Look before you leap: is risk of injury a foraging cost?

Theory states that an optimal forager should exploit a patch so long as its harvest rate of resources from the patch exceeds its energetic, predation, and missed opportunity costs for foraging. However, for many foragers, predation is not the only source of danger they face while foraging. Foragers also face the risk of injuring themselves. To test whether risk of injury gives rise to a foraging cost, we offered red foxes pairs of depletable resource patches in which they experienced diminishing returns. The resource patches were identical in all respects, save for the risk of injury. In response, the foxes exploited the safe patches more intensively. They foraged for a longer time and also removed more food (i.e., had lower giving up densities) in the safe patches compared to the risky patches. Although they never sustained injury, video footage revealed that the foxes used greater care while foraging from the risky patches and removed food at a slower rate. Furthermore, an increase in their hunger state led foxes to allocate more time to foraging from the risky patches, thereby exposing themselves to higher risks. Our results suggest that foxes treat risk of injury as a foraging cost and use time allocation and daring—the willingness to risk injury—as tools for managing their risk of injury while foraging. This is the first study, to our knowledge, which explicitly tests and shows that risk of injury is indeed a foraging cost. While nearly all foragers may face an injury cost of foraging, we suggest that this cost will be largest and most important for predators.     

Integrated modelling of foraging behaviour, energy budget and memory properties

A predator's foraging performance is related to its ability to acquire sufficient information on environmental profitability. This process can be affected by the patchy distribution and clustering of food resources and by the food intake process dynamics.We simulated body mass growth and behaviour in a forager acting in a patchy environment with patchy distribution of both prey abundance and body mass by an individual-based model. In our model, food intake was a discrete and stochastic process and leaving decision was based on the estimate of net energy gain and searching time during their foraging activities. The study aimed to investigate the effects of learning processes and food resource exploitation on body mass and survival of foragers under different scenarios of intra-patch resource distribution.The simulation output showed that different sources of resource variability between patches affected foraging efficiency differently. When prey abundance varied across patches, the predator stayed longer in poorest patches to obtain the information needed and its performance was affected by the cost of sampling and the resulting assessment of the environment proved unreliable. On the other hand, when prey body mass, but not abundance, varied among the patches the predator was quickly able to assess local profitability. Both body mass and survival of the predator were greatly affected by learning processes and patterns of food resource distribution.     

Patch departure rules in Bumblebees: evidence of a decremental motivational mechanism

The patch living rules of a pollinator, the bumblebee Bombus terrestris L., are studied here in the framework of motivational models widely used for parasitoids: The rewarding events found during the foraging process are supposed to increase or decrease suddenly the tendency of the insect to stay in the current patch and therefore to adjust the patch residence time to the patch profitability. The foraging behaviour of these pollinators was observed in two environment types to determine their patch-leaving decisions. The rich environment was composed of male-fertile flowers, offering pollen and nectar, and the poor one of male-sterile flowers, offering little nectar and no pollen. The experimental design consisted of a patch system in which inflorescences were evenly arranged in two rows (1 m distance). Residence times of foragers inside inflorescences and rows were analysed by a Cox proportional hazards model, taking into account recent and past experience acquired during the foraging bout. Most of the results showed a decremental motivational mechanism, that is, a reduction in the residence time on the inflorescence or in the row related to exploitation of flowers within inflorescences and inflorescences within rows These results indicate that bumblebees tend to leave the patch using departure rules similar to those found in parasitoids. The results also provide information on the memory, learning and evaluating capabilities of bumblebees especially when rich and poor environments were compared. The patch-leaving mechanism suggested by this study is consistent with the central place foraging theory.     

Behavioral resource depression and decaying perceived risk of predation in two species of coexisting gerbils

