Games spiders play: Behavioral variability in territorial disputes

Summary

• 1. The agonistic behavior of the funnel-web building spider, Agelenopsis aperta (Gertsch), was studied using induced encounters between adult females at natural web sites. All behavior exhibited by either individual during the course of an encounter was recorded. The results were analyzed through the use of transition matrices and the following multivariate treatments: factor analysis, ordination, and multiple regression. These latter methods were used to provide insight into possible sources of variability and their underlying causes.

• 2. Sequence outcome is primarily determined by the relative weight of the two contestants engaged in a territorial dispute. If the size difference is large, the larger of the two individuals wins in a significant number of cases. Home bias is evidenced in cases where body weights are close.

• 3. A stereotypy measure is devised that reflects the percent similarity of each sequence of events to an expected sequence (average). The frequency distribution, presence-absence or duration of all behavior patterns observed during the course of the disputes are utilized in measuring stereotypy. The territorial disputes of A. aperta exhibit low stereotypy, averaging 43% on a scale from 0 to 100%.

• 4. Low stereotypy is, in part, related to the utilization of 33 different action patterns by spiders in these disputes. Factor analysis is used to express these action patterns in terms of five functional groups including locating behavior, signaling behavior, threat behavior, contact behavior, and a multiple function category. The order of these categories represents increasing cost based on relative estimates of the energy expenditure necessary to complete a specific behavior pattern and the potential for injury through the use of it.

• 5. A Bray and Curtis ordination shows the important sources of between sequence variability to be the total energetic cost of the dispute and the complexity of behavior exhibited in it. The factors, in turn, depend on the resident-visitor status of the losing spider and on the relative size of the two contestants. The energetic cost of a dispute is markedly higher in those disputes in which the resident loses her territory. The number of action patterns and total frequency of acts observed are also greater in encounters in which the resident is the losing spider. Behavioral complexity is higher as well in cases where the weights of the two contestants are close.

• 6. The variation in the pathways through which the sequences progress is shown to reflect the operation of assessment strategies by A. aperta. Initial assessment of the relative weights of the opponent is made through movements on the web at a distance (locating behavior). Subsequent activities depend on the results of this assessment, the predominant strategy being ‘retaliator’ (Maynard Smith and Price, 1973) in which an individual responds to escalation with further escalation. Spiders with a large weight advantage over the opponent tend to escalate directly to threat and contact behavior (‘hawk’ strategy). The corresponding strategy for a much smaller visiting spider is immediate retreat (‘mouse’). A much smaller resident spider, however, will exhibit the ‘retaliator’ strategy to the ‘hawk’ rather than the more conservative ‘mouse’ strategy. The particular stategy exhibited, then, also depends on the energetic investment a particular individual has in the contended resource.

• 7. Within functional group variability is shown to significantly affect the outcome of territorial disputes. Winning spiders exhibit an average of 20% less stereotypy than losing spiders. Unpredictable behavior possibly confuses the opponent, causing it to make inaccurate assessments of the weights of the opponent relative to it. Retreat follows. This behavior is linked to the ‘protean displays’ exhibited in defense against predators (Humphries and Driver, 1967).