maj ors, 0.6 g), Pogonomyrmex badius (Latreille)(2. 1-2.4 mm, 1.4 g), Pseudomyrmex

pallidus (F. Smith)(0.6 mm, 0.1 g), Trachymyrmex septentrionalis (McCook)(0.8-1.0

mm, 0.2 g), and 6 colonies of S. invicta (0.6-1.4 mm, 1.5 g) workers. In the laboratory,

P. curvatus successfully parasitizes Solenopsis ants with head widths of 0.6-1.1 mm

(median of 0.74 mm; Morrison et al. 1997 and SDP unpublished data). All ant species

used in these tests were collected near Gainesville, Florida (September 2003).

 Trays with the 15 non-Solenopsis ants were set out first. Trays were 40 x 26 x 8 cm

in size and contained only one species of ant. The non-Solenopsis ants were then

removed after 30 min and replaced with the 6 trays of S. invicta. At the conclusion of 30

min, the S. invicta trays were replaced with the 15 trays of non-Solenopsis ants to

determine if the flies originally attracted from the S. invicta trials would exploit the other

genera in the absence of its primary host (no-choice). Established Pseudacteon curvatus

flies observed hovering in attack mode over each tray were collected at 5 min intervals

for 30 min. All flies were aspirated with an Allen-type double chamber aspirator and

retained in vials until the conclusion of each 30 min trial when they were identified to

species using a hand lens. Aspiration of flies normally does not change attack behavior

once flies are released (Morrison et al. 1997). Collection and identification for presence

of P. curvatus flies was necessary since P. tricuspis flies were present at the study site

from a release in Gainesville, Florida, in the summer and fall of 1997 (Porter et al. 2004).

Sampled flies were then released prior to setting up additional trays. These methods were

replicated on two consecutive days.

 Further tests ofP. curvatus host specificity were conducted with five trays ofS.

invicta and five trays of the native fire ant, S. geminate. Each tray contained 2 g of