Summary Behavioral resource depression occurs when the behavior of prey individuals changes in response to the presence of a predator, resulting in a reduction of the encounter rate of the predator with its prey. Here I present experimental evidence on the response of two species of gerbils (Gerbillus allenbyi and G. pyramidum) to the presence of barn owls. I conducted the experiments in a large aviary. Both gerbils responded to the presence of barn owl predators by foraging in fewer resource patches (seed trays) and by quitting foraged resource patches at a higher resource harvest rate (giving-up density of resource; GUD). This reduced the amount of time gerbils were exposed to owl predation, and hence the encounter rate of owls with gerbils, i.e., behavioral resource depression. Thus, the presence of owls imposes a foraging cost on gerbils due to risk of predation, and also on the owls themselves due to resource depression. I then examined how resource depression relaxed over time following exposure to owls. In the days following an encounter with the predator, the reduction in foraging activity for both gerbil species eased. Increasing numbers of trays were foraged each day, and GUDs in seed trays declined. The two gerbils differed in their rate of recovery, with G. pyramidum returning to prepredator levels of foraging after 1 or 2 nights and G. allenbyi taking 5 nights or longer. Interspecific differences in recovery rates may be based on differences between the species in vulnerability to predation and/or ability to detect the presence of predators. The differences in recovery rates may be due to optimal memory windows or decay rates, where differences between species are based on risk of predation or on how perceived risk changes with time since a predator was last encountered. Finally, differences between or among competitors in recovery from resource depression may provide foraging opportunities in time for the species which recover most quickly and may have implications for species coexistence.     

A field investigation of scrounging in semipalmated sandpipers

Animals that forage in groups can produce their own food patches or scrounge the food discoveries of their companions. Mean tactic payoffs are expected to be the same at equilibrium for phenotypically equal foragers. Scrounging is also typically viewed as a risk-averse foraging strategy that provides a more even food intake rate over time. The occurrence of scrounging and the payoffs from different foraging modes have rarely been investigated in the field. Over two field seasons, I examined patch sharing in semipalmated sandpipers (Calidris pusilla) foraging on minute food items at the surface of the substrate. Birds could find patches on their own, a producing event, or join the food patches discovered by others, a scrounging event. I found that the average search time per patch did not differ between producing and scrounging but that the average time spent exploiting a patch was reduced nearly by half when scrounging. As a result, the proportion of time spent exploiting a patch, a measure of foraging payoffs, was significantly lower when scrounging. The variance in payoffs was similar for producing and scrounging. When producing their own patches, individuals that scrounged spent the same proportion of time exploiting a patch as those that only produced. However, within the same individuals, the search time for a scrounged patch was longer than the search time for a produced patch. The results show unequal payoffs for producing and scrounging in this system and suggest that low success in finding patches elicited scrounging.     

Resource distributions affect social learning on multiple timescales

We study how learning is shaped by foraging opportunities and self-organizing processes and how this impacts on the effects of “copying what neighbors eat” on multiple timescales. We use an individual-based model with a rich environment, where group foragers learn what to eat. We vary foraging opportunities by changing local variation in resources, studying copying in environments with pure patches, varied patches, and uniform distributed resources. We find that copying can help individuals explore the environment by sharing information, but this depends on how foraging opportunities shape the learning process. Copying has the greatest impact in varied patches, where local resource variation makes learning difficult, but local resource abundance makes copying easy. In contrast, copying is redundant or excessive in pure patches where learning is easy, and mostly ineffective in uniform environments where learning is difficult. Our results reveal that the mediation of copying behavior by individual experience is crucial for the impact of copying. Moreover, we find that the dynamics of social learning at short timescales shapes cultural phenomena. In fact, the integration of learning on short and long timescales generates cumulative cultural improvement in diet. Our results therefore provide insight into how and when such processes can arise. These insights need to be taken into account when considering behavioral patterns in nature.

Optimal movement strategies for social foragers in unpredictable environments

Spatial movement models often base movement decision rules on traditional optimal foraging theories, including ideal free distribution (IFD) theory, more recently generalized as density-dependent habitat selection (DDHS) theory, and the marginal value theorem (MVT). Thus optimal patch departure times are predicted on the basis of the density-dependent resource level in the patch. Recently, alternatives to density as a habitat selection criterion, such as individual knowledge of the resource distribution, conspecific attraction, and site fidelity, have been recognized as important influences on movement behavior in environments with an uncertain resource distribution. For foraging processes incorporating these influences, it is not clear whether simple optimal foraging theories provide a reasonable approximation to animal behavior or whether they may be misleading. This study compares patch departure strategies predicted by DDHS theory and the MVT with evolutionarily optimal patch departure strategies for a wide range of foraging scenarios. The level of accuracy with which individuals can navigate toward local food sources is varied, and individual tendency for conspecific attraction or repulsion is optimized over a continuous spectrum. We find that DDHS theory and the MVT accurately predict the evolutionarily optimal patch departure strategy for foragers with high navigational accuracy for a wide range of resource distributions. As navigational accuracy is reduced, the patch departure strategy cannot be accurately predicted by these theories for environments with a heterogeneous resource distribution. In these situations, social forces improve foraging success and have a strong influence on optimal patch departure strategies, causing individuals to stay longer in patches than the optimal foraging theories predict.     

