INTRODUCTION

The European social wasp, Vespula germanica (Fab.), is a successful invasive species which has established in many parts of the world (e.g. Archer, 1998). It was first seen in Northwestern Patagonia, Argentina, around 1980, becoming in few years, one of the most abundant insects in the Andean region (D’Adamo et al., 2002; Farji-Brener and Corley, 1998). An important characteristic of this species that enables workers to forage highly efficiently, is the ability to collect form large, stationary food sources in addition to hunting for small, active ones (Greene, 1991). These wasps have one of the most eclectic diets known for insects. In addition to preying on many invertebrates, they also take flesh and produce wounds to imbibe flood from living vertebrates, scavenge on soft tissues of vertebrate and invertebrate carrion, feed on many different fruits as well as other natural carbohydrates and consume a wide array of processed human food, both carbohydrate and proteinaceous (for review see Greene, 1991). It has been suggested that their foraging behavior is probably one of the most important factors favoring rapid colony growth in new areas (Farji-Brener and Corley, 1998). Several studies on V. germanica foraging behavior and social communication (D’Adamo et al., 2000, 2001, 2003, 2004) concluded that behavioral mechanisms might be crucial for its successful establishment. These mechanisms would be related both to the social and the individual level. In the first case, it is known that these wasps are able to attract nestmates to highly rewarding food sources either by local enhancement processes (D’Adamo et al., 2000, 2001, 2003, 2004) or by inducing nestmates to search for an odor sampled inside the nest (Overmyer and Jeanne, 1998; Jandt and Jeanne, 2005). At the individual level, studies regarding the re-localization of a food reward have begun to be conducted (D’Adamo and Lozada, 2003). In the present work we will further analyze this latter aspect of foraging behavior in V. germanica wasps.

Vespids often return to rich food sites (Takagi et al., 1980; Raveret Richter and Jeanne, 1985). Therefore, they must learn how to navigate from their nest to the feeding place and then locate the food source. Global and local landmarks are used for guiding wasps to re-locate these feeding sites (Collet and Zeil, 1998). The acquisition of landmark information occurs through specialized learning flights during which the wasp examines the location and stores necessary information of the place to which it will return (for review see Zeil et al., 1996). This is an important aspect of V. germanica foraging behavior, since scavengers frequently return to non-depleted food sources.

Associative learning, involves the establishment of a temporal or spatial link between two stimuli or a stimulus and a response. In previous studies we found that V. germanica wasps trained to feed from a certain array, persisted visiting the feeder after removing the food. Thus, we demonstrated that wasps learned to associate some food reward (unconditioned stimulus) with certain visual, spatial and odor cues (conditioned stimuli) (D’Adamo and Lozada, 2003, and unpublished data). However, there are no studies on this species analyzing the conditioned response decrease after the reward has been removed. This behavior, called memory extinction (e.g. Pavlov, 1927; Bouton, 2004), implies the reduction of the conditioned response (CR) by presenting the conditioned stimulus (CS) without reinforcement (i.e. food reward).

In the present study we investigate how long wasps continue searching for a meat food source that is no longer available. We first trained wasps to feed on a dish surrounded by four yellow cylinders, and then, during the testing phase, we removed the food, and recorded foragers’ behavior until wasps stopped visiting the array. We compared wasps’ responses when receiving one or three feeding trials. As the acquisition of a conditioned response is a function of the number of reinforcements (e.g. Hall, 1994; Gallistel et al., 2004), we hypothesize that wasps visiting more frequently a food source, will spend more time searching for it once this is no longer available. Then, we predict that 1) wasps trained with one feeding trial will hover and land less frequently over the array than wasps trained with three trials, 2) wasps trained with one feeding trial will spend less time searching over the array than wasps trained with three trials, 3) initial level of response (i.e. number of hovers and landings in the first visit during the testing phase) will be higher for wasps receiving three feeding trials than for wasps receiving one.

