1981] 
Levings & Traniello — Territoriality in Ants 
279 
with colony foundation of potential competitors. Few studies have 
examined the pattern of species mingling. Brian and his co-workers 
have shown that nest sites probably limit many species in England, 
which apparently is a rather poor habitat for ants. The pattern of 
dominance over nest sites determines the location and abundance of 
many species (Brian 1952, 1956b, Brian et ah 1965). Nest density 
could be increased by providing new nest sites, and once established, 
populations remained relatively stable over long periods (Brian et 
al. 1966). Competition between species where nest sites were not as 
limiting tended to restrict individual species to areas which were 
close to optimal species requirements (Brian et al. 1966, Elmes 1971, 
1974). These studies indicate that the details of species biology and 
physical tolerances may be critical, even in very simple habitats like 
heath. However, even in these systems, species are definitely not 
distributed independently of their competitors. 
Tropical canopy dominants are associated with certain canopy 
conditions, and tend to be found mostly in shade or under certain 
other limited environmental states (Majer 1976a). Removal experi- 
ments indicate that colony foraging areas are competitively com- 
pressed; when a dominant is removed, surrounding colonies expand 
to fill the available space. Species usually found in one kind of 
canopy may expand into other types of foliage if adjacent domi- 
nants are extirpated (Majer 1976a,b). This pattern is similar to that 
found in far simpler grassland communities. 
In a complex tropical ground ant community with at least 16 
ecologically similar species, Levings and Franks (1982) have shown 
that new nests are not added at random to the nest array. Grouping 
all species, nests are overdispersed from each other. Each common 
species considered independently was also overdispersed. This is 
interpreted as evidence that species are interacting more strongly 
intra- than interspecifically, but that interspecific effects were still 
important in determining nest distributions. Similar patterns in 
simpler communities indicate that this may be common (Table 1). 
The worst neighbor in a competitive sense should be identical to 
oneself. In any case, the simplifying assumption that species have 
identical requirements is almost infinitely unlikely to apply; even 
small differences in requirements or tolerances can be important in 
determining colony distributions. However, few adequate tests have 
been done and, in one published case, two closely related congeners, 
