James K. Wetterer.
Forager polymorphism and foraging ecology in the leaf-cutting ant, Atta colombica.
Psyche 102:131-145, 1995.

Full text (searchable PDF, 1112K)
Durable link: http://psyche.entclub.org/102/102-131.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.

FORAGER POLYMORPHISM AND FORAGING ECOLOGY IN THE LEAF-CUTTING ANT. ATTA COLOMBICA
Center for Environmental Research and Conservation Columbia University
1200 Amsterdam Avenue
New York, NY 10027
I compare forager size and foraging selectivity of the leaf-cut- ting ant Atta colombica and that of its close relative Atta cephalotes. In both species, larger foragers cut fragments of greater mass and area, and at vegetation sources of greater specific density (masslarea). However, the size-range of A. colombica for- agers (1.5-56.8 mg) was wider than the range typical for A. cephalotes (1.4-32.1 mg). In A. colombica, the maxima workers (24-60 mg) commonly participate in foraging, making up 13% of all foragers in this study and in a previous study. In contrast, A. cephalotes maxima workers (24-100 mg) rarely forage (less than 1% of all foragers in two previous studies), but instead serve pri- marily as soldiers defending the nest. Thus, A. colombica maxima workers are smaller and do not appear to be so specialized as sol- diers as are A. cephalotes maxima workers. The broader size-range of workers participating in foraging appears to allow A. colombica to exploit a wider range of resources than A. cephalotes, including tougher, denser vegetation and fallen fruits. Leaf-cutting ants of the genus Atta are the most important agri- cultural pests of the New World tropics (Wilson, 1986). Three species of Atta occur in Central America: Atta sexdens (L.), Atta colombica Gukrin, and Atta cephalotes (L.) (Weber 1972). A. sex- dens is fairly easy to distinguish from the other two Atta species using morphological characters (Weber 1968, 1969; Mackay and Manuscript received 17 April 1995.
131




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

132 Psyche [vo~. 102
Mackay, 1986). A. colombica and A. cephalotes, however, are very similar in appearance and often difficult to distinguish. Emery (1913) considered A. colombica to be a subspecies of A. cephalotes, saying that the differences between the two taxa were simply related to the degree of polymorphism, with A. colombica having soldiers that were smaller than A. cephalotes soldiers. Later, however, Emery (1922) separated A. colombica and A. cephalotes as two distinct species.
Currently, the most reliable morphological character for distin- guishing A. colombica workers from A. cephalotes workers is the luster of their integument: A. colombica workers are entirely matte, while A. cephalotes workers have shiny areas, particularly on the head and pronotum (Weber 1968, 1969; J. Longino, pers. comm.). Another character often used for distinguishing these two species in the field is that A. colombica is the only species of Atta reported in the literature to have external refuse dumps for exhausted fungus garden substrate (Haines, 1978). The presence of an external refuse dump indicates that the A. "cephalotes" colony studied by Hodg- son (1955) in Panama was almost certainly an A. colombica colony.
A. colombica is known from Colombia, Panama, Costa Rica, and Guatemala, whereas A. cephalotes is widespread from Brazil to Mexico (Kempf, 1972; Weber, 1972). Where colonies of these two species of leaf-cutters come into direct contact, prolonged ter- ritorial battles may ensue (J. Longino, pers. comm.). Several stud- ies have compared the ecology of A. colombica and A. cephalotes. Weber (1969) described both A. colombica and A. cephalotes as "forest or forest-grassland ecotone species," stating that "the eco- logical relations between cephalotes and colombica remain to be determined." Several authors found that when the two species occur together in the same area, A. colombica tends to be more common in younger forest, with A. cephalotes more common in the older forest (Rockwood, 1973; Feener and Moss, 1990; Orr, 1992). Mackay and Mackay (1986) stated that A. colombica usually occurs in drier areas, while A. cephalotes is found in a diversity of habitats. Rockwood (1973), however, proposed that A. colombica occurs in wetter areas than A. cephalotes. The foraging ecology of A. colombica and A. cephalotes appears to be fairly similar. Mature colonies of both species primarily



