- Email: [email protected]
Foraging strategy switching in an antlion larva
Behavioural Processes 91 (2012) 1–7
Contents lists available at SciVerse ScienceDirect
Behavioural Processes journal homepage: www.elsevier.com/locate/behavproc
Foraging strategy switching in an antlion larva Yu-Jen Tsao a,b , Toshinori Okuyama a,∗ a b
Department of Entomology, National Taiwan University, Taipei, Taiwan Division of Management, Endemic Species Research Institute, Nantou, Taiwan
a r t i c l e
i n f o
Article history: Received 30 January 2012 Received in revised form 10 March 2012 Accepted 22 April 2012 Keywords: Optimal foraging Dynamic optimization Predation risk Trap-building Ambush
a b s t r a c t Antlion larvae are typically considered as trap-building predators, but some species of antlions always forage without using pits or only sometimes use pits to capture prey; they can ambush prey without pits. This study examined a species that switches its strategy between pit-trapping and ambushing and asked the mechanism behind the switching behaviour. A dynamic optimization model incorporating tradeoffs between the two strategies was built. The tradeoffs were prey capture success and predation risk (both are higher when pit-trapping). The model predicted that antlions should use the trap-building strategy when their energy status is low and should use the ambush strategy when their energy status is high. These predictions as well as an assumption (i.e., predation risk associated with pit-trapping is higher than that associated with ambushing) of the model were empirically confirmed. The results suggest that antlions flexibly switch between pit-trapping and ambushing to maximize their fitness by balancing the costs and benefits of the two strategies. © 2012 Elsevier B.V. All rights reserved.
1. Introduction Foraging strategies vary widely among organisms. For example, sit-and-wait foraging and active (or widely) foraging may be perceived as two end points of a continuum, and their comparative economics have been widely studied in a variety of ecological contexts (Huey and Pianka, 1981; Helfman, 1990; Perry, 1999; Scharf et al., 2006). Understanding how animals make their foraging decisions is important not only for the study of behaviour (Stephens and Krebs, 1986; Stephens et al., 2007) but also for examining large scale ecological processes such as population and community level dynamics (Bolker et al., 2003; Abrams, 2010). For example, foraging modes of predators can affects ecosystem level processes (e.g., nutrient cycling) by altering adaptive strategies of their prey that propagates through ecosystems through direct and indirect interactions (Schmitz, 2010). Foraging strategies are often described in dichotomous manners. For example, many birds and lizards exhibit either sit-and-wait or widely foraging tactics (McLaughlin, 1989). In lizards, this bimodal distribution is shaped both by historical and current selection pressures (Schwenk, 1993; Vitt and Pianka, 1994). Similarly, spider species may also be categorized dichotomously into species that use or do not use webs to capture prey. In these situations, comparative studies describing general patterns
∗ Corresponding author at: Department of Entomology, National Taiwan University, No. 1, Sec. 4, Roosevelt Road, Taipei 106, Taiwan. Tel.: +886 2 3366 5282; fax: +886 2 2732 5017. E-mail address: [email protected] (T. Okuyama). 0376-6357/$ – see front matter © 2012 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.beproc.2012.04.012
associated with each strategy is useful (Anderson and Karasov, 1981; Huey and Pianka, 1981; Porges et al., 2003). However, species that use mixed strategies (e.g., exhibit both sit-and-wait and widely foraging tactics) are particularly useful for directly examining hypotheses (Balent and Andreadis, 1998). For example, with a species with a mixed strategy, theoretical predictions (e.g., conditions that selects sit-and-wait vs. widely foraging) can be directly examined by observing foraging expressions under respective conditions (Grant and Noakes, 1987). Trap-foraging organisms are ecologically important (Wise, 1993), but the ecology and evolution of trap-foraging (e.g., ecological cost) is still not well understood (see Ruxton and Hansell, 2009 for review). Although antlion larvae (Neuroptera: Myrmeleontidae) are commonly known as sit-and-wait predatory insects that capture small invertebrates in sandy environments using conical pits (Wilson, 1974; Griffiths, 1980; Lucas, 1982; Griffiths, 1986; Devetak, 2005; Devetak et al., 2005; Fertin and Casas, 2006; Nardi, 2007; Ruxton and Hansell, 2009), foraging behaviour of antlion larvae is much more diverse than the common perception. Some antlions species do not use pits to capture prey (we call this strategy ambush strategy). Furthermore, some antlion larvae exhibit a mixed strategy (switch between foraging with and without a trap) rather than specializing on one strategy (Miller et al., 1999), which is uncommon in spiders (a major taxonomic group in which some species exhibit trap-foraging). To understand the costs and benefits associated with trap-foraging, we use an antlion species with a mixed strategy. When considering why an antlion sometime forages with a pit and sometime ambushes prey, a natural starting place is a potential tradeoff between the two strategies. First, the function of pits
2
Y.-J. Tsao, T. Okuyama / Behavioural Processes 91 (2012) 1–7
implies that the pit-trapping strategy enjoys higher prey capture successes than the ambush strategy (Heinrich and Heinrich, 1984; Griffiths, 1986; Cain, 1987; van Zyl et al., 1997; Elimelech and Pinshow, 2008). On the other hand, a potential cost of pit-foraging is the metabolic cost. However, although the metabolic expenditure during pit constructions is 10 times resting metabolic rate (Lucas, 1985), there is no known difference in metabolic rate between individuals with and without pits when they are awaiting for prey once pits are constructed (van Zyl et al., 1997), suggesting that the pit maintenance cost may be negligible. Furthermore, non-pit building antlions have been suggested to have higher resting metabolic rates than pit-building antlions because pits allow them to capture prey more easily (Lucas, 1985), which would diminish the potential metabolic cost of pit-trapping strategy. Predation risk is an important factor that affects foraging behaviour of many organisms (Kats and Dill, 1998; Lima, 1998; McPeek and Peckarsky, 1998; Caro, 2005; Ruxton and Hansell, 2009). With respect to trap foragers, orb-weaving spiders decrease their web sizes in the presence of predators to reduce predation risk (Li and Lee, 2004). In antlions, predation is also hypothesized as a selection factor against pit-trapping (Ruxton and Hansell, 2009). Hauber (1999) discusses that Florida scrub-jays visually locate Myrmeleon carolinus from pits and preferentially prey on larvae with large pits. Loria et al. (2008) found that antlions decreased pit-building activity when potential predators were present. Despite this evidence, there are few studies that explicitly examined the effect of predation risk on the foraging activity of antlions (Scharf et al., 2011). In our question, it is also clear that predation risk cannot fully explain the difference in foraging strategy because antlion larvae exhibit large behavioural variation, e.g., under an enemy free environment in the laboratory condition and within the same area in the field (personal observation). In other words, some individuals forage with pits while others ambush even in the same environment. Thus, although predation risk appears to be an important factor in antlions’ decision to switch their foraging strategies, the behavioural rule is not well understood. In this study, we examined the mechanism in which the antlion larva Myrmeleon persimilis switches its foraging strategies. The key factor we focused was the tradeoff between the pit-trapping and ambush strategies (discussed above) where the pit-trapping strategy results in a higher prey intake rate but also faces higher predation risk (i.e., a tradeoff between energy intake and predation risk). We present a hypothetical mechanism describing how antlions switch between the two foraging strategies using an optimal foraging model whose predictions and an assumption (predation risk of pit-trapping is higher than that of ambushing) were empirically validated. As will be discussed below, because the results imply that the physiological status of the antlion is an important factor for its behavioural expression, supplementary field observations were also conducted to evaluate the natural physiological conditions of the antlion.
