:HAr. VI] ANIMALS 139 



lumps. Besides this, in opposition to those of the Atta-fungus they 

 regularly develop in a nutritive solution into ordinary hyphae and thus 

 show that they ha\e lost less of their filamentous character. At a still 

 lower stage, in spite of the better definition of their shape as a whole, 

 ive find the 'kohlrabi-clumps' of Cyphomyrmex strigatus (Fig. 75). 

 The lowest stage of development, however, is shown in the fungus-gardens 

 )f Cyphomyrmex auritus, Apterostigma pilosum, A. Molleri, and A. IV, 

 \-here the swellings show no strict localization in their arrangement on 

 he hyphae and no constancy in their dimensions. 



A factor so destructive as the parasol-ants in tropical America, in 

 jarticular near the Equator and north of it, cannot have persisted without 

 nfluencing the character of the vegetation. The fate of introduced plants 

 s very instructive from this point of view. Many plants are so sought 

 ifter that their cultivation is quite impossible where parasol-ants are 

 :ommon ; such are rose, orange, coffee, chicory, mango, cabbage ; other 

 slants are comparatively unmolested, 



such as Eucalyptus, the ramie plant fi ^ 



Boehmeria), grasses, heliotrope, mag- 

 lolia, bay, Cucurbitaceae, wormwood, 

 ■adish, parsley, celery'. We must , 



issume that a similar condition pre- ji^^ 



mailed in regard to the vegetation of 

 ropical America before the appear- 

 nce of the parasol-ants. The vege- 

 ation consisted on the one hand of .^^.^ 



pecies that were very frequently 

 ttacked, and on the other of those /■■"•, 75- 'KohUabi-clumps' of the fungus of 



a South Brazuian hump-backed ant, Cvphomyr- 

 hat were seldom or not at all mex strigatus. Magnified 270. After Alf. MoUer. 



ttacked. The former, unless they 



lelonged to the most rapidly growing and commonest species, were 

 ither completely annihilated, or persisted only in such examples as 

 iwed their individual immunity to some characteristic or other. This 

 haracteristic was further selected in the struggle against the parasol-ants. 

 The protective characteristics in many cases may be of a histological 

 '.ature, as in very fibrous plants, such as grasses, palms, Bromeliaceae, 

 .'hich are included among the plants very rarely or not at all attacked, 

 n other cases, it is probably due to substances that have a sharp taste 

 nd smell, or are poisonous, or to very viscous latex rich in caoutchouc, 

 .•hich however is not always protective (Manihot). The number of 

 romatic plants is relatively very great, as well among those that are 

 ought after as among those that are almost always avoided, a fact 



' Alf. ^toiler, op. cit. p. 83. These data refer to South Brazil only. The species of 

 tta near the Equator may have other preferences. 