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

19951 Wetterer 133
attack the leaves of trees, preferring the younger, softer leaves (Shepherd, 1985; Vasconcelos, 1990). My preliminary observa- tions, however, indicate that in A. colombica the largest "maxima" workers (24-60 mg) regularly participate in foraging, whereas A. cephalotes maxima workers (24-100 mg) seldom forage. In thl- present study, I examine forager size and foraging selectivity of A. colombica, and evaluate ecological similarities and differences between A. colombica and A. cephalotes.
I conducted this study with several large colonies of Atta colom- bica on Barro Colorado Island (BCI), Panama in July 1992. In addition, I considered more informal observations of A. colombica that I made on BCI in August 1990.
At the main A. colombica study colony (colony A), I collected 25 laden ants from each of three foraging trails in order to evaluate the match between forager mass and vegetation resource types. I selected foraging trails coming from three very different vegetation sources: 1) a tree with thin, soft leaves, 2) a tree with tough, veiny leaves, and 3) a crop of thick fig leaf stipules scattered on the for- est floor.
I placed each ant with her load in a separate 2.5 ml vial, and anesthetized the ants by placing the vials in a refrigerator for sev- eral minutes. I then measured the mass of each ant (mA) and each load (mi) to the nearest 0.1 mg on a Mettler balance. In addition, I measured the area (a) of each leaf and stipule fragment using a Licor area meter.
From mass and area data I calculated the following measures: burden ((3) = the ratio of load mass to ant mass (mL/mi); individual load specific density (6) = m,/a; overall specific density (D) for each vegetation source = the sum of the load masses divided by the sum of the 25 areas.
To normalize distributions, I performed all regression analyses on log-transformed data. I calculated "residual log load mass" val- ues by subtracting the log load mass values predicted by the log ant mass versus log load mass regressions (for the combined data from all 75 ants) from the actual observed log load mass values. I calculated "residual log load area" values in an analogous manner.



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

134 Psyche [vo~. 102
At a second A. colombica colony (colony B) where foragers along one trail were carrying whole berries and berry fragments, I collected 10 ants laden with whole berries. I did not collect for- agers with berry fragments.
At several additional A. colombica colonies on BCI, I collected the largest unladen workers I could find (along trails and exiting disturbed nests) in an effort to determine the maximum worker size for this species. I weighed each of these ants and measured their head widths under a dissecting microscope using an ocular micrometer. I also measured head widths of A. colombica and A. cephalotes specimens in the collection at the Museum of Compara- tive Zoology (MCZ) at Harvard University. Any specimens in the MCZ collection that appeared to be incorrectly identified I sent to John T. Longino for verification. A. colombica voucher specimens from this study have been deposited in the MCZ. Forager mass and load size
Overall, the A. colombica foragers from colony A (n = 75: three collections of 25 ants each) weighed from 1.5 to 56.8 mg, a 38- fold range of mass (Table 1). Ten of the 75 foragers (13%) weighed more than 24 mg (Table 1).
Specific density varied greatly among the three vegetation sources attacked by colony A (Table 1). The fragments harvested from the thin leaf tree had an overall specific density (D) of 0.12 mg/mm2 (range of specific density for individual fragments, 6 = 0.104 to 0.155 mg/mm2), whereas the fallen fig leaf stipules had an overall specific density of 0.42 mg/mm2 (range of 8 = 0.263 to 0.726 mg/mm2). The fragments from the tree with tough veiny leaves consisted of 22 leaf fragments and 3 leaf stipule fragments and were variable in specific density (Table 1; overall D = 0.28 mg/mm2, range of 8 for 22 leaf fragments = 0.187 to 0.361 mg/mm2 and for the 3 stipule fragments = 0.351 to 1.206 mg/mm2).
Mean forager mass also varied greatly among the three vegeta- tion sources (Table l), from 4.9 mg at the thin leaf tree to 20.6 mg at the fig leaf stipule source. Mean ant mass correlated positively with specific density (Table 1).




