field manipulative experiments as it did not feed on mosquito larvae.
Figure 4. Effects of spider's absence and presence on dipteran larval abundance and spider's corresponding weight. Effects of absence (days 1 and 3) and presence (day 5) of the red crab spider Misumenops nepenthicola (filled circles) and the yellow crab spider Thomisus nepenthiphilus (unfilled circles) and their corresponding weight (median) (a) relating to the abundance of the mosquito larvae Tripteroides spp. (b). Effects of absence (days 1 and 3) and presence (day 5) of M. nepenthicola and its corresponding weight (median) (c, e) relating to the abundance of the larvae of the scuttle fly Endonepenthia schuitemakeri (squares) (d) and gall midge Lestodiplosis spp. (triangles) (f). All spiders were introduced on day 3 only after experimental larval abundances were recorded for that day. For all data, n = 10 replicates. Different letters represent significant difference within experiments/species at P < 0.001, applying post hoc Friedman test.
Significant reductions in the larval abundance of E. schuitemakeri (χ2 = 15.0, df = 2, P < 0.001; Figure 4d) and Lestodiplosis spp. (χ2 = 16.8, df = 2, P < 0.001; Figure 4f) were observed after M. nepenthicola was introduced. Although M. nepenthicola had significant weight changes throughout the experiment (feeding on E. schuitemakeri: χ2 = 18.6, df = 2, P < 0.001 (Figure 4c); feeding on Lestodiplosis spp.: χ2 = 18.7, df = 2, P < 0.001 (Figure 4e)), a significant mass increment after the spider's introduction was only observed when feeding on E. schuitemakeri (Figure 4c).
Field experiments
A total of 42 pitchers were involved in statistical analyses. Over 4 wk (i.e. wk 3 to 6), dipteran larval abundance significantly changed when M. nepenthicola was introduced or removed (Tripteroides spp.: χ2 = 20.7, df = 3, P < 0.001; Culex spp.: χ2 = 19.0, df = 3, P < 0.001; E. schuitemakeri: χ2 = 33.7, df = 3, P < 0.001; Lestodiplosis spp.: χ2 = 24.2, df = 3, P < 0.001), with a general decrease in prey abundance in the spider's presence and concomitant increase after the spider's removal (Figure 5).
Figure 5. Box-plots on predation of Misumenops nepenthicola on dipteran larvae. Effects of absence (wk 3 and 5) and presence (wk 4 and 6) of M. nepenthicola (presence indicated by spider inserts) on larval abundance of mosquitoes Culex spp. (a) and Tripteriodes spp. (b), scuttle fly Endonepenthia schuitemakeri (c) and gall midge Lestodiplosis spp. (d). Central bar: median; hinges: 25 and 75%; whiskers: 5 and 95%. For all data, n = 42 replicates. Different letters represent significant difference within species at P < 0.001, applying post hoc Friedman test.
DISCUSSION
Our study is one of the first to demonstrate the influence of a terrestrial phytotelm forager on key phytotelm organisms via direct predation: Misumenops nepenthicola, but not T. nepenthiphilus, significantly influences phytotelm dipteran larval abundance in N. gracilis. This supports the ecological importance of cross-habitat-capable predators in influencing phytotelm insect larvae numbers (Greeney Reference GREENEY2001, Romero & Srivastava Reference ROMERO and SRIVASTAVA2010), suggesting this spider's role in regulating larval abundance in N. gracilis.
