Content uploaded by Scott Powell
Author content
All content in this area was uploaded by Scott Powell on Jul 27, 2015
Content may be subject to copyright.
RESEARCH ARTICLE (C.W. RETTENMEYER MEMORIAL PAPER)
How much do army ants eat? On the prey intake of a neotropical
top-predator
S. Powell
Received: 7 August 2010 / Revised: 14 January 2011 / Accepted: 17 January 2011 / Published online: 30 January 2011
Ó International Union for the Study of Social Insects (IUSSI) 2011
Abstract New World army ants (Ecitoninae) are nomadic
group-predators that are widely thought to have a sub-
stantial impact on their prey. Nevertheless, quantitative data
on prey intake by army ants is scarce and mostly limited to
chance encounters. Here, I quantify the prey intake of the
army ant Eciton hamatum at the contrasting scales of raid,
colony (sum of simultaneous raids), and population. Like
most army ants, E. hamatum conducts narrow ‘column
raids’ and has a specialized diet of ant prey. I show that
individual raids often had periods of no prey intake, and raid
intake rates, calculated in g/min, differed significantly
among colonies. Moreover, neither mean nor peak raid
intake rates were correlated with colony size. Similarly,
colony intake rates differed significantly among colonies,
and mean colony intake rates were not correlated with
colony size. However, mean colony intake rates were sig-
nificantly higher than mean raid intake rates, and peak
colony intake rate was correlated with colony size. Having
multiple raids thus improves colony-level intake rates, and
larger colonies can harvest more prey per unit time. Mean
colony intake rate across colonies was 0.067 g/min dry
weight and mean daily colony intake was calculated at
38.2 g. This intake is comparable to that of Eciton bur-
chellii, which has a more generalized diet and conducts
spectacular ‘swarm raids’ that are seen as having a greater
impact on prey than column raids. Population size on Barro
Colorado Island, Panama, was estimated to be 57 colonies,
which extrapolates to a daily population intake of nearly
2 kg of prey dry weight, or 120 g/km
2
. Broadly, these
findings demonstrate that column raiding army ants expe-
rience considerable variation in prey intake for individual
raids, but can still achieve notable impact at the larger scales
of colony and population. Furthermore, they challenge the
idea that swarm-raiding species necessarily have greater
intake and thus impact on prey. Instead, I suggest that
conducting multiple column raids may be a strategy that
allows for comparable intake from a more specialized diet.
Keywords Army ants Predation Diet Column raid
Swarm raid Biomass
Introduction
New World army ants (Ecitoninae; army ants hereafter) are
nomadic group-predators that hunt in soil, leaf litter, and
low vegetation throughout the Neotropics (Schneirla, 1971;
Rettenmeyer, 1963; Rettenmeyer et al., 1983). Early
researchers were quick to recognize the potentially cata-
strophic nature of army ant raids for animals caught in their
path (e.g., Darwin, 1839; Belt, 1874; Wheeler, 1910;
Schneirla, 1955). Nevertheless, it was Rettenmeyer’s (1963)
classic treatise that provided the first detailed insights into
the diet of army ants, and their potential impact on their
prey. Contrary to the popular view that army ants consume
all they encounter, Rettenmeyer showed that they are pre-
dominantly specialist predators of other ants (Rettenmeyer,
1963; Rettenmeyer et al., 1983). This only underscores their
potential significance as top-predators, because of the
diversity, abundance, and wide-reaching ecological
Present Address:
S. Powell (&)
Department of Ecology and Evolutionary Biology, University of
Arizona, BioSciences West room 310, 1041 E. Lowell St.,
Tucson, AZ 85721, USA
e-mail: scottpowell@mac.com
S. Powell
Smithsonian Tropical Research Institute, Apartado 2072, Balboa,
Ancon, Republic of Panama
Insect. Soc. (2011) 58:317–324
DOI 10.1007/s00040-011-0152-3
Insectes Sociaux
123
importance of ants in tropical systems (e.g., Ho
¨
lldobler and
Wilson, 1990; Rico-Gray and Oliveira, 2007; Lach et al.,
2010). Despite these early insights into diet, we still have
only a superficial understanding of the impact that army ants
have on their ant prey.
Direct measurements of prey intake are crucial for
understanding the impact of army ants. Data on how much
army ant colonies eat can then be considered with local raid
rates (e.g., Kaspari and O’Donnell, 2003; O’Donnell et al.,
2007), raid signatures in the litter (e.g., Franks and Bossert,
1983; Otis et al., 1986), and geographic variation in species
richness and abundance (O’Donnell et al., 2007) to give a
complete picture of the ecological importance of this
predatory group. Nevertheless, prey intake data are limited
to a handful of species, and even for these we have much
still to learn.