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

19951 Wetterer 135
Table 1. Atta colombica forager mass at different sources compared with data from earlier studies with Atta cephalotes. A. colombica: foragers from one colony (colony A) collected coming from three different vegetation sources (thin leaf tree, veiny leaf tree, and fallen fig stipules), from a second colony (colony B) carrying whole berries, and data from Lighton et al. (1987). The data from the veiny leaf tree are further divided into data for ants with leaf fragments and for ants with leaf stip- ule fragments. Overall specific density (D = total masdtotal area), mean and range of ant mass (mA in mg), coefficient of variation of forager mass (CV = mean/SD), and percent of foragers larger than 24 mg (% > 24). A. cephalotes: foraging from two earlier field studies (Wetterer 1990a; 1994). In Wetterer (1990a) collections were made during the day and night along a single trail (n = 400 for each) and com- ing down a grapefruit tree (n = 150 for each). In Wetterer (1994) collections were made along trails coming down 18 trees (n = 50 for each tree; total n = 900). Density Forager Mass (mg)
Source n (mg/mm2) mean å SD range CV % >24 Atta colombica
Colony A (total)
thin leaf tree
veiny leaf tree
leaf
stipule
fig stipule
Colony B: berries
Lighton et al.
Atta cephalotes
Wetterer
Combining data from all 75 ants, I found that logarithmic rela- tionships between ant mass (mA) and load mass (my), load area (a), and specific density (8) were all significantly increasing func- tions (Fig. 1; log m = 0.09 + 1.16 log mA, R2 = 0.67, n = 75; P < 0.001; Fig. 2; log a = 1.08 + 0.74 log mA, R2 = 0.42, n = 75, P < 0.001; Fig. 3; log 8 = -0.99 + 0.42 log mA, R2 = 0.36, n = 75, P < 0.001). Larger ants cut fragments of greater mass, area, and den- sity.
Examining each vegetation source separately, there was a signif- icant (P < 0.05) relationship between ant mass and load mass and between ant mass and load area at all three sources (Table 2). There was no significant relationship between specific density and ant mass at the thin leaf tree or the fig stipules (thin leaf: log 6 = -0.92 + 0.04 log mA, R2 = 0.03, n = 25, P > 0.05; stipules: log 8 = -0.38 + 0.01 log mA, R~ = 0.00, n = 25, P > 0.05). There was a



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

Psyche [vo~. 102
Thin leaf
0 Veiny leaf
A Fig stipules
0 + Whole berries
å
1.3 2.5 5 10 20 40
Ant Mass (mg)
Figure 1. Load mass selection by Atta colombica foragers. The logarithmic rela- tionship between ant mass (mA in mg) and load mass (my in mg).The line is the least squares linear regression for the 75 foragers from colony A with vegetation fragments (values for foragers from colony B carrying berries not included in regression calculation). Equation given in text. Closed squares = thin leaf tree; open squares = veiny leaf tree (with dot = leaf stipule); closed triangles = fig stipules; crosses = whole berries.
Table 2. Load size for Atta colombica foragers from colony A at three different vegetation sources (thin leaf, veiny leaf, and fig stipule). Burden (13 = mLlmA), residual log load mass (log mL), and residual log load area (log a). For each col- lection, n = 25.
Burden residual residual
Source mean å 1 sd log m regression R~ log m log a regression R~ log a thinleaf 1.8k0.9
0.2+0.910gmA
0.44
0.002 1.2+0.910gmA
0.44 0.175
veiny leaf 1.5 å 0.8 -0.1 + 1.2 log % 0.49 -0.125 0.8 + 0.9 log mA
0.36 -0.170
fig stipule 3.2 å 2.2 0.5 + 0.9 log mA 0.55 0.123 0.9 + 0.9 log mA 0.50 -0.004
significant relationship between specific density and ant mass at the tree with veiny leaves (log 8 = -0.87 + 0.33 log mA, R~ = 0.17, n = 25, P < 0.05), primarily because the three foragers with dense leaf stipule fragments were much larger than the 22 foragers with leaf fragments (Table 1).




