Behavioral Ecology Advance Access originally published online on October 12, 2005
Behavioral Ecology 2006 17(1):34-40; doi:10.1093/beheco/ari094
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Jumping spiders attend to context during learned avoidance of aposematic prey
a Neuroscience and Behavior Program, Department of Psychology, Tobin Hall, 135 Hicks Way, University of Massachusetts, Amherst, MA 01003, USA and b Department of Psychology, Tobin Hall, 135 Hicks Way, University of Massachusetts, Amherst, MA 01003, USA
Address correspondence to C.D. Skow. E-mail: cds{at}nsm.umass.edu.
Received 23 October 2003; revised 6 September 2005; accepted 9 September 2005.
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
A large number of studies on both animals and humans have demonstrated that learning is influenced by context or secondary cues that are present when an association is formed. Few studies, however, have examined the functional value of attending to context. We first demonstrated that jumping spiders, Phidippus princeps, could be trained to avoid aposematic, distasteful milkweed bugs, Oncopeltus fasciatus. Spiders readily attacked bugs on first exposure but were significantly less likely to do so after eight trials, although they subsequently attacked and ate crickets. Spiders exposed to nontoxic milkweed bugs reared on sunflower seeds did not show the same decline in attack rate. We next examined the effects of secondary contextual cues on spider learning by training spiders to avoid milkweed bugs in one of two environments. When spiders were tested in an environment different from the one in which they were trained, attack rates increased, and spiders no longer demonstrated retention of the association. Spiders tested in the same environment in which they were trained continued to avoid attacking the bugs. These results have potential consequences for the evolution of both predator and prey and point to the importance of studying context-dependent learning.
Key words: aposematism, avoidance, context, experience, learning, Salticidae, spiders.
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Cognitive traits such as learning and memory can limit present-day behavior as well as be under selection themselves. Cognitive ethologists (sensu, Kamil, 1998
) have benefited from adapting ideas from the rich psychological literature concerning the mechanisms of learning and memory and applying them to ecologically relevant problems in a variety of taxa not traditionally studied by psychologists (e.g., Dukas, 1998; Kamil, 1998; Real, 1993; Shettleworth, 2001; Stamps, 1991). For example, we now know that the adaptive decisions animals make while foraging include cognitive mechanisms for sampling prey (Dall et al., 1999), forming search images (Blough, 2002; Bond and Kamil, 1999; Langley, 1996), and handling constraints caused by attending to other stimuli that affect fitness (limited attention, Dukas, 2002, 2004). This blend of approaches has enriched our understanding of the evolution of behavior.
An area that is well studied by psychologists but relatively neglected by behavioral ecologists is the role of context in learning. Learning occurs in a dynamic environment in the presence of many cues, some of which can be peripheral to the associative task of primary interest. These contextual cues can include environmental background cues, time of day, and internal motivational states like hunger and stress levels (Bouton, 1993; Pearce and Bouton, 2001). When contextual cues remain consistent during the acquisition phase of learning, they can become encoded along with the primary reinforced stimuli (Tulving and Thomson, 1973). When tested in the absence of contextual cues, performance of the animal may be disrupted because the memory is not recalled or acted on (Balsam, 1984; Lubow et al., 1976; Odling-Smee, 1975). In other words, the animal learns not only the main association but also the context in which the association occurs. The precise nature of the mental representation of relationships between the available cues can take different forms, consisting of various separate elemental associations or a global configuration of elements (González et al., 2003; Riccio et al., 1992). However, it is generally held that if a change in contextual cues between training and testing for memory disrupts performance, then it is reasonable to assume that the animal attended to and encoded contextual cues in some way (Zhou and Riccio, 1996).
Context-dependent learning has been widely demonstrated in the psychology literature using both animal and human models (reviewed in Bouton, 2002). Most studies have been done with classic laboratory animal models like rats, pigeons, and rabbits. In addition, they focus on tasks (e.g., bar or button pressing) and stimuli (e.g., tones, flashing lights, and odors) that are not representative of naturally occurring events. For example, González et al. (2003) compared the effects of either addition or subtraction of contextual components on memory recollection in rats. The authors conditioned freezing behavior using footshocks and added or subtracted visual, auditory, or olfactory contextual cues to the experimental chamber during testing. Rats that received additions to context did not show impaired recollection, while the removal of contextual components resulted in impaired performance, that is, a decrease in freezing behavior. While highly controlled experiments allow us to understand the associative interactions among stimuli and contribute to models of learning and memory, they often do not address the functional value of these mechanisms. However, there are a few experiments addressing how context is used during ecologically relevant tasks (Colborn et al., 1999; Collett et al., 1997; Fauria et al., 2002). For example, butterflies can associate different colors with different behavioral contexts. After being trained to associate a particular color with either a food source (sucrose) or an oviposition site (host plant extracts), they perform the correct behavior based on color cues alone (Weiss and Papaj, 2003). This functional approach to contextual cue used during learning and memory is important for understanding how natural selection shapes mechanisms underlying behavior and how cognitive processes produce adaptive behaviors.
