Phorid parasitoids affect foraging activity of Solenopsis richteri under different availability of food in Argentina

Phorid parasitoids affect foraging activity ofSolenopsis richteri under different availability of foodin Argentina

P A T R I C I A J . F O L G A R A I T and L A W R E N C E E . G I L B E R T *IFEVA,

Facultad de AgronomõÂa, University of Buenos Aires, Argentina, and *Department of Zoology and Brackenbridge Field

Laboratory, University of Texas at Austin, U.S.A.

Abstract. 1. In Argentina, six species of Pseudacteon parasitoids (Phoridae) attack

Solenopsis richteri, one of the two species of South American ®re ant that are

exotic pests in North America.

2. The presence of these Pseudacteon species signi®cantly reduces the number of

ants at food resources in the ®eld, as well as foraging activity generally.

3. Some Pseudacteon not only attack ants walking on trails or at feeding sites,

but also at mound entrances, inhibiting workers from leaving to forage.

4. The average size of foraging ants (which prescribes their suitability as hosts)

decreased in the presence of phorids.

5. The number of attacking phorids was correlated positively with the number of

ants walking towards the food on the trail before the attack.

6. Solenopsis richteri workers responded to manipulations of food size and

presence or absence of parasitoids in a risk-adjusting way, i.e. although more

foragers were recruited to larger food items, attacking phorids reduced ant foraging

activity by the same factor regardless of the size of the food offered.

7. The data suggest that S. richteri colonies juggle the needs to harvest food

ef®ciently, reduce competition, and avoid excess risks from parasitoids in complex

ways.

Key words. Argentina, biological control, ®re ants, food availability, parasitoid±

host interactions, Pseudacteon, Solenopsis richteri.

Introduction

Parasitoids are recognized as important regulators of many

insect species in natural, as well as agricultural, systems

(Hawkins & Sheehan, 1994), however most studies have

concentrated on the effects of hymenopteran parasitoids

(Price, 1975; Waage & Greathead, 1986; Mackauer et al.,

1990; La Salle & Gauld, 1993). Some members of the

dipteran family Phoridae are specialized parasitoids of ants

(reviewed by Disney, 1994). Only recently, after initial

recognition of the ecological effects of phorids (Feener,

1981), have studies addressed their role as possible

regulators of host ants (Orr et al., 1995; Porter et al.,

1995a). Phorid parasitoids, genus Pseudacteon, have been

shown to kill infected Solenopsis ants (Porter et al., 1995b),

and to reduce their diurnal foraging time (Feener & Brown,

1992; Orr et al., 1995; Porter et al., 1995c). Solenopsis

responses to Pseudacteon attacks range among remaining

frozen in place; elevating gasters, probably to spray venom

(Obin & Vander Meer, 1985); or hiding in the mound or

underneath a piece of litter or rock (Orr et al., 1995; Porter

et al., 1995c). In cases when foraging does not cease

completely, the size of foragers may be reduced to smaller

ants that are not preferred Pseudacteon hosts, as happens

with Solenopsis geminata (Feener, 1987).

Fire ants, Solenopsis richteri and Solenopsis wagneri

(= invicta), are native to Argentina and Brazil and were

introduced accidentally (free of their natural enemies) into

North America during the 1920s and 1940s, respectively.

R

Correspondence: Patricia J. Folgarait, Centro de Estudios e Investiga-

ciones, Universidad Nacional de Quilmes, Avenida Roque Saenz PenÄa

180, Bernal 1876, Buenos Aires, Argentina. E-mail: pfolgarait

@unq.edu.ar

Ecological Entomology (1999) 24, 163±173Ecological Entomology (1999) 24, 163±173

# 1999 Blackwell Science Ltd 163

They are yet to be controlled successfully. Recent

laboratory studies have investigated the interaction between

different species of phorid parasitoids and S. wagneri

(Gilbert & Morrison, 1997; Morrison et al., 1997). Other

experimental studies of interactions between imported ®re

ants and parasitoids in their natural habitats have been

conducted in Brazil (Orr et al., 1995; Porter et al., 1995c)

and/or involved the closely related S. saevissima that has

not been introduced into North America (Campiolo et al.,

1994; Porter et al., 1995a).