For Eciton burchellii, Franks (1982b) recorded a mean
item delivery rate of 48 items/min (N = 100). By combin-
ing this with measurements of mean prey-item dry weight
(0.0021 g), this gives a mean prey-intake rate of 0.10 g/min
(recalculated in different units from data in Franks (1980)).
It is important to note, however, that E. burchellii is atypi-
cal. Colonies conduct a single ‘swarm raid’ each day, which
has a unified and densely populated raid front that can be up
to 10 m wide (Rettenmeyer, 1963; Schneirla, 1971). In
addition to the typical ant prey, these special raids harvest a
diversity of non-ant arthropods, and result in a high and
seemingly continuous flow of prey into the nest (Schneirla,
1971; Franks, 1982b; Powell and Franks, 2006).
The majority of army ant species conduct narrow ‘col-
umn raids’, with raid fronts that are rarely wider than 1 m
(Rettenmeyer, 1963). This more focused raid structure is
associated with specialization on ant prey, and has been
generally characterized as having lower and more sporadic
prey delivery (e.g., Rettenmeyer, 1963; Schneirla, 1971;
Rettenmeyer et al., 1983; LaPolla et al., 2002; Powell and
Clark, 2004; Powell and Franks, 2006). Quantitative prey
intake data for column raiders has frequently been limited to
opportunistic encounters with raids against individual prey
colonies (e.g., Swartz, 1998; LaPolla et al., 2002; Powell
and Clark, 2004; Breton et al., 2007). For instance, LaPolla
et al. (2002) documented that Neivamyrmex rugulosus raids
can deplete colonies of Trachymyrmex arizonensis of as
much as 75% of their brood. Raids by Nomamyrmex esen-
beckii against massive colonies of Atta leaf-cutting ants
can be similarly devastating, resulting in up to a kilogram of
brood loss and even colony death (Powell and Clark, 2004).
These examples show that a significant amount of prey can
be harvested from individual prey colonies, but provide no
estimate of sustained prey intake and thus general impact of
column raiders. For this, raid intake needs to be recorded
over hours and days. This is usually difficult for most
column raiders because their activities are often nocturnal
and partly subterranean (Rettenmeyer, 1963).
Eciton hamatum is exceptional among column raiders in
that it forages diurnally, and raids and emigrates entirely
aboveground (Rettenmeyer, 1963). A colony will also
conduct multiple simultaneous raids each day. This differs
from the single-raid strategy of E. burchellii, but may not be
uncommon for column-raiding species (Rettenmeyer, 1963;
Schneirla, 1971). Prey intake has been estimated for
E. hamatum to between 1 and 32 g dry weight per day
(Rettenmeyer et al., 1983). While the methods used to arrive
at this broad range are unclear (Rettenmeyer et al., 1983), it
does demonstrate that such estimates are possible for this
species. The unusually accessible biology of E. hamatum
should allow for further insights into the prey intake of
column raiding army ants. Specifically, E. hamatum pro-
vides the opportunity to understand prey-intake rates and
values at a number of important scales, and for contrasts
with the atypical but ecologically important swarm-raider
E. burchellii.
Here, I investigate prey intake by E. hamatum at the
scales of individual raid, colony (sum of all simultaneous
raids), and population. I address the general hypothesis that
the column-raid strategy does experience considerable
variation in intake for individual raids, but that it still results
in notable impact at the larger scales of colony and popu-
lation. I focus on mean prey-intake rates and values as
estimates of sustained intake, but contrast this to peak intake
when informative. One common prey species was also
distinguished form all others, to provide insight into the
intake of preferred prey taxa. The relationships between
colony size and prey intake at the scales of raid and colony
are addressed using new quantitative colony size estimates.
The calculations for the population-level intake use the first
quantitative estimates of population size for E. hamatum.
Methods
Study site, colony discovery, and colony tracking
Work was conducted on Barro Colorado Island (BCI),
Panama (9° 09
0
N, 79°50
0
W) between March and June 2001.
BCI is a 15.6 km
2
forested island in the central part of the
Panama Canal (see Leigh (1999) for details). Colonies were
located by walking established trails during daylight hours,
until raid columns were encountered. Raids were then
tracked in the direction of prey flow, to locate the nest or
‘bivouac’. Colonies were tracked nightly, by following the
emigration traffic to the new bivouac site. This ensured
the location of the colony was known for data collection the
following day (see below).