Foraging in a tidally structured environment by Red Knots (Calidris canutus): ideal, but not free

Besides the "normal" challenge of obtaining adequate intake rates in a patchy and dangerous world, shorebirds foraging in intertidal habitats face additional environmental hurdles. The tide forces them to commute between a roosting site and feeding grounds, twice a day. Moreover, because intertidal food patches are not all available at the same time, shorebirds should follow itineraries along the best patches available at a given time. Finally, shorebirds need additional energy stores in order to survive unpredictable periods of bad weather, during which food patches are covered by extreme tides. In order to model such tide-specific decisions, we applied stochastic dynamic programming in a spatially explicit context. Two assumptions were varied, leading to four models. First, birds had either perfect (ideal) or no (non-ideal) information about the intake rate at each site. Second, traveling between sites was either for free or incurred time and energy costs (non-free). Predictions were generated for three aspects of foraging: area use, foraging routines, and energy stores. In general, non-ideal foragers should feed most intensely and should maintain low energy stores. If traveling for such birds is free, they should feed at a random site; otherwise, they should feed close to their roost. Ideal foragers should concentrate their feeding around low tide (especially when free) and should maintain larger energy stores (especially when non-free). If traveling for such birds is free, they should feed at the site offering the highest intake rate; otherwise, they should trade off travel costs and intake rate. Models were parameterized for Red Knots (Calidris canutus) living in the Dutch Wadden Sea in late summer, an area for which detailed, spatially explicit data on prey densities and tidal heights are available. Observations of radio-marked knots (area use) and unmarked knots (foraging routines, energy stores) showed the closest match with the ideal/non-free model. We conclude that knots make state-dependent decisions by trading off starvation against foraging-associated risks, including predation. Presumably, knots share public information about resource quality that enables them to behave in a more or less ideal manner. We suggest that our modeling approach may be applicable in other systems where resources fluctuate in space and time.     

The use of waggle dance information by honey bees throughout their foraging careers

We studied the extent to which worker honey bees acquire information from waggle dances throughout their careers as foragers. Small groups of foragers were monitored from time of orientation flights to time of death and all in-hive behaviors relating to foraging were recorded. In the context of a novice forager finding her first food source, 60% of the bees relied, at least in part, on acquiring information from waggle dances (being recruited) rather than searching independently (scouting). In the context of an experienced forager whose foraging has been interrupted, 37% of the time the bees resumed foraging by following waggle dances (being reactivated) rather than examining the food source on their own (inspecting). And in the context of an experienced forager engaged in foraging, 17% of the time the bees initiated a foraging trip by following a waggle dance. Such dance following was observed much more often after an unsuccessful than after a successful foraging trip. Successful foragers often followed dances just briefly, perhaps to confirm that the kind of flowers they had been visiting were still yielding forage. Overall, waggle dance following for food discovery accounted for 12–25% of all interactions with dancers (9% by novice foragers and 3–16% by experienced foragers) whereas dance following for reactivation and confirmation accounted for the other 75–88% (26% for reactivation and 49–62% for confirmation). We conclude that foragers make extensive use of the waggle dance not only to start work at new, unfamiliar food sources but also to resume work at old, familiar food sources.     

Division of foraging labor in ants can mediate demands for food and safety

Solitary foragers can balance demands for food and safety by varying their relative use of foraging patches and their level of vigilance. Here, we investigate whether colonies of the ant, Formica perpilosa, can balance these demands by dividing labor among workers. We show that foragers collecting nectar in vegetation near their nest are smaller than are those collecting nectar at sites away from the nest. We then use performance tests to show that smaller workers are more likely to succumb to attack from conspecifics but feed on nectar more efficiently than larger workers, suggesting a size-related trade-off between risk susceptibility and harvesting ability. Because foragers that travel away from the nest are probably more likely to encounter ants from neighboring colonies, this trade-off could explain the benefits of dividing foraging labor among workers. In a laboratory experiment, we show that contact with aggressive workers results in an increase in the mean size of recruits to a foraging site: this increase was not the result of more large recruits, but rather because fewer smaller ants traveled to the site. These results suggest that workers particularly susceptible to risk avoid dangerous sites, and suggest that variation in worker size can allow colonies to exploit profitably both hazardous and resource-poor patches.Communicated by L. Sundström     

