Materials & Methods
I conducted experiments on mature lateral traps of two aquatic species from section Utricularia, Utricularia australis and U. gibba. U. australis is a rhizoidless, freely-suspended macrophyte with filamentous, multi lobed leaves organised in whorls on non-differentiated monomorphic shoots. During the growing season shoots have continuous, very rapid, apical growth with corresponding basal decay (Taylor, 1989; Adamec and Kovářová, 2006). Traps are dimorphic with i) Numerous lateral traps (0.5-2.5 mm long) borne on leaves from capillary leaf segments (Taylor, 1989) or in place of a leaf filament in a dichotomy (depending on perspective), with a single leaf incorporating a range of trap sizes while smaller, basal traps, with truncated antennal trunks, uniform in size, grow in the angle between the shoot and primary leaf segment (Taylor, 1989) (Fig. 2.2A). U. gibba is an affixed aquatic. Groups of highly ramified individuals, densely intertwined, form mats of vegetation (ramified colonies). U. gibba grows more robustly on and just below substrate as an amphibious plant, but can grow as a suspended aquatic. Taylor (1989) refers to this suspended form as sterile, as flowering will not normally take place unless U. gibba is anchored, although flowering can occur in deeper water when plants use floating mats of detritus as scaffolding. I used this suspended form in all experiments. The traps of U. gibba are uniform in shape, being of the lateral kind found in U. australis, although less variable in size (1-2.5 mm long). They arise on sparsely lobed leaves singularly, in pairs or occasionally more numerous (Taylor, 1989; Guiral and Rougier, 2007) (Fig. 2.2B). Traps of both species share the branched, filiform antennae characteristic of section Utricularia, although U. gibba has considerably more supplementary bristles (Taylor, 1989; Guiral and Rougier, 2007) including a set arising on the dorsal margin of the trap door which is absent from U. australis.
I conducted this experiment using one clone each of U. australis and U. gibba. Stolons of both species were collected from the Upper North Island, New Zealand. Prior to experimentation plant material used in this experiment was pre-cultivated from the field collected plants for no less than 12 months in an indoor, rooftop greenhouse at the School of Biological Sciences, University of Auckland. U. gibba was cultivated in a 60 l plastic tank with a water column of 400 mm. The tank contained one clone derived from a single stolon fragment. Multiple stolon lengths of U. australis were to produce a single cultivation in a 300 l glass aquarium, filled almost to the brim. As the U. australis stolons used were collected from a single population and U. australis seldom flowers and produces sterile seed (Taylor, 1989), this cultivation was assumed to be a single clone. All tanks/aquarium contained a 30 mm base layer of 1:1 peat moss/sand mix. Water depth was maintained through the addition of deionised water, steeped in identical substrate and filtered prior to addition. The pH within the tanks varied between 4.8 and 6.8. Light in the greenhouse was approximately 40% of the outside light in the open. All traps used in the experiments were positioned near the base of the leaf, arising either from primary or initial secondary leaf filaments (see Chapter One for detail on leaf anatomy). For these traps, the two species had different, non-overlapping size distributions, with U. gibba having smaller traps. Mean trap bladder lengths were determined from a randomly selected subsample of the traps used in the experiments: U. gibba 1.32 mm (0.97–1.69 mm, n= 23), U. australis 2.03 mm (1.81–2.38 mm, n=22).
Five species of micro-crustaceans from three orders were in the appendage removal experiments, the cladoceran C. sphaericus (Chydoridae, Diplostraca), the ostracod C. vidua (Cyprididae, Podocopida), the daphnid S. kingi (Daphniidae, Diplostraca) and two cyclopoid copepods, A. robustus (Cyclopidae, Cyclopoida) and M. cf. leuckarti (Cyclopidae, Cyclopoida). All prey species used are of a cosmopolitan distribution and were cultivated in the rooftop greenhouse of the School of Biological Sciences, University of Auckland (Fig. 2.3).