318 S. Powell
123
Observation bouts and prey intake
Prey-intake data were recorded at the bivouac of seven
E. hamatum colonies. Data were collected on four consecutive
days of the nomadic phase (approx. 16-day period of the
migratory cycle when colonies move nightly; see Schneirla
(1971)) for six colonies, and three consecutive days for the
seventh (nomadic phase ended before the fourth observation
day). Observations were started at dawn, mid-morning,
noon, or early afternoon, with the order randomized for each
colony. Observations were made for 3 h each day, unless
interrupted by heavy rain or emigration. Three-minute
observation bouts were made for each active raid every
20 min. I follow the standard definition of a raid as a con-
tiguous system of foraging columns, with a distinct raid
front that remains connected to the bivouac by a ‘principal
trail’ (Schneirla, 1971). During each observation bout, all
visible prey items were counted as prey-laden foragers past
a pre-determined point on the principal trail near the biv-
ouac. Army ants carry prey items beneath the body, making
them conspicuous when viewed from the side (Fig. 1). A
subcount was maintained for the distinctive larvae and
pupae of Acromyrmex octospinosus, a common prey species
of E. hamatum on BCI (Powell and Franks, 2006; Retten-
meyer et al., 1983). A. octospinosus larvae are nearly
spherical (Wheeler and Wheeler, 1976), and the pupae have
distinct tubercles (raised bumps) covering the dorsal surface
of the gaster (Fig. 1).
To allow prey-item counts to be converted to prey intake
in units of weight, which is more useful for interspecific
comparisons, a random sample of 30 prey items was taken
from each of six E. hamatum colonies. Item dry weight was
recorded in the lab. Raid intake rates were calculated by
multiplying the prey-item count for a bout by mean prey-
item weight, and dividing by the 3 min bout duration, giving
intake rates in g/min. Colony intake rates were calculated by
summing the prey-item counts from all active raids within
the same 20-min period (see above), multiplying by mean
prey-item weight, and dividing by the 3 min bout duration.
Colony intake rates thus assume that the sum of the counts
from each active raid, taken in quick succession, is equiv-
alent to the simultaneous intake by the colony from those
same raids. True simultaneous counts on all active raids
were not possible logistically.
Daily colony intake was calculated by multiplying the
overall mean colony intake rate by mean raid duration. The
time at which the first prey item was delivered to the biv-
ouac, when emigrations began, and the number of raids
remaining at emigration time were recorded throughout the
study. These data provided estimates of mean raid duration,
as well as emigration start time and number of columns at
emigration time for population estimates (see below). Daily
population intake was calculated by multiplying the esti-
mated population size by daily colony intake, and the
probability that a colony raids on any given day. The
probability of raiding was calculated by dividing the num-
ber of days with raids per migratory cycle by cycle length
[31 and 36 days, respectively; data from Schneirla (1971)
and Teles Da Silva (1982)]. These calculations assume that
daily colony intake is equivalent for raids conducted during
the nomadic and stationary phases of the migratory cycle.
This is reasonable, because although stationary-phase raids
can be lower intensity than some nomadic-phase raids, they
can also extend further from the bivouac and last longer
(Rettenmeyer, 1963; Schneirla, 1971).
Bivouac size estimates
Bivouac size was estimated for six of seven colonies using
non-destructive image sampling during emigrations [fol-
lowing Franks (1985)]. Bivouac size (B) was calculated
using B = TNV, where T is the duration of the emigration,
N is the mean number of ants per unit length, and V is the
mean velocity of emigrating ants. The emigration start was
recorded as the time when the largely incoming afternoon
traffic near the bivouac reversed to strong outgoing traffic
(Rettenmeyer, 1963). The emigration end was recorded as
the time when the last ant left the old bivouac site. To
asses average ant density (N), a photograph of 1 m of
column was taken every 15 min throughout the emigration.
Mean ant velocity (V) was calculated from the running
speed of 90 individuals. Bivouac size in E. hamatum is
total colony size minus those ants that are already on the
raid column used to emigrate. However, foragers from
other raids return to the bivouac before leaving in the
emigration traffic, and the column that the emigration uses
experiences strong inbound forager traffic before the
emigration starts. Thus, bivouac size does not include all
ants, but it is likely to represent a good approximation of
overall colony size in E. hamatum.