Evidence of aquatic dipteran larval predation
The decrease in mosquito larval abundance and increase in spider mass reported here support earlier claims on the aquatic foraging capability of M. nepenthicola (Clarke Reference CLARKE1998, Moran Reference MORAN1993). A small pit on the ventral abdomen of M. nepenthicola allows storage of a small reserve supply of air that facilitates aquatic foraging and possibly predator avoidance. Like all crab spiders, M. nepenthicola possesses eyes that provide excellent spatial resolution (Land Reference LAND and Barth1985). Its relatively longer legs possibly allow swift and safe locomotion into and out of the pitcher fluid since the pitcher inner wall is usually lined with numerous draglines. In contrast, T. nepenthiphilus did not affect mosquito larval abundance; a significant weight decline meant it is not capable of aquatic foraging. No cannibalism in mosquito larvae was recorded, since their numbers did not differ after 3 and 5 d during laboratory experiments. Although the aquatic foraging ability of M. nepenthicola is further established with lower abundances of E. schuitemakeri and Lestodiplosis spp. larvae after its introduction, results of the latter's abundance did not concur with spider weight change: spiders lost weight with decreased Lestodiplosis spp. abundances. We believe that the total amount of Lestodiplosis biomass consumed was inadequate to sustain an increase in spider mass, given that they are half the size (2 mm) of E. schuitemakeri and Tripteroides (both 4 mm).
Future directions
Though M. nepenthicola is described as a nepenthebiont, empirical data supporting its symbiotic relation with N. gracilis is lacking. Clarke (Reference CLARKE1997) proposed that the entire in-fauna of Nepenthes pitchers is in a symbiotic interaction with the plant as these organisms contribute to the more efficient breakdown of prey items within the pitchers. Additionally, Phillipps et al. (Reference PHILLIPPS, LAMB and LEE2008) explained that while M. nepenthicola feeds on insects, the plant may benefit from the spider's waste products, suggesting this spider's symbiotic interaction with N. gracilis (also see Clarke et al. Reference CLARKE, BAUER, LEE, TUEN, REMBOLD and MORAN2009, Romero et al. Reference ROMERO, MAZZAFERA, VASCONCELLOS-NETO and TRIVELIN2006). In the mutualistic interaction between the ant C. schmitzi and its host pitcher plant, N. bicalcarata, Clarke & Kitching (Reference CLARKE and KITCHING1995) reported that, while providing this ant with a domicile within the swollen tendrils of the pitchers, the host in fact benefits from the comminution of larger prey items by the ants, which the plant extracts from the pitcher fluid. Without this ant-assisted breakdown, the pitcher will likely become anoxic as the rate of decay outruns that of digestion (Clarke & Kitching Reference CLARKE and KITCHING1995). Also, these ants prey on organisms within the pitcher, possibly acting as a top predator within the contained food web (Kitching Reference KITCHING2001). Likewise, by regulating the abundance of dipteran larvae in the pitcher fluid via direct predation, M. nepenthicola can reduce the potential amount of prey putrefaction in N. gracilis that can disrupt the plant's digestive system.
The predation of M. nepenthicola on various dipteran species suggests that it can regulate dipteran larval populations and indirectly affect the food chain and ecosystem within pitchers. Several well-studied food webs of N. gracilis (Clarke & Kitching Reference CLARKE and KITCHING1993, Phillipps et al. Reference PHILLIPPS, LAMB and LEE2008, Tan Reference TAN1997) have proposed M. nepenthicola as a higher trophic level consumer and possibly an apex predator in pitcher phytotelmata (Figure 1). Our results support this possibility: M. nepenthicola can influence the population of key organisms in N. gracilis. Future research should focus on this spider's potential to alter food web and community structure (e.g. altering the balance between aquatic detritivores and predators) and the ecosystem functions these dipteran larvae provide. We also propose that future ecological studies of phytotelm communities include investigating the potential of terrestrial phytotelm dwellers, particularly those with aquatic foraging ability, to influence the aquatic organisms’ populations and hence food web and ecosystem functions. Finally, with global warming altering predator–prey interactions (Traill et al. Reference TRAILL, LIM, SODHI and BRADSHAW2010), we urge that future phytotelm dipteran studies should take into consideration the effect of abiotic factors, in particular temperature (Hoekman Reference HOEKMAN2010), in influencing pitcher community structure and ecosystem function.
ACKNOWLEDGEMENTS
We thank National Parks Board Singapore for their permission to carry out this research under permit number NP/RP950. We also wish to express our appreciation to Charles Clarke for his help in the identification of mosquito larvae, and Wan Jean Lee and Tien Ming Lee for their comments and suggestions on the manuscript. This project was supported financially by a final-year project grant from the National University of Singapore.