Spatio-temporal ranging behaviour and its relevance to foraging strategies in wide-ranging wolverines

Conservation of carnivores in an increasingly changing environment is greatly helped by understanding the decision-making processes underlying habitat patch choice. Foraging theory may give us insight into spatio-temporal search patterns and consequent foraging decisions that carnivores make in heterogeneous and fluctuating environments. Constraints placed on central-place foragers in particular are likely to influence both foraging decisions and related spatio-temporal movement patterns. We used discrete choice models to investigate the spatio-temporal ranging behaviour of GPS collared female wolverines (Gulo gulo) with dependent cubs in south-central Norway. Activity patterns, home range use and selection for elevation were analyzed in relation to spatial and temporal covariates (daily and seasonal) and related to different foraging behaviours. In spring, wolverines showed restricted movement patterns around rendezvous sites at high elevations by day, whereas during the night animals were active at lower elevations. Over the summer, this daily pattern in intensity of use diminished and their overall selectiveness for elevation decreased as cubs grow more mobile and independent. At the onset of autumn, wolverines showed intensive use of the profitable forest-alpine tundra ecotone. We argue that reproducing wolverines deployed a foraging strategy attuned to altering their movement patterns throughout the summer to address a continuous, but diminishing, trade-off between providing both food and shelter for their offspring. Incorporating spatially and temporally explicit activity patterns and home range use in discrete choice resource selection models thus enhances the understanding of the motives behind wolverine resource utilization in space and time. Such knowledge may provide guidance to managers designing regional-scale zoning, in order to facilitate carnivore recovery and to minimize conflicts with human activities.     

Optimal within-patch movement strategies for optimising patch residence time: an agent-based modelling approach

Several optimisation models, like the marginal value theorem (MVT), have been proposed to predict the optimal time foraging animals should remain on patches of resources. These models do not clearly indicate, however, how animals can follow the corresponding predictions. Hence, several proximate patch-leaving decision rules have been proposed. Most if not all of these are based on the animals’ motivation to remain on the patches, but the real behaviours involved in such motivation actually still remain to be identified. Since animals are usually exploiting patches of resources by walking, we developed a model simulating the intra-patch movement decisions of time-limited animals exploiting resources distributed in delimited patches in environments with different resource abundances and distributions. The values of the model parameters were optimised in the different environments by means of a genetic algorithm. Results indicate that simple modifications of the walking pattern of the foraging animals when resources are discovered can lead to patch residence times that appear consistent with the predictions of the MVT. These results provide a more concrete understanding of the optimal patch-leaving decision rules animals should adopt in different environments.     

Comparisons of forager distributions from matched honey bee colonies in suburban environments

We conducted experiments designed to examine the distribution of foraging honey bees (Apis mellifera) in suburban environments with rich floras and to compare spatial patterns of foraging sites used by colonies located in the same environment. The patterns we observed in resource visitation suggest a reduced role of the recruitment system as part of the overall colony foraging strategy in habitats with abundant, small patches of flowers. We simultaneously sampled recruitment dances of bees inside observation hives in two colonies over 4 days in Miami, Florida (1989) and from two other colonies over five days in Riverside, California (1991). Information encoded in the dance was used to determine the distance and direction that bees flew from the hive for pollen and nectar and to construct foraging maps for each colony. The foraging maps showed that bees from the two colonies in each location usually foraged at different sites, but occasionally they visited the same patches of flowers. Each colony shifted foraging effort among sites on different days. In both locations, the mean flight distances differed between colonies and among days within colonies. The flight distances observed in our study are generally shorter than those reported in a similar study conducted in a temperate deciduous forest where resources were less dense and floral patches were smaller.     