MATERIALS AND METHODS

All experiments were carried out under natural conditions near San Carlos de Bariloche (41°S, 71°W), Argentina, during the period of major activity of V. germanica wasps (February–April) in 2005. The experiments were conducted near the pebbly shoreline of a lake, under similar weather conditions (sunny and still). All experiments involved training individual forager wasps to feed on a white plastic dish (diameter=7 cm) containing 20 g of minced fresh meat, which was surrounded by four yellow cylinders 2 cm diameter and 60 cm height. These cylinders were distributed in the shape of a square of 25 cm side, in the centre of which the feeding dish was placed. An observer sat in front of this array, at 0.5 m distance. When one forager spontaneously arrived at the dish and was feeding, it was distinguishably marked with a dot of washable paint on the abdomen for further identification. This procedure minimally disturbed wasps as they were not captured for marking. Any other wasp visiting the dish was removed, in order to work with only one experimental individual per training session. At each training session, the studied wasp fed and collected food from the dish, then departed to the nest and returned a few minutes later. An individual wasp was used for only one experiment and one treatment.

Training sessions consisted on either one or three feeding trials (depending on the treatment), during which an individually marked wasp was allowed to feed on a dish with protein bait. In this sense, one feeding trial consisted on one discrete visit by a forager to the feeding dish, demarcated from the next feeding trial by a return to its nest. For the first treatment, with one feeding trial, the feeding dish was immediately replaced by an empty one following a forager’s first departure to the nest with its load, and for the second treatment, this was done following three visits (feeding trials) to the feeding station and returns to the nest with its load by the forager. During the testing phase, a clean dish (without food) replaced the original feeding dish. When the wasp returned in its second visit (or fourth, depending on the training received), the number of hovers and landings over the array were recorded. We considered a hovering episode occurred when the flying wasp remained in the same place, beating its wings, over the arrangement (either cylinders or feeder) but did not land on it. If the wasp moved away and then returned hovering again, this was recorded as another hovering episode. But if the wasp remained hovering on the same place during, for example 1 min, this was recorded as one discrete incidence of hovering. On the other hand, landing occurred when the wasp touched, with its six legs, the area delimited by the four cylinders or the empty feeder. If the wasp landed on the array, flied away, and then returned landing again, this was considered another discrete incidence of landing. But if the wasp remained landed or walking on the array, this was recorded as one landing episode. Forager behavior was scored by recording during the testing phase, the number of hovers and landings during all the time the wasp visited the experimental set up, until it did not come for 60 min (cut off period). Assays corresponding to each treatment were done by the same person. Therefore, one conducted the experiments of one feeding trial and another person recorded those corresponding to three feeding trials.

We did 17 replicates for the treatment consisting on one feeding visit during training and 15 replicates for the treatment consisting on three feeding visits. Paired comparisons between landing and hovering responses were conducted by Wilcoxon matched-pair test and comparisons between groups with different treatments were conducted by Mann Whitney U test.

RESULTS

Number of visits (hovers and landings) over the array, composed by the feeder and the four cylinders, was significantly higher when wasps were trained with three feeding trials than when trained with one (Mann Whitney U test, Z=4.22, p < 0.0001, N 1,2=15, 17; Z=4.57, p < 0.0001, N 1,2=15, 17 for hovers and landings respectively), (Fig. 1). Moreover, the total time wasps spent searching for food once this was removed differed according to the number of trials received during the training. Mean time spent searching over the original array was x=80.13 min. (±12.55 min.) in wasps trained with three trials, while x=26.72 min. ±2.71 min.) in wasps that had received one trial. This difference is significantly different (Mann Whitney U test, Z=4.248, p < 0.0001, N 1,2=15, 17).

Fig. 1.
figure 1

Mean number of wasp visits (i.e. hovers and landings), ± standard error, during the testing phase, over the array where wasps were previously trained with three or one feeding trials, *** p < 0.0001, ** p < 0.001, N=15 for wasps trained with three trials and N=17 for wasps trained with one trial.

Full size image

Hovering behavior was significantly more frequent than landing behavior both in wasps trained with three feeding trials (Wilcoxon matched-pair test, Z=3.408, p < 0.0007, N=15) and wasps trained with one trial (Wilcoxon matched-pair test, Z=3.621, p < 0.0003, N=17). This reflects that searching was the dominant behavior wasps performed once food was not available. Fists hovers occurred more frequently in wasps trained with three feeding trials than in wasps trained with one (Mann Whitney U test, Z=2.47, p < 0.014, N 1,2=16, 17). Furthermore, as previously mentioned, wasps trained with one trial stopped hovering over the array before than wasps trained with three trials (Fig. 2(a)).