Learned avoidance of aposematic prey
For a generalist predator, learning to avoid toxic aposematic prey is a highly adaptive foraging strategy. Aposematic prey display conspicuous colors or patterns that warn potential predators of toxicity (Guilford and Dawkins, 1991). Typical warning signals include bold patterns of red, yellow, or blue, sometimes paired with black or white (Cott, 1940; Poulton, 1898). For both vertebrates and invertebrates, the physiological consequences of feeding on toxic prey are many. For example, predators may regurgitate the prey item (Berenbaum and Miliczky, 1984; Brower, 1958; Brower et al., 1968; Brown and Neto, 1976; Chai, 1986). Longer lasting effects of toxic prey consumption include inhibition of further feeding (Toft and Wise, 1999), as well as decreased nutrient absorption (Toft and Wise, 1999), weight gain (Strohmeyer and Stamp, 1998), or growth (Paradise and Stamp, 1993). Predators may respond to aposematic prey in several ways. First, generalist predators sometimes innately avoid colors associated with aposematic prey (Gamberale-Stille, 2000; Lindstrom et al., 1999; Mastrota and Mench, 1995). Second, naive predators exhibiting dietary conservatism can be reluctant to attack novel prey (Thomas et al., 2003). Finally, predators can learn to differentiate aposematic prey from palatable prey. In such cases, conspicuous coloration and pattern cues can increase the learning rate (Guilford, 1990; Paradise and Stamp, 1991; Roper, 1990; Speed, 2000; Whittman, 1988).
In our study, we focused on whether peripheral contextual cues affect learning about aposematic prey. Our model predator is a jumping spider, Phidippus princeps (family Salticidae). Jumping spiders are particularly well suited for learning studies because of their excellent vision, which makes experiments easy to design. They rely on vision to navigate in their environment, capture prey, recognize and signal to mates and rivals, and detect conspecifics and predators (Clark and Uetz, 1990; Foelix, 1996; Richman and Jackson, 1992). They learn a variety of tasks, including the avoidance of particular colors associated with heat (Nakamura and Yamashita, 2000), the avoidance of unpalatable ants (Edwards and Jackson, 1994), and associating landmarks or beacons with prey (Popson, 1999) and nest sites (Hoefler and Jakob, in press). In addition, invertebrate predators have largely been ignored in studies addressing how predator psychology contributes to aposematism (Kauppinen and Mappes, 2003). Context-dependent learning has been explored in only a handful of invertebrate species, including ants (Bisch-Knaden and Wehner, 2003), bees (Collett et al., 1997; Fauria et al., 2002), butterflies (Weiss and Papaj, 2003), nematodes (Rankin, 2000), pond snails (Haney and Lukowiak, 2001), and vinegar flies (Liu et al., 1999).
Our study addresses three main points. First, we document learned avoidance in spiders of aposematic juvenile milkweed bugs (Oncopeltus fasciatus). Secondly, we present behavioral evidence that contextual cues (external background cues) are encoded along with associations between prey and taste. Finally, we address the potential adaptive value of context-dependent learning and discuss the implications it holds for optimal prey behavior.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Maintenance of study organisms
We collected adult and penultimate P. princeps from fields devoid of milkweed plants to reduce the chance that spiders had recently encountered milkweed bugs. We maintained spiders at 27°C on a 13:11 h light:dark cycle on a mixed diet of crickets (Achaeta domestica) and vinegar flies (Drosophila melanogaster) and provided ad libitum water in cotton-stoppered test tubes. We housed spiders individually in plastic cages (31.5 x 23.5 x 10 cm high) with screened lids, a painted green dowel and plastic plants to mitigate the effects of captivity (Carducci and Jakob, 2000
). We conducted the learned avoidance experiment in the winter and spring of 2001–2002 on spiders collected as penultimate or adults in the fall of 2001 and the context experiment in the winter and spring of 2002–2003 on spiders captured in the fall of 2002. Spiders were mature at the time of testing. We tested spiders with milkweed bug nymphs. We used both unpalatable and palatable bugs in the learning experiment and only unpalatable bugs in the context experiment. Unpalatable bugs were offspring of a laboratory-reared colony maintained for multiple generations on milkweed seeds, Asclepias syriaca. Palatable bugs were purchased from Carolina Biological Supply Company, where they have been maintained for generations on sunflower seeds, Helianthus annus. We reared nymphs from these lines on shelled H. annus seeds. All bugs had ad libitum access to seeds and water. Nymphs were one-third to one-half the length of the spider to which they were presented. Palatable and unpalatable strains did not appear to differ in either size or color.
Learned avoidance of distasteful prey
We randomly assigned 16 male and 16 female adult P. princeps to be tested either with unpalatable or palatable bugs, keeping the sexes balanced between treatments. Spiders were deprived of food for 9 days prior to testing. All tests for any given individual were conducted on the same day from 0830 to 1830 h. Test arenas were two large stacked petri dish bottoms, 14 cm in diam and 2.5 cm deep, one right side up and one inverted on top. To confine individuals to the arena's floor, we coated the walls of the bottom petri dish with Fluon and those of the top dish with petroleum jelly. We videotaped trials with a camcorder located 45 cm above the test arena.
We placed spiders individually in a test arena and allowed them to acclimate for 5 min. An active milkweed bug nymph was then placed in the arena for 10 min. At the end of the trial, the milkweed bug was removed if the spider was not feeding. Feeding spiders were allowed to consume the prey. Spiders remained in the test arena for 50 min between trials. This procedure was repeated for seven additional trials, minus the acclimation period, for a total of eight trials per spider. Each spider was then immediately given one cricket nymph (8–10 mm long) to test for satiation.
From the videotape, we scored the number of orientations and number of attacks for each trial. The scorer was blind to treatment group. An "orientation" was defined as an initial swivel of the spider's cephalothorax to direct its image-forming anterior median eyes toward the prey. Subsequent movements of the cephalothorax to continuously track moving prey were not counted as new orientations. "Attacks" consisted of either predatory lunges toward the prey without visible feeding or lunges that ended in grappling with and feeding on the prey. We combined these because it was not always possible to discern from the video whether the chelicerae were inserted into the prey.