Fig. 1 Typical prey retrieval in the army ant Eciton hamatum. Prey
items shown are pupae of the preferred prey species Acromyrmex
octospinosus. The distinct gasteral tubercles of this species are visible
below the mandibles of the two laden E. hamatum foragers
Prey intake of army ants 319
123
Population estimate
Population size was estimated using a modified version of
the footpath encounter-rate method developed by Franks
(1982a) and tested by Vidal-Riggs and Chaves-Campos
(2008). The summed length of army ant columns (S) in the
study area was calculated using S = pNA/2L, where N is the
number of encounters with army ant columns, A is the area
of the study site, and L is the total distance walked. This
summed length was then divided by the estimated length of
column per colony at the time the encounters were made, to
give number of colonies. For E. hamatum, the average
column length must be multiplied by the average number of
columns at emigration time. This gives the full equation
C = (pNA/2L)/EM, where C is the number of colonies, E is
the length of a column at emigration, and M is the mean
number of columns. Three non-overlapping footpath loops
of 4.4, 5.6 and 6.7 km were covered on consecutive days.
They were repeated four times, with 3 weeks between each
replicate. All walks were started at 15:00, the time of day
when the first signs of colony emigration are typically
observed in E. hamatum (Rettenmeyer, 1963), and each
encounter with a column was recorded. Following the
recommendations of Vidal-Riggs and Chaves-Campos
(2008), total distance walked (66.8 km) and total number of
encounters were used in these calculations. The area used
for BCI was 15.6 9 10
6
m
2
[following Franks (1982a)].
Results
Observation bouts, prey-item counts, and A.
octospinosus prey
Across all raids, prey-item counts during 3-min observation
bouts ranged from 0 to 463 individual prey items, with a
mean of 56 items (SD ± 63, N = 704 bouts). 628 obser-
vation bouts (89%) had at least one prey item. The
observation bouts with positive prey counts delivered
a mean of 63 prey items (SD ± 63). Subcounts of
A. octospinosus prey items ranged from 0 to 400 within the
3 min bouts, with a mean of 13 items per bout (SD ± 37,
N = 704). 179 observation bouts (25%) had at least one
A. octospinosus prey item, with a mean of 27 A. octospinosus
items (SD ± 37) per bout for these positive subcounts.
Table 1 shows the observation bout and prey data by col-
ony. While A. octospinosus constitutes an impressive 23%
of the diet of E. hamatum when summed across colonies
(Powell and Franks, 2006), the proportional intake of this
prey was significantly different among colonies (Pearson’s
v
2
test, v
2
= 46.67, df = 6, P \0.0001; Table 1). How-
ever, there was no correlation between the proportion of a
colony’s bouts with prey and the proportion of bouts that
included A. octospinosus prey (Pearson’s product-moment
correlation, t = 1.54, df = 5, P = 0.19).
Raid intake rates, colony intake rates, and colony size
Mean prey-item dry weight was 0.0015 g (SD ± 0.0019,
N = 176). Figure 2a shows that raid intake rates varied
considerably in all colonies, ranging from 0 to as high as
0.23 g/min. However, raid intake rates were significantly
different among colonies (Kruskal–Wallis rank sum test,
v
2
= 34.05, df = 6, P \ 0.0001; Fig. 2a). Similarly, col-
ony intake rates varied from 0 to as high as 0.27 g/min
within colonies, and were also significantly different among
colonies (Kruskal–Wallis rank sum test, v
2
= 34.56,
df = 6, P \0.0001; Fig. 2b). Despite these similarities,
having multiple simultaneous raids improved the average
prey delivery rate, because mean colony intake rate was
significantly higher than mean raid intake rate across colo-
nies (paired t test, t = 6.52, df = 6, one-tailed hypothesis,
P = 0.0003; Fig. 2). The observed number of simultaneous
raids from a colony ranged from one to four, with a mode of
2.
Bivouac size differed by more than a factor of two
(Colony B = 121,587 ants; Colony C = 125,085; Colony
D = 71,182; Colony E = 63,278; Colony F = 48,340;
Colony G = 115,071). However, colony size does not
appear to influence sustained raid intake, because mean raid
Table 1 The observation bouts and prey-count data for seven colonies of the army ant Eciton hamatum
Colony Total observation
bouts
% observation
bouts with prey
% observation bouts
with A. octospinosus prey
Total prey
items
% prey items
A. octospinosus
A 157 83 21 7,279 27
B 98 84 6 6,610 12
C 141 87 33 7,138 36
D 68 94 13 3,535 9
E 87 95 33 6,158 30
F 79 94 29 3,753 25
G 74 97 43 5,207 19
Within the prey counts, a subcount was maintained for the leaf-cutting ant Acromyrmex octospinosus, a preferred prey species of E. hamatum
320 S. Powell
123
intake rate was not correlated with bivouac size (Pearson’s
product-moment correlation, t = 0.63, df = 4, P = 0.56).