Various species of Pseudacteon might not only differ in

respect of which species of ®re ants they attack (Gilbert &

Morrison, 1997; Morrison et al., 1997), but also in their

range of tolerance to climatic conditions (Disney, 1994).

Because climatic conditions where Argentinian S. richteri

are found are comparable to those in the range of the

introduced ®re ants in south-eastern North America,

associated phorid populations may be more promising

candidates for introduction and biocontrol of imported ®re

ants than those from Brazil. With this in mind, the effect

of Pseudacteon spp. on S. richteri was tested in a native

habitat in Argentina.

Because one negative effect of Pseudacteon on Solenop-

sis is to reduce worker foraging activity (Feener & Brown,

1992; Orr et al., 1995; Porter et al., 1995c), it is important

to ask whether these parasitoids are effective only at certain

levels of food availability for the ants. Foraging decisions

should be a compromise of at least two variables: resources

and predation risk. In this system, worker ants emerging

from a mound to forage are in danger of being attacked by

phorid ¯ies, which means early death for the ant if the

oviposition is successful. Fraser and Huntingford (1986)

proposed four models (risk-reckless, risk-avoiding, risk-

balancing, risk-adjusting) and predicted the results of an

ANOVA of predation and food quality for prey that deal with

the con¯icting goals of foraging for food and avoiding

predation.

The interpretation of their model for the phorid±

Solenopsis system would be as follows (Fig. 1). At the

extremes where no trade-offs are found, ants may continue

foraging according to the value of the food despite the

presence of phorids (risk-reckless) or, conversely, ants may

stop foraging despite the quality/quantity of the food if the

parasitoid is present (risk-avoiding). The risk-reckless

response is expected in the case of hosts whose parasitoids

are very inef®cient in attacking. In contrast, the risk-

avoiding response would be expected in cases where

parasitoids are highly effective. When compromise between

food value and parasitism risk exists, two situations could

be expected: ants may risk foraging more for food of better

quality and less for food of lower quality in the presence of

the parasitoids (risk-balancing), or ants may forage

proportionally to the risk of parasitism in such a way that

the amount of foraging activity reduction is related to the

size of the hazard (risk-adjusting). In the system investi-

gated in this paper, some of these alternatives are

potentially confused in ANOVA tables but are easily

distinguished biologically. In one case of risk-adjusting

(Fig. 1, Risk-adjusting panel, bottom graph), a signi®cant

food±phorid interaction could be obtained, but only if the

parasitoids require a minimum threshold level of ants to be

present in order to become attracted and attack the ants.

Such attacks could lower ant activity to levels comparable

to those obtained for smaller quantities of food. Conversely,

in the risk-balancing model (Fig. 1, Risk-balancing panel,

bottom graph), an exponential recruitment of phorids at

larger baits could lead to a subsequent reduction in the rate

of increase in ant activity. In an experiment with only two

food states this could result in what appears to be risk-

adjusting. With these situations in mind, an experiment was

designed to explore whether, and to what extent, the effects

of phorid ¯ies on foraging by worker ®re ants depend on

food availability, using terms de®ned by Fraser and

Huntingford (1986).

The phorid±Solenopsis system seems to be appropriate

to test Fraser and Huntingford's (1986) model because:

(1) Solenopsis forages in a patchy way, because food resources

are distributed patchily (e.g. dead insects, aggregated honey-

dew-producers, plants with extra-¯oral nectaries). (2) Sole-

nopsis response varies with resource availability (Wilson,

1962a). (3) The presence of phorids has a negative effect on

Solenopsis (Feener & Brown, 1992; Orr et al., 1995; Porter

et al., 1995b). (4) Phorids elicit antiparasitoid responses that

include reducing food-harvesting by Solenopsis (Feener &

Brown, 1992; Orr et al., 1995; Porter et al., 1995c).