Social foraging by honeybees: how colonies allocate foragers among patches of flowers

Summary To understand how a colony of honeybees keeps its forager force focussed on rich sources of food, and analysis was made of how the individual foragers within a colony decide to abandon or continue working (and perhaps even recruit to) patches of flowers. A nectar forager grades her behavior toward a patch in response to both the nectar intake rate of her colony and the quality of her patch. This results in the threshold in patch quality for acceptance of a patch being higher when the colonial intake rate of nectar is high than when it is low. Thus colonies can adjust their patch selectivity so that they focus on rich sources when forage is abundant, but spread their workers among a wider range of sources when forage is scarce. Foragers assess their colony's rate of nectar intake while in the nest, unloading nectar to receiver bees. The ease of unloading varies inversely with the colonial intake rate of nectar. Foragers assess patch quality while in the field, collecting nectar. By grading their behavior steeply in relation to such patch variables as distance from the nest and nectar sweetness, foragers give their colony high sensitivity to differences in profitability among patches. When a patch's quality declines, its foragers reduce their rate of visits to the patch. This diminishes the flow of nectar from the poor patch which in turn stimulates recruitment to rich patches. Thus a colony can swiftly redistribute its forager force following changes in the spatial distribution of rich food sources. The fundamental currency of nectar patch quality is not net rate of energy intake, (Gain-Cost)/Time, but may be net energy efficiency, (Gain-Cost)/Cost.     

The role of internal and external information in foraging decisions of Melipona workers (Hymenoptera: Meliponinae)

Social insect foragers have to make foraging decisions based on information that may come from two different sources: information learned and memorised through their own experience (“internal” information) and information communicated by nest mates or directly obtained from their environment (“external” information). The role of these sources of information in decision-making by foragers was studied observationally and experimentally in stingless bees of the genus Melipona. Once a Melipona forager had started its food-collecting career, its decisions to initiate, continue or stop its daily collecting activity were mainly based upon previous experience (activity on previous days, the time at which foraging was initiated the day(s) before, and, during the day, the success of the last foraging flights) and mediated through direct interaction with the food source (load size harvested and time to collect a load). External information provided by returning foragers advanced the start of foraging of experienced bees. Most inexperienced bees initiated their foraging day after successful foragers had returned to the hive. The start of foraging by other inexperienced bees was stimulated by high waste-removal activity of nest mates. By experimentally controlling the entries of foragers (hence external information input) it was shown that very low levels of external information input had large effect on the departure of experienced foragers. After the return of a single successful forager, or five foragers together, the rate of forager exits increased dramatically for 15 min. Only the first and second entry events had large effect; later entries influenced forager exit patterns only slightly. The results show that Melipona foragers make decisions based upon their own experience and that communication stimulates these foragers if it concerns the previously visited source. We discuss the organisation of individual foraging in Melipona and Apis mellifera and are led to the conclusion that these species behave very similarly and that an information-integration model (derived from Fig. 1) could be a starting point for future research on social insect foraging. Received: 16 April 1997 / Accepted after revision: 30 August 1997     

Assessing the benefits of cooperation in honeybee foraging: search costs,forage quality,and competitive ability

Summary The foragers in honeybee colonies cooperate by sharing information about rich sources of food. This study examines three hypotheses about the benefits of this cooperation: (H1) it decreases foragers' costs in finding new food sources, (H2) it increases the quality of the food sources located by foragers, and (H3) it increases the ability of a colony's foragers to compete for high-quality food sources. To test each hypothesis, we identified a critical pattern in the foraging process which, if observed, would cast doubt on that hypothesis, and then gathered data to check for these patterns. Our observations do not support the first hypothesis, but do support the second and third. These results, in addition to helping us understand the functional significance of the honeybee's dance language, provide insights into the colonial organization of foraging by honeybees.     

Vicarious sampling: the use of personal and public information by starlings foraging in a simple patchy environment