Colony size also does not appear to influence peak raid
intake, because the highest, non-outlier raid intake rate for
each colony (defined by the limit of the upper boxplot
whisker; Fig. 2a) was not correlated with bivouac size
(Pearson’s product-moment correlation, t = 1.93, df = 4,
P = 0.13). Similarly, mean colony intake rate was not
correlated with bivouac size (Pearson’s product-moment
correlation, t =-0.11, df = 4, P = 0.92). However, the
highest, non-outlier colony intake rate for each colony (limit
of the upper boxplot whisker; Fig. 2b) was correlated with
bivouac size (Pearson’s product-moment correlation,
t = 3.48, df = 4, P value = 0.03), suggesting that colony
size does influence peak intake by a colony as a whole (i.e.,
summed intake from all simultaneous raids).
Prey intake at the colony and population levels
The overall mean colony intake rate (mean of means for all
colonies) was 0.067 g/min dry weight of prey (N = 7,
SD ± 0.019). The mean start time of prey intake was 6:46
(N = 10, SD ± 19 min), and the mean end time (i.e.,
emigration start time) was 16:16 (N = 15, SD ± 82 min).
Mean raid duration was therefore 570 min. From these
values, daily colony intake was calculated to be 38.2 g dry
weight of prey.
Thirty-three E. hamatum columns were encountered
while walking trail-loops that totaled 66.8 km. Mean col-
umn length at emigration time was calculated as 152 m
(N = 50, SD ± 56 m; data from (Schneirla, 1949). Emi-
gration distance provides a good estimate of the length of
the column used to emigrate at emigration time. This is
because the new bivouac site is typically established before
the emigration starts at the old bivouac, and ants on that
column that are beyond the new bivouac site will join the
new bivouac. The chances of encountering ants beyond the
new bivouac site during the time of emigration are therefore
minimal. The mean number of columns at the start of an
emigration was 1.4 (N = 15, SD ± 0.6). However, for all
colonies, only the emigration column remained within 1 h
of the start of the emigration. Table 2 shows the calculations
for population size, daily population prey intake, and daily
prey harvested by the population per km
2
using a realistic
range of values for column length and number.
Discussion
Here I have addressed prey intake in the army ant
E. hamatum. The focus was on understanding intake at the
scales of raid, colony, and population. I have shown that for
individual raids, periods of no prey intake are not unusual,
and periods with no intake of the preferred prey A. octo-
spinosus are common. Moreover, overall proportional intake
of A. octospinosus varied significantly among colonies, but
was not correlated with the overall intake success of raids.
Raid intake rates, in g/min, also differed significantly
among colonies, but neither mean nor peak raid intake rates
were correlated with colony size. Similarly, colony intake
rates (simultaneous intake from all active raids) differed
significantly among colonies, but mean colony intake rates
were also not correlated with colony size. Nevertheless,
mean colony intake rates were significantly higher than
mean raid intake rates across colonies, and peak colony
intake rate was correlated with colony size. The overall
mean colony intake rate across colonies was 0.067 g/min,
and daily colony intake was calculated at 38.2 g dry weight
of prey. These values are comparable to published values for
the swarm-raiding army ant E. burchellii. Using the most
likely parameters, population size on BCI was estimated to
be 57 colonies, which extrapolates to a daily population
intake of nearly 2 kg of prey, or 0.12 kg/km
2
. Broadly, these
findings demonstrate that although column-raiding species
can experience considerable variation in prey intake for
individual raids, they still achieve notable impact at the
(a) (b)
ABCDEFG
0.00
0.05
0.10
0.15
0.20
0.25
0.30
Colony
Raid intake rate (g/min)
ABCDEFG
0.00
0.05
0.10
0.15
0.20
0.25
0.30
Colony
Colony intake rate (g/min)
Fig. 2 Boxplots of a raid prey-
intake rates and b colony prey-
intake rates for seven Eciton
hamatum colonies. In both plots,
the box encompasses the inter-
quartile range, a line is drawn at
the median, a circle is drawn at
the mean, whiskers extend to the
nearest value within 1.5 times
the inter-quartile range, and
outliers are marked with plus
signs
Prey intake of army ants 321
123
larger scales of colony and population. Consequently, they
challenge the common perception that swarm-raiding spe-
cies necessarily have greater intake and thus impact on prey.