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

19951 Wetterer 137
Burden (p = mL/mA) ranged from 0.2 to 11.0. Of the three vege- tation sources, mean burden was lowest at the veiny leaf source and highest at the fig leaf stipules (Table 2). The load mass cut by ants of a given size varied among the three vegetation sources; the mean residual log load mass was lowest at the very thin leaf source and highest at the stipules (Table 2). The fragment area cut by ants of a given size also varied among the three sources; the mean residual log load area was lowest at the veiny leaf source and high- est at the thin leaf (Table 2). At all three sources, the workers with loads of the lowest residual area values were among the smallest workers attacking the source (Fig. 2).
Foragers with whole berries
At A. colombica colony B, the 10 foragers that I collected carry- ing whole berries ranged from 16.8 to 52.7 mg (Table 1). Four of
these foragers weighed more than 24 mg; nine weighed more than 20 mg (Fig. 1). The mass of the berries ranged from 177.9 to 247.8 1.3 2.5 5 10 20 40
Ant Mass (me)
Figure 2. Load area selection by Atta colombica foragers. The logarithmic rela- tionship between ant mass (mÌ in mg) and load area (a in mm2). The line is the least squares linear regression for the 75 foragers from colony A with vegetation fragments. Equation given in text. Symbols as in Fig. 1.



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

Psyche
[Vol. 102
mg. Mean burden was 8.3 å 3.0. Burden ranged from 3.8 for the largest ant to 12.3 for the smallest ant. The maximum burden of 12.3 is higher than the maximum observed for A. colombica foragers carrying vegetation fragments. Foragers appear to be capable of carrying such heavy burdens only when the load is compact like a berry or a flower bud; very large leaf fragments appear to be more difficult to balance (pers. obs.). If the maximum burden possible for A. colombica foragers is approx- imately 12 times body mass, then only workers larger than 14 mg would be able to carry even the smallest of the whole berries. Workers smaller than 14 mg would have to cut fragments from the berries in order to be able to carry the load. Largest maxima workers
Of the very large unladen A. colombica workers I collected from several colonies around BCI, the largest weighed 58.9 mg and had a head width of 4.6 mm. This mass was only slightly greater than the largest laden forager I collected at colony A (56.8 mg). These maximum values are approximately equal to or greater than those previously reported (35 mg - Weber 1968; 60 mg - Lighton et al. 1987; 4.2 mm - Borgmeier 1959). The largest A. colombica in the MCZ collection had a head width of 4.2 mm. The largest A. cephalotes worker I have measured weighed approximately 110 mg (Wetterer unpublished) and the largest in the MCZ collection had a head width of 6.4 mm. These maximum values are equal to or greater than those previously reported (103 mg - Weber 1968; 6.1 mm - Stradling 1978; 6.4 mm - Borgmeier 1959).
General observations
I found A. colombica on BCI foraging almost exclusively during the day, with very few ants returning to the nest with loads after early evening. All A. colombica colonies I observed had external refuse dumps. I found that the ants removed refuse from the nests day and night, as reported by Hodgson (1955). A. colombica appeared to have foraging trails that were more branching than A. cephalotes, simultaneously attacking greater numbers of different vegetation sources. Trails often fanned out through the forest floor to cut crops of fallen leaves, flowers, stip- ules, sepals, and fruit. I commonly found large A. colombica foragers