We examined whether toxic and nontoxic bugs differed in movement behavior to test whether prey presented different stimuli to the spiders. Sixteen bugs from each group were randomly selected from the tapes using a random number table. We scored the percentage of time bugs spent moving prior to capture. We compared arcsine square root–transformed data (Sokal and Rohlf, 1995) with an unpaired t test.
The effect of context
Twenty-five spiders were randomly assigned to be trained in either a complex environment or a simple environment. The complex environment consisted of a 3-cm-high transparent plastic ring coated with silicone and placed in a clear plastic (31.5 x 23.5 x 10 cm) cage with no lid. This cage was decorated to mimic a natural setting, with artificial grass (consisting of 15-cm-long plastic "leaves" glued to two 4-cm blocks of wood), three plastic leafy plants 18–30 cm high, and scattered pebbles. To reduce visual distractions in the room, we placed the cage in a plastic tub (40.5 x 56 x 22 cm) with inner walls covered in sky-patterned contact paper (light blue with white clouds) and the floor covered in green construction paper. The simple environment lacked any noticeable secondary environmental cues and consisted of a transparent ring coated with silicon in an empty cage, placed in a tub with walls and floor covered in white contact paper. In both cases, spiders were confined to the plastic ring, so that the environments provided background cues only.
For each trial, the test spider was placed in its environment and allowed to acclimate for 5 min. An active milkweed bug nymph was then placed in the silicone-coated ring with the spider for a 10-min trial. At the end of the trial, spiders were removed from their training environment and kept in small (19 x 13.5 x 10.5 cm high) clear plastic resting cages for 50 min. This procedure was repeated for seven more trials, including the acclimation period, for a total of eight training trials per spider.
Next, spiders were given a test trial. Half the spiders (group 1) were tested in the same environment in which they were trained. The other half (group 2) were tested in the environment opposite of their training environment: spiders trained in complex environments were tested in simple environments and vice versa. The 50-min intertrial interval remained the same between trial 8 and the test trial. Spiders had a 5-min acclimation period prior to testing. After the test trial, each spider was given a cricket to test for satiation, and data from spiders that did not eat a cricket were discarded. Trials were taped and scored as in the first experiment; the observer was blind to treatment group. In addition, we were interested in whether differences between the groups could be attributed to a failure of spiders to notice the bug or to differences in deciding whether to attack it. Thus, we compared latency to initial orientation to the bug.
Data were analyzed with JMP (SAS Institute, 2002) on a Macintosh computer. To estimate the power of Mann-Whitney U tests, we used JMP to calculate the power of the t test and then assumed a conservative relative efficiency of 0.864 (Daniel, 1978). For goodness-of-fit tests, we followed Cohen (1988). Power calculators for paired Wilcoxon signed-rank tests are not readily available; Thomas and Juanes (1996) provide an example, but for data with a binomial outcome. We therefore calculated power via the following procedure. For each data set that yielded a nonsignificant p value, we estimated the noncentral distribution (the distribution of the test statistic under a given effect size 
and sample size N) numerically using a modified bootstrap resampling scheme. We will call the original data x, a vector of length N containing observed differences between before and after training and an observed mean value of 
. We assumed that the sampling variance around the expected effect size was constant and given by the original data. This permitted us to resample N spiders from the original data, with replacement, to obtain a bootstrapped replicate, x'. To adjust the effect size of these x', we subtracted the original data's mean and added the expected effect size to get a modified vector x'' = x' – µ + . The elements of x'' are distributed as real numbers, whereas the original data in our tests were integers (counts of orients or attacks). We therefore took the integer value of each element and added 1 if a uniform pseudorandom number was less than its fractional part (e.g., if an element of x'' was 3.3, we scored it as 4 if the pseudorandom number was less than 0.3 and as 3 otherwise.) We obtained 1000 replicates of each modified x'' for each tested. Each x'' was run through the Wilcoxon signed-rank test and its significance level (p value) obtained from a z approximation. The power at that effect size is the proportion of these 1000 replicates that yield p < .05. We replicated this in steps of = 0.25. A Mathematica (Wolfram) notebook is available.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Learned avoidance of distasteful prey
We detected no behavioral differences between bug nymphs fed on milkweed versus sunflower seeds (unpaired t test on arcsine square root–transformed data; t = 0.96, df = 30, n = 32 per group, p > .3). Bugs fed on milkweed moved during 50.1% (SE ±3.70) of the observation time, and bugs fed on sunflower seeds moved 41.6% (SE ±4.94). Our sample size would allow us to detect large effect sizes 93% of the time and medium effect sizes 63% of the time (Cohen, 1988
). In general, bugs did not appear to notice spiders or take evasive action, even after an attack. Spiders did not innately avoid milkweed bugs: 94% of spiders presented with toxic bugs (n = 16) and 88% of those presented with nontoxic bugs (n = 16) attacked them on the first trial. These two groups did not significantly differ (G test of independence, G2 = 0.374, df = 2, p > .5). Our sample size would give us the power to detect large differences 80% of the time and to detect medium differences approximately 40% of the time. Clearly, spiders did not avoid toxic bugs on initial encounter as attack rates were higher on toxic versus nontoxic bugs.