2. Methods 2.1. Study animals The antlion species used in this study was Myrmeleon persimilis (Neuroptera: Myrmeleontidae), an endemic and the most widespread coastal species of genus Myrmeleon in Taiwan (Stange et al., 2002). M. persimilis larva, like other antlion species, has three instar stages and usually eclose in February to May (unpublished data). M. persimilis larvae were collected in a coastal habitat in Shimen District, New Taipei City (25◦ 18 N,
121◦ 32 E). It is a sandy region with abundant rainfall. When it rains, M. persimilis larvae burrow a few centimeters beneath and come out when the sand is dried. Common plant species in the habitat include wormwoods (Artemisia capillaries), tree heliotropes (Tournefortia argentea), Hosobawadan (Crepidiastrum lanceolatum), and Indian Blankets (Gaillardia pulchella) (personal observations). Pit-trapping individuals of the species abruptly abandon their pits and assume an ambush strategy rather than gradually reducing the diameter of pits. Thus, instead of considering the pit-trapping and ambush strategies as a continuum of the same strategy (i.e., pit-diameter), they were considered as distinct strategies in this study. 2.2. Dynamic optimization model A dynamic optimization model was constructed to examine how an antlion with the mixed strategy chooses its foraging expression based on its state variables (i.e., energy level, time, and the current strategy) to maximize its fitness. The model considers a situation where an antlion larva aims to survive for a specific time duration T (e.g., the larval stage). In each time step (discussed in detail below), a larva makes a decision about its foraging strategy. There are two behavioural options: pit-trapping and ambushing. When the larva adopts the pit-trapping strategy, it enjoys greater prey capturing success (Griffiths, 1991; Nardi, 2007; Elimelech and Pinshow, 2008) but also suffers higher predation risk (Hauber, 1999) with respect to the ambush strategy. Parameters associated with each strategy are fQ (foraging success), dQ (predation risk), cQ (metabolic cost) and z (pit construction cost). The subscript (Q = P or A) describes the behavioural option. For example, fP and fA are the probability of foraging success for an individual with the pit-trapping strategy and with the ambush strategy, respectively. A larva is associated with three state variables: energy state x, time t, and behavioural states, QC . (Q is the behavioural option to be chosen while QC is the currently exhibiting strategy.) The energy state varies from 1 to xmax , and the time step varies from 1 to T (both take integer values only). Suppose, at time t, a larva whose energy state is x chooses the behavioural option Q, its expected fitness is WQ = (1 − dQ )[fQ F(x + Y − cQ − z, Q, t + 1) + (1 − fQ )F(x − cQ , Q, t + 1)] where F(x, t) is the maximum expected fitness of a larva at time t whose energy state is x. (How to obtain F is discussed below.) If the larva is predated (the probability is dQ ), its fitness is zero. If the larva survives one time step (the probability is 1 − dQ ), it may forage successfully (the probability is fQ ) or fail to capture a prey (the probability is 1 − fQ ). The larva which forages successfully improves its energy state by Y (prey profitability). The pit-construction cost z depends on the current behavioural state QC while all other parameters are independent of QC . In particular, z takes a positive value only when the current behavioural state is ambush (QC = A) and the antlion decides to build a pit (Q = P). In all other behavioural sequences, z = 0, because z is the cost associated with building a pit. The model predicts that the option P (pit-trapping) is the optimal strategy if WP > WA ; the option A (ambushing) is the optimal strategy when WA > WP . It has been reported that the pit-construction cost is 10 times of the resting metabolic rate (e.g., z = 10cQ ) (Lucas, 1985), but the relationship between cA and cP are not well known. As discussed above, previous studies have argued that metabolic rate should be lower for individuals with pits (i.e., cA > cP ) because they can capture prey more easily (Lucas, 1985) while others may argue for the opposite because there may be a pit-maintenance cost. In
Y.-J. Tsao, T. Okuyama / Behavioural Processes 91 (2012) 1–7
this study, we assume the difference is negligible (i.e., cA = cP ) and set these values 1. This determines z = 10. Y depends on types of prey. It is worth mentioning that the metabolic rate of antlions is higher after they capture a prey (van Zyl et al., 1997). Thus, Y should be discounted for the metabolic cost, but it will not affect this analysis. We are interested in how the ecological tradeoff influences the behavioural decision making in antlions. Thus dP > dA (i.e., pit-trapping experiences higher predation risk) and fP > fA (i.e., pittrapping experiences higher foraging success) were assumed. To analyze the model, we need to determine four more parameter values: dQ , fQ (Q = P or A), and the state conditions: xmax and T. In addition, the terminal fitness must be decided (Mangel and Clark, 1988). The terminal fitness describes the fitness of an individual with energy level x at the last time step. For the results we describe below, the terminal fitness is assumed to be the energy state x [i.e., F(x,Q,T) = x]. In other words, the greater the energy state, the greater the fitness is. However, changing the detail (e.g., the fitness increases acceleratingly or deceleratingly with the energy state) does not alter the qualitative conclusion discussed below. We set xmax = 200 and T = 50, but the qualitative results are also robust to these settings (details are mentioned in Section 3). Once the details are specified, the optimal solutions of the model can be derived using the backward iteration method (Mangel and Clark, 1988). To validate that results are not specific to the particular parameter values, the tradeoff parameters were varied in their values. The probability of death due to predation for pit-trapping individuals was set to dA = 0.01 (1% death per time step), and different degrees of the tradeoff were examined (dP = 0.02,0.04,0.06), which encompasses the odds ratio from 2.02 to 6.32. The foraging success probability for the ambush was set to fA = 0.1 (10% foraging success per time step), and different degrees of the tradeoff was examined (fP = 0.2,0.4,0.6), which encompasses the odds ratio from 2.25 to 13.5. The model parameters are summarized in Table 1. 2.3. Laboratory experiments As will be described below, the dynamic optimization model predicts that the optimal strategy is the pit-trapping strategy when energy status is low (e.g., starved) and is the ambush strategy when energy status is high. Laboratory experiments were conducted to examine whether antlions forage according to the model predictions. In addition, an assumption of the model (i.e., dP > dA , the probability of death due to predation is higher for the pit-trapping strategy) was also tested to strengthen the validity of the model. The average (± SD) weight of the prey larva (Tribolium confusum) used in the experiments was 2.637 ± 0.38 mg. Each individual was used only once in the experiments. 2.3.1. Effects of prey on pit-trapping individuals In this experiment, whether pit-trapping antlions change their strategy to the ambush strategy when their energy states improve Table 1 Parameter definitions and their values.
Prey capture success (pit-trap) Prey capture success (ambush) Predation risk (pit-trap) Predation risk (ambush) Metabolism (pit-trap) Metabolism (ambush) Prey profitability Pit construction cost (ambush)
Symbol
Value
fP fA dP dA cP cA Y z
0.2,0.4,0.6 0.1 0.02,0.04,0.06 0.01 1 1 20 10
3
was examined. The ambush strategy was defined as hiding under sand surface with only mandibles exposing outside, sometimes with a shallow cavity (Heinrich and Heinrich, 1984). The pittrapping strategy was defined as having a conical pitfall trap with the diameter greater than 1 cm. Pit-trapping individuals usually stayed at the bottom of pits, opening their mandibles and waiting for falling prey. If antlions behave according to the model prediction (see Section 3 for details), pit-trapping individuals should switch to ambush once they acquire enough prey (i.e., an experimental treatment). Antlions used in the experiment were individually housed in containers filled with sands (44.5 cm by 34.7 cm, sand depth 4 cm) that were maintained in a controlled environment: temperature (25 ± 2 ◦ C), relative humidity (40 ± 5%), and L:D 14:10 cycle. Second instar antlions (n = 37) were fed with flour beetle larvae, Tribolium confusum, until they stopped accepting a new prey. Subsequently they were starved for four weeks to standardize the starvation level of the study subjects. At this point, all individuals were pit-trapping. The starved larvae were randomly allocated to three treatment levels: 0, 1, or 6 prey items. The low treatment levels (i.e., 1 and 0 prey) were intended to affect the satiation levels of the larvae only little to none, while the high prey level (i.e., 6 prey) was intended to increase the energy level substantially. Prey larvae were placed near the edge of pits in such a way that they walked into the pits by themselves. In the multiple prey treatment level, one prey was given at a time. Whether the pit-trapping individuals switched their strategies to the ambush strategy was recorded after 24 h. 2.3.2. Effects of prey on ambushing individuals The purpose of this experiment was to test another prediction of the model; ambushing individuals change their foraging strategy to the pit-trapping strategy in response to a shortage of prey (i.e., low energy status in the model). In other words, if antlions behave according to the model prediction, ambushing individuals should switch to pit-foraging as they experience prey shortage (of an experimental treatment). The study organisms were maintained in the same way (described above). To conduct the experiment, we needed to prepare antlions exhibiting the ambush strategy. Because the first experiment described above showed that an abundant prey supply makes them exhibit the ambush strategy (see Section 3 for details), second instar antlions were fed until exhibiting the ambush strategy. Ambushing individuals were randomly allocated to two treatment levels: no prey (n = 35), two prey items (n = 35) (T. confusum larvae were used as the prey). Unlike the previous experiment, the ambushing larvae were generally more satiated so that large treatment levels could not be established. Furthermore, in the two prey treatment level, three (of the 35 replications) antlions ate only one larvae even though two larvae were provided (i.e., rejected the second prey item). Thus for the data analysis, these three replications were considered as a new treatment level (i.e., one prey). Subsequently, whether the ambushing antlions switched their strategies to the pit-trapping strategy was recorded after 24 h. 2.3.3. Effects of foraging strategy on cannibalism risk The purpose of this experiment was to test an assumption of the model; the probability of death due to predation is higher with the pit-trapping strategy than the ambushing strategy. Previous studies have shown that smaller antlion larvae were more likely to be cannibalized by larger larvae especially when the individual density was high (Matsura and Takano, 1989; Griffiths, 1992; Gotelli, 1997; Barkae et al., 2010). Therefore, in this experiment, starved third instar antlions (starved more than one month) were used as predators.
4
Y.-J. Tsao, T. Okuyama / Behavioural Processes 91 (2012) 1–7
In a plastic cup (diameter: 8.4 cm; sand depth: 3.5 cm), three equal pie-shaped areas were created with plastic plate partitions. One trap-building second instar larva, one ambushing second instar larva, and one third instar larva were randomly introduced: one in each partition. Second instar larvae were introduced first. The foraging strategy of the second instar larvae was manipulated by controlling their starvation status as described above. After each larva assumed the respective foraging strategy, a third instar larva was introduced, and the partitions were removed. Antlions were observed every 30 min for 270 min. If cannibalism did not occur within 270 min, the experimental trial was checked again after 24 h. If cannibalism still did not occur within 48 h, then the experiment was terminated regardless of the outcome. If the third instar larva prefers to attack pit-trapping individuals, it confirms the model assumption, i.e., pit-trapping strategy is associated with higher predation risk. 2.3.4. Data analysis The effects of prey on pit-trapping and ambushing individuals were tested with Binomial Generalized Linear Model (GLM) where the probability of switching to the other strategy (either pit-trapping or ambush strategy) was modeled. The predictor variable was the number of consumed prey item (i.e., treatment), and whether a larva switched to another strategy was the binary response variable. The Binomial test was used to test whether third instar larvae preferentially forage on trap-building or ambushing second instar larvae in the cannibalism experiment. The data were checked for overdispersion (Gelman and Hill, 2007), and no data were found overdispersed. 2.4. Field observation The purpose of this field observation was to describe the starvation status of antlions in the field because energy status is considered as an important factor in the model and its predictions. All field observations were conducted between 0900 h and 1800 h. When we observed a pit in the field, the diameter of the pit was recorded with a caliper. Then, the larva was dug out, and its weight and head width were recorded in the laboratory within a few hours of collection. These measurements represent the conditions of larvae in the field. Subsequently, larvae were kept in plastic cups (5.2 cm in diameter which is enough for larvae in each stage to construct a pit) filled with sand collected from the same habitat and were maintained in a controlled room as described above. Immediately after the initial measurement, each larva was fed flour beetle larvae (T. confusum) until it stopped consuming them. This procedure usually took 7 h, and antlion larvae consumed 3–6 flour beetle larvae to attain satiation. After they reached satiation, their weights were recorded again. These weights represent the weights at satiation. At this point, we have the field observed weights and satiation weights from each larva. Subsequently, the larvae were starved for four weeks and their weights during starvation were recorded weekly. Initially, we planned to quantify the starvation status of antlions in the field based on the weight loss profile characterized in the laboratory (Jakob et al., 1996; Bilde and Toft, 1998). However, because the majority of antlions in the field had much smaller weights than the weights after 28 days of starvation (details are described in Section 3), the extrapolation was not conducted. Although pit size is not a focus of this study, how pit size correlated with the energy state of antlions was examined with auxillary data. Body weight was used as the proxy of energy state, but larger individuals would have greater body weight; size correction was necessary (Berner, 2011). For this, body weight was
first mean-scaled (i.e., devided by the mean of a population where a population was defined by each instar group). Then, the meanscaled body weight was regressed against body size (i.e., head capsule width). The residuals were treated as the size and scale corrected energy status index (Berner et al., 2010). If antlions enlarge their pits over days (e.g., over unsuccessful prey captures) (Heinrich and Heinrich, 1984), antlions in good condition showing positive large residuals (i.e., large ratio of body weight to body size) should show smaller pit sizes (i.e., a negative correlation is expected). 3. Results 3.1. Dynamic optimization model The result of the dynamic optimization model is shown in Fig. 1. Whether an antlion should ambush or pit-forage depends on the three state variables: time, energy, and current behavioural state. Given that an antlion is already foraging with a pit, the model predicts that it should switch to ambush once acquiring enough energy. Given that an antlion is ambushing, it should switch to pit-trapping if its energy status becomes lower than some threshold level. This threshold changes with time and peaks at some time steps before the terminal time step. In other words, when the time step is very close to the end or very early, ambushing antlions are less likely to switch to pit-foraging. Similarly, trap-foraging individuals will switch to ambush when their energy states become above threshold, but the threshold increases with time. That is, trap-foraging individuals are less likely to switch to ambush near the terminal time step. In Fig. 1, two specific combinations (i.e., weak tradeoff and strong tradeoff) are shown, but the qualitative results hold for other parameter combinations. One main prediction of the model is that foragers should exhibit the pit-trapping strategy when their energy state is low. This result is clear for currently pit-trapping individuals (Fig. 1). Although for ambushing individuals, very low energy states are associated with the ambush strategy, this is mostly because individuals with low energy cannot build a pit because of the pit-building cost, z. In summary, the list of model predictions to be tested are: (1) pit-trapping antlions switch to ambushing when they capture enough prey, and (2) ambushing antlions switch to pittrapping when their energy state decreases (e.g., due to failure to capture prey). 3.2. Laboratory experiments 3.2.1. Effects of prey on pit-trapping individuals There was a positive relationship between the probability of the strategy switch and the number of prey given to them (Fig. 2). The estimated model is logit(p) = –1.944 + 0.482x where p is the probability and x is the number of prey (Binomial GLM, p-values for the intercept and the slope are 0.001 and 0.002, respectively based on the Wald test). The positive and significant treatment coefficient of 0.482 suggests that increasing one prey will increase the odds of switching to the ambush strategy by 61%. 3.2.2. Effects of prey on ambushing individuals The prey supply had a negative effect on the probability of switching to the pit-trapping strategy (Fig. 3). In other words, when prey were not given to the ambushing antlions, those antlions were more likely to switch to the pit-trapping strategy. The estimated model is logit(p) = −0.208 − 1.092x where p is the probability of switching, and x is the number of prey (Binomial GLM, p-values for the intercept and the slope are 0.536 and 0.002, respectively based on the Wald test). The negative and significant
Y.-J. Tsao, T. Okuyama / Behavioural Processes 91 (2012) 1–7
5
Fig. 1. Optimal behavioural solution obtained by the dynamic optimization model. The colum panel (ambush vs. pit-trap) shows the current behavioural strategy (QC ) of an antilion. The row panel shows parameter combinations. It shows what the optimal strategy (Q, indicated by color) for given antlion with a specific comibination of the state variables (time, energy, and behavioural states). fP and fA are foraging success for pit-trapping and ambushing, respectively. dP and dA are predation risk for pit-trapping and ambushing, respectively.