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

199.51 Wetterer 139
cutting crops of fallen fig fruits (also see Weber, 1969). This con- trasts with Shepherd's (1985) finding of little ground foraging in A. colombica in Colombia.
Species identification
Several dozen A. cephalotes and A. colombica specimens in the ant collection at the Museum of Comparative Zoology (MCZ) at Harvard University were found to be misidentified (pers. obs.; J. Longino, pers. comm.). It seems very likely that some of the pub- lished field studies of A. cephalotes and A. colombica may have also confused the identification of these two species. I found that the size-range of workers participating in foraging was broader for A. colombica (1.5-56.8 mg) than for A. cephalotes (1.4-32.1 mg; Wetterer 1990a, 1994). Overall, 13% of the A. colombica foragers attacking the three vegetation sources weighed 24 mg or more (Table 1). Whether this small sample of vegetation sources is representative of what A. colombica foragers generally harvest is unknown. However, Lighton et al. (1987), studying the energetics of A. colornbica foragers on BCI, similarly found 13% (201154) of the foragers weighed 24 mg or more (Table 1). In Lighton et al.'s (1987) study, mean forager mass was 15.2 å 7.1 mg (range 3.6 to 35.6 mg; n = 154). This size-distribution is intermedi- ate for A. colombica attacking the two denser vegetation sources in the present study.
In A. cephalotes, the largest workers (24-100 mg) rarely forage (Wetterer, l99Oa; 1994). Instead, the largest A. cephalotes workers appear to function primarily as soldiers, guarding the nest. In two field studies of A. cephalotes, I found less than 1% of the foragers weighed more than 24 mg (Table 1; data from Wetterer 1990a, n = 1100; and Wetterer 1994, n = 900). In my 1994 study, the range of densities of 18 trees attacked by A. cephalotes was 0.15 to 0.37 mg/mm2. This appears to be representative of the range of densi- ties of tree leaves typically cut by A. cephalotes, though on occa- sion they attack denser sources, e.g., Rudolph and Loudon (1986) observed A. cephalotes attacking the leaves of an epiphyte with a density of 0.58 mg/mm2. The maxima workers of A. cephalotes are larger and appear to be more specialized as soldiers than the maxima



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

140 Psyche [VOI. 102
workers of A. colombica. Wilson (1984) found a similar trend in species of Pheidole where increased specialization of maxima workers was correlated with increased maxima to minima size- ratio, as predicted by caste optimization theory (see Oster and Wil- son, 1978).
In both A. colornbica and A. cephalotes, larger foragers cut at denser, tougher vegetation sources and cut the denser, tougher parts of individual sources (Cherrett, 1972; Rudolph and Loudon, 1986; Nichols-Orians and Schultz, 1989; Wetterer 1990b, 199 la, 1994, present study: Table 1, Fig. 3). This match of ant mass to vegetation toughness and specific density appears to increase the efficiency of both cutting (Wilson, 1980) and load delivery (Wet- terer, 1994). High specific density and toughness can prevent suc- cessful foraging in A. cephalotes (Nichols-Orians and Schultz, 1989, 1990). The broader size-range of A. colombica foragers may allow them to exploit efficiently a wider range of resources than Ant Mass (me)
Figure 3. Load density selection by Atta colombica foragers. The logarithmic relationship between ant mass (mÌ in mg) and load specific density (5 in mg/mm2). The line is the least squares linear regression for the 75 foragers from colony A with vegetation fragments. Equation given in text. Symbols as in Fig. 1.