(5) Different phorid species have different ef®ciencies as

parasitoids (Gilbert & Morrison, 1997; Orr et al., 1997).

The particular interpretation in this study of Fraser and

Huntingford's (1986) model for the Solenopsis±Pseudacteon

system involves two key assumptions. (1) Increasing the

amount of food available increases the perceived quality of a

patch through a feedback system between foragers and the

colony. The evolution of such systems is due to the greater

harvest and defence ef®ciency allowed by food concentrated in

one place (Detrain & Deneubourg, 1997); foraging workers

can communicate both quality and quantity of food in

recruiting colony members to a resource located away from

the nest (Nonacs, 1990). (2) Changing the numbers of foraging

workers going to and from a patch or maintained at the food

source in a eusocial species is similar to changing the time

spent in a patch by a genetic individual. In Solenopsis, most

workers leaving a food source contribute to a trail odour, the

intensity of which is important in attracting a worker force

(Wilson, 1962b; Hangartner, 1969). Because changes in

worker numbers in this study were measured as a function of

food supply interacting with parasitoid presence, the number of

workers foraging on a trail or at a patch with food quantity x

and parasitoid number y is considered to be comparable to

Fraser and Huntingford's empirical system ± number of bites

by a stickleback in a patch of Daphnia density x and trout

density y.

Thus, this study had two major goals: (1) to quantify the

effect of Pseudacteon species on foraging rate and behaviour

of S. richteri in a native habitat in Argentina, and (2) to

determine whether the effect of Pseudacteon spp. on S. richteri

changes as the quantity of food available to the ants changes,

L164 Patricia J. Folgarait and Lawrence E. Gilbert164 Patricia J. Folgarait and Lawrence E. Gilbert

# 1999 Blackwell Science Ltd, Ecological Entomology, 24, 163±173

i.e. the ®rst test of Fraser and Huntingford's (1986) model for

an ant±parasitoid system.

Methods

This study was carried out at Reserva Costanera Sur in

Buenos Aires, Argentina (35.5°N, 58.5°W), a naturally

colonized 350-ha area of river terrace. The reserve parallels

the La Plata River and is characterized by a mosaic of

grasses and herbaceous growth interspersed with small

patches of forests and eutrophic wetlands.

A minimum of one observation per week was made

throughout the year (April 1995±June 1996), although there

were occasions on which there was virtually no ant activity

(below 9 °C or above a wind speed of 3.8 m s±1 at 10 cm from

the soil), or no phorid activity (below 14 °C or above a wind

speed of 1.5 m s±1 at 10 cm from the soil; all of August and

September 1995). Because of human disturbance that caused

frequent movement of ant colonies to new mound sites, some

of the 50 mounds found may have been repeat samples of the

same colonies.

Ants and phorids were collected for identi®cation purposes

throughout the year. All ®re ants were Solenopsis richteri

R

Fig. 1. Fraser and Huntingford's (1986) model adapted for an ant±parasitoid system with the predicted results from a two-factor ANOVA for ant

foraging responses (y, dependent variable) under two situations of food availability (small and large bait) and phorid hazard (absence and

presence). (See text for details.)

Parasitoid suppression of ®re ant activity 165Parasitoid suppression of ®re ant activity 165


(black ®re ant) and phorids were Pseudacteon curvatus

Borgmeier, P. tricuspis Borgmeier, P. borgmeieri Schmitz,

P. nudicornis Borgmeier, P. litoralis Borgmeier, and P.

comatus Borgmeier.

Experiment 1

This experiment tested the hypothesis that Pseudacteon spp.

affects foraging rate and behaviour of S. richteri negatively.

