Sally C. Levings and James F. A. Traniello.
Territoriality, Nest Dispersion, and Community Structure in Ants.
Psyche 88:265-320, 1981.

Full text (searchable PDF, 3704K)
Durable link: http://psyche.entclub.org/88/88-265.html

The following unprocessed text is extracted from the PDF file, and is likely to be both incomplete and full of errors. Please consult the PDF file for the complete article.

TERRITORIALITY, NEST DISPERSION,
AND COMMUNITY STRUCTURE IN ANTS.
BY SALLY C. LEVINGS~ AND JAMES F. A. TRANIELLO~ The dispersion patterns of ant colonies have been reported for a variety of species having very different ecological characteristics (Pontin 1961; Yasuno 1963, 1964a,b, 1965; Brian 1964; Brian et al. 1965, 1966; Greenslade 1971; Room 1971, 1975a,b; Bernstein and Gobbel 1979; Levings and Franks 1982), and typically, spacing studies involve discussions of territoriality. Recently, Holldobler and Lumsden (1980), using a cost/ benefit approach, examined the importance of the economic defensibility of territories and its influence on the use of space and dispersion patterns. Holldobler (1974, 1976a, 1979a,b) demonstrated the relationship between re- source distribution, territory shape and nest spacing. These studies also emphasize that in order to understand thoroughly territoriality and other intra- and interspecific relationships, it is necessary to comprehend the role of social design in the establishment and maintenance of territory. Without such a combined approach of behavior and ecology, it is difficult to assess accurately the signifi- cance of territoriality in social species such as ants. In many studies there have been problems in the application of the term territoriality and discrepancies in the identification of territorial phenomena. Terms describing the use of foraging area such as territory and home range have been rather poorly defined and vary in meaning between authors. Territory to some authors denotes a defended area (Baroni-Urbani 1979; Holldobler 1974, 1976a; Holldober and Wilson 1977a,b; Holldobler and Lumsden 1980) whereas to others it is synonymous with home range or is casually used (Dobrzanska 1958, 1966). There are also problems with the application of information on territoriality in the interpre- tation of spacing patterns. For example, mathematical evidence of Museum of Comparative Zoology Laboratories, Harvard University, Cambridge, Mass. 02138.
~e~artment of Biology, Boston University, Boston, Mass. 02215 [To whom reprint requests should be sent].
Manuscript received by the editor June 19, 1981.



================================================================================

Psyche [Vol. 88
nest overdispersion is frequently confused with, or taken as evidence for, territoriality although crucial behavioral patterns are not considered. However, sufficient information is available in the literature to suggest some of the behavioral and ecological factors important in the regulation of nest distribution. With the above cautions in mind, we here present a simple model of predicted spatial distributions of colonies under different ecologi- cal conditions. We then survey the literature to examine the fit of available data to our predictions. Finally we discuss the general problem of the form of interactions between colonies and some of the implications of this for both field and theoretical considerations. We would first like to develop a set of biologically realistic predicted spatial distributions of colonies. We begin by positing some simple assumptions about a hypothetical ant population: 1. Nest sites are unlimited.
2. The habitat is homogenous and inhabited by a single species. 3. Each colony forages symmetrically around the nest to some distance r, which forms the radius of a circle. Within this circle, no other colonies can forage or become established. Simberloff (1979) derives the maximum foraging distance, r, as where p is the density of nests. In this case, nests are hexagonally packed and the array of nests is overdispersed (more regularly spaced than expected if random; Figure 1, case 1). Nests are spaced 2 r apart and have 6 equidistant nearest neighbors. Under different ecological conditions, the expected spatial dis- tribution of nests will change. In low density populations, nest distribution should reflect the best foraging or nest sites; nests may be
dispersed in any way and should tend towards a random distribution (Figure 1, case 2). Internest distance should on the average be at least twice r and usually more; its variance should be high. If nest sites are not uniformly available, then nest spacing will