treatment coefficient of −1.092 suggests that increasing one prey will decrease the odds of switching to the pit-trapping strategy by 66%.
3.3. Field observation
3.2.3. Effects of foraging strategy on cannibalism risk Cannibalism events were rare. Among 106 experimental trials, only nine trials resulted in cannibalism events. However, given that cannibalism occurred, third instar individuals cannibalized pit-trapping individuals (eight cannibalism events) more than ambushing individuals (one cannibalism event). The result suggests that the pit-trapping strategy is more risky than the ambush strategy (Binomial test: p = 0.045).
Even when antlions were starved for four weeks in the laboratory, body weights generally did not fall below their initial capture weights in the field. The average (± SD) weights of the second instars (n = 49) and third instars (n = 46) after 28 days of starvation were 9.00 ± 2.36 mg and 23.38 ± 6.84 mg, respectively. The average (± SD) weights of the second instars and the third instars in the field were 7.96 ± 2.50 mg and 19.05 ± 6.20 mg, respectively. Of 49 second instar individuals, the field weights were lower than the weights at 28th days of starvation for 37 individuals (76%). Of 46 third instar individuals, the same was true for 40 individuals (87%). These results suggest that antlions in the field are unlikely to attain satiation.
Fig. 2. Proportion of pit-trapping individuals switched to exhibit the ambush strategy. The ratio numbers indicate the number of individuals switched the strategy (numerator) to the total number (denominator). The trend line is based on logit(p) = –1.944 + 0.482x where p is the probability of switching to the ambush strategy, and x is the number of prey given.
Fig. 3. Proportion of ambushing individuals switched to exhibit the pit-trapping strategy. The ratio numbers indicate the number of individuals switched the strategy (numerator) to the total number (denominator). The trend line is based on logit(p) = –0.208 − 1.092x where p is the probability to switch to the pit-trapping strategy, and x is the number of prey consumed.
6
Y.-J. Tsao, T. Okuyama / Behavioural Processes 91 (2012) 1–7
Fig. 4. Relationship between scale and size corrected weights and pit diameters of antlions. Black and grey points show second instar individuals and third instar individuals, respectively. The lines are the best fit straight lines to the data: second instar (y = 1.585 − 0.279x) and third instar (y = 2.524 + 1.182x). Size (represented by different point sizes indexed in the legend) is the head width of the antlion. The second instar and the third instar larvae are distinct in their sizes (the largest second instar larva is 1.03 mm and the smallest third instar larva is 1.32 mm in the samples).
The scale and size corrected weights of antlions and their pit diameters in the field were positively correlated for the third instar (Linear regression, Wald test, p < 0.001) (Fig. 4), but no significant trend was observed for the second instar (Wald test, p = 0.393). 4. Discussion In this study, theoretical and empirical approaches were used to understand when an antlion changes its foraging strategy. The model predicted that the optimal strategy is to ambush when energy status is high and to pit-trap strategy when energy status is low. The model predictions as well as an assumption of the model were confirmed by laboratory experiments. These results suggest that the model captures important dynamics of antlion foraging; antlions flexibly switch between the two strategies to maximize their fitness by balancing the cost and benefit of the strategies. In a model validation experiment, M. persimilis larvae switched its strategy from the pit-trapping strategy to the ambush strategy after consuming more prey (Fig. 2). Similar results were found in other studies (but see Segoli et al., 2004). For example, a desert burrowing spider decreased its investment in web construction after consuming enough prey items (Lubin and Henschel, 1996). Elimelech and Pinshow (2008) found that an antlion species Myrmecaelurus sp. that can adopt both the pit-trapping and ambush strategies decreased the period of maintaining a pit in response to an increase in prey encounter rate. These results suggest that the results found in this study may be general, and the similar selective pressure (e.g., risk of predation) operates on a variety of organisms with trap-foraging strategies. However, although foragers decreasing foraging effort with an increased prey intake rate may be common as discussed above, this pattern may be explained by a simpler hypothesis, e.g., foragers stop foraging after consuming enough prey because they do not need to forage more (e.g., Okuyama, 2011). In other words, predation risk is not a factor. In fact, optimal foraging models that do not consider predation risk can predict that foraging effort decreases with resource level (e.g., Abrams, 1992). There are a few observations that suggest this simple hypothesis does not apply to the current study. First, the field observation revealed that all individuals were far from satiation. Thus, it is unlikely for them to attain
satiation or near satiation in the field. Second, ambushing individuals readily attacked prey in an experiment. These observations as well as the result of the cannibalism experiment suggest that predation risk plays an important role in their behavioural decisions. A model assumption that pit-foraging individuals face higher predation risk than ambushing ones was tested experimentally. It is an issue being discussed but seldom tested experimentally (but see Loria et al., 2008). We found that the pit-trapping strategy resulted in higher cannibalism risk than the ambush strategy although cannibalism rate was generally low. In the experiment, we observed that third instar larvae stayed near the pits of pittrapping larvae and sometimes even ruined or replaced the pits. These observations suggest that third instar larvae can perceive the presence of pits, e.g., based on vibratory cues (Devetak, 1998; Devetak et al., 2007). However, we cannot conclude that whether third instar larvae attacked pit-trapping individuals more because pit-trapping individuals are easier to be detected and/or easier to be attacked (e.g., third instar larvae detected the presence of both pittrapping and ambushing individuals). Although cannibalism was considered in this study for convenience, there are many other potential natural enemies of M. persimilis larvae, which include the larvae of other antlion species (e.g., ambushing species, Distoleon littoralis; personal observation) and parasitoids of antlions (Stange et al., 2002). Although our model does not make any prediction about the diameter of pits, it has been reported that antlions enlarge pits over days (Heinrich and Heinrich, 1984). If this were the case, the field observation and the model prediction appear contradictory at least for the third instar. Based on this expectation, we should see a negative correlation between the body condition (after size correction) and their pit sizes. This is because individuals with better body condition (higher energy levels) tend to ambush more and thus should have shorter periods of pit maintenance (smaller pits). This contradictory result may be explained by the differences in their life stages. For example, the model predicts that individuals at the end of the larval period should exhibit more pit-trapping at a given energy state than earlier stages (Fig. 1). Third instar larvae may keep enlarging their pits to increase energy gain prior to pupation because higher body mass before pupation results in higher adult body size (Gotelli, 1997; Scharf et al., 2008). In fact, the same positive correlation was not observed for the second instar (Fig. 4). However, to understand results with respect to pit size, future studies need to reveal how pit size actually relates to predation risk, which we know very little and is likely to substantially vary among predators (e.g., birds vs. ground dwelling predators). In addition, starvation tolerance should be characterized. Even though antlions can survive many days without feeding, overly starved individuals may not be able to maintain large pits. Then starvation can still be a factor for the observed pattern. Factors influencing pit-size is another important axis for future studies (Scharf et al., 2011). When considering predation risk, future studies should characterize predators of antlions in the field. For example, although pit-trapping may be associated with high risk for some predators, this is not necessarily the case. For example, if predators detect prey based on mechanical cues, ambushing individuals may face higher predation risk if they move more frequently. In addition, this study did not consider potential intraspecific interactions. However, antlion pits are often aggregated (Farji-Brener, 2003). In fact, with respect to some predators, pit density would also influence predation risk (e.g., predators can more easily detect pits in aggregation than an isolated pit). Potential game-theoretical interactions among antlions should also be considered. Our study suggests that the antlion, M. persimilis, increases its fitness by balancing the cost and benefit of ambushing and pit-trapping, which results in context-dependent flexible foraging behaviour. Although the laboratory experiments and the