The design of the phorid exclusion experiments (April 1995±

March 1996) was modi®ed from that developed by Orr et al.

(1997). It was ®rst necessary to encourage the establishment of

foraging trails from which phorids were excluded. Bait was

offered in a 3-cm3 syringe, the barrel of which was cut to

expose a surface of 0.64 cm2 of tuna. A constant food supply

was maintained by the observer pushing the plunger forward,

releasing more tuna as it was consumed. The trail, from its

origin to the bait (» 15±20 cm), was covered by a

23 3 17 3 4 cm plastic transparent box. A ®ne mesh cloth

glued to the ends of the walls of the box excluded phorids but

allowed air circulation. By disturbing the mound slightly at one

point and by leaving tiny pieces of tuna along a route that ants

could follow to the syringe, the observer was able to encourage

a foraging trail that ®tted inside the box. The phorid exclusion

box or the whole length of the foraging trail (after the box was

removed) was shaded by a black umbrella to minimize

variation in temperature, humidity and wind during each set

of measurements and among mounds in different positions and

on different days.

To evaluate the impact of phorids, several counts of ants

were made. When trail activity was high (ants going back and

forth), two or three measures of ant activity were taken every

30 min. After 1±1.5 h, the box was removed to allow phorid

attacks. If, after being attacked, ants made an opening

underneath the bait or a new entrance closer to the bait, these

holes were plugged with small rocks in order to force the ants

to continue foraging along the same length of trail in all

observations. This artefact could overestimate the effect of

phorids on ant foraging activity, because Solenopsis can forage

underground (Markin et al., 1975) or tunnel underground to

shorten above-ground trails. On the other hand, this bias may

be countered by an opposing bias: the likelihood that a large

proportion of natural foraging trails in the ®eld is longer than

the one used in this study, because the foraging trails in this

study were very near the mounds. Ant activity was measured at

30-min intervals as: (a) the number of ants at the bait (referred

to in the Results section as ants at bait) ± the maximum number

of ants counted was 25, despite the fact that in many cases

more than 25 ants were present at the bait; (b) the number of

ants moving on the trail from the mound to the bait at one

instant (referred to as ants moving); and (c) the number of ants

(during a 4-min period) passing a ®xed point on the trail. For

(c), numbers approaching (referred to as ants approaching) and

leaving (ants leaving) the bait were tabulated separately.

Sample mounds at which phorids did not appear (post hoc

controls) were used to evaluate whether ant activity changed

over time or after the box was removed, by comparing the

activity of ants before and after the removal of the box.

The fact that phorids sometimes appeared near the end of an

observation interval or, conversely, may have disappeared just

prior to an interval in which their effects were scored was

taken into account when considering how to analyse the effect

of phorids on worker foraging. In each of these cases, foragers

may not have responded fully to the presence or absence of

phorids. Therefore, for each mound sampled, foraging counts

from a second interval (from each sampling mound on each

day) after phorids had arrived or disappeared, were used,

unless phorids were present only at one interval, in which case

that single data point was used. Summing the data for all the

observations made in the absence or in the presence of

parasitoids, or using the most extreme numbers found for each

situation would certainly overestimate the effect of phorids on

ant activity, and averaging the data would underestimate the

effect of phorids, so these approaches were not used. Data were

analysed using nonparametric Mann±Whitney (for two com-

parisons) or Wilcoxon (for more than two comparisons) tests,

both expressed as chi-squared approximations (JMP 3

Statistical Package, SAS Institute Inc., Cary, North Carolina).

To test the effect of Pseudacteon spp. on differential activity

by S. richteri castes, the average size of workers foraging in

the presence and absence of phorids was quanti®ed. Every ®fth

ant that crossed a ®xed point was collected from each sampled

mound and during each 30-min time interval. The ants

collected during periods with no phorids were kept in different

vials from those collected during or immediately after an

attack by phorids. Head width and total body length were

measured for each ant from a randomly chosen subset of 14

from all the paired vials (which represented samples of ants

taken before and after phorid attacks). The size distribution of

ants in the presence and absence of phorids was compared

using nonparametric Kolmogorov±Smirnov tests. The average

size of ants foraging in the presence and absence of phorids

was analysed using the Mann±Whitney U-test.