We tested for a change in spider behavior over the course of the trials in two ways. First, we examined the number of orientations and attacks per trial. To test for changes in responsiveness of spiders across trials, we conducted an interaction contrast. We regressed the number of orientations and attacks against trial number separately for each spider. Each regression generated a single slope for each spider (Figure 1a). A negative slope indicates that a spider's responsiveness toward prey decreased across trials. Slopes were normally distributed and analyzed with parametric statistics. This method is preferable to examining the trial x treatment interaction term in a repeated-measures test, which tests the null hypothesis that the effect of trial is the same for both treatments. The interaction contrast that we conducted asks the more focused question of whether the linear trend over trials is the same across treatments and is appropriate when sufficiently motivated by theory, as in our case (Meyers and Well, 1991; Well, personal communication).
|
Spiders fed unpalatable bugs had significantly lower slopes than spiders fed palatable bugs for both orientations (unpaired t test, t = 2.44, df = 30, p < .03) and attacks (t = –3.92, df = 30, p < .0005; Figure 1b). We also conducted pairwise tests of the number of orientations and attacks in the first versus the last trial. These data were not normally distributed and were analyzed with Wilcoxon signed-rank tests. We found no significant change in the number of orientations between the first and last trial for either group of spiders (unpalatable bugs: tied z = –0.642, p > .5; palatable bugs: tied z = –1.474, p > .1). The number of attacks significantly decreased for the spiders fed unpalatable bugs (tied z = –2.901, p < .004) but not for those fed palatable bugs (tied z = 0.258, p > .7). For each nonsignificant test, our power simulations suggest that we are likely to detect an effect size of 1.5 orientations or attacks 85% of the time and 2 orientations or attacks 90% of the time.
In nature, repeated attacks may be less likely than in our experiment because either predator or prey may move away after the initial encounter. Therefore, we ran a second set of analyses in which we scored spiders as attacking if they did so at least once during the trial. (Spiders nearly always oriented at least once, so we did not perform this test for orientations.) We compared the first and last trials with a McNemar test for the significance of changes (Sokal and Rohlf, 1995). By this measure, both sets of spiders decreased the probability of attack in the final trial compared to the first trial (palatable prey: 
2 = 5.143, df = 2, p < .03; unpalatable prey: 2 = 7.111, df = 2, p < .008). We suspect that decreased responsiveness may have different roots in the two groups. By the last trial, spiders fed palatable prey were probably reaching satiation: all but one fed on at least one bug for at least 10 s. The average cumulative time spent feeding over the eight trials was 22.7 ± 4.1 min (range 10–61.5 min), and 7 of 16 failed to attack a cricket at the end of the experiment. In contrast, spiders presented with unpalatable prey attacked them but quickly released them, feeding on average only 12.4 ± 3.8 s over the course of eight trials (range 0–47 s). All these spiders immediately attacked and fed on on a cricket presented at the end of testing. Thus, the decrease in responsiveness to unpalatable prey was not because of satiation.
The effect of context
As in the first experiment, we examined the data in two ways. We analyzed the number of orientations and attacks with nonparametric statistics. We also used categorical classifications of orienting or attacking at least once per trial versus not at all and analyzed these with contingency table tests.
First, we examined whether the environment in which spiders were trained influenced the likelihood of orientation or attack: we hypothesized that spiders might find it more difficult to notice prey against the complex background. However, this appeared not to be the case. Although spiders trained in a simple background (n = 13) had slightly but not significantly more orientations to prey in trial 1 than did those trained in a complex background (n = 12) (Mann-Whitney U, tied z = –1.732, p = .083; power to detect effect sizes of 1 and 1.5 are, respectively, 0.75 and 0.86), there were no differences in number of attacks (tied z = –.357, p > .7; power to detect effect sizes of 1 and greater is 0.86). After the final training trial was complete, we found no differences between spiders trained in the two backgrounds (orients: tied z = –0.057, p > .9; attacks: tied z = –0.372, p > .7; power to detect effect sizes of 1 and 1.5 are, respectively, 0.74 and 0.85 for orientations, and 0.86 and 0.86 for attacks). The categorical analysis revealed no significant differences for either orientations or attacks in either the first or last training trial (G tests of independence, p > .4 in all cases), but sample sizes (as in all subsequent G tests in this section) allowed us to detect only large effects with confidence (Cohen, 1988).
It is important to confirm that the spiders that we randomly assigned to be tested either in the same environment (n = 14) or different environment (n = 11) did not differ by chance. We expected no difference in performance between these groups in the training trials and found none for either the first training trial (orients: Mann-Whitney U, tied z = –1.029, p > .3; attacks: tied z = –1.556, p > .1; power to detect effect sizes of 1 and 1.5 are 0.73 and 0.86, respectively) or the last training trial (orients: tied z = –1.092, p > .2; attacks: tied z = –0.156, p > .8; power to detect effect sizes of 1 or 1.5 are 0.86).
In the test trial, spiders tested in a different context than the training context both oriented (Mann-Whitney U, tied z = –3.058, p < .003) and attacked (tied z = –2.063, p < .04) more often than spiders tested in the same context in which they were trained. Thus, spiders moved to a new context were more responsive to unpalatable prey.
In the categorical analyses, we found no effect of testing regime on orientation to the prey in training or test trials. During the first training trial, all spiders oriented to the prey at least once. Most spiders also oriented to the prey in the last training trial (same environment: 85.7% oriented [n = 14]; different environment: 90.9% oriented [n = 11]; G2 = .161, df = 2, p > .6) and in the test trial (same environment: 78.6% oriented; different environment: 90.9% oriented; G2 = 0.733, p > .3). We also found no difference between the treatment groups in their propensity to attack prey in either their first (G2 = 3.798, df = 2, p > .05) or last (G2 = .109, df = 2, p > .7) training trials (Figure 2).
|
However, in the test trial, only 29% of spiders tested and trained in the same environment attacked the unpalatable prey at least once, whereas 73% of those tested in a different environment did so (G2 = 4.975, df = 2, p < .03; Figure 2). We found no difference in latency to orient to the unpalatable bugs in spiders that had been shifted to a new environment (n = 11) versus spiders tested in the same environment (n = 14) (Mann-Whitney U, tied z = –0.679, p > .4). Most spiders oriented to the bug within the first 20 s of the trial (71% of spiders tested in the same environment and 67% of spiders tested in a new environment).