High variation in raid intake rates, the lack of a rela-
tionship between intake of all prey and A. octospinosus
prey, and the lack of a correlation between colony size and
raid-level intake rates all suggest that prey patchiness
strongly influence prey intake in E. hamatum. The higher
raid intake rates (whiskers and outliers in Fig. 2a) are con-
sistent with documented successes from individual prey
colonies (e.g., LaPolla et al., 2002; Powell and Clark, 2004).
What this dataset shows, however, is that these high raid
rates are frequently contrasted against periods of low or zero
prey intake. Given the documented specialization on ant
prey by E. hamatum and other column raiders (e.g.,
Rettenmeyer et al., 1983; Powell and Franks, 2006), and the
likelihood that preferred prey are distributed unpredictably,
these dynamics fit well with existing knowledge. It is sur-
prising, however, that larger colonies cannot overcome
these raid dynamics to some extent. It is, for instance, rea-
sonable to expect that raids from larger colonies may have
greater success against individual prey colonies, on average.
At the level of the colony, having multiple simultaneous
raids seems to overcome some of issues with prey intake by
individual raids. The mean intake rate for whole colonies
(summing intake for all active raids) was significantly
higher than that for individual raids (Fig. 2). Although this
result may seem obvious, it would not be the case if only one
raid experienced any real success at any given time. This
identifies an important practical property of E. hamatum’s
foraging strategy: multiple column raids, each covering
different ground, provide sustained colony intake, even
though individual raids frequently experience low or zero
intake. In addition, peak intake was correlated with colony
size for whole colonies, but not individual raids. This
suggests that larger colonies may be generally better at
overcoming larger prey colonies, when summing across all
active raids. Large prey colonies that actively defend
against army ants can successfully limit brood loss or
completely repel raids (Powell and Clark, 2004). Larger
E. hamatum colonies may therefore suffer these kinds of
prey-intake limitations less often, on average. Further work
will be needed, however, to understand how success in
attacks against individual prey colonies is related to colony
size, and how this scales to differences in colony-level intake.
Historically, swarm-raiding army ant species have been
perceived as having the greatest impact on their prey (e.g.,
Rettenmeyer, 1963; Schneirla, 1971). This dataset suggests
that the impact of column raiding species may not be so
different. Mean colony intake rate for the swarm-raiding
army ant E. burchellii has been estimated at 0.10 g/min
(Franks, 1980). Here, I have shown that mean colony intake
rate for E. hamatum is 0.067 g/min, and that the highest
mean intake rate for an individual colony was over 0.10 g/
min (Fig. 2b, Colony E). Moreover, emphasis was placed on
sampling throughout the daily foraging cycle of E. hamatum
(including low activity period in the afternoon and before
the emigration; Schneirla, 1971), which is likely to make for
a conservative estimate of mean colony intake rate. Daily
colony intake also seems to be very similar between species,
calculated at 38.2 g for E. hamatum (present study) and a
daily average of 42 g for E. burchellii, with 62 g reached
during larger nomadic raids (Franks, 1982b).
Swarm raid and multiple column raids may therefore be
different strategies to achieve similar success from different
prey types, and differences in success may be partly errors
in perception. The single conspicuous raid of a swarm-rai-
der delivers a seemingly continuous stream of diverse prey
(Franks,
1982b; Schneirla, 1971). In comparison, a column
raid appears to provide spotty intake of a specialized suite of
Table 2 Estimated colony number, daily prey intake, and daily prey harvested per km
2
for the Eciton hamatum population on Barro Colorado
Island (BCI), Panama
No. of columns at
emigration time (M)
Column length at emigration
time (E) in meters
No. of colonies
(C)
Daily pop. prey
intake in grams
Daily prey harvested
per km
2
in grams
1 96 126 4,148 266
1 152 80 2,620 168
1 208 58 1,914 123
1.4 96 90 2,963 190
1.4 152 57 1,871 120
1.4 208 42 1,367 88
The estimated number of colonies (C) was calculated using the formula C = (pNA/2L)/EM. N was the 33 encounters with E. hamatum columns,
A was the area of BCI (15.6 9 10
6
m), and L was the total distance walked (66,800 m). The values for column length are the mean (152 m) minus
and plus one standard deviation. The number of columns at emigration time (M) are the mean (1.4) and 1, the typical number of columns remaining
an hour after an emigration begins. Daily population prey intake was calculated by multiplying colony number by the daily colony prey intake
(38.2 g) and the probability that a colony is raiding on a given day (0.86; see ‘‘Methods’’). The row with the bold typeface uses the mean column
length and number at emigration time
322 S. Powell
123
ant prey (present study). Nevertheless, these perceived
differences in raid impact do not account the cumulative
intake from multiple column raids, which can be hard to
determine in the field. The vast majority of army ants biv-
ouac and forage partly or exclusively below ground
(Rettenmeyer, 1963). The raiding dynamics of this majority
are therefore largely unknown. Nevertheless, nocturnal
foraging Eciton species, such as Eciton dulcium and Eciton
mexicanum, frequently have multiple column raids leaving
the bivouac (S. Powell, pers. obs.). Moreover, widely sep-
arated raid fronts of a number of Neivamyrmex species can
often be tracked back to the same subterranean location,
suggesting multiple raids from the same colony (S. Powell,
pers. obs.). I have made similar observations for the column
raiding Nomamyrmex esenbeckii, which appears to have
significantly larger colonies than E. hamatum and E. bur-
chellii, despite similar worker size (Rettenmeyer, 1963;
Franks, 1985; present study). Multiple column raids may
therefore be a more widespread strategy among column
raiding army ants than previously thought, and it may not
necessarily limit prey intake and colony size below that of
swarm raiders.