Y.-J. Tsao, T. Okuyama / Behavioural Processes 91 (2012) 1–7
model were clearly consistent with each other, the field data suggested other interesting patterns that call for future investigations such as stage-dependence and pit size variation. The relatively sessile nature of antlions and their variation in foraging strategy make them good subjects for studying foraging strategies. Acknowledgements We thank Drs. Wen-San Huang and Chaun-Chan Wang for their valuable comments. Dr. Inon Scharf and an anonymous reviewer also provided a number of constructive criticisms that improved the manuscript. We gratefully acknowledge the National Science Council of Taiwan’s support (97-2321-B-002036-MY2, 99-2628-B-002-051-MY3) during the completion of this project. References Abrams, P.A., 1992. Predators that benefit prey and prey that harm predators—unusual effects of interacting foraging adaptations. Am. Nat. 140, 573–600. Abrams, P.A., 2010. Implications of flexible foraging for interspecific interactions: lessons from simple models. Funct. Ecol. 24, 7–17. Anderson, R.A., Karasov, W.H., 1981. Contrasts in energy intake and expenditure in sit-and-wait and widely foraging lizards. Oecologia 49, 67–72. Balent, K.L., Andreadis, P.T., 1998. The mixed foraging strategy of Juvenile Northern water snakes. J. Herpet. 32, 575–579. Barkae, E.D., Scharf, I., Subach, A., Ovadia, O., 2010. The involvement of sand disturbance, cannibalism and intra-guild predation in competitive interactions among pit-building antlion larvae. Zoology 113, 308–315. Berner, D., 2011. Size correction in biology: how reliable are approaches based on (common) principal component analysis? Oecologia 166, 961–971. Berner, D., Stutz, W.E., Bolnick, D.I., 2010. Foraging trait (co)variances in stickleback evolve deterministically and do not predict trajectories of adaptive diversification. Evolution 64, 2265–2277. Bilde, T., Toft, S., 1998. Quantifying food limitation of arthropod predators in the field. Oecologia 115, 54–58. Bolker, B., Holyoak, M., Kˇrivan, V., Rowe, L., Schmitz, O., 2003. Connecting theoretical and empirical studies of trait-mediated interactions. Ecology 84, 1101–1114. Cain, M.L., 1987. Prey capture behavior and diel movement of Brachynemurus (Neuroptera: Myrmeleontidae) antlion larvae in south central Florida. Fl. Entomol. 70, 397–400. Caro, T.M., 2005. Antipredator Defenses in Birds and Mammals. University of Chicago Press, Chicago, IL. Devetak, D., 1998. Detection of substrate vibration in Neuropteroidea: a review. Acta Zool. Fenn. 209, 87–94. Devetak, D., 2005. Effects of larval antlions Euroleon nostras (Neuroptera, Myrmeleontidae) and their pits on the escape-time of ants. Physiol. Entomol. 30, 82–86. ˇ A., 2007. Sand Devetak, D., Mencinger-Vracko, B., Devetak, M., Marhl, M., Spernjak, as a medium for transmission of vibratory signals of prey in antlions Euroleon nostras (Neuroptera: Myrmeleontidae). Physiol. Entomol. 32, 268–274. ˇ Devetak, D., Spernjak, A., Janˇzekoviˇc, F., 2005. Substrate particle size affects pit building decision and pit size in the antlion larvae Euroleon nostras (Neuroptera: Myrmeleontidae). Physiol. Entomol. 30, 158–163. Elimelech, E., Pinshow, B., 2008. Variation in food availability influences preycapture method in antlion larvae. Ecol. Entomol. 33, 652–662. Farji-Brener, A.G., 2003. Microhabitat selection by antlion larvae, Myrmeleon crudelis: effect of soil particle size on pit-trap design and prey capture. J. Insect Behav. 16, 783–796. Fertin, A., Casas, J., 2006. Efficiency of antlion trap construction. J. Exp. Biol. 209, 3510–3515. Gelman, A., Hill, J., 2007. Data Analysis Using Regression and Multilevel/Hierarchical Models. Cambridge University Press, Cambridge. Gotelli, N.J., 1997. Competition and coexistence of larval ant lions. Ecology 78, 1761–1773. Grant, J.W.A., Noakes, D.L.G., 1987. Movers and stayers: foraging tactics of youngof-the-year brook charr, Salvelinus fontinalis. J. Anim. Ecol. 56, 1001–1013. Griffiths, D., 1980. The feeding biology of ant-lion larvae: prey capture, handling and utilization. J. Anim. Ecol. 49, 99–125. Griffiths, D., 1986. Pit construction by ant lion larvae: a cost–benefit analysis. J. Anim. Ecol. 55, 39–57. Griffiths, D., 1991. Intraspecific competition in larvae of the ant-lion Morter sp. and interspecific interactions with Macroleon quinquemaculatus. Ecol. Entomol. 16, 193–201.
7
Griffiths, D., 1992. Interference competition in ant-lion (Macroleon quinquemaculatus) larvae. Ecol. Entomol. 17, 219–226. Hauber, M.E., 1999. Variation in pit size of antlion (Myrmeleon carolinus) larvae: the importance of pit construction. Physiol. Entomol. 24, 37–40. Heinrich, B., Heinrich, M.J.E., 1984. The pit-trapping foraging strategy of the ant lion, Myrmeleon immaculatus DeGeer (Neuroptera: Myrmeleontidae). Behav. Ecol. Sociobiol. 14, 151–160. Helfman, G.S., 1990. Mode selection and mode switching in foraging animals. Adv. Study Behav. 19, 249–298. Huey, R.B., Pianka, E.R., 1981. Ecological consequences of foraging mode. Ecology 62, 991–999. Jakob, E.M., Marshall, S.D., Uetz, G.W., 1996. Estimating fitness: a comparison of body condition indices. Oikos 77, 61–67. Kats, L.B., Dill, L.M., 1998. The scent of death: chemosensory assessment of predation risk by prey animals. Ecoscience 5, 361–394. Li, D.Q., Lee, W.S., 2004. Predator-induced plasticity in web-building behaviour. Anim. Behav. 67, 309–318. Lima, S.L., 1998. Stress and decision making under the risk of predation: recent developments from behavioral, reproductive, and ecological perspectives. Adv. Study Behav. 27, 215–290. Loria, R., Scharf, I., Subach, A., Ovadia, O., 2008. The interplay between foraging mode, habitat structure, and predator presence in antlions. Behav. Ecol. Sociobiol. 62, 1185–1192. Lubin, Y., Henschel, J., 1996. The influence of food supply on foraging behaviour in a desert spider. Oecologia 105, 64–73. Lucas, J.R., 1982. The biophysics of pit construction by ant lion larvae (Myrmeleon, Neuroptera). Anim. Behav. 30, 651–664. Lucas, J.R., 1985. Metabolic rates and pit-construction costs of two antlion. J. Anim. Ecol. 54, 295–309. Mangel, M., Clark, C.W., 1988. Dynamic Modeling in Behavioral Ecology. Princeton University Press, Princeton, NJ. Matsura, T., Takano, H., 1989. Pit-relocation of antlion larvae in relation to their density. Res. Popul. Ecol. 31, 225–234. McLaughlin, R.L., 1989. Search modes of birds and lizards: evidence for alternative movement patterns. Am. Nat. 133, 654–670. McPeek, M.A., Peckarsky, B.L., 1998. Life histories and the strengths of species interactions: combining mortality, growth, and fecundity effects. Ecology 79, 867–879. Miller, R.B., Stange, L.A., Wang, H.Y., 1999. New species of antlions from Taiwan (Neuroptera: Myrmeleontidae). J. Natl. Taiwan Museum 52, 47–48. Nardi, J.B., 2007. Antlions, Life in the Soil: A Guide for Naturalists and Gardeners, Life in the Soil: A Guide for Naturalists and Gardeners. University of Chicago Press, Chicago, IL. Okuyama, T., 2011. Biphasic activity of a jumping spider. Naturwissenschaften 98, 15–22. Perry, G., 1999. The evolution of search modes: ecological versus phylogenetic perspectives. Am. Nat. 153, 98–109. Porges, S.W., Riniolo, T.C., McBride, T., Campbell, B., 2003. Heart rate and respiration in reptiles: contrasts between a sit-and-wait predator and an intensive forager. Brain Cogn. 52, 88–96. Ruxton, G.D., Hansell, M.H., 2009. Why are pitfall traps so rare in the natural world? Evol. Ecol. 23, 181–186. Scharf, I., Filin, I., Golan, M., Buchshtav, M., Subach, A., Ovadia, O., 2008. A comparison between desert and Mediterranean antlion populations: differences in life history and morphology. J. Evol. Biol. 21, 162–172. Scharf, I., Lubin, Y., Ovadia, O., 2011. Foraging decisions and behavioural flexibility in trap-building predators: a review. Biol. Rev. 86, 626–639. Scharf, I., Nulman, E., Ovadia, O., Bouskila, A., 2006. Efficiency evaluation of two competing foraging modes under different conditions. Am. Nat. 168, 350–357. Schmitz, O.J., 2010. Resolving Ecosystem Complexity. Princeton University Press, Princeton, NJ. Schwenk, K., 1993. The evolution of chemoreception in squamate reptiles: a phylogenetic approach. Brain Behav. Evol. 41, 124–137. Segoli, M., Maklakov, A., Gavish, E., Tsurim, I., Lubin, Y., 2004. The effect of previous foraging success on web-building behaviour in the sheet-web spider. Frontinellina cf. frutetorum (Araneae Linyphiidae). Ethol. Ecol. Evol. 16, 291–298. Stange, L.A., Miller, R.B., Wang, H.Y., 2002. Identification and Biology of Myrmeleontidae (Neuroptera) in Taiwan. Education Centre of Yilan Natural History, Yilan, Taiwan. Stephens, D.W., Brown, J.S., Ydenberg, R.C., 2007. Foraging: Behavior and Ecology. University of Chicago Press, Chicago, IL. Stephens, D.W., Krebs, J.R., 1986. Foraging Theory. Princeton University Press, Princeton, NJ. van Zyl, A., van der Linde, T.C.D.K., Grimbeek, R.J., 1997. Metabolic rates of pitbuilding and non-pitbuilding antlion larvae (Neuroptera: Myrmeleontidae) from southern Africa. J. Arid Environ. 37, 355–365. Vitt, L.J., Pianka, E.R. (Eds.), 1994. Lizard Ecology: Historical and Experimental Perspectives. Princeton University Press, Princeton, NJ. Wilson, D.S., 1974. Prey capture and competition in the ant lion. Biotropica 6, 187–193. Wise, D.H., 1993. Spiders in Ecological Webs. Cambridge University Press, Cambridge.