C. sphaericus and C. vidua are both phytophilous and benthic feeding but unlike C. sphaericus, C. vidua does not filament feed (Roca and Danielopol, 1991). While adept at penetrating the interstitial spaces of substrate, C. vidua preferentially seeks out and scrapes periphyton on plant surfaces (Roca et al., 1993). S. kingi is associated with weedy littoral; areas but is an epineustic feeder, applying its ventral side to the water surface and moving about upside-down (Chapman et al., 2011). A. robustus and M. cf. leuckarti are both predatory, similar sized, cyclopoid copepods, occupying a wide range of habitats in sympatry, primarily littoral and benthic (Maier, 1990).
C. sphaericus, A. robustus and M. cf. leuckarti have previously been recorded in content surveys of field collected traps (Meyers and Strickler, 1979; Andrikovics, 1988; Mette et al., 2000; Kurbatova and Yershov, 2009). Although C. vidua has never been specifically identified, ostracods are preyed upon by aquatic Utricularia (Guisande et al., 2004; Walker, 2004; Kurbatova and Yershov, 2009) but accurate identification beyond class is infrequent. Scapholeberis mucronata, a cladoceran with similar morphology and behaviour to S. kingi has been surveyed with Utricularia (Pokyi; Marazanof, 1967; Mahoney et al., 1990; Kuczyńska-Kippen and Nagengast, 2006)
Antennae and bristle removal
I performed a series of three appendage manipulation experiments in the rooftop greenhouse at of the School of Biological Sciences, University of Auckland. All experiments were conducted Nov – Jan (summer). I tested the effect of appendage (bristles and antenna) ablation on the following combinations of aquatic Utricularia and prey species: 1) U. gibba and U. australis with C. sphaericus, to confirm the role of appendages in enhancing the capture rate of phytophilous, epibiont grazers and assess the effect of trap size on the ability to enhance capture probability, 2) U. gibba and U. australis, with C. sphaericus, C. vidua , S. kingi to compare the effects of differing prey locomotory and feeding behaviour on capture probability, and 3) U. australis with a mixture of A. robustus and M. cf. leuckarti to test whether appendages affected the rate of capture of phytophilous but predatory animals, and to test if appendages interact with the matrix of leaf filaments surrounding traps to enhance capture of a prey species.
In all experiments I pipetted animals into 25 ml cell culture dishes (60x15mm) under a stereoscope. Dishes were filled with 20 ml of plant cultivation water filtered through 150 µm nylon mesh. I used the trap door sizes of randomly selected exemplar traps for both Utriculaia species to help select only animals small enough to be successfully trapped. The exemplar traps were taken from the same stolon-position as those selected for ablation (specified above), from additional randomly selected stolon lengths, at nodes as specified for each experiment (below). The size distributions of animals in the U. australis treatments were larger than those in U. gibba treatments. The number of animals varied among experiments and between each dish (see below for counts for each experiment). Animals were occasionally added during the experiment by way of live birth as it was not always feasible to use only non-gravid animals. All dishes were inoculated to the point of saturation, however, so animals remained available for capture throughout the duration of the experiment (see below for ranges in the number of animals per dish for each experiment).
I removed trap-bearing plant modules (the type of modules, the number of traps per dish and stolon positions from which that traps were derived varied among experiments, for details see below) from cultivation tanks to an additional set of filtered cultivation water filled 25 ml culture dishes and randomly allocated each dish to a treatment. To reduce the likelihood of air bubbles forming in the traps I moved plant material using a cut-down 3 ml Pasteur pipette. Stolon fragments and excised leaves are relatively autonomous organs able to survive for weeks in ambient water with traps continuing to function for at least four days (Sirová et al., 2003). The use of multiple fragments allowed for up to ten traps per treatment while ensuring they were of a similar age, as Friday (1989) showed trapping efficiency declines rapidly in older traps. The position of the trap bearing node on the stolon, in relation to the growth tip, served as a crude approximation for trap age. As U. australis and U. gibba were each represented by a single clone, for both species I considered traps taken from different stolon lengths to come from a single plant.