Like E. burchellii, E. hamatum’s daily intake of prey
of 38.2 g/colony, or 120 g/km
2
, is lower than common
vertebrate insect herbivores on BCI (Franks, 1982b) and
mammalian top-predators in the Neotropics (Leigh, 1999).
Nevertheless, army ants are eating other ants, which
themselves have a dominant ecological footprint (reviewed
in Lach et al., 2010). The ecological significance of army
ant predation may therefore be much larger than the raw
biomass they consume. For instance, here I report that the
leaf-cutting ant A. octospinosus constitutes a significant
proportion of E. hamatum’s prey on BCI. Like most leaf-
cutting ants, this species is responsible for harvesting
substantial quantities of live plant material in its native
habitat (Wetterer et al., 1998). Moreover, A. octospinosus
colony size and population density can explode in heavily
disturbed areas and non-native locations where army
ants do not occur, becoming a major agricultural pest
(Cherrett and Peregrine, 1976; Cherrett, 1986). While raw
biomass intake is one important metric in determining the
impact of army ant predation, it will also be necessary to
consider the ecological importance of the prey. This will
be a challenging but vital topic for future army ant
research.
Acknowledgments This manuscript is dedicated to Carl W.
Rettenmeyer. His positive influence on my career and life are
immeasurable. I also thank Ellie Clark, Egbert G. Leigh Jr., and
William T. Wcislo for valuable discussion, as well as the staff of the
Smithsonian Tropical Research Institute for logistical support. This
research was funded primarily by a Short-Term Fellowship from the
Smithsonian Tropical Research Institute, Panama, with additional
support from Carl Rettenmeyer.
References
Belt T. 1874. The Naturalist in Nicaragua. John Murray, London.
442 pp
Breton J.L., Dejean A., Snelling G. and Orivel J. 2007. Specialized
predation on Wasmannia auropunctata by the army ant species
Neivamyrmex compressinodis. J. Appl. Entomol. 131: 740–743
Cherrett J.M. and Peregrine D.J. 1976. A review of the status of leaf-
cutting ants and their control. Ann. Appl. Biol. 84: 124–128
Cherrett J. 1986. The economic importance and control of leaf-cutting
ants. In Economic Impact and Control of Social Insects (Vinson
S.B., Ed) Praeger, New York. pp 165–192
Darwin C. 1839. Journal of Researches into the Natural History and
Geology of the Countries visited during the Voyage of the H. M. S.
‘Beagle’ round the World, under the Command of Capt. Fitz Roy,
R. N. Henry Colburn, London. 551 pp
Franks N.R. 1982a. A new method for censusing animal populations:
the number of Eciton burchelli army ant colonies on Barro
Colorado Island, Panama. Oecologia 52: 266–268
Franks N.R. 1982b. Ecology and population regulation in the army ant
Eciton burchelli. In: The Ecology of a Tropical Forest: Seasonal
Rhythms and Long-Term Changes (Leigh E.G., Rand A.S. and
Windsor D.M., Eds), Oxford University Press, Oxford.
