CARNIVOROUS BUTTERFLIES CLARK 459 



finish the meal, for a tiny ant, which had been dragging persist- 

 ently at the remaining morsel, managed to get it away. 



When a new shelter containing Pheidole rotundata var. and 

 membracid larvae was put into a tube an ant immediately seized the 

 Megalopalpus larva ventrally just behind the mandibles. This 

 larva was not successfully reared. ^ 



Other caterpillars of the same species were found in other shel- 

 ters on Triu7nfetta, and once two were found in a single shelter. 



The larva is dark brown, a tint approximating very closely to 

 the color of the debris out of which the ant shelters are constructed. 



The food seems to range within certain limits; but though larvas 

 have been found eating both jassids and membracids, a larva accus- 

 tomed to take jassids will refuse membracid;!! and vice versa, and 

 there is some evidence that a larva which habitually eats one form 

 of membracid will refuse a closely allied species. 



One larva found ate a species of membracid closely allied to Gar- 

 gara variegata, although it refused that species. Another was 

 found to feed upon Leptocentrus altifrons. 



An ant shelter on Triumfetta often contains, in addition to imma- 

 ture examples of Gargara, young membracids of several other spe- 

 cies, apparently belonging to the genera Anchon and Beninia. 

 Imagines of the genera Anchon, Beninia, or Gargara are frequently 

 found in the open on a stem close to an ant shelter and tended by 

 ants from it. These shelters are constructed by two species of ants. 



Not only does Megalopalpus feed in the larval state on the Homop- 

 tera, but the butterfly seems frequently to flourish also at their ex- 

 pense, probing them with its proboscis and obtaining food material 

 direct from their surface as well as from the plant on which they 

 happen to be resting. As this habit is as characteristic of males 

 as of females it can not be interpreted as bearing any relation to 

 oviposition. 



Mr. Lamborn has not found that the ants derive any benefit from 

 the presence of this larva, or that they are of service to it. There 

 is, on the contrary, some evidence to show that their attitude toward 

 it is distinctly one of hostility, in connection with which it is note- 

 worthy that the larva is not of the smooth, soft, onisciform type 

 characteristic of the Lycsenidse, but is protected by a hard skin 

 studded wdth tubercles which are surmounted by coarse sparse hairs. 



The ants in attendance on the jassids frequently run over these 

 caterpillars and stroke them with their antennae, but are not so at- 

 tentive as they are to other lycaenid larva\ Mr. Lamborn was under 

 the impression that the attitude of the small black house ants to- 

 ward this caterpillar was distinctly hostile. Besides the one men- 

 tioned above another caterpillar was attacked by house ants, one 

 of which fastened on to one of its legs so that it had to be removed. 