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

19951 Wetterer 141
A. cephalotes. Indeed, Rockwood (1976) found that A. colombica harvested a wider range of plant species than did A. cephalotes. The larger foragers of A. colombica may allow more successful exploitation of tough resources such as leaf stipules. Large A. colombica foragers also may be important in cutting through the thick tough skin of some fruits, such as figs, making the flesh of the fruit available to smaller foragers (pers. obs.). A. colombica foragers cutting from thick and veiny leaf sources showed a reduction in fragment area cut (Table 2, Fig. 2). A. cephalotes foragers show similar reductions under some circum- stances, particularly when there is a mismatch between forager size and vegetation source, i.e., when small ants cut very dense vegeta- tion (Wetterer, l99Oa; 1994; Roces and Holldobler, 1994). In A. cephalotes, this reduction can be due to many factors including large workers finishing cuts abandoned by smaller workers, large workers taking oversize loads away from smaller workers (Rudolph and Loudon, 1986; Wetterer, 1990a; 1994), and active changes in cutting behavior (Roces and Holldobler, 1994). In the present study, there is no information as to the mechanism of the reduction.
The maxima foragers of A. colombica not only appear to have greater cutting abilities, they also can carry heavier loads (Fig. 1). This can be a particular advantage when harvesting resources such as berries (Table 1, Fig. 1). For possibly similar reasons, Acromyrmex octospinosus and Acromyrmex volcanus, two species of leaf-cutting ants that depend largely on cutting or collecting fallen fruit, flowers and other vegetation, have very large foragers, large enough to cut or collect any desirable vegetation encountered (Wetterer, 199 1b; 1993). In addition to advantages that may result from the use of very large workers as foragers, there are also sub- stantial costs. Larger workers are energetically more expensive to produce than smaller workers, and foragers are generally exposed to greater mortality risks than workers that remain in the nest. Also, parasitic phorid flies preferentially attack larger Atta foragers (Orr, 1992), making the maxima foragers particularly vulnerable. A. colombica and A. cephalotes show other ecological differ- ences. On BCI and on the Osa Peninsula in Costa Rica, A. colom- bica are primarily diurnal foragers year-round, whereas A. cephalotes forage primarily during the day in the wet season and



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

142 Psyche [Vol. 102
primarily during the night in the dry season (Lugo et al., 1973; Feener and Moss, 1990; Orr, 1992). (The observation of diurnal foraging during the dry season suggests that the A. "cephalotes" colony studied by Lutz (1929) on BCI may have been actually an A. colombica colony.) Rockwood (1975), however, found that although A. colombica in Guanacaste, Costa Rica, foraged primar- ily during the day, they also showed significant foraging at night. A. cephalotes at La Selva Biological Station in Costa Rica also shift from primarily diurnal foraging in the wet season to primarily nocturnal foraging in the dry season (Wetterer, 1990a). Wetterer (1990a) proposed that this shift could be a mechanism for avoid- ance of parasitism by diurnal phorid flies that are active in the dry season. At La Selva, however, phorids known to attack A. cephalotes appear to be, if anything, more active in the wet season (B. Brown pers. comm.; also see Feener and Brown, 1993). Orr (1992) similarly proposed that avoidance of phorids could explain the differences between the foraging of A. colombica and A. cephalotes on the Osa Peninsula, because at this site phorids were found only attacking A. cephalotes. However, the same pattern of foraging also occurs on BCI, where diurnal phorids heavily attack A. colombica foragers (Feener and Moss, 1990). The phorid flies' differential attack of A. colombica and A. cephalotes suggests that the pheromones of the two Atta species differ. Interestingly, A. sexdens may be easily distinguished from A. colombica and A. cephalotes by the strong lemon-like smell of crushed A. sexdens workers (pers. obs.). Brown and Feener (1991) proposed that phorids that use olfactory cues to find their hosts may be important in the evolution of pheromone diversity in ants. What role parasitic phorids might play in the distribution, ecology, and evolution of A. colombica and A. cephalotes deserves further attention.
A. colombica is the only Atta species for which external refuse dumps have been reported in the literature (Haines, 1978). Stefan Cover (pers. comm.), however, has noted such refuse dumps out- side Atta colonies in Trinidad (the Trinidad species is currently classified as A. cephalotes, but may be a separate species, J. Longino pers. comm.; also see Borgmeier, 1950). Fowler (1985) found external refuse dumps in the Paraguayan leaf-cutting ants