Experiment 2

This experiment was a modi®cation of experiment 1. In

order to test the null hypothesis of no differences in the effect

of Pseudacteon on Solenopsis foraging under different

availability of resources, the same type of exclusion experi-

ment was conducted from January to June 1996, but two

different quantities of food were offered. Following the model

of Fraser and Huntingford (1986), this system consisted of a

prey (S. richteri), a predator (in this case a parasitoid,

Pseudacteon), and two alternative quantities of food resource.

To standardize the food offered, two tuna-®lled syringes, one a

2.5-cm3 syringe (surface of 0.78 cm2 exposed), the other a 10-

cm3 syringe (surface of 2.00 cm2 exposed), were presented

simultaneously. Each syringe was placed 15±20 cm from the

colony, permitting both syringes to ®t into the plastic container

(as in experiment 1), but suf®ciently far apart (10±15 cm) to

allow two distinct recruiting columns to form from different

holes/entrances in the mound.



Adjacent foraging trails were independent with respect to

phorid activity. In general, different phorids attacked ants

along each trail simultaneously. However, if trails were

attacked alternately, the foraging rates/behaviour of ants from

the trail without attacking phorids remained unchanged until

phorids attacked that trail. Pseudacteon attracted to baited

trails were not manipulated arti®cially. No more than 40 ants

on the large syringe and 20 on the smaller syringe could be

counted accurately, but most of the time there were

undoubtedly more than these numbers of ants, especially at

the larger resource.

As in experiment 1, sample mounds at which phorids did not

appear (post hoc controls) were used to evaluate whether ant

activity changed over time or after the box was removed, by

comparing the activity of ants before and after the removal of

the box. Also, daily ant counts from each sampled mound

where phorids appeared were used to measure foraging activity

in the absence and presence of phorids. Most of the data used

to test the effect of phorids and food size on ant foraging were

transformed by the square root of the natural logarithm, except

for the number of ants moving away from the bait, which was

transformed by the natural logarithm of the square root of that

variable, in order to obtain normality and homoscedasticity.

These transformed data were analysed using a parametric two-

way factorial ANOVA (JMP 3 Statistical Package, SAS Institute

Inc., Cary, North Carolina). In this experiment, phorids were

found at only 13 mounds, which were used as replicates in the

ANOVAs. Data gathered in the absence of phorids (during

exclusion and before box removal) were used as controls, and

those gathered in the presence of phorids (after the box was

removed and phorids appeared) as the experimental group.

Because there were also two food variables, the total number

of observations was 52.

Results

Phorids

Total attack rate (all six species combined) averaged

2.07 6 1.1 min±1 (n = 79). These parasitoids did not attack

other species of ants such as Crematogaster spp., Acromyrmex

spp., Pheidole spp., or Monomorium spp., even though these

ant species appeared on baits or nearby. Pseudacteon spp.

associated with S. richteri included mostly the same set as

those reported for Brazilian S. saevissima (Porter et al., 1995a)

and S. wagneri (Orr et al., 1997), and those reported in historic

host records (Disney, 1994).