I ablated appendages from submerged traps under a stereoscope (Leica) between x10 to x25 magnifications with a pair of 2mm cutting edge spring scissors (Vannass). Micro-surgical ablation of appendages does not noticeably affect trap function (Meyers and Strickler, 1979), but care is still required not to damage trigger hairs, the trapdoor, or its margin. After 24 hours I checked dishes and replaced any dead or injured animals (< 5% in all cases). I also replaced any traps not obviously resetting (< 1%). Finally, between 10:00 and 14:00 hrs, I relocated the plant fragments to the dishes containing prey, beginning the experiment. Traps were exposed to prey for 24 hrs, which for all experiments included a light period of c.a. 15 h. Water temperatures in the culture dishes across all three experiments ranged between 20.5 – 26.2 °C (only one temperature reading taken per experiment, from a randomly selected culture dish), reflecting the ambient temperature of the greenhouse on the day each experiment was conducted. At the termination of the experiment I added c. 2 ml of 98% ethanol to all treatments and recorded both captured and uncaptured animals, pooling results for each dish.
Experiment one: U. gibba and U. australis with C. sphaericus.
Treatments consisted of 20-30 C. sphaericus and 10 traps of either, U. gibba or U. australis. I subjected traps attached to shoot fragments to one of four appendage manipulations: i) All appendages removed,
ii) Antennae removed, iii) Bristles removed, iv) All appendages intact (control). I replicated each treatment block six times (6x2x4=48 dishes in total). For the U. gibba treatments I used 10 traps attached to the 7th to 12th nodes of two shoots. For U. australis the 10 traps came from the excised 10th nodes of two shoots. I removed any basal traps and all but the five largest lateral traps. I trimmed the capillary leaf segments on the U. australis leaves until the ratio of leaf to traps was similar to that on the U. gibba nodes. Water in the culture dishes was diluted to 50:50 mix with deionised water.
Experiment two: U. gibba and U. australis with C. sphaericus, C. vidua and S. kingi.
Treatments consisted of 15-25 animals from one crustacean species and 10 traps of either U. gibba or U. australis. I treated traps in one of two ways: i) All appendages intact, ii) All appendages removed. Animals and traps were selected and prepared as for experiment one. I replicated each treatment block six times (6x2x3x2=72 dishes in total). I conducted a subsequent trial using only U. gibba and S. kingi with each treatment block replicated 12 times (12×2=24 dishes in total). Water in the culture dishes was undiluted.
Experiment three: U. australis with A. robustus and M. cf. leuckarti.
Treatments consisted of six lateral traps and 12-20 copepods drawn from mixed cultivation of A. robustus and M. cf. leuckarti. I used the largest trap from the 7th to 13th nodes of a single stolon fragment. Animals were selected as for experiment one. Traps had either: i) all appendages intact, ii) all appendages removed. In order to assess the effect of the surrounding leaf matrix on the effectiveness of appendages, leaves were also manipulated in one of two ways: i) capillary leaf segments around the traps trimmed away, ii) capillary leaf segments around the trap intact. In the latter case I trimmed the tips of any capillary leaf segments not directly surrounding the traps to ensure all nodes were subject to similar types of damage. I replicated each treatment block 6 times (6x2x2=24 dishes in total). Water in the culture dishes was undiluted.
I conducted all analyses in R 3.1.1 (Team, 2014).
For all experiments I analysed the proportion of available prey captured after 24 hrs (termed rate of capture) using Generalised Linear Models (GLMs) with binomial distributions and logit link functions. The logit link function is the canonical link function for the Bernoulli distribution. The response variables for all experiments were two column matrices of the number of captured and remaining animals. Experiment one had two factors: appendage ablation at four levels and plant species at two levels. Experiment two had three factors: appendage ablation at two levels, plant species at two levels, and prey species at three levels; the supplementary experiment had one factor, appendage ablation at two levels. Experiment three had two factors: appendage ablation at two levels and leaf ablation at two levels. When model fitting, I removed non-significant interactions providing the subsequent model (the model without the non-significant interaction) had a smaller Akaike Information Criterion (AIC) than the previous one (the model including the non-significant interaction). For chosen models I assessed the statistical significance of model terms (main effects and interactions) for factors with more than two levels using a chi-squared test based on the reduction of residual deviance. Tukey contrasts for multiple comparisons were performed using the multcomp package (Hothorn et al., 2008). As the logit function gives the logarithm of the odds p/(1-p), where p = the probability of an event (in this case, a capture), I reported odds ratios in addition to probabilities. Graphs of model estimated capture probabilities (effects sizes) and 95% confidence intervals, created using the effects package (Fox, 2003) are also provided. These provide Tabulated logistic regression, ANOVA and Tukey outputs for all analyses are provided in Appendix 1. For all experiments the critical value for rejecting the null hypothesis was taken as α = 0.05.
