374 
Psyche 
[Vol. 85 
distributions were not significantly different among colonies 
(ANOVA, F = .09, 2 df, P < .05). Head widths were relatively 
invariant in this species (x = .552 mm, s = .043 mm) such that the 
largest individual was less than 1.5X the smallest (Figure 3). Thus 
the potential for size-biased polyethism was quite restricted. 
Head widths of workers displaying different ethogram behaviors 
were considered. Data were again pooled over three colonies since 
mean head widths did not differ significantly for any behavior 
(ANOVA with variable df; in all cases P < .05). Figure 4 illustrates 
statistics of head width for workers displaying each behavior. No 
obvious differentiation of head width according to roles can be 
discerned. Analysis of data in Figure 4 is summarized in Table 7, 
which is based on ANOVA’s for differences between mean head 
widths (LSD tests). Most of the significant differences separate 
workers regurgitating with larvae from other behaviors (row and 
column headed RL in Table 7). That workers displaying RL were on 
average larger than others is evident from Figure 4. Starred entries 
of Table 7 are sporadic; certainly patterns of differences in mean 
head width showed no clear segregation by roles. Workers exhibit- 
ing a brood care behavior were not more similar in size to those 
displaying other brood care behaviors than they were to workers 
involved in social interactions or provisioning. Thus there was no 
apparent morphological bias underlying polyethism for L. am- 
biguus. 
The overall picture that emerges of L. ambiguus social organ- 
ization is short-term specialization of individual on task according 
to four roles: brood care, social interactions, physical nest mainte- 
nance, and provisioning (Table 3). However, the division of labor 
was rather loose, since switching between roles was often observed 
over 30-minute periods (Table 4). The nonrandom co-occurrence of 
sets of behaviors (Figure 2) statistically reinforced inferences about 
polyethism from the transition matrix. Finally, no strong morpho- 
logical bias was demonstrated for ants specializing on specific tasks. 
Comparison with L. longispinosus 
Results of this study were compared with data from its closely- 
related congener L. longispinosus; such comparisons were valid 
since all observations were conducted in the same laboratory using 
standard husbandry techniques. The major difference in culture 
conditions between species was the addition of fruitflies to L. ambi- 
guus diets. The earlier study had not incorporated feeding insect prey 