Group foragers may be able to assess patch quality more efficiently by paying attention to the sampling activities of conspecifics foraging in the same patch. In a previous field experiment, we showed that starlings foraging on patches of hidden food could use the successful foraging activities of others to help them assess patch quality. In order to determine whether a starling could also use another individual’s lack of foraging success to assess and depart from empty patches more quickly, we carried out two experimental studies which compared the behaviour of captive starlings sampling artificial patches both when alone and when in pairs. Solitary starlings were first trained to assess patch quality in our experimental two-patch system, and were then tested on an empty patch both alone and with two types of partner bird. One partner sampled very few holes and thus provided a low amount of public information; the other sampled numerous holes and thus provided a high amount of public information. In experiment 1, we found no evidence of vicarious sampling. Subjects sampled a similar number of empty holes when alone as when with the low and high information partners; thus they continued to rely on their own personal information to make their patch departure decisions. In experiment 2, we modified the experimental patches, increasing the ease with which a bird could watch another’s sampling activities, and increasing the difficulty of acquiring accurate personal sampling information. This time, subjects apparently did use public information, sampling fewer empty holes before departure when with the high-information partner than when with the low-information partner, and sampling fewer holes when with the low-information partner than when alone. We suggest that the degree to which personal and public information are used is likely to depend both on a forager’s ability to remember where it has already sampled and on the type of environment in which foraging takes place. Received: 31 January 1995/Accepted after revision: 11 September 1995     

Locating food in a spatially heterogeneous environment: Implications for fitness of the macrodecomposer Armadillidium vulgare (Isopoda: Oniscidea)

To assess the fitness consequences of foraging on patchy resources, consumption rates, growth rates and survivorship of Armadillidium vulgare were monitored while feeding in arenas in which the spatial distribution of patches of high quality food (powdered dicotyledonous leaf litter) was varied within a matrix of low quality food (powdered grass leaf litter). Predictions from behavioural experiments that these fitness correlates would be lower when high quality food is more heterogeneously distributed in space were tested but not supported either by laboratory or field experiments. To investigate whether A. vulgare can develop the ability to relocate high quality food patches, changes in foraging behaviour, over a comparable time period to that used in the fitness experiments, were monitored in arenas in which there was a high quality food patch in a low quality matrix. A. vulgare increased its ability to relocate the position of high quality food over time. It reduced time spent in low quality food matrices and increased time spent in high quality food patches with time after the start of the experiment. When the position of a high quality food patch was moved, the time spent in the low quality food matrix increased and less time was spent in high quality food patches, compared to arenas in which the food was not moved. The fitness benefits for saprophages of developing the ability to relocate high quality patches while foraging in spatially heterogeneous environments are discussed.     

Nonlinear effects of food aggregation on interference competition in mallards

Previous studies of interference competition have shown an asymmetric effect on intake rate of foragers on clumped resources, with only subordinate individuals suffering. However, the food distributions in these studies were uniform or highly clumped, whereas in many field situations, food aggregation is intermediate. Here we investigated whether food distribution (i.e., uniform, slightly clumped, and highly clumped) affects the behavioral response of mallards foraging alone or competing with another. Although the amount of food was the same in all distributions, the mallards reached higher intake rates, visited fewer patches, and showed longer average feeding times in the highly clumped distribution. Competing mallards had lower intake rates on the slightly clumped than on the uniform or highly clumped food distributions. Subordinates generally visited more patches and had shorter feeding times per patch, but their intake rates were not significantly lower than those of dominants. Therefore, we propose that subordinates do not necessarily suffer from interference competition in terms of intake rate, but do suffer higher search costs. In addition, although dominants had significantly higher average feeding times on the best quality patches of the highly clumped food distribution, such an effect was not found in the slightly clumped distribution. These findings indicate that in environments where food is aggregated to a lesser extent, monopolization is not the best strategy for dominants. Our results suggest that interference experiments should use food distributions that resemble the natural situation animals are faced with in the field.     

Worker piping in honey bees (<Emphasis Type="Italic">Apis mellifera</Emphasis>): the behavior of piping nectar foragers

This study investigates the brief piping signals ("stop signals") of honey bee workers by exploring the context in which worker piping occurs and the identity and behavior of piping workers. Piping was stimulated reliably by promoting a colony's nectar foraging activity, demonstrating a causal connection between worker piping and nectar foraging. Comparison of the behavior of piping versus non-piping nectar foragers revealed many differences, e.g., piping nectar foragers spent more time in the hive, started to dance earlier, spent more time dancing, and spent less time on the dance floor. Most piping signals (approximately 99%) were produced by tremble dancers, yet not all (approximately 48%) tremble dancers piped, suggesting that the short piping signal and the tremble dance have related, but not identical, functions in the nectar foraging communication system. Our observations of the location and behavior of piping tremble dancers suggest that the brief piping signal may (1) retard recruitment to a low-quality food source, and (2) help to enhance the recruitment success of the tremble dance.