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Learned avoidance of aposematic prey
Before assessing whether jumping spiders can learn to avoid aposematic juvenile milkweed bug nymphs, we tested for innate avoidance of the prey. Innate avoidance of toxic prey should occur when predators can directly assess toxicity before attack, especially when an attack results in severe illness or death. Neither situation seems to be the case in our system. P. princeps did not exhibit an innate avoidance of milkweed bugs. High attack rates in trial 1 for both experimental and control groups suggest that spiders did not detect noxious airborne chemical cues emitted by milkweed bugs. This is in line with Hoefler et al. (2002)
, who found that a congener, Phidippus audax, does not respond to chemical cues of prey. While we matched bugs for size, and did not detect statistically significant differences in the time that different types of bugs spent moving, it is possible that there were differences we could not detect. However, given the high percentage of attack rates on both bug types in trial 1, it suggests that spiders did not perceive differences in palatability based on appearance or movement. The lack of innate avoidance of aposematic prey may indicate that a lost opportunity to forage is more costly than attacking and releasing a toxic prey item (Wise and Wagner, 1992).
Learned avoidance of aposematic prey should be favored by selection when there is a high probability of repeated encounters with distasteful prey of similar appearance. Bernays (1992) and Bernays et al. (1994) suggest that even short-term retention of the memory of toxic prey is adaptive if such prey occur in patches. This is likely to be the case in this spider species. Milkweed bugs aggregate as a defense mechanism (Riipi et al., 2001) and A. syriaca plants tend to be found in patches (McCauley, 1991). Thus, there is a high probability of a jumping spider coming into repeated contact with milkweed bugs in a particular location. Jumping spiders learn to avoid milkweed bug nymphs over the course of eight trials spanning 8 h. This is evident by the significant decrease in the number of attacks on unpalatable bugs compared to attacks on palatable bugs. Learned avoidance of aposematic prey has been demonstrated in other visually hunting invertebrate predators, including mantids (Berenbaum and Miliczky, 1984; Bowdish and Bultman, 1993) and dragonflies (Kauppinen and Mappes, 2003).
We controlled for potential detrimental physiological effects of attacking toxic prey by excluding individuals who did not readily consume a cricket at day's end. While we cannot entirely exclude illness as a cumulative factor affecting performance, it seems highly unlikely that recovery from illness spontaneously occurs in the brief time between the end of the last milkweed bug trial and an immediate cricket offering. Milkweed bugs sequester toxic cardiac glycosides from seeds of the milkweed plant in thoracic and abdominal cavities (Duffey and Scudder, 1972; Duffey et al., 1978; Ralph, 1976). Along with toxins, secondary chemical defenses include bitter-tasting alkyl methoxypyrazines, which contribute to unpalatability (Aldrich et al., 1997). Because these spiders digest their food externally by injecting digestive enzymes into prey and sucking nutrients back in, P. princeps may primarily detect the bitter taste and quickly release prey before ingesting high levels of toxins. Thus, predation on milkweed bug nymphs does not appear to be extremely costly in P. princeps. We may have seen cumulative negative effects if we had carried out the experiment over a longer period. Two species of wolf spiders (Schizocosa ocreata and Schizocosa stridulans) fed toxic prey over several weeks exhibited inhibition of feeding rates, impaired use of nutrients from higher quality prey, and even death (Toft and Wise, 1999). P. audax forced to consistently feed on toxic prey for several days lost weight (Strohmeyer and Stamp, 1998).
The effect of context
Classic models of learning and memory first brought attention to the conditioning of context along with explicitly reinforced cues (Estes, 1973, 1975; Rescorla and Wagner, 1972). Since then, much progress has been made in understanding the detailed relationships between cues reinforced during learning and how they affect memory retention and task performance (Bills et al., 2003; Bouton, 2002; Bouton and Swartzentruber, 1989; González et al., 2003; Riccio et al., 1992). As this is the first known demonstration of context-dependent learning in spiders, the particular details of associative interactions among the available stimuli are beyond the scope of this study. We do not yet know whether the relationship between stimuli consists of elemental associations between individual components like context-taste, taste-prey, or context-prey or if it is a global configuration of prey-context-taste. However, we do know the overall association spiders form while learning to avoid toxic prey includes both the prey and background environmental cues. First, they decrease attack rates on unpalatable prey while continuing to attack crickets under constant environmental conditions, suggesting an association between prey attributes and taste. If the spiders were only learning context, then one would expect to see avoidance of all subsequent prey items presented in the same context, which is not the case. This suggests that specific associations are formed between prey attributes and taste. Second, jumping spiders increase attacks when contextual cues shift, suggesting that context plays an integral role in the association between prey and taste. From an ecological and evolutionary standpoint, there are several functional implications of context-dependent learning for both the predator and prey.
Either generalizing associations across contexts (context-independent learning) or the loss of an association when secondary cues change (context-dependent learning) might be beneficial, depending on circumstance. For example, a predator that learns to avoid aposematic prey in one environment would benefit by retaining that information in a new environment, as long as similar prey were also toxic. If, however, prey toxicity varies across environments, context-dependent learning would be advantageous. Context dependence may also be favored by selection in other circumstances, such as when learning the location of nest sites or good foraging areas. We do not yet know whether spiders' attention to context in this situation is beneficial or detrimental in terms of current adaptive value. On one hand, if prey are reliably distasteful across environments, attention to cues that are not features of the prey itself seems inefficient because it results in renewed attention to, and attacks on, prey the spider already learned to avoid. This could be costly both in terms of exposure to toxins and in lost time that could be spent foraging for profitable prey and learning positive prey associations. In addition, learning itself is likely to bear energetic costs (Dukas, 1999). On the other hand, if similar prey in different locations is sometimes palatable and the costs of making an error are low, attending to context may be advantageous. For example, plant cues might be particularly informative. Herbivorous prey that acquire toxins from host plants may remain in the vicinity of their plants for substantial periods of time (McCauley, 1991, and references therein). Jumping spiders often move many meters in the course of a single day (Hoefler and Jakob, 2004), and thus are likely to contact different species of plants.