Contents lists available at SciVerse ScienceDirect
Behavioural Processes journal homepage: www.elsevier.com/locate/behavproc
Foraging strategy switching in an antlion larva Yu-Jen Tsao a,b , Toshinori Okuyama a,∗ a b
Department of Entomology, National Taiwan University, Taipei, Taiwan Division of Management, Endemic Species Research Institute, Nantou, Taiwan
a r t i c l e
i n f o
Article history: Received 30 January 2012 Received in revised form 10 March 2012 Accepted 22 April 2012 Keywords: Optimal foraging Dynamic optimization Predation risk Trap-building Ambush
a b s t r a c t Antlion larvae are typically considered as trap-building predators, but some species of antlions always forage without using pits or only sometimes use pits to capture prey; they can ambush prey without pits. This study examined a species that switches its strategy between pit-trapping and ambushing and asked the mechanism behind the switching behaviour. A dynamic optimization model incorporating tradeoffs between the two strategies was built. The tradeoffs were prey capture success and predation risk (both are higher when pit-trapping). The model predicted that antlions should use the trap-building strategy when their energy status is low and should use the ambush strategy when their energy status is high. These predictions as well as an assumption (i.e., predation risk associated with pit-trapping is higher than that associated with ambushing) of the model were empirically confirmed. The results suggest that antlions flexibly switch between pit-trapping and ambushing to maximize their fitness by balancing the costs and benefits of the two strategies. © 2012 Elsevier B.V. All rights reserved.
1. Introduction Foraging strategies vary widely among organisms. For example, sit-and-wait foraging and active (or widely) foraging may be perceived as two end points of a continuum, and their comparative economics have been widely studied in a variety of ecological contexts (Huey and Pianka, 1981; Helfman, 1990; Perry, 1999; Scharf et al., 2006). Understanding how animals make their foraging decisions is important not only for the study of behaviour (Stephens and Krebs, 1986; Stephens et al., 2007) but also for examining large scale ecological processes such as population and community level dynamics (Bolker et al., 2003; Abrams, 2010). For example, foraging modes of predators can affects ecosystem level processes (e.g., nutrient cycling) by altering adaptive strategies of their prey that propagates through ecosystems through direct and indirect interactions (Schmitz, 2010). Foraging strategies are often described in dichotomous manners. For example, many birds and lizards exhibit either sit-and-wait or widely foraging tactics (McLaughlin, 1989). In lizards, this bimodal distribution is shaped both by historical and current selection pressures (Schwenk, 1993; Vitt and Pianka, 1994). Similarly, spider species may also be categorized dichotomously into species that use or do not use webs to capture prey. In these situations, comparative studies describing general patterns
∗ Corresponding author at: Department of Entomology, National Taiwan University, No. 1, Sec. 4, Roosevelt Road, Taipei 106, Taiwan. Tel.: +886 2 3366 5282; fax: +886 2 2732 5017. E-mail address: [email protected] (T. Okuyama). 0376-6357/$ – see front matter © 2012 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.beproc.2012.04.012
associated with each strategy is useful (Anderson and Karasov, 1981; Huey and Pianka, 1981; Porges et al., 2003). However, species that use mixed strategies (e.g., exhibit both sit-and-wait and widely foraging tactics) are particularly useful for directly examining hypotheses (Balent and Andreadis, 1998). For example, with a species with a mixed strategy, theoretical predictions (e.g., conditions that selects sit-and-wait vs. widely foraging) can be directly examined by observing foraging expressions under respective conditions (Grant and Noakes, 1987). Trap-foraging organisms are ecologically important (Wise, 1993), but the ecology and evolution of trap-foraging (e.g., ecological cost) is still not well understood (see Ruxton and Hansell, 2009 for review). Although antlion larvae (Neuroptera: Myrmeleontidae) are commonly known as sit-and-wait predatory insects that capture small invertebrates in sandy environments using conical pits (Wilson, 1974; Griffiths, 1980; Lucas, 1982; Griffiths, 1986; Devetak, 2005; Devetak et al., 2005; Fertin and Casas, 2006; Nardi, 2007; Ruxton and Hansell, 2009), foraging behaviour of antlion larvae is much more diverse than the common perception. Some antlions species do not use pits to capture prey (we call this strategy ambush strategy). Furthermore, some antlion larvae exhibit a mixed strategy (switch between foraging with and without a trap) rather than specializing on one strategy (Miller et al., 1999), which is uncommon in spiders (a major taxonomic group in which some species exhibit trap-foraging). To understand the costs and benefits associated with trap-foraging, we use an antlion species with a mixed strategy. When considering why an antlion sometime forages with a pit and sometime ambushes prey, a natural starting place is a potential tradeoff between the two strategies. First, the function of pits
2
Y.-J. Tsao, T. Okuyama / Behavioural Processes 91 (2012) 1–7
implies that the pit-trapping strategy enjoys higher prey capture successes than the ambush strategy (Heinrich and Heinrich, 1984; Griffiths, 1986; Cain, 1987; van Zyl et al., 1997; Elimelech and Pinshow, 2008). On the other hand, a potential cost of pit-foraging is the metabolic cost. However, although the metabolic expenditure during pit constructions is 10 times resting metabolic rate (Lucas, 1985), there is no known difference in metabolic rate between individuals with and without pits when they are awaiting for prey once pits are constructed (van Zyl et al., 1997), suggesting that the pit maintenance cost may be negligible. Furthermore, non-pit building antlions have been suggested to have higher resting metabolic rates than pit-building antlions because pits allow them to capture prey more easily (Lucas, 1985), which would diminish the potential metabolic cost of pit-trapping strategy. Predation risk is an important factor that affects foraging behaviour of many organisms (Kats and Dill, 1998; Lima, 1998; McPeek and Peckarsky, 1998; Caro, 2005; Ruxton and Hansell, 2009). With respect to trap foragers, orb-weaving spiders decrease their web sizes in the presence of predators to reduce predation risk (Li and Lee, 2004). In antlions, predation is also hypothesized as a selection factor against pit-trapping (Ruxton and Hansell, 2009). Hauber (1999) discusses that Florida scrub-jays visually locate Myrmeleon carolinus from pits and preferentially prey on larvae with large pits. Loria et al. (2008) found that antlions decreased pit-building activity when potential predators were present. Despite this evidence, there are few studies that explicitly examined the effect of predation risk on the foraging activity of antlions (Scharf et al., 2011). In our question, it is also clear that predation risk cannot fully explain the difference in foraging strategy because antlion larvae exhibit large behavioural variation, e.g., under an enemy free environment in the laboratory condition and within the same area in the field (personal observation). In other words, some individuals forage with pits while others ambush even in the same environment. Thus, although predation risk appears to be an important factor in antlions’ decision to switch their foraging strategies, the behavioural rule is not well understood. In this study, we examined the mechanism in which the antlion larva Myrmeleon persimilis switches its foraging strategies. The key factor we focused was the tradeoff between the pit-trapping and ambush strategies (discussed above) where the pit-trapping strategy results in a higher prey intake rate but also faces higher predation risk (i.e., a tradeoff between energy intake and predation risk). We present a hypothetical mechanism describing how antlions switch between the two foraging strategies using an optimal foraging model whose predictions and an assumption (predation risk of pit-trapping is higher than that of ambushing) were empirically validated. As will be discussed below, because the results imply that the physiological status of the antlion is an important factor for its behavioural expression, supplementary field observations were also conducted to evaluate the natural physiological conditions of the antlion.