================================================================================

19811 Levings & Traniello - territorial it^ in Ants 267 depend upon whether or not nest sites are farther apart or closer together than this distance. We predict one of 2 patterns: (1) nests will be more overdispersed than potential nest sites (Case 3a) or (2) although nests may be clumped in space, foraging ranges which are asymmetric and which partition foragers will develop (Case 3b). If potential nest sites are farther apart than twice r, then nests will be distributed only with respect to potential nest sites. The effects of habitat heterogeneity will depend upon the scale and extent of the patchiness in relation to the foraging range of a species. If patches hold several to many colonies, then clumps of nests which are overdispersed within the clump are predicted. Smaller patches in complex mosaics will not generate predictable nest distributions unless the arrays of patches are very regularly distributed. The effect of adding more species to the system will depend upon the species. Generally, in multi-species systems, the level of repul- sion observed between co-occurring species should be directly proportional to the amount of overlap in resource use. Species utilization curves can range in overlap from 0 to essentially com- plete ecological identity (100% overlap). Predicted spatial patterns will clearly depend on the actual distribution of species. If two or more species with identical requirements and foraging radii occur in the same area, interactions within and between species should be equally strong. In this case, the pattern of nest distribution predicted is random for any one species (Franks 1980; Levings and Franks 1982). Nests should be overdispersed, but each species is distributed with respect to every other species (i.e., nests of all species are treated as equivalents). In addition, there should be no pattern in the species identity of nearest neighbors (Case 4). Removal of any one species should have the effect of the removal of a nest at random from an overdispersed array; the degree of observed overdispersion should decrease. The spatial dispersion of any one species in such an array should tend to look like a low density nest population (Case 2), but the history of the area may cause any type of pattern under different conditions.
If two or more species have the same foraging radius but do not overlap 100% in resource requirements, intraspecific interactions should be stronger than interspecific interactions (Case 5). We predict that (1) the entire array will be overdispersed and (2) each species will also be overdispersed from itself. Franks (1980) and



================================================================================

Psyche
[Vol. 88
FIGURE 1
Case 1 High density population
Assumptions: 1. Single species
2. All nests have the same r
3. Unrestricted nest sites
Predictions: 1. Overdispersed nest array 2. Nest to nest distance = 2 r
Case 2 Low density population
Assumptions: 1, 2, 3
Predictions: 1. Nest distribution will tend to randomness 2. Average nest to nest distance > 2 r
3. High variance in nest to nest distance Case 3 Limited nest sites
a. Assumptions: 1, 2
Predictions:
1. Nests more overdispersed than potential nest sites 2. Nest spacing will vary with nest site location, minimum nest to nest distance = 2 r, average nest to nest distance > 2 r
3. High variance in nest to nest distance b. Assumptions: 1
Predictions: 1. Nests distributed as nest sites 2. Asymmetric foraging ranges
Case 4
Intraspecific = interspecific interactions Assumptions: 2, 3
Predictions:
1. Entire nest array overdispersed
2. Individual species are more randomly dispersed than the total array
3. No pattern in the identity of nearest neighbor 4. High variance in nest to nest distances within a species, average nest to nest distance > 2 r
5. Low variance in nest to nest distances for the entire array, average nest to nest distance = 2 r
Case 5
Intraspecific interactions > interspecific interactions Assumptions: 2, 3
Predictions:
1. Entire nest array overdispersed
2. Individual species within the array are also overdispersed 3. Nearest neighbors tend to be members of other species 4. Low variance in nest to nest distances within species, average nest to nest distance > 2 r
5. Low variance in internest distances for the entire array, average nest to nest distance = 2 r




================================================================================

19811
Levings & Traniello - Territoriality in Ants CASE I CASE 2
CASE 3a CASE 3b
nest
+R
0 potential nest site
CASE 4 CASE 5
Figure 1. Theoretical nest dispersion patterns under different ecological condi- tions. Additional details in text.