Context-dependent learning also holds implications for the optimal behavior of the prey. Aposematism is thought to drive predator learning and enhance memory (Servedio, 2000; Speed, 2001), thus decreasing predation attempts by experienced predators. Many animals aggregate in the presence of predators because of the "safety in numbers" or dilution effect (Hamilton, 1964). If predator learning is frequently context dependent, aposematic prey that aggregate may be favored over those that disperse into different microhabitats. This idea complements other predator learning theories favoring the evolution of aposematic aggregations (Riipi et al., 2001).
This study further illustrates the need for integrated work addressing both cognitive mechanism and its functional value (Healy and Jones, 2002). Similar to other tested species, learning in spiders appears to be context dependent. Given its prevalence, the mechanisms underlying context dependence may not be species specific but rather may be inherent properties of informational processing. Aside from the potential comparative value, understanding how contextual cues affect spiders' task performance in more complex and realistic situations is the next step in characterizing both the costs and benefits of attending to context. To this end, we are currently conducting more mechanistic studies with the ultimate goal of understanding the functional value of contextual cue use.
|
ACKNOWLEDGEMENTS |
|---|
A. Porter wrote the code and performed the simulation for the bootstrap power analyses, and we are very grateful. A. Well and A. Porter provided statistical advice. We thank J. Podos, J. Ayres, G. Wyse, C. Hoefler, M. Baker, and A. Porter for comments on various drafts of the manuscript and G. Uetz for helpful discussion. R. Souza, A. Nichols, and A. Plourde assisted with spider care. This work was supported by an American Arachnological Society student research award to C.D.S and a National Science Foundation Small Grants for Exploratory Research grant to E.M.J.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Aldrich JR, Leal WS, Nishida R, Khrimian AP, Lee CJ, Sakuratani Y, 1997. Semiochemistry of aposematic seed bugs. Entomol Exp Appl 84:127–135.
Balsam PD, 1984. Bringing the background to the foreground: the role of contextual cues in autoshaping. In: Quantitative analyses of behavior: Volume 3: acquisition (Commons ML, Hernstein RJ, Wagner AR, eds). Cambridge: Ballinger; pp. 145–171.
Berenbaum MR, Miliczky E, 1984. Mantids and milkweed bugs: efficacy of aposematic coloration against invertebrate predators. Am Midl Nat 1:64–68.
Bernays EA, 1992. Interaction of insects and plants. Sci Prog 76:247–271.
Bernays EA, Bright KL, Gonzalez N, Angel J, 1994. Dietary mixing in a generalist herbivore—tests of 2 hypotheses. Ecology 75:1997–2006.[CrossRef][Web of Science]
Bills C, Smith S, Myers N, Schachtman TR, 2003. Effects of context exposure during conditioning on conditioned taste aversions. Learn Behav 31:369–377.[Web of Science][Medline]
Bisch-Knaden S, Wehner R, 2003. Local vectors in desert ants: context-dependent landmark learning during outbound and homebound runs. J Comp Physiol A Neuroethol Sens Neural Behav Physiol 189:181–187.[Web of Science][Medline]
Blough DS, 2002. Measuring the search image: expectation, detection, and recognition in pigeon visual search. J Exp Psychol Anim Behav Process 28:397–405.[CrossRef][Web of Science][Medline]
Bond AB, Kamil AC, 1999. Searching image in blue jays: facilitation and interference in sequential priming. Anim Learn Behav 27:461–471.[Web of Science]
Bouton ME, 1993. Context, time and memory retrieval in the interference paradigms of Pavlovian learning. Psychol Bull 144:80–99.
Bouton ME, 2002. Context, ambiguity, and unlearning: sources of relapse after behavioral extinction. Biol Psychiatry 52:976–986.[CrossRef][Web of Science][Medline]
Bouton ME, Swartzentruber D, 1989. Slow reacquisition following extinction—context, encoding, and retrieval mechanisms. J Exp Psychol Anim Behav Process 15:43–53.[CrossRef][Web of Science]
Bowdish TI, Bultman TL, 1993. Visual cues used by mantids in learning aversion to aposematically colored prey. Am Midl Nat 129:215–222.[CrossRef]
Brower JVZ, 1958. Experimental studies of mimicry in some North American butterflies. I. The monarch, Danaus plexipus, and viceroy, Limenitis archippus archippus. Evolution 12:32–47.
Brower LP, Ryerson WN, Coppinger LL, Glazier SC, 1968. Ecological chemistry and the palatability spectrum. Science 161:1349–1351.
Brown KS, Neto JV, 1976. Predation on aposematic ithomiine butterflies by tanagers (Pipraeidea melanonota). Biotropica 8:136–141.[CrossRef]
Carducci J, Jakob E, 2000. The effect of rearing environments on the behavior of a jumping spider. Anim Behav 59:39–46.[CrossRef][Web of Science][Medline]
Chai P, 1986. Field observations and feeding experiments on the responses of rufous-tailed jacamars (Galbula ruficauda) for free-flying butterflies in a tropical rainforest. Biol J Linn Soc 29:161–189.[CrossRef]
Clark DL, Uetz GW, 1990. Video image recognition by the jumping spider Maevia inclemens (Araneae: Salticidae). Anim Behav 40:884–890.[CrossRef]
Cohen J, 1988. Statistical power analyses for the behavioral sciences, 2nd ed. Hillsdale, NJ: Lawrence Erlbaum Associates.
Colborn M, Ahmad-Annuar A, Fauria K, Collett TS, 1999. Contextual modulation of visuomotor associations in bumble-bees (Bombus terrestris). Proc R Soc Lond B 266:2413–2418.
Collett TS, Fauria K, Dale K, Baron J, 1997. Places and patterns—a study of context learning in honeybees. J Comp Physiol A 181:343–353.[CrossRef]
Cott HB, 1940. Adaptive coloration in animals. London: Methuen.
Dall SRX, McNamara JM, Cuthill IC, 1999. Interruptions to foraging and learning in a changing environment. Anim Behav 57:233–241.[CrossRef][Web of Science][Medline]
Daniel W, 1978. Applied nonparametric statistics. Boston: Houghton Mifflin.
Duffey SS, Blum MS, Isman MB, Scudder GGE, 1978. Cardiac glycosides: a physical system for their sequestration by the milkweed bug. J Insect Physiol 24:639–645.[CrossRef]
Duffey SS, Scudder GGE, 1972. Cardiac glycosides in Oncopeltus fasciatus (Hemiptera: Lygaeidae). I. The uptake and distribution of natural cerdenolides in the body. Can J Zool 52:283–290.[CrossRef]
Dukas R (ed), 1998. Cognitive ecology. Chicago: University of Chicago Press.
Dukas R, 1999. Costs of memory: ideas and predictions. J Theor Biol 197:41–50.[CrossRef][Web of Science][Medline]
Dukas R, 2002. Behavioural and ecological consequences of limited attention. Phil Trans R Soc Lond B 357:1539–1547.
Dukas R, 2004. Causes and consequences of limited attention. Brain Behav Evol 63:197–210.[CrossRef][Web of Science][Medline]
Edwards GB, Jackson RR, 1994. The role of experience in the development of predatory behaviour in Phidippus regius, a jumping spider (Araneae, Salticidae) from Florida. NZ J Zool 21:269–277.
Estes WK, 1973. Contemporary approaches to conditioning and learning. Washington, District of Columbia: Winston.
Estes WK, 1975. Handbook of learning and cognitive processes: conditioning and behavior theory. Hillsdale, New Jersey: Lawrence Erlbaum Associates.
Fauria K, Dale K, Colburn M, Collett TS, 2002. Learning speed and contextual isolation in bumblebees. J Exp Biol 205:1009–1018.
Foelix R, 1996. The biology of spiders, 2nd ed. New York: Oxford University Press.
Gamberale-Stille G, 2000. Decision time and prey gregariousness influence attack probability in naive and experienced predators. Anim Behav 60:95–99.[CrossRef][Web of Science][Medline]
González F, Quinn JJ, Fanselow MS, 2003. Differential effects of adding and removing components of a context on the generalization of conditional freezing. J Exp Psychol Anim Behav Process 29:78–83.[Web of Science][Medline]
Guilford T, 1990. Evolutionary pathways to aposematism. Acta Oecol 11:835–841.
Guilford T, Dawkins MS, 1991. Receiver psychology and the evolution of animal signals. Anim Behav 42:1–14.
Hamilton WD, 1964. The evolution of social behavior. J Theor Biol 7:1–52.[CrossRef][Web of Science][Medline]
Haney J, Lukowiak K, 2001. Context learning and the effect of context on memory retrieval in Lymnaea. Learn Mem 8:35–43.
Healy SD, Jones CM, 2002. Animal learning and memory: an integration of cognition and ecology. Zoology 105:321–327.
Hoefler C, Jakob EM, 2004. Jumping spiders in space: movement patterns, nest site fidelity, and the use of beacons. Anim Behav (in press).
Hoefler C, Taylor M, Jakob EM, 2002. Chemosensory respone to prey in Phidippus audax (Araneae: Salticidae) and Pardosa milvina (Araneae: Lycosidae). J Arachnol 30:155–158.[CrossRef]
Kamil AC, 1998. On the proper definition of cognitive ethology. In: Animal cognition in nature (Balda RP, Pepperberg IM, Kamil AC, eds). San Diego: Academic Press; 1–28.
Kauppinen J, Mappes J, 2003. Why dragonflies are so intimidating: field experiments on hunting dragonflies (Odonata: Aeshna grandis). Anim Behav 66:505–511.[CrossRef]
Langley CM, 1996. Search images: selective attention to specific visual features of prey. J Exp Psychol Anim Behav Process 22:152–163.[CrossRef][Web of Science][Medline]
Lindstrom L, Alatalo RV, Mappes J, 1999. Reactions of hand-reared and wild-caught predators toward warningly colored, gregarious, and conspicuous prey. Behav Ecol 10:317–322.
Liu L, Wolf R, Ernst R, Heisenberg M, 1999. Context generalization in Drosophila visual learning requires the mushroom bodies. Nature 400:753–756.[CrossRef][Medline]
Lubow RE, Rifkin B, Alek M, 1976. The context effect: the relationship between stimulus pre-exposure and environmental pre-exposure determines subsequent learning. J Exp Psychol Anim Behav Process 2:38–47.[CrossRef][Web of Science]
Mastrota NF, Mench JA, 1995. Colour avoidance in northern bobwhites: effects of age, sex, and previous experience. Anim Behav 50:519–526.[CrossRef]
McCauley DE, 1991. The effect of host plant patch size variation on the population structure of a specialist herbivore insect, Tetraopes tetraophthalmus. Evolution 45:1675–1684.[CrossRef][Web of Science]
Meyers JL, Well AD, 1991. Research design and statistical analysis. New York: HarperCollins Publishers Inc.
