colony member more than any member chased it; rank 2 chased each colony member 

 more than any other member chased it, except for rank 1. 



For analysis, data on chases were ordered by rank, and all data for each 

 rank were summed across all 7 groups. The data were then divided into three 

 categories: (a) the number of chases received by each rank; (b) the number of 

 chases given by each rank; and (c) the number of chases given by each rank to 

 each other rank. Each of these categories was analyzed using the Kruskal -Wal lis 

 one-way ANOVA to determine whether or not there was a significant difference 

 among ranks, i.e., an H-value giving p < 0.05. If so, then the data were 

 subjected to a nonparametric version of the Newman-Keuls Multiple Range test 

 (Zar, 1974) to determine how the distribution of chases varied among ranks 

 (significance level, p < 0.05). 



RESULTS 



For each colony, a distinct size-dependent social hierarchy was apparent; 

 it was highly linear with proportionally very few reversals (4.5%). A reversal 

 is a chase directed at an individual higher in the hierarchy than the chaser. 



Figure 1 shows the distribution of chases received by each rank (category 

 a). This distribution is not homogeneous (Kruskal -Wal 1 is, p < 0.001). The 

 nonparametric Multiple Range test showed that ranks 3 and 4 received the most 

 chases; ranks 2, 5, and 6 received less than 3 and 4 but more than rank 1 (p < 

 0.05 in each case). 



Figure 2 shows the distribution of chases given by each rank (category b). 

 This distribution is not homogeneous (Kruskal -Wal 1 is, p < 0.001). The 

 nonparametric Multiple Range test showed that ranks 1, 2, and 3 gave the most 

 chases; rank 4 gave fewer chases but more than ranks 5 and 6 (p < 0.05 in each 

 case). 



Figure 3 shows the distribution of chases from each rank to each rank 

 below it in the hierarchy (category c). The distribution is not homogeneous 

 for ranks 1 to 4 (Kruskal -Wal 1 is , p < 0.01). In most cases, each rank directed 

 most chases to its closest ranking subordinate and least to the individual 

 ranked farthest from itself. See figure 3 for the significant differences 

 shown by the nonparametric Multiple Range test (p < 0.05 in each case). 



CONCLUSIONS 



The analyses of the distribution of chases given and received show that 

 there is a distinct pattern to intracolony aggression: 



(a) there is a strong size-dependent social hierarchy which is linear 

 with few reversals; 



(b) individuals chase those closest to and just below themselves in rank 

 most frequently and those furthest in rank least frequently; 



(c) the highest ranking individuals (1, 2, and 3) chase more than other 

 ranks. Lower ranking individuals chase progressively less the lower their rank; 



131 