================================================================================

Psyche
[Vol. 88
Levings and Franks (1982) have reviewed the relevant statistical literature and give a suggested procedure for examining this problem.
In addition to changes in the observed spatial array of any one species, in multi-species populations, there should be correlated changes in expected internest distances under different competitive regimes. If intra- and interspecific interactions are equally strong, the average internest distance within any one species should be longer than twice the species' average r and the variance in between nest distances within any one species should be high (essentially a low density population, Case 2). If intraspecific interactions are more important than interspecific interactions, then internest dis- tance within any one species should be greater than twice the species' average r and their variance should be relatively low. The exact predicted distance will be a function of the number of interacting species and their relative abundances. It may be possible to use the degree of departure from predicted intraspecific spacing patterns as a measure of competition between species in homog- enous habitats. If intranest distances within a species are 2 r, then it does not appear to be interacting significantly with sympatric species, at least not in ways which affect its spatial distribution. DETECTION OF OVERDISPERSION AND METHODOLOGICAL PROBLEMS There are certain methodological difficulties in applying any sort of spatial analysis to previously published data on nest distribu- tions. In particular, the complicated structure of the nests of many speties has confused workers, especially when many nest entrances are present. In Lasius neoniger, Headley (1941) assumed that the species was unicolonial, since he could only occasionally elicit aggression between adjacent nest entrances. In fact, L. neoniger colonies are distinct and well organized, but extensive field tests are required to delineate colony boundaries (Traniello 1980). Simple mapping of nest openings may reflect the distribution of colonies fairly well (as it does for many species in the ground ant community in Panama, Levings and Franks 1982; Levings, personal observa- tions), but may lead to confusion unless sufficient data on the species are available (see, for example, Brough 1976). Whitford et al, (1980) assumed that workers of Novomessor cockerelli were entering an alien nest because they did not return to the same nest



================================================================================

19811 Levings & Traniello - Territoriality in Ants 27 1 entrance from which they departed. However, Holldobler et al. (1978) and Davidson (1980) documented that this species has nests with multiple entrances.
Although there are several methods for the detection of over- dispersion (Pielou 1977), we have chosen to apply Clark and Evans' (1954) nearest neighbor (NN) technique wherever possible. It is based upon the ratio between the observed mean nearest neighbor distance and the expected distance when a population is distribution at random. The index R can range from 0 (perfect aggregation) to 2.1491 (perfect hexagonal overdispersion). A value of 1 indicates a random dispersion pattern. The significance of R is tested using the z transformation. In an overdispersed population, the observed mean nearest neighbor distance is larger and the variance in nest to nest distance is lower than it would be in a randomly distributed population. Thus a population which is significantly overdispersed using this measure confirms 2 of our predictions (overdispersion and low variance in NN distance). Other methods do not have this property.
In our evaluation of spacing information in the literature, if we were unable to apply nearest neighbor methods, but complete quadrat counts were published, we calculated variance/ mean ratios and tested them for significance using X* statistics (Pielou 1977). A VIM ratio of less than 1 indicates overdispersion while values greater than 1 indicate clumping. Cases are included in which data are not sufficient to test for statistical overdispersion, but informa- tion on partitioning of resources or area was published. We have organized the available data by geographic region, habitat and food types (Table 1). Methods used in gathering previously unpublished data will be described with the specific set of data. In testing our model and spatial predictions from the literature, we are limited by the previous interests and focus of other authors. We are able to test the spatial predictions far more thoroughly than the hypotheses about the actual expected distances between nests, but there is no empirical reason that they cannot be experimentally verified in the field (see discussion).
Data are discussed by subdividing reported cases into groups according to foraging type: (1) species which do not defend re- sources although they may or may not recruit to food, (2) species which defend randomly and unpredictably distributed resources (e.g., dead insects, which are patchy in both space and time), (3)



