present at none of the five S. geminate trays during the first day and at 1 of 5 and 3 of 5

trays on the second day (Table 4-1).

 Discussion

 Established P. curvatus individuals were attracted to S. invicta over S. geminata by

a ratio of about 30 to 1 (1 19 to 4 total flies, Table 4-1). These results were better than

results predicted from quarantine tests where P. curvatus hovered over S. invicta versus

S. geminata at a ratio of 1.3 to 1 in no-choice tests (Vazquez et al. 2004). Perhaps this

difference was because P. curvatus flies in the laboratory tests were confined in small test

containers leading to higher rates of hovering. Furthermore, attacks on S. geminata were

very rare to non-existent in the field confirming laboratory choice tests where attack rates

were 16 times higher for females hovering over S. invicta than for flies hovering over S.

geminata (7.02 & 1.41 (mean & SE) versus 0.44 & 0.28 attacks/min, respectively; Vazquez

et al. 2004). I demonstrated in quarantine tests (no-choice and choice) that the Formosa

biotype of P. curvatus does not complete development in S. geminate (Vazquez et al.

2004).

 Post-release populations of P. curvatus were not attracted to any of the 15 non-host

ant genera. In host-specificity tests with a biotype from Las Flores, Argentina, P.

curvatus hovered over most of 19 non-host genera in quarantine conditions (Porter 2000);

however, they generally hovered without attacking and no parasitism occurred in any of

the 19 non-host genera (Porter 2000). Results from this study demonstrate that host

specificity ofP. curvatus is restricted to S. invicta and poses no realistic threat to the

congener S. geminata or ants in other genera.