Nakamura T, Yamashita S, 2000. Learning and discrimination of colored papers in jumping spiders (Araneae: Salticidae). J Comp Physiol A Sens Neural Behav Physiol 186:897–901.[CrossRef][Web of Science][Medline]
Odling-Smee FJ, 1975. The role of background stimuli during Pavlovian conditioning. Q J Exp Psychol 27:201–209.[Web of Science][Medline]
Paradise CJ, Stamp NE, 1991. Prey recognition time of praying mantids (Dictyoptera: Mantidae) and consequent survivorship of unpalatable prey (Hemiptera: Lygaeidae). J Insect Behav 4:265–273.[CrossRef]
Paradise CJ, Stamp NE, 1993. Episodes of unpalatable prey reduce consumption and growth of juvenile praying mantids. J Insect Behav 6:155–166.[CrossRef]
Pearce JM, Bouton ME, 2001. Theories of associative learning in animals. Annu Rev Psychol 52:111–139.[CrossRef][Web of Science][Medline]
Popson M, 1999. Finding the way: learning in Phidippus audax (MS thesis). Ohio: Bowling Green State University.
Poulton EB, 1898. Natural selection: the cause of mimetic resemblance and common warning colours. Linn Soc J Zool 26:558–612.[CrossRef]
Ralph CP, 1976. Natural food requirements of the large milkweed bug, Oncopeltus fasciatus (Hemiptera: Lygaeidae) and their relations to gregariousness and host plant morphology. Oecologia 26:157–175.[CrossRef][Web of Science]
Rankin C, 2000. Context conditioning in habituation in the nematode C. elegans. Behav Neurosci 114:496–505.[CrossRef][Web of Science][Medline]
Real LA, 1993. Toward a cognitive ecology. Trends Ecol Evol 8:413–417.[CrossRef]
Rescorla RA, Wagner AR, 1972. A theory of Pavlovian conditioning: variations in the effectiveness of reinforcement and nonreinforcement. In: Classical conditioning: II. Current research and theory (Black AH, Prokasy WF, eds). New York: Appleton-Century-Crofts; 64–99.
Riccio DC, Ackil J, Burch-Vernon A, 1992. Forgetting of stimulus attributes: methodological implications for assessing associative phenomena. Psychol Bull 112:433–445.[CrossRef][Web of Science][Medline]
Richman D, Jackson RR, 1992. A review of the ethology of jumping spiders (Araneae, Salticidae). Bull Br Arachnol Soc 9:33–37.
Riipi M, Alatalo RV, Lindstrom L, Mappes J, 2001. Multiple benefits of gregariousness cover detectability costs in aposematic aggregations. Nature 413:512–514.[CrossRef][Medline]
Roper TJ, 1990. Responses of domestic chicks to artificially coloured insect prey—effects of previous experience and background colour. Anim Behav 39:466–473.[CrossRef]
Servedio MR, 2000. The effects of predator learning, forgetting, and recognition errors on the evolution of warning coloration. Evolution 54:751–763.[CrossRef][Web of Science][Medline]
Shettleworth SJ, 2001. Review: animal cognition and animal behaviour. Anim Behav 61:277–286.[CrossRef]
Sokal RR, Rohlf FJ, 1995. Biometry. New York: W. H. Freeman.
Speed MP, 2001. Can receiver psychology explain the evolution of aposematism? Anim Behav 61:205–216.[CrossRef][Web of Science][Medline]
Speed MP, 2000. Review: warning signals, receiver psychology, and predator memory. Anim Behav 60:269–278.[CrossRef][Web of Science][Medline]
Stamps JA, 1991. Why evolutionary issues are reviving interest in proximate behavioral mechanisms. Am Zool 31:338–348.[Web of Science]
Strohmeyer HH, Stamp NE, 1998. Prey species and prey diet affect growth of invertebrate predators. Ecol Entomol 23:68–79.
Thomas L, Juanes F, 1996. The importance of statistical power analysis: an example from animal behaviour. Anim Behav 52:856–859.[CrossRef]
Thomas RJ, Marples NM, Cuthill IC, Takahashi M, Gibson EA, 2003. Dietary conservatism may facilitate the initial evolution of aposematism. Oikos 101:458–466.[CrossRef][Web of Science]
Toft S, Wise DH, 1999. Behavioral and ecophysiological responses of a generalist predator to single- and mixed-species diets of different quality. Oecologia 119:198–207.[CrossRef][Web of Science]
Tulving E, Thomson DC, 1973. Encoding specificity and retrieval processes in episodic memory. Psychol Rev 80:352–373.[CrossRef][Web of Science]
Weiss MR, Papaj DR, 2003. Colour learning in two behavioural contexts: how much can a butterfly keep in mind? Anim Behav 65:425–434.[CrossRef]
Whittman DW, 1988. Allelochemical interaction among plants, herbivores, and their predators. In: Contemporary learning theory and research. New York: McGraw-Hill Companies, Inc.
Wise DH, Wagner JD, 1992. Evidence of exploitative competition among young stages of the wolf spider Schizocosa ocreata. Oecologia 91:7–13.[Web of Science]
Zhou Y, Riccio DC, 1996. Manipulation of components of context: the context shift effect and forgetting of stimulus attributes. Learn Motiv 27:400–407.[CrossRef][Web of Science][Medline]
CiteULike 
Connotea 
Del.icio.us What's this?
This article has been cited by other articles:
|
|
 |
|
  N. D. VanderSal and E. A. Hebets Cross-modal effects on learning: a seismic stimulus improves color discrimination learning in a jumping spider J. Exp. Biol., October 15, 2007; 210(20): 3689 - 3695. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||