================================================================================

272 Psyche [Vol. 88
species which defend predictable and persistent resources (e.g., honeydew from aphids, resources which are patchy in space but not in time) and (4) truly territorial species which defend area which has potential food resources. These divisions mark some ecologically important foraging types within communites. 1. Nest Defense
The data suggest that species which display only nest defense fall into 4 major groups, depending upon the details of their foraging biology. First, some species forage only as solitary individuals for food items which a single forager can subdue and retrieve (Group I foragers, Oster and Wilson, 1978). Examples of this group include most Dacetini, many Ponerinae, and some of the non-leaf cutting Attini (Brown and Wilson 1959, Wilson 1971, Oster and Wilson 1978).
There is very little applicable data on this group. The frequency of dacetine nests in extensive Berlese sampling of a tropical deciduous forest fit a Poisson distribution indicating a random distribution (Levings, unpublished data), but this sort of data does not differen- tiate between the suitability of the site or other important factors in the distribution of nests. Certainly there was no indication that nests were clumped. The maximum number of nests found was 6 in 84 0.25 rn* samples. When a truncated Poisson was fit (0 class excluded), the distribution did not differ from Poisson expectation (p > 0.5, x2 test).
Second, some species may recruit nestmates to food resources, but make no attempt to defend them, decamping if another, more aggressive, species arrives before the food is retrieved (Group 11, in part, Oster and Wilson 1978). These species specialize in the rapid location and removal of food. Examples include Paratrechina longicornis and Tapinoma melanocephalum (Wilson 197 1). No data on their nest distribution is available, but many are known to form small fragmented colonies which move frequently between ephem- eral nest sites.
The third set of species have developed mechanisms for feeding at the same resources as other, more aggressive ants, without eliciting defensive reactions (Groups I & 11, in part, Oster and Wilson 1978, Wilson 1971). It is not known how much of a colony's food intake



================================================================================

198 11 Levings & Traniello - Territoriality in Ants - results from such theft and how much is independently g a t h e y - Examples include Leptothorax acervorum and various C a r e condyla species (Brian 1955; Wilson 1959a, 1971). These spe<<: usually recruit few other workers to the food item; many of t-- species recruit only one other nestmate using tandem r u n n (Wilson 1959a). No spacing information is available for t- species.
The fourth set of species include the legionary ants (true gr- foragers) and most of the specialists on extremely difficult p (Groups IV and V, Oster and Wilson 1978). These species defe- only their nest sites (which may move often) and forage in varZ - sized groups. The most spectacular examples of this type of foraj- are the army ants (Schneirla 1971). Specialists on difficult F-- occur in several genera (examples, Pachycondyla (=Termitopc^ Leptogenys, Gnamptogenys); specialized retrieval methods :- involve extensive cooperative foraging (Wilson 1971). Little 7 spacing information is recorded about these groups. Army antr 7 several genera have been observed to avoid each other when - meet in the field, but no similar information is available for r e l a - groups (Schneirla 197 1). Other legionary groups are relatively on BCI and, in 4 years of field work, no interactions were obse 7 (Levings, personal observation).
In general, information on spacing patterns of ants which d e F Ì only their nests is extremely difficult to gather, since the investig- - must usually depend upon luck to locate colonies and will neve; v certain that all colonies in an area have been found. B e c Ì - information on foraging ranges for most species is unavailable - are unable to test those aspects of our hypotheses. Many s p e which are now assumed to defend only their nest sites may we= - found to defend either resources or a foraging territory. 2. Resource defense
a. short term
The defense of unpredictable resources occurs on varying scales. Resources which persist for very short periods (i.e., min L for most dead insects) are defended by many generalist or scavenp-- ants during the recruitment1 retrieval process (Groups I1 & I1 E part, Oster and Wilson 1978). Spatial overdispersion in den- populated areas has been shown in one complex tropical = - munity (Levings and Franks 1982). It is probably typical of n-