Fig. 2.
figure 2

Time course of wasps’ visits over the array without food, when trained with three or one feeding trials; a) extinction of hovering responses, b) extinction of landing responses. Means and standard errors are shown.

Full size image

Landing responses also differed between treatments. First landings occurred more frequently in wasps trained with three feeding trials than in wasps trained with one trial (Fig. 2(b)) (Mann Whitney U test, Z=4.18, p < 0.0001, N 1,2=16, 17). Furthermore, landing frequency in wasps trained with one trial was very low.

DISCUSSION

The present study shows that, as predicted, when V. germanica foragers are feeding on a non-depleted food source, they search for it for a certain period of time once this has been removed. The time spent searching for this resource was approximately three times longer in wasps receiving three consecutive trials than in those receiving one. Furthermore, wasps trained with three feeding trials hovered and landed more frequently over the array than wasps trained with one. Initial level of response during the testing phase was significantly higher in wasps that had received three feeding trials than in wasps receiving one. This initial difference occurs both for hovering and landing responses. We like to highlight that, once food was not available, hovering behavior was more frequent than landing responses. This occurred in both treatments and probably indicates that searching is the dominant behavior wasps perform when looking for a previously visited food source.

It is important to study learning processes related to foraging behavior under natural circumstances (Shettlewoth, 1994; Menzel, 1999). Our work was carried out under natural conditions and involved feasible circumstances for V. germanica wasps. In our experimental design, we removed meat after wasps had fed on it, and in nature, wasps exploiting carrion may find that suddenly the resource has been removed by another predator. Therefore, we think that it might be adaptive to extinguish the association between a particular place with certain food. In fact, memory extinction is an important phenomenon that allows the organism to adapt its behavior to a changing environment (Bouton, 2004). In the present study, we analyzed short term memory using an appetitive simple paradigm. It is interesting to note that wasps did not return to the conditioned stimulus 24 h later after three training sessions (data not shown) suggesting that, under these experimental conditions, long term memory does not seem to occur. However, it would be interesting to study memory consolidation after a stronger acquisition treatment.

In the present experimental design, V. germanica wasps learnt to associate the dish with the food with only one trial. We consider that this is very interesting and we do not know about any study showing one trial learning in this species. In honeybees, a single learning trial initiates time-dependent processes, leading to high retention immediately after the trial, low retention 2–4 min later, consolidation to a high level during the next 10–15 min and vanishing retention overall several days (Menzel and Müller, 1996). In our experiment, wasps were tested immediately after the training, for a maximum of 120 min. Therefore, if memory processes in V. germanica are similar to those found in honeybees, our observations were conducted during the early phase of memory consolidation. This early phase depends on the strength of the US and is highly sensitive to extinction (Menzel and Müller, 1996). Our results agree with this, as wasps receiving stronger US (i.e. three feeding trials) were less sensitive to extinction than wasps receiving a weaker US (i.e. one feeding trial).

In conclusion, wasps exploiting a rich food source which suddenly disappears, continue visiting the site where they had been feeding for a certain period of time. How long wasps continue searching in this place depends on the number of feeding visits wasps had previously experienced. As natural environments imply uncertainty, and for V. germanica foragers, food may suddenly be removed by any other predator, it would be adaptive for the species to have the plasticity to extinguish an association previously established which is no longer beneficial. It has biological significance that this memory extinction, as also the initial level of response, depend on the number of experiences foragers have with food, since this is related to the uncertainty level (i.e. if food disappears after one feeding visit, uncertainty is higher than if it disappears after three feeding visits). The ability to extinguish differently an association between a stimulus and a food resource could be one of the various behavioral mechanisms in V. germanica that had allowed this species to establish in new areas of the world. However, from another perspective, one could have expected that it should be more adaptive to spend less time searching for food as learning processes become consolidated. This hypothesis sounds suitable, although our results do not support it but rather point out that as more contacts wasps have with food, more time will be spent searching for it once it was removed. This could be a consequence of wasp’s nervous system constrains or might indicate that coming back to sites where foragers have found food repeatedly is an adaptive strategy. In fact, it has been shown that vespids often return to sites where they have been successful (Takagi et al., 1980; Raveret Richter and Jeanne, 1985).