2. Methods 2.1. Study animals The antlion species used in this study was Myrmeleon persimilis (Neuroptera: Myrmeleontidae), an endemic and the most widespread coastal species of genus Myrmeleon in Taiwan (Stange et al., 2002). M. persimilis larva, like other antlion species, has three instar stages and usually eclose in February to May (unpublished data). M. persimilis larvae were collected in a coastal habitat in Shimen District, New Taipei City (25◦ 18 N,
121◦ 32 E). It is a sandy region with abundant rainfall. When it rains, M. persimilis larvae burrow a few centimeters beneath and come out when the sand is dried. Common plant species in the habitat include wormwoods (Artemisia capillaries), tree heliotropes (Tournefortia argentea), Hosobawadan (Crepidiastrum lanceolatum), and Indian Blankets (Gaillardia pulchella) (personal observations). Pit-trapping individuals of the species abruptly abandon their pits and assume an ambush strategy rather than gradually reducing the diameter of pits. Thus, instead of considering the pit-trapping and ambush strategies as a continuum of the same strategy (i.e., pit-diameter), they were considered as distinct strategies in this study. 2.2. Dynamic optimization model A dynamic optimization model was constructed to examine how an antlion with the mixed strategy chooses its foraging expression based on its state variables (i.e., energy level, time, and the current strategy) to maximize its fitness. The model considers a situation where an antlion larva aims to survive for a specific time duration T (e.g., the larval stage). In each time step (discussed in detail below), a larva makes a decision about its foraging strategy. There are two behavioural options: pit-trapping and ambushing. When the larva adopts the pit-trapping strategy, it enjoys greater prey capturing success (Griffiths, 1991; Nardi, 2007; Elimelech and Pinshow, 2008) but also suffers higher predation risk (Hauber, 1999) with respect to the ambush strategy. Parameters associated with each strategy are fQ (foraging success), dQ (predation risk), cQ (metabolic cost) and z (pit construction cost). The subscript (Q = P or A) describes the behavioural option. For example, fP and fA are the probability of foraging success for an individual with the pit-trapping strategy and with the ambush strategy, respectively. A larva is associated with three state variables: energy state x, time t, and behavioural states, QC . (Q is the behavioural option to be chosen while QC is the currently exhibiting strategy.) The energy state varies from 1 to xmax , and the time step varies from 1 to T (both take integer values only). Suppose, at time t, a larva whose energy state is x chooses the behavioural option Q, its expected fitness is WQ = (1 − dQ )[fQ F(x + Y − cQ − z, Q, t + 1) + (1 − fQ )F(x − cQ , Q, t + 1)] where F(x, t) is the maximum expected fitness of a larva at time t whose energy state is x. (How to obtain F is discussed below.) If the larva is predated (the probability is dQ ), its fitness is zero. If the larva survives one time step (the probability is 1 − dQ ), it may forage successfully (the probability is fQ ) or fail to capture a prey (the probability is 1 − fQ ). The larva which forages successfully improves its energy state by Y (prey profitability). The pit-construction cost z depends on the current behavioural state QC while all other parameters are independent of QC . In particular, z takes a positive value only when the current behavioural state is ambush (QC = A) and the antlion decides to build a pit (Q = P). In all other behavioural sequences, z = 0, because z is the cost associated with building a pit. The model predicts that the option P (pit-trapping) is the optimal strategy if WP > WA ; the option A (ambushing) is the optimal strategy when WA > WP . It has been reported that the pit-construction cost is 10 times of the resting metabolic rate (e.g., z = 10cQ ) (Lucas, 1985), but the relationship between cA and cP are not well known. As discussed above, previous studies have argued that metabolic rate should be lower for individuals with pits (i.e., cA > cP ) because they can capture prey more easily (Lucas, 1985) while others may argue for the opposite because there may be a pit-maintenance cost. In
Y.-J. Tsao, T. Okuyama / Behavioural Processes 91 (2012) 1–7
this study, we assume the difference is negligible (i.e., cA = cP ) and set these values 1. This determines z = 10. Y depends on types of prey. It is worth mentioning that the metabolic rate of antlions is higher after they capture a prey (van Zyl et al., 1997). Thus, Y should be discounted for the metabolic cost, but it will not affect this analysis. We are interested in how the ecological tradeoff influences the behavioural decision making in antlions. Thus dP > dA (i.e., pit-trapping experiences higher predation risk) and fP > fA (i.e., pittrapping experiences higher foraging success) were assumed. To analyze the model, we need to determine four more parameter values: dQ , fQ (Q = P or A), and the state conditions: xmax and T. In addition, the terminal fitness must be decided (Mangel and Clark, 1988). The terminal fitness describes the fitness of an individual with energy level x at the last time step. For the results we describe below, the terminal fitness is assumed to be the energy state x [i.e., F(x,Q,T) = x]. In other words, the greater the energy state, the greater the fitness is. However, changing the detail (e.g., the fitness increases acceleratingly or deceleratingly with the energy state) does not alter the qualitative conclusion discussed below. We set xmax = 200 and T = 50, but the qualitative results are also robust to these settings (details are mentioned in Section 3). Once the details are specified, the optimal solutions of the model can be derived using the backward iteration method (Mangel and Clark, 1988). To validate that results are not specific to the particular parameter values, the tradeoff parameters were varied in their values. The probability of death due to predation for pit-trapping individuals was set to dA = 0.01 (1% death per time step), and different degrees of the tradeoff were examined (dP = 0.02,0.04,0.06), which encompasses the odds ratio from 2.02 to 6.32. The foraging success probability for the ambush was set to fA = 0.1 (10% foraging success per time step), and different degrees of the tradeoff was examined (fP = 0.2,0.4,0.6), which encompasses the odds ratio from 2.25 to 13.5. The model parameters are summarized in Table 1. 2.3. Laboratory experiments As will be described below, the dynamic optimization model predicts that the optimal strategy is the pit-trapping strategy when energy status is low (e.g., starved) and is the ambush strategy when energy status is high. Laboratory experiments were conducted to examine whether antlions forage according to the model predictions. In addition, an assumption of the model (i.e., dP > dA , the probability of death due to predation is higher for the pit-trapping strategy) was also tested to strengthen the validity of the model. The average (± SD) weight of the prey larva (Tribolium confusum) used in the experiments was 2.637 ± 0.38 mg. Each individual was used only once in the experiments. 2.3.1. Effects of prey on pit-trapping individuals In this experiment, whether pit-trapping antlions change their strategy to the ambush strategy when their energy states improve Table 1 Parameter definitions and their values.
Prey capture success (pit-trap) Prey capture success (ambush) Predation risk (pit-trap) Predation risk (ambush) Metabolism (pit-trap) Metabolism (ambush) Prey profitability Pit construction cost (ambush)
Symbol
Value
fP fA dP dA cP cA Y z
0.2,0.4,0.6 0.1 0.02,0.04,0.06 0.01 1 1 20 10
3
was examined. The ambush strategy was defined as hiding under sand surface with only mandibles exposing outside, sometimes with a shallow cavity (Heinrich and Heinrich, 1984). The pittrapping strategy was defined as having a conical pitfall trap with the diameter greater than 1 cm. Pit-trapping individuals usually stayed at the bottom of pits, opening their mandibles and waiting for falling prey. If antlions behave according to the model prediction (see Section 3 for details), pit-trapping individuals should switch to ambush once they acquire enough prey (i.e., an experimental treatment). Antlions used in the experiment were individually housed in containers filled with sands (44.5 cm by 34.7 cm, sand depth 4 cm) that were maintained in a controlled environment: temperature (25 ± 2 ◦ C), relative humidity (40 ± 5%), and L:D 14:10 cycle. Second instar antlions (n = 37) were fed with flour beetle larvae, Tribolium confusum, until they stopped accepting a new prey. Subsequently they were starved for four weeks to standardize the starvation level of the study subjects. At this point, all individuals were pit-trapping. The starved larvae were randomly allocated to three treatment levels: 0, 1, or 6 prey items. The low treatment levels (i.e., 1 and 0 prey) were intended to affect the satiation levels of the larvae only little to none, while the high prey level (i.e., 6 prey) was intended to increase the energy level substantially. Prey larvae were placed near the edge of pits in such a way that they walked into the pits by themselves. In the multiple prey treatment level, one prey was given at a time. Whether the pit-trapping individuals switched their strategies to the ambush strategy was recorded after 24 h. 2.3.2. Effects of prey on ambushing individuals The purpose of this experiment was to test another prediction of the model; ambushing individuals change their foraging strategy to the pit-trapping strategy in response to a shortage of prey (i.e., low energy status in the model). In other words, if antlions behave according to the model prediction, ambushing individuals should switch to pit-foraging as they experience prey shortage (of an experimental treatment). The study organisms were maintained in the same way (described above). To conduct the experiment, we needed to prepare antlions exhibiting the ambush strategy. Because the first experiment described above showed that an abundant prey supply makes them exhibit the ambush strategy (see Section 3 for details), second instar antlions were fed until exhibiting the ambush strategy. Ambushing individuals were randomly allocated to two treatment levels: no prey (n = 35), two prey items (n = 35) (T. confusum larvae were used as the prey). Unlike the previous experiment, the ambushing larvae were generally more satiated so that large treatment levels could not be established. Furthermore, in the two prey treatment level, three (of the 35 replications) antlions ate only one larvae even though two larvae were provided (i.e., rejected the second prey item). Thus for the data analysis, these three replications were considered as a new treatment level (i.e., one prey). Subsequently, whether the ambushing antlions switched their strategies to the pit-trapping strategy was recorded after 24 h. 2.3.3. Effects of foraging strategy on cannibalism risk The purpose of this experiment was to test an assumption of the model; the probability of death due to predation is higher with the pit-trapping strategy than the ambushing strategy. Previous studies have shown that smaller antlion larvae were more likely to be cannibalized by larger larvae especially when the individual density was high (Matsura and Takano, 1989; Griffiths, 1992; Gotelli, 1997; Barkae et al., 2010). Therefore, in this experiment, starved third instar antlions (starved more than one month) were used as predators.
4
Y.-J. Tsao, T. Okuyama / Behavioural Processes 91 (2012) 1–7
In a plastic cup (diameter: 8.4 cm; sand depth: 3.5 cm), three equal pie-shaped areas were created with plastic plate partitions. One trap-building second instar larva, one ambushing second instar larva, and one third instar larva were randomly introduced: one in each partition. Second instar larvae were introduced first. The foraging strategy of the second instar larvae was manipulated by controlling their starvation status as described above. After each larva assumed the respective foraging strategy, a third instar larva was introduced, and the partitions were removed. Antlions were observed every 30 min for 270 min. If cannibalism did not occur within 270 min, the experimental trial was checked again after 24 h. If cannibalism still did not occur within 48 h, then the experiment was terminated regardless of the outcome. If the third instar larva prefers to attack pit-trapping individuals, it confirms the model assumption, i.e., pit-trapping strategy is associated with higher predation risk. 2.3.4. Data analysis The effects of prey on pit-trapping and ambushing individuals were tested with Binomial Generalized Linear Model (GLM) where the probability of switching to the other strategy (either pit-trapping or ambush strategy) was modeled. The predictor variable was the number of consumed prey item (i.e., treatment), and whether a larva switched to another strategy was the binary response variable. The Binomial test was used to test whether third instar larvae preferentially forage on trap-building or ambushing second instar larvae in the cannibalism experiment. The data were checked for overdispersion (Gelman and Hill, 2007), and no data were found overdispersed. 2.4. Field observation The purpose of this field observation was to describe the starvation status of antlions in the field because energy status is considered as an important factor in the model and its predictions. All field observations were conducted between 0900 h and 1800 h. When we observed a pit in the field, the diameter of the pit was recorded with a caliper. Then, the larva was dug out, and its weight and head width were recorded in the laboratory within a few hours of collection. These measurements represent the conditions of larvae in the field. Subsequently, larvae were kept in plastic cups (5.2 cm in diameter which is enough for larvae in each stage to construct a pit) filled with sand collected from the same habitat and were maintained in a controlled room as described above. Immediately after the initial measurement, each larva was fed flour beetle larvae (T. confusum) until it stopped consuming them. This procedure usually took 7 h, and antlion larvae consumed 3–6 flour beetle larvae to attain satiation. After they reached satiation, their weights were recorded again. These weights represent the weights at satiation. At this point, we have the field observed weights and satiation weights from each larva. Subsequently, the larvae were starved for four weeks and their weights during starvation were recorded weekly. Initially, we planned to quantify the starvation status of antlions in the field based on the weight loss profile characterized in the laboratory (Jakob et al., 1996; Bilde and Toft, 1998). However, because the majority of antlions in the field had much smaller weights than the weights after 28 days of starvation (details are described in Section 3), the extrapolation was not conducted. Although pit size is not a focus of this study, how pit size correlated with the energy state of antlions was examined with auxillary data. Body weight was used as the proxy of energy state, but larger individuals would have greater body weight; size correction was necessary (Berner, 2011). For this, body weight was
first mean-scaled (i.e., devided by the mean of a population where a population was defined by each instar group). Then, the meanscaled body weight was regressed against body size (i.e., head capsule width). The residuals were treated as the size and scale corrected energy status index (Berner et al., 2010). If antlions enlarge their pits over days (e.g., over unsuccessful prey captures) (Heinrich and Heinrich, 1984), antlions in good condition showing positive large residuals (i.e., large ratio of body weight to body size) should show smaller pit sizes (i.e., a negative correlation is expected). 3. Results 3.1. Dynamic optimization model The result of the dynamic optimization model is shown in Fig. 1. Whether an antlion should ambush or pit-forage depends on the three state variables: time, energy, and current behavioural state. Given that an antlion is already foraging with a pit, the model predicts that it should switch to ambush once acquiring enough energy. Given that an antlion is ambushing, it should switch to pit-trapping if its energy status becomes lower than some threshold level. This threshold changes with time and peaks at some time steps before the terminal time step. In other words, when the time step is very close to the end or very early, ambushing antlions are less likely to switch to pit-foraging. Similarly, trap-foraging individuals will switch to ambush when their energy states become above threshold, but the threshold increases with time. That is, trap-foraging individuals are less likely to switch to ambush near the terminal time step. In Fig. 1, two specific combinations (i.e., weak tradeoff and strong tradeoff) are shown, but the qualitative results hold for other parameter combinations. One main prediction of the model is that foragers should exhibit the pit-trapping strategy when their energy state is low. This result is clear for currently pit-trapping individuals (Fig. 1). Although for ambushing individuals, very low energy states are associated with the ambush strategy, this is mostly because individuals with low energy cannot build a pit because of the pit-building cost, z. In summary, the list of model predictions to be tested are: (1) pit-trapping antlions switch to ambushing when they capture enough prey, and (2) ambushing antlions switch to pittrapping when their energy state decreases (e.g., due to failure to capture prey). 3.2. Laboratory experiments 3.2.1. Effects of prey on pit-trapping individuals There was a positive relationship between the probability of the strategy switch and the number of prey given to them (Fig. 2). The estimated model is logit(p) = –1.944 + 0.482x where p is the probability and x is the number of prey (Binomial GLM, p-values for the intercept and the slope are 0.001 and 0.002, respectively based on the Wald test). The positive and significant treatment coefficient of 0.482 suggests that increasing one prey will increase the odds of switching to the ambush strategy by 61%. 3.2.2. Effects of prey on ambushing individuals The prey supply had a negative effect on the probability of switching to the pit-trapping strategy (Fig. 3). In other words, when prey were not given to the ambushing antlions, those antlions were more likely to switch to the pit-trapping strategy. The estimated model is logit(p) = −0.208 − 1.092x where p is the probability of switching, and x is the number of prey (Binomial GLM, p-values for the intercept and the slope are 0.536 and 0.002, respectively based on the Wald test). The negative and significant
Y.-J. Tsao, T. Okuyama / Behavioural Processes 91 (2012) 1–7
5
Fig. 1. Optimal behavioural solution obtained by the dynamic optimization model. The colum panel (ambush vs. pit-trap) shows the current behavioural strategy (QC ) of an antilion. The row panel shows parameter combinations. It shows what the optimal strategy (Q, indicated by color) for given antlion with a specific comibination of the state variables (time, energy, and behavioural states). fP and fA are foraging success for pit-trapping and ambushing, respectively. dP and dA are predation risk for pit-trapping and ambushing, respectively.
treatment coefficient of −1.092 suggests that increasing one prey will decrease the odds of switching to the pit-trapping strategy by 66%.
3.3. Field observation
3.2.3. Effects of foraging strategy on cannibalism risk Cannibalism events were rare. Among 106 experimental trials, only nine trials resulted in cannibalism events. However, given that cannibalism occurred, third instar individuals cannibalized pit-trapping individuals (eight cannibalism events) more than ambushing individuals (one cannibalism event). The result suggests that the pit-trapping strategy is more risky than the ambush strategy (Binomial test: p = 0.045).
Even when antlions were starved for four weeks in the laboratory, body weights generally did not fall below their initial capture weights in the field. The average (± SD) weights of the second instars (n = 49) and third instars (n = 46) after 28 days of starvation were 9.00 ± 2.36 mg and 23.38 ± 6.84 mg, respectively. The average (± SD) weights of the second instars and the third instars in the field were 7.96 ± 2.50 mg and 19.05 ± 6.20 mg, respectively. Of 49 second instar individuals, the field weights were lower than the weights at 28th days of starvation for 37 individuals (76%). Of 46 third instar individuals, the same was true for 40 individuals (87%). These results suggest that antlions in the field are unlikely to attain satiation.
Fig. 2. Proportion of pit-trapping individuals switched to exhibit the ambush strategy. The ratio numbers indicate the number of individuals switched the strategy (numerator) to the total number (denominator). The trend line is based on logit(p) = –1.944 + 0.482x where p is the probability of switching to the ambush strategy, and x is the number of prey given.
Fig. 3. Proportion of ambushing individuals switched to exhibit the pit-trapping strategy. The ratio numbers indicate the number of individuals switched the strategy (numerator) to the total number (denominator). The trend line is based on logit(p) = –0.208 − 1.092x where p is the probability to switch to the pit-trapping strategy, and x is the number of prey consumed.
6
Y.-J. Tsao, T. Okuyama / Behavioural Processes 91 (2012) 1–7
Fig. 4. Relationship between scale and size corrected weights and pit diameters of antlions. Black and grey points show second instar individuals and third instar individuals, respectively. The lines are the best fit straight lines to the data: second instar (y = 1.585 − 0.279x) and third instar (y = 2.524 + 1.182x). Size (represented by different point sizes indexed in the legend) is the head width of the antlion. The second instar and the third instar larvae are distinct in their sizes (the largest second instar larva is 1.03 mm and the smallest third instar larva is 1.32 mm in the samples).
The scale and size corrected weights of antlions and their pit diameters in the field were positively correlated for the third instar (Linear regression, Wald test, p < 0.001) (Fig. 4), but no significant trend was observed for the second instar (Wald test, p = 0.393). 4. Discussion In this study, theoretical and empirical approaches were used to understand when an antlion changes its foraging strategy. The model predicted that the optimal strategy is to ambush when energy status is high and to pit-trap strategy when energy status is low. The model predictions as well as an assumption of the model were confirmed by laboratory experiments. These results suggest that the model captures important dynamics of antlion foraging; antlions flexibly switch between the two strategies to maximize their fitness by balancing the cost and benefit of the strategies. In a model validation experiment, M. persimilis larvae switched its strategy from the pit-trapping strategy to the ambush strategy after consuming more prey (Fig. 2). Similar results were found in other studies (but see Segoli et al., 2004). For example, a desert burrowing spider decreased its investment in web construction after consuming enough prey items (Lubin and Henschel, 1996). Elimelech and Pinshow (2008) found that an antlion species Myrmecaelurus sp. that can adopt both the pit-trapping and ambush strategies decreased the period of maintaining a pit in response to an increase in prey encounter rate. These results suggest that the results found in this study may be general, and the similar selective pressure (e.g., risk of predation) operates on a variety of organisms with trap-foraging strategies. However, although foragers decreasing foraging effort with an increased prey intake rate may be common as discussed above, this pattern may be explained by a simpler hypothesis, e.g., foragers stop foraging after consuming enough prey because they do not need to forage more (e.g., Okuyama, 2011). In other words, predation risk is not a factor. In fact, optimal foraging models that do not consider predation risk can predict that foraging effort decreases with resource level (e.g., Abrams, 1992). There are a few observations that suggest this simple hypothesis does not apply to the current study. First, the field observation revealed that all individuals were far from satiation. Thus, it is unlikely for them to attain
satiation or near satiation in the field. Second, ambushing individuals readily attacked prey in an experiment. These observations as well as the result of the cannibalism experiment suggest that predation risk plays an important role in their behavioural decisions. A model assumption that pit-foraging individuals face higher predation risk than ambushing ones was tested experimentally. It is an issue being discussed but seldom tested experimentally (but see Loria et al., 2008). We found that the pit-trapping strategy resulted in higher cannibalism risk than the ambush strategy although cannibalism rate was generally low. In the experiment, we observed that third instar larvae stayed near the pits of pittrapping larvae and sometimes even ruined or replaced the pits. These observations suggest that third instar larvae can perceive the presence of pits, e.g., based on vibratory cues (Devetak, 1998; Devetak et al., 2007). However, we cannot conclude that whether third instar larvae attacked pit-trapping individuals more because pit-trapping individuals are easier to be detected and/or easier to be attacked (e.g., third instar larvae detected the presence of both pittrapping and ambushing individuals). Although cannibalism was considered in this study for convenience, there are many other potential natural enemies of M. persimilis larvae, which include the larvae of other antlion species (e.g., ambushing species, Distoleon littoralis; personal observation) and parasitoids of antlions (Stange et al., 2002). Although our model does not make any prediction about the diameter of pits, it has been reported that antlions enlarge pits over days (Heinrich and Heinrich, 1984). If this were the case, the field observation and the model prediction appear contradictory at least for the third instar. Based on this expectation, we should see a negative correlation between the body condition (after size correction) and their pit sizes. This is because individuals with better body condition (higher energy levels) tend to ambush more and thus should have shorter periods of pit maintenance (smaller pits). This contradictory result may be explained by the differences in their life stages. For example, the model predicts that individuals at the end of the larval period should exhibit more pit-trapping at a given energy state than earlier stages (Fig. 1). Third instar larvae may keep enlarging their pits to increase energy gain prior to pupation because higher body mass before pupation results in higher adult body size (Gotelli, 1997; Scharf et al., 2008). In fact, the same positive correlation was not observed for the second instar (Fig. 4). However, to understand results with respect to pit size, future studies need to reveal how pit size actually relates to predation risk, which we know very little and is likely to substantially vary among predators (e.g., birds vs. ground dwelling predators). In addition, starvation tolerance should be characterized. Even though antlions can survive many days without feeding, overly starved individuals may not be able to maintain large pits. Then starvation can still be a factor for the observed pattern. Factors influencing pit-size is another important axis for future studies (Scharf et al., 2011). When considering predation risk, future studies should characterize predators of antlions in the field. For example, although pit-trapping may be associated with high risk for some predators, this is not necessarily the case. For example, if predators detect prey based on mechanical cues, ambushing individuals may face higher predation risk if they move more frequently. In addition, this study did not consider potential intraspecific interactions. However, antlion pits are often aggregated (Farji-Brener, 2003). In fact, with respect to some predators, pit density would also influence predation risk (e.g., predators can more easily detect pits in aggregation than an isolated pit). Potential game-theoretical interactions among antlions should also be considered. Our study suggests that the antlion, M. persimilis, increases its fitness by balancing the cost and benefit of ambushing and pit-trapping, which results in context-dependent flexible foraging behaviour. Although the laboratory experiments and the
Y.-J. Tsao, T. Okuyama / Behavioural Processes 91 (2012) 1–7
model were clearly consistent with each other, the field data suggested other interesting patterns that call for future investigations such as stage-dependence and pit size variation. The relatively sessile nature of antlions and their variation in foraging strategy make them good subjects for studying foraging strategies. Acknowledgements We thank Drs. Wen-San Huang and Chaun-Chan Wang for their valuable comments. Dr. Inon Scharf and an anonymous reviewer also provided a number of constructive criticisms that improved the manuscript. We gratefully acknowledge the National Science Council of Taiwan’s support (97-2321-B-002036-MY2, 99-2628-B-002-051-MY3) during the completion of this project. References Abrams, P.A., 1992. Predators that benefit prey and prey that harm predators—unusual effects of interacting foraging adaptations. Am. Nat. 140, 573–600. Abrams, P.A., 2010. Implications of flexible foraging for interspecific interactions: lessons from simple models. Funct. Ecol. 24, 7–17. Anderson, R.A., Karasov, W.H., 1981. Contrasts in energy intake and expenditure in sit-and-wait and widely foraging lizards. Oecologia 49, 67–72. Balent, K.L., Andreadis, P.T., 1998. The mixed foraging strategy of Juvenile Northern water snakes. J. Herpet. 32, 575–579. Barkae, E.D., Scharf, I., Subach, A., Ovadia, O., 2010. The involvement of sand disturbance, cannibalism and intra-guild predation in competitive interactions among pit-building antlion larvae. Zoology 113, 308–315. Berner, D., 2011. Size correction in biology: how reliable are approaches based on (common) principal component analysis? Oecologia 166, 961–971. Berner, D., Stutz, W.E., Bolnick, D.I., 2010. Foraging trait (co)variances in stickleback evolve deterministically and do not predict trajectories of adaptive diversification. Evolution 64, 2265–2277. Bilde, T., Toft, S., 1998. Quantifying food limitation of arthropod predators in the field. Oecologia 115, 54–58. Bolker, B., Holyoak, M., Kˇrivan, V., Rowe, L., Schmitz, O., 2003. Connecting theoretical and empirical studies of trait-mediated interactions. Ecology 84, 1101–1114. Cain, M.L., 1987. Prey capture behavior and diel movement of Brachynemurus (Neuroptera: Myrmeleontidae) antlion larvae in south central Florida. Fl. Entomol. 70, 397–400. Caro, T.M., 2005. Antipredator Defenses in Birds and Mammals. University of Chicago Press, Chicago, IL. Devetak, D., 1998. Detection of substrate vibration in Neuropteroidea: a review. Acta Zool. Fenn. 209, 87–94. Devetak, D., 2005. Effects of larval antlions Euroleon nostras (Neuroptera, Myrmeleontidae) and their pits on the escape-time of ants. Physiol. Entomol. 30, 82–86. ˇ A., 2007. Sand Devetak, D., Mencinger-Vracko, B., Devetak, M., Marhl, M., Spernjak, as a medium for transmission of vibratory signals of prey in antlions Euroleon nostras (Neuroptera: Myrmeleontidae). Physiol. Entomol. 32, 268–274. ˇ Devetak, D., Spernjak, A., Janˇzekoviˇc, F., 2005. Substrate particle size affects pit building decision and pit size in the antlion larvae Euroleon nostras (Neuroptera: Myrmeleontidae). Physiol. Entomol. 30, 158–163. Elimelech, E., Pinshow, B., 2008. Variation in food availability influences preycapture method in antlion larvae. Ecol. Entomol. 33, 652–662. Farji-Brener, A.G., 2003. Microhabitat selection by antlion larvae, Myrmeleon crudelis: effect of soil particle size on pit-trap design and prey capture. J. Insect Behav. 16, 783–796. Fertin, A., Casas, J., 2006. Efficiency of antlion trap construction. J. Exp. Biol. 209, 3510–3515. Gelman, A., Hill, J., 2007. Data Analysis Using Regression and Multilevel/Hierarchical Models. Cambridge University Press, Cambridge. Gotelli, N.J., 1997. Competition and coexistence of larval ant lions. Ecology 78, 1761–1773. Grant, J.W.A., Noakes, D.L.G., 1987. Movers and stayers: foraging tactics of youngof-the-year brook charr, Salvelinus fontinalis. J. Anim. Ecol. 56, 1001–1013. Griffiths, D., 1980. The feeding biology of ant-lion larvae: prey capture, handling and utilization. J. Anim. Ecol. 49, 99–125. Griffiths, D., 1986. Pit construction by ant lion larvae: a cost–benefit analysis. J. Anim. Ecol. 55, 39–57. Griffiths, D., 1991. Intraspecific competition in larvae of the ant-lion Morter sp. and interspecific interactions with Macroleon quinquemaculatus. Ecol. Entomol. 16, 193–201.
7
Griffiths, D., 1992. Interference competition in ant-lion (Macroleon quinquemaculatus) larvae. Ecol. Entomol. 17, 219–226. Hauber, M.E., 1999. Variation in pit size of antlion (Myrmeleon carolinus) larvae: the importance of pit construction. Physiol. Entomol. 24, 37–40. Heinrich, B., Heinrich, M.J.E., 1984. The pit-trapping foraging strategy of the ant lion, Myrmeleon immaculatus DeGeer (Neuroptera: Myrmeleontidae). Behav. Ecol. Sociobiol. 14, 151–160. Helfman, G.S., 1990. Mode selection and mode switching in foraging animals. Adv. Study Behav. 19, 249–298. Huey, R.B., Pianka, E.R., 1981. Ecological consequences of foraging mode. Ecology 62, 991–999. Jakob, E.M., Marshall, S.D., Uetz, G.W., 1996. Estimating fitness: a comparison of body condition indices. Oikos 77, 61–67. Kats, L.B., Dill, L.M., 1998. The scent of death: chemosensory assessment of predation risk by prey animals. Ecoscience 5, 361–394. Li, D.Q., Lee, W.S., 2004. Predator-induced plasticity in web-building behaviour. Anim. Behav. 67, 309–318. Lima, S.L., 1998. Stress and decision making under the risk of predation: recent developments from behavioral, reproductive, and ecological perspectives. Adv. Study Behav. 27, 215–290. Loria, R., Scharf, I., Subach, A., Ovadia, O., 2008. The interplay between foraging mode, habitat structure, and predator presence in antlions. Behav. Ecol. Sociobiol. 62, 1185–1192. Lubin, Y., Henschel, J., 1996. The influence of food supply on foraging behaviour in a desert spider. Oecologia 105, 64–73. Lucas, J.R., 1982. The biophysics of pit construction by ant lion larvae (Myrmeleon, Neuroptera). Anim. Behav. 30, 651–664. Lucas, J.R., 1985. Metabolic rates and pit-construction costs of two antlion. J. Anim. Ecol. 54, 295–309. Mangel, M., Clark, C.W., 1988. Dynamic Modeling in Behavioral Ecology. Princeton University Press, Princeton, NJ. Matsura, T., Takano, H., 1989. Pit-relocation of antlion larvae in relation to their density. Res. Popul. Ecol. 31, 225–234. McLaughlin, R.L., 1989. Search modes of birds and lizards: evidence for alternative movement patterns. Am. Nat. 133, 654–670. McPeek, M.A., Peckarsky, B.L., 1998. Life histories and the strengths of species interactions: combining mortality, growth, and fecundity effects. Ecology 79, 867–879. Miller, R.B., Stange, L.A., Wang, H.Y., 1999. New species of antlions from Taiwan (Neuroptera: Myrmeleontidae). J. Natl. Taiwan Museum 52, 47–48. Nardi, J.B., 2007. Antlions, Life in the Soil: A Guide for Naturalists and Gardeners, Life in the Soil: A Guide for Naturalists and Gardeners. University of Chicago Press, Chicago, IL. Okuyama, T., 2011. Biphasic activity of a jumping spider. Naturwissenschaften 98, 15–22. Perry, G., 1999. The evolution of search modes: ecological versus phylogenetic perspectives. Am. Nat. 153, 98–109. Porges, S.W., Riniolo, T.C., McBride, T., Campbell, B., 2003. Heart rate and respiration in reptiles: contrasts between a sit-and-wait predator and an intensive forager. Brain Cogn. 52, 88–96. Ruxton, G.D., Hansell, M.H., 2009. Why are pitfall traps so rare in the natural world? Evol. Ecol. 23, 181–186. Scharf, I., Filin, I., Golan, M., Buchshtav, M., Subach, A., Ovadia, O., 2008. A comparison between desert and Mediterranean antlion populations: differences in life history and morphology. J. Evol. Biol. 21, 162–172. Scharf, I., Lubin, Y., Ovadia, O., 2011. Foraging decisions and behavioural flexibility in trap-building predators: a review. Biol. Rev. 86, 626–639. Scharf, I., Nulman, E., Ovadia, O., Bouskila, A., 2006. Efficiency evaluation of two competing foraging modes under different conditions. Am. Nat. 168, 350–357. Schmitz, O.J., 2010. Resolving Ecosystem Complexity. Princeton University Press, Princeton, NJ. Schwenk, K., 1993. The evolution of chemoreception in squamate reptiles: a phylogenetic approach. Brain Behav. Evol. 41, 124–137. Segoli, M., Maklakov, A., Gavish, E., Tsurim, I., Lubin, Y., 2004. The effect of previous foraging success on web-building behaviour in the sheet-web spider. Frontinellina cf. frutetorum (Araneae Linyphiidae). Ethol. Ecol. Evol. 16, 291–298. Stange, L.A., Miller, R.B., Wang, H.Y., 2002. Identification and Biology of Myrmeleontidae (Neuroptera) in Taiwan. Education Centre of Yilan Natural History, Yilan, Taiwan. Stephens, D.W., Brown, J.S., Ydenberg, R.C., 2007. Foraging: Behavior and Ecology. University of Chicago Press, Chicago, IL. Stephens, D.W., Krebs, J.R., 1986. Foraging Theory. Princeton University Press, Princeton, NJ. van Zyl, A., van der Linde, T.C.D.K., Grimbeek, R.J., 1997. Metabolic rates of pitbuilding and non-pitbuilding antlion larvae (Neuroptera: Myrmeleontidae) from southern Africa. J. Arid Environ. 37, 355–365. Vitt, L.J., Pianka, E.R. (Eds.), 1994. Lizard Ecology: Historical and Experimental Perspectives. Princeton University Press, Princeton, NJ. Wilson, D.S., 1974. Prey capture and competition in the ant lion. Biotropica 6, 187–193. Wise, D.H., 1993. Spiders in Ecological Webs. Cambridge University Press, Cambridge.