================================================================================

19811 Levings & Traniello - Territoriality in Ants 273 results from such theft and how much is independently gathered. Examples include Leptothorax acervorum and various Cardio- condyla species (Brian 1955; Wilson 1959a, 197 1). These species usually recruit few other workers to the food item; many of these species recruit only one other nestmate using tandem running (Wilson 1959a). No spacing information is available for these species.
The fourth set of species include the legionary ants (true group foragers) and most of the specialists on extremely difficult prey (Groups IV and V, Oster and Wilson 1978). These species defend only their nest sites (which may move often) and forage in various sized groups. The most spectacular examples of this type of foraging are the army ants (Schneirla 1971). Specialists on difficult prey occur in several genera (examples, Pachycondyla (=Termitopone), Leptogenys, Gnamptogenys); specialized retrieval methods may involve extensive cooperative foraging (Wilson 1971). Little nest spacing information is recorded about these groups. Army ants of several genera have been observed to avoid each other when they meet in the field, but no similar information is available for related groups (Schneirla 197 1). Other legionary groups are relatively rare on BCI and, in 4 years of field work, no interactions were observed (Levings, personal observation).
In general, information on spacing patterns of ants which defend only their nests is extremely difficult to gather, since the investigator must usually depend upon luck to locate colonies and will never be certain that all colonies in an area have been found. Because information on foraging ranges for most species is unavailable, we are unable to test those aspects of our hypotheses. Many species which are now assumed to defend only their nest sites may well be found to defend either resources or a foraging territory. 2. Resource defense
a. short term
The defense of unpredictable resources occurs on varying time scales. Resources which persist for very short periods (i.e., minutes for most dead insects) are defended by many generalist or scavenging ants during the recruitmentlretrieval process (Groups I1 & 111, in part, Oster and Wilson 1978). Spatial overdispersion in densely populated areas has been shown in one complex tropical com- munity (Levings and Franks 1982). It is probably typical of many



================================================================================

274 Psyche [vol. 88
reported cases of overdispersion in temperate ground ant com- munities dominated by relatively few generalist species (most species of the genera Myrmica, Tetramorium, Lasius, Aphaenogaster, some Formica; Table 1). Some species are placed here somewhat ar- bitrarily because good foraging ecology data are not available. In more complex (i.e., non-uniform) habitats, the pattern of nest spacing is reported to be directly related to environmental condi- tions. Lasius flavus, which has been intensively studied in several European habitats, displays different nest distributions between locations. Waloff and Blackith (1962; Table 1) found that nests were overdispersed in a high density population and tended toward randomness in a low density population. In a wet, low pasture with limited nest sites, nests were also overdispersed (Blackith et al., 1963, Table 1). With Myrmica rubra present in a low density population, L. flavus
was randomly distributed (Elmes 1974).
However, the partial segregation of species indicated that both intra- and interspecific interactions were present; M. rubra nests were more overdispersed than potential nest sites (Table 1). Similar patterns have been noted in other species. Petal (1972) showed that the pattern of distribution in Myrmica laevinodis depended upon the scale with which the species was examined. Within the habitat, nests were clumped, but within clumps of nests on a small scale, nests were either overdispersed or randomly distributed. In another study, Petal (1977) linked observed nest distribution and the avail- able food supply in Myrmica lemanica. In a year with low food abundance, nests were overdispersed; when food was abundant, nest distribution was random, tending to aggregation. Petal did not state if she distinguished between nests and nest openings by testing aggressive responses between colonies. However, overall nest density remained approximately the same. Most other studies have assumed but not demonstrated the correlation between food abundance and nest dispersion patterns.
Within colonies with multiple nest entrances, the distance be- tween nest entrances should be approximately 2 r and nest entrances should be overdispersed if avoiding redundant search is the under- lying cause of polydomy. This appears to be the case in Lasius neoniger. Each nest is composed of a series of nest entrances which are overdispersed within a colony (Traniello 1980). L. neoniger is unable to retrieve prey effectively further than approximately 15 cm from any given nest opening due to interference from other species



================================================================================

19811 Levings & Traniello - Territoriality in Ants 275 or congeners (Traniello, 1980). Inter-opening distances are not statistically different from 30 cm in a set of 12 nests with varying numbers of nest openings (P > 0.10, t test, 11 / 12 cases; range 2-27 nest entrances), fitting our predictions quite well. The only nest with consistently closer inter-opening spacing was hemmed in by 3 larger nests; its openings occupied essentially the entire available area (18 cm between entrances, 4 entrances). Although this species fits our predictions, we are unable to test them further with other species, either within species between nest openings or between separate nests. Nest entrance patterns of Palto thyreus tarsatus, which is also a polydomous species, appear to be similar in function to those of L. neoniger (Holldobler, personal communication). However, in poly- domous species of Camponotus, Atta and Pheidole, nest entrances are often much less than 2 r apart (Yasuno 1964a; Holldobler and Moglich 1980). Therefore,the association between foraging ecology and nest structure probably depends on the details of the biology of