302 
Psyche 
[Vol. 95 
paired presentations across groups can be analysed using the Wil- 
coxon paired sample test (Zar 1974). Occasionally more than one 
ant was involved in the retrieval of larvae, and in some instances 
both larvae were judged to be taken simultaneously. For the pur- 
pose of analysis those ties were omitted and the number of presenta- 
tions for that group consequently reduced. 
Following the series of paired presentations, marked and un- 
marked larvae which had not been taken were re-introduced in one 
large pile on a glass coverslip outside the nest entrance. The amount 
of time taken to retrieve the remaining larvae was recorded (if time 
allowed on the day of the trial) and the number of each type of larva 
remaining in the nest was recorded daily until the internal mark 
could no longer be distinguished (6 to 16 days). For each day, data 
were analysed using a heterogeneity Chi-Square test (Zar 1974) that 
compared the observed number of each type of larva in the nest tc 
an expected null difference within each group (i.e. if no preference 
were exhibited then we would expect equal numbers of the two 
types inside the nest) and for the pooled difference across groups. 
The experiment was replicated using 16 L. ambiguus worker 
groups and larvae from their own maternal colonies. For L. ambi- 
guus, the number of initial paired presentations ranged between five 
and 17 (9.0+ 1.1 days [mean±SE]). Data from these presentations 
are included and analysed as outlined above. However, the marked 
larvae used in these trials were marked seven to 10 days (7.7 + 0.9 
[mean±SE]) in advance of the trials (i.e. these were marked for a 
pilot study addressing the feasibility of obtaining marked larvae). 
Consequently the record of the number of marked and unmarked 
larvae remaining in the nest became increasingly confounded over 
time as the internal mark faded or disappeared. This trend became 
noticeable on the third day of observations and thus for L. ambi- 
guus data are only presented on acceptance over time for the first 
two days. 
Results 
Reliable data on the persistence of the internal mark are available 
only for the larvae offered to L. longispinosus workers in the “mark 
bias” experiment. In those trials, the mark duration ranged from six 
to 16 days (11.1 + 1.0 days [mean + SE], mode = 12 days) across 
groups. 