Experiment one: Utricularia gibba and U. australis with Chydorus sphaericus.
U. gibba trapped C. sphaericus in higher proportions than U. australis irrespective of the presence or absence of appendages (z = 4.105, p < 0.0001) (Fig. 2.4). Appendage ablation significantly reduced capture in the U. gibba treatments only. Only the removal of all appendages resulted in a significant reduction in capture rate compared to the control treatments that had all appendages intact (z = 5.047, p< 0.0001). For the pairwise comparisons between the other levels of the ablation factor (e.g. bristles only-all appendage and antenna only-all appendages) for U. gibba and all pairwise comparisons for U. australis, see, Table 1.1.4 in Appendix 1. The odds of C. sphaericus being captured by traps with a full complement of appendages (control) is 5.11 times higher than those with none.
Experiment two: U. gibba and U. australis with C. sphaericus, Cypridopsis vidua and Scapholebris kingi.
As there was no significant three-way interaction between plant species, prey species and appendages, or two-way interaction between plant species and appendages, these two terms were removed from the model. The interaction between plant species and prey species was also insignificant (χ2 test, df = 2,65; p = 0.056), but retained as the removal of this term did not lower the AIC of the subsequent model. U. gibba captured all prey species in higher proportions than U. australis (z = 8.658, p < 0.0001) (Fig. 2.5). Appendages impacted prey capture, but their effect differed significantly among prey species (χ2 test, df = 3,63; p= 0.0001) (Fig. 2.6). Appendages had no effect on the capture of C. vidua in either U. gibba (z= 0.912, p = 0.740) or U. australis (z = -1.615, p = 0.286) treatments. When considering both plant species collectively, the odds of C. sphaericus being caught by traps with appendages (without consideration of plant species) was 2.04 times higher than that of being caught by traps without them. This difference was significant only in the U. gibba treatments (z =- 2.915, p = 0.011), but not in the U. australis treatments (z = 2.222, p = 0.077). Appendages of both Utricularia species negatively affected the capture of S. kingi. In the U. gibba treatments the odds of S. kingi being captured by traps without appendages was 2.23 higher than those with, although this difference was not significant (U. gibba: z = -2.190, p =0.083). In the subsequent trial, conducted only using U. gibba and S. kingi with increased replication, this effect was significant (z = -4.815, p < 0.0001). Here, the odds of S. kingi being captured by traps without appendages was 3.87 times higher than those with them.
Experiment three: U. australis with Acanthocyclops robustus and Mesocyclops cf. leuckarti
The leaf matrix surrounding traps did not significantly interact with appendages to affect the capture of the cyclopoid copepods so this term was dropped from the model. Neither did appendages nor a surrounding matrix of leaf segments have any individual effect on the probability of prey capture (appendages: z = -1.558, p = 0.119; leaf matrix: z = -0.230, p=0.818).
It seems reasonable that traps of aquatic Utricularia would be under selection for mechanisms that enhance trapping success. In aquatic species, traps are in competition with other plant modules for the attention of phytophilous microcrustaceans engaged in feeding and ovipositing, and taking refuge from other predators. Prey species who are themselves predatory, such as carnivorous cyclopoid copepods, hunt throughout the plant. Lures or funnels steering animals through the competing plant matrix to a trap encounter should confer an adaptive advantage (Darwin, 1875; Meyers and Strickler, 1979; Guisande et al., 2007).