pp 389–395
Franks N.R. 1985. Reproduction, foraging efficiency and worker
polymorphism in army ants. In: Experimental Behavioral Ecol-
ogy and Sociobiology: in Memoriam Karl von Frisch, 1886-1982
(Ho
¨
lldobler B. and Lindauer M., Eds), Sinauer Associates,
Sunderland Massachusetts. pp 91–107
Franks N.R. 1980. The evolutionary ecology of the army ant Eciton
burchelli on Barro Colorado Island, Panama. Ph.D. dissertation,
The University of Leeds, Leeds, England. pp 242
Franks N.R. and Bossert W.H. 1983. The influence of swarm raiding
army ants on the patchiness and diversity of a tropical leaf litter
ant community. In: Tropical Rain Forest: Ecology and Manage-
ment. Special Publication of the British Ecological Society No. 2
(Sutton S.L., Whitmore T.C. and Chadwick A.C., Eds), Blackwell
Scientific Publications, Oxford. pp 151–163
Ho
¨
lldobler B. and Wilson E.O. 1990. The Ants. Harvard University
Press, Cambridge Massachusetts. pp. 752
Kaspari M. and O’Donnell S. 2003. High rates of army ant raids in the
Neotropics and implications for ant colony and community
structure. Evol. Ecol. Res. 5: 933–939
Lach L., Parr C.L. and Abbott K.L. 2010. Ant Ecology. Oxford
University Press, Oxford. pp 432
LaPolla J.S., Mueller U.G., Seid M. and Cover S.P. 2002. Predation by
the army ant Neivamyrmex rugulosus on the fungus-growing ant
Trachymyrmex arizonensis. Insect. Soc. 49: 251–256
Leigh E.G. 1999. Tropical Forest Ecology: a View from Barro
Colorado Island. Oxford University Press, USA. pp 264
O’Donnell S., Lattke J., Powell S. and Kaspari M. 2007. Army ants
in four forests: geographic variation in raid rates and species
composition. J. Anim. Ecol. 76: 580–589
Otis G.W., Santana C.E., Crawford D.L. and Higgins M.L. 1986. The
effect of foraging army ants on leaf-litter arthropods.
Biotropica
18: 56–61
Powell S. and Clark E. 2004. Combat between large derived societies:
a subterranean army ant established as a predator of mature leaf-
cutting ant colonies. Insect. Soc. 51: 342–351
Powell S. and Franks N.R. 2006. Ecology and the evolution of worker
morphological diversity: a comparative analysis with Eciton army
ants. Funct. Ecol. 20: 1105–1114
Rettenmeyer C.W. 1963. Behavioral studies of army ants. Univ. Kans.
Sci. Bull. 44: 281–465
Prey intake of army ants 323
123
Rettenmeyer C.W., Chadab-Crepet R., Naumann M.G. and Morales L.
1983. Comparative foraging by neotropical army ants. In: Social
Insects in the Tropics. Volume 2 (Jaisson P., Ed). Universite
´
Paris-
Nord, Paris. pp 59–73
Rico-Gray V. and Oliveira P.S. 2007. The Ecology and Evolution of
Ant-Plant Interactions. University of Chicago Press, Chicago.
320 pp
Schneirla T.C. 1971. Army Ants. A Study in Social Organization
(Topoff H.R., Ed). W.H. Freeman and Co., San Francisco. pp 369
Schneirla T.C. 1949. Army-ant life and behavior under dry-season
conditions. 3. The course of reproduction and colony behavior.
Bull. Am. Mus. Nat. His. 94: 3–81
Schneirla T.C. 1955. The army ants. Rep. Smith. Inst. 1955: 379–406
Swartz M.B. 1998. Predation on an Atta cephalotes colony by an army
ant Nomamyrmex esenbecki. Biotropica 30: 682–684
Teles Da Silva M. 1982. Behaviour of army ants Eciton burchelli and
E. hamatum (Hymenoptera, Formicidae) in the Belem region III.
Raid activity. Insect. Soc. 29: 243–267
Vidal-Riggs J.M. and Chaves-Campos J. 2008. Method review:
estimation of colony densities of the army ant Eciton burchellii
in Costa Rica. Biotropica 40: 259–262
Wetterer J.K., Gruner D.S. and Lopez J.E. 1998. Foraging and nesting
ecology of Acromyrmex octospinosus (Hymenoptera: Formici-
dae) in a Costa Rican tropical dry forest. Fla. Entomol. 81: 61–67
Wheeler G.C. and Wheeler J. 1976. Ant larvae: review and synthesis.
Mem. Entomol. Soc. Wash. 7: 1–108
Wheeler W.M. 1910. Ants: their Structure, Development and Behav-
ior. The Columbia University Press, New York. pp 663
324 S. Powell
123
