Nest architecture of group-living wasps Microstigmus ( Hymenoptera : Sphecidae ) : role of the pedicel as a def�nse against ants

Abstraet: The nest pedicel of a social sphecid wasp Microstigmus nr. thripoctenuslplays an important role in protecting the brood against ant predators, as inferred froq¡ data on artificial nests .. � pair oC artificial nests with or without pedicels was suspended Cromihe undersides oC leaves (Carludovica pa'mata) for 30 rnin to assess rates of discovery and removal of food (tuna bait) by ants (N = 36 pairs). Threehundred lmd forty-thlee individuals oC five ant species removed food at a higher frequency from nests wihout pedicels (83.3% oC 36 nests) than from nests with pedicels (11.1 % of 36 nests). I '

Tbe thPead was ioserted ioto a small hoIe made witl!_a needle, and the artificial nest was suspended from the underside of a leaf. Non pendant nests were pinned directIy to the undersides of leaves. Each of 36 nest pairs were observed for· 30 min, and 3 replicates were observed during one day.
Differences in fre.quencies of attacks.by ants on artificial nests with and without pedicels were compared with a X2 test.

DISCUSSION
In our study, ants discovered and removed food more frequently from artificial nests without pedicels than those with pedicels. These findings suggest that tbis architectural feature' can impede ant predation, witbout a chemically-defended pedicel like that kl10wn for sorne paper wasps (Vespidae) (e.g., Jeanne 1972Jeanne , 1979Kojima 1993). Instead, this deterrence may result from a decreásed liklihóod that' ants detect potential food that is farther from tbe substrate, or from a decreased probability of encountering a nest because the slender pedicel reduces the point of contact between the nest and substrate. Alternatively, the pedicel might be too difficult to descend.
During several hundred hours of observations of nests of M. comes in Costa Rica, Matthews and Starr (unpubl. data, cited in Matthews 1991) rarely observed foraging ants crawl onto a nest pedicel, and ant predation does not seem to be a major source of . brood mortality in M. comes (Mátthews 1991). Matthews and Starr (op. cit.) showed that Solenopsis ants routinely walked down pedicels during experimental manipulations, and thus are capable of reaching a Microstigmus nest. Matthews (1991) suggested that low rates of predation are related to the small biomass in Microstigmus nests, and the small payoff to predators. A slender pedicel further reduces the payoff.

ACKNOWLEDGMENTS
This studybegan as GGG'sproject in a field course sponsored by the Smithsonian Tropical Research Institute (STRI), and was continued with support from STRI general research funds to WTW. Gabriel Melo kindly helped with identifying the Microstigmus, and Mary Jane West-Eberhard provided helpful ilnd incisive cornments on an earlier draft.