Appendages lure filament feeders…
Appendage removal resulted in a significant reduction in the capture of the filament-feeding cladoceran C. sphaericus. Unlike Meyers and Stickler (1979), who used a similar level of replication, I was unable to demonstrate an increase in appendage effectiveness with successive additions of bristle sets and antenna. Only fully ablated traps captured lower numbers of C. sphaericus than those with a full complement of appendages.
…but do not act as funnels
I found no evidence that appendages funnel in a stricter sense, acting as a locomotory pathway that guides pelagic animals to the trap door. The absence of appendages affected the capture of C. sphaericus (Fig. 2.5), but not Cypridopsis vidua, an animal also phytophilous and associated with the benthos, but without the filament feeding behaviour of C. sphaericus (Figs. 2.6-2.7). Possessing appendages actually deterred the capture of the pelagic cladoceran, Scapholebris kingi whose interactions with the plants were limited to passing through them between episodes of epineustic feeding. While C. vidua and C. sphaericus move among plants, S. kingi’s vertical traversal of the water column results in movement perpendicular to the stolon. It is possible that with this angle of approach to traps appendages deflect these animals from contact with trigger hairs.
Plant matrix has no effect on the ability of appendages to enhance prey capture
Appendages do not exist in isolation. The traps of U. australis are surrounded by a leaf matrix whose filamentous nature resembles the bristles and antennae branches themselves. While appendage ablation did not reduce the capture rate of capture of the cyclopoid copepods Acanthocyclops robustus and Mesocyclops cf. leuckarti, irrespective of the presence or absence of a surrounding plant matrix, these animals were being tested in isolation from prey species.
Copepods do not predate blindly, searching in an undirected fashion (Kerfoot, 1978; Williamson, 1983) and change their behaviour in relation to prey density (Williamson, 1981). Copepod behaviour (locomotion) may change in the presence of their own prey, in such a way as to increase the likely hood of a trap encounter. The same also applies to the prey of copepods and other predatory microcrustaceans. The presence of predatory animals causes their prey to seek refuge (Lima, 1990) and appendages may offer such shelter, increasing the chance of an encounter with the trap.
Prey also aggregate in response to nearby predators (Lima, 1990; Sparrevik and Leonardsson, 1995) and some microcrustaceans aggregate in response to changes I their population density (Harms and Johansson, 2000). Changes in density lead in turn to changes in locomotion which can effect encounter rates with Utricularia traps (Harms and Johansson, 2000). Harms and Johansson (2000) demonstrated that patterns in prey selection in a two prey system can reverse with changes in density of the respective prey species. In the future ablation trials should be conducted with species assemblages and where prey density is varied.
The effectiveness of appendages varied between aquatic Utricularia species.
While the general pattern of capture probabilities (the central tendency) was the same for both Utricularia species, appendages had a statistically significant effect on capture rates in only plant species, U. gibba. The U. australis traps used in my experiments, however, were more similar to those of U. vulgaris used in experiments by Meyers and Stickler (1979), who demonstrated larger effects than those of U. gibba. The reason for this disparity is unclear; why I failed to observe capture enhancement by U. australis appendages, while similar work on the similarly large traps of U. vulgaris demonstrated such a clear affect. The trap resetting rates of aquatic Utricularia are too similar to be responsible for a difference in capture rate, moreover, Adamec (2011c) recorded U. australis having slightly faster resetting rates than other aquatic Utricularia (including U. vulgaris and U. floridana, an affixed aquatic species closely related to U. gibba).
1. General Introduction
1.1 Carnivorous plants
1.2 The Genus Utricularia
1.3 Thesis content and structure
2. Do appendages enhance prey capture? The role of antennae and bristles in two aquatic Utricularia species
2.2 Materials & Methods
3. The effect of light and water level on trap size and appendage expression of the amphibious Utricularia gibba
3.2 Materials & Methods
4. The effect of feeding and fertilisation on growth, trap size and appendage expression of the suspended aquatic Utricularia australis
4.2 Materials and Methods
5. General Discussion
5.1 Aquatic appendages and prey attraction
5.2 Investment in carnivory and appendage plasticity
5.3 Future directions
GET THE COMPLETE PROJECT