Experiment 1

Comparisons of ant activity across dates (n = 16) revealed no

signi®cant differences and were therefore pooled (for ants at

the bait: c2 = 18.31, d.f. = 15, P = NS; for ants moving:

c2 = 19.04, d.f. = 15, P = NS; for ants approaching the bait:

c2 = 16.79, d.f. = 15, P = NS; for ants leaving the bait:

c2 = 17.79, d.f. = 15, P = NS). For those sample mounds

(n = 25) at which no phorids appeared (post hoc controls),

there were no differences in ant activity after the box was

removed (for ants at the bait: c2 = 0.16, d.f. = 1, P = NS; for

ants moving: c2 = 0.70, d.f. = 1, P = NS; for ants approaching

the bait: c2 = 2.83, d.f. = 1, P = NS; for ants leaving the bait:

c2 = 0.08, d.f. = 1, P = NS). This result shows that neither the

presence of the box nor time had an effect on ant activity.

At the other mounds where Pseudacteon appeared after

removal of the box, however, the median number of ants at the

bait, moving along the trail, approaching the bait, and leaving

the bait was signi®cantly smaller (Mann±Whitney test:

P < 0.01 for each variable, n = 22) than in the absence of these

parasitoids (median number of ants in the presence and

absence of phorids, respectively, at the bait: 11.0 and 25.0,

moving along the trail: 6.0 and 31.0, approaching the bait: 7.7

and 72.0, leaving the bait: 8.0 and 45.5). The only signi®cant

correlation found was between the number of phorids attacking

and the number of ants approaching the bait before the attack

(r = 0.63, P < 0.001).

In 57.6% of all cases (n = 59), phorids did not arrive at the

experimental trails, while in 25.5% of the cases, phorids found

the ants rapidly (within 30 min) after the exclusion box was

removed (Fig. 2). An index of change in foraging activity (the

ratio between minimum and maximum ant activity for each

mound and day) was calculated for the presence and absence

R

Fig. 2. Cumulative percentage of all cases of baited S. richteri

discovered by Pseudacteon spp. plotted against observed phorid

arrival time after the exclusion box was removed in experiment 1

(top panel) and experiment 2 (bottom panel). For experiment 2, data

are shown separately for small and large bait sizes.



of parasitoids. Values near zero indicate that either the

environmental conditions (low temperatures or high winds)

or phorids had suppressed ant activity almost completely;

values near one indicate no change in activity over the

observation period. The median ratio of ant activity in the

presence of phorids was signi®cantly smaller than in the

absence of phorids (Mann±Whitney U-test = 95.5, 121.5, 77.5,

and 84.0, respectively, for the number of ants at the bait,

moving, approaching, and leaving the bait; all P < 0.05),

despite having a similar distribution (Kolmogorov±Smirnov

test for each dependent variable, Kp > 3.6, P > 0.05; Fig. 3).

Although the size distribution of ants foraging in the

presence and absence of phorids did not differ signi®cantly

(Kp = 4, P > 0.05; Fig. 4), the median size of ants foraging in

the presence of phorids was signi®cantly smaller (U = 9.5,

n = 14, P < 0.01) than in the absence of phorids (Fig. 4). There

was a signi®cant positive linear relationship between head

width and total length of ants collected in the absence

(y = 1.7805 + 4.3553x, r2 = 0.97, P < 0.001, n = 205) and in the

presence (y = 0.74396 + 4.9897x, r2 = 0.96, P < 0.001, n = 149)

of phorids.

Experiment 2

Comparisons of ant activity across dates (n = 11) revealed no

signi®cant differences for any of the dependent variables, and

data were therefore pooled (for ants at the bait: c2 = 9.77,

d.f. = 10, P = NS; ants moving: c2 = 9.16, d.f. = 10, P = NS; ants

approaching the bait: c2 = 6.28, d.f. = 10, P = NS; ants leaving

the bait: c2 = 7.22, d.f. = 10, P = NS).

For the post hoc control mounds (n = 9), signi®cantly more

ants remained at the larger bait (c2 = 17.00, d.f. = 1, P < 0.001),

and more were moving on the trail that connected to the larger

syringe (c2 = 5.25, d.f. = 1, P < 0.05). There were no signi®cant

differences, however, between the number of ants approaching

the larger bait (c2 = 1.89, d.f. = 1, P = NS) or leaving the larger

bait (c2 = 1.89, d.f. = 1, P = NS), and the bait offered in the

smaller syringe. Ant activity remained the same before and

after the box was removed in the post hoc controls (for each

dependent variable and size of bait, all P > 0.05), indicating

that neither the box nor time had an effect on ant activity.

In 42±46% of all cases (n = 26), phorids did not appear at the

experimental setting. In 30% of cases (at both food sizes),

phorids found the ants rapidly (within 30 min) after removing

the box (Fig. 2).

ANOVA showed that food quantity was signi®cant for ants

at the bait, marginally signi®cant for ants approaching the

bait, and nonsigni®cant for the number of ants leaving the

bait and moving on the trail (Table 1). Phorid presence was

signi®cant for each of the dependent variables measured,

indicating the negative effect of these parasitoids on ant

foraging. None of the interaction terms was signi®cant,

suggesting that reduction in foraging by ants was propor-

tional to the risk of attack by phorids, but independent of

the amount of food available. Overall, presence of phorids

reduced ant activity, although the greater availability of

food recruited a greater number of ants, but in a risk-

adjusting manner sensu Fraser and Huntingford (1986)

(Figs 1 and 5). Although numbers of parasitoids were not

controlled experimentally, they did not differ signi®cantly

L

Fig. 3. Frequency distributions of changes in ant foraging activity

(minimum/maximum foraging activity registered per mound and

day) in the absence and presence of phorids, from experiment 1.

Measurement of foraging activity included: ants at the bait, ants

moving, ants approaching, and ants leaving. The ratio of minimum

to maximum rates of foraging ranked from its smallest change (left)

to its greatest reduction (right). Diamonds indicate the category at

which the median is located for each dependent variable. Minimum

activity fell below 40% of maximum activity in only two to eight

cases (across the four dependent variables) out of 18 cases in phorid

absence, while in the presence of phorids it dropped below 40% of

maximum activity in 10±18 out of 22 cases.

168 Patricia J. Folgarait and Lawrence E. Gilbert168 Patricia J. Folgarait and Lawrence E. Gilbert


among the 13 colonies tested (c2 = 19.58, d.f. = 12, P = NS),

or between baits of different sizes (c2 = 0.0012, d.f. = 1,

P = NS). On average, there were 1.58 6 1.79 Pseudacteon

individuals present at each ant foraging trail or at the bait.

Discussion

Several species of Pseudacteon have been identi®ed that,

singly or as a group, have a negative effect on the foraging

behaviour of black imported ®re ants in Argentina, a habitat

comparable to that of South-eastern North America. Ant

activity is reduced by 77% (average for the four variables

measured), which is comparable to that shown for the red

imported ®re ant, S. wagneri from Brazil (Orr et al., 1995), S.

saevissima from Brazil (Porter et al., 1995c), and S. geminata

from Costa Rica (Feener & Brown, 1992). This reduction is

due to the presence of phorids (Table 1) and not to the effect of

time or correlated environmental changes after removal of the

box. Furthermore, the average size of foraging ants declined

after phorid attacks (Fig. 4), resulting in fewer available ants

suitable as phorid hosts. The lower density of Solenopsis

mounds found in natural sites in Argentina, compared to those

in North America (Porter et al., 1997), suggests that a

regulating factor is missing from the exotic environment into

which ®re ants have spread. The results of this study add

further support to the idea that phorids may be important in

regulating the ants.

Phorid abundance was correlated positively with ant activity

prior to attack in the experimental settings, suggesting that

Pseudacteon may have a numerical response to S. richteri

availability. However, phorids did not arrive at approximately

half of the experimental foraging trails (Fig. 2) despite the

presence of stable foraging trails and baits saturated by ants.

This pattern persisted throughout the year (experiment 1) and

even during the period of greater phorid activity (experiment

2). Whenever the same nest was sampled more than once, if

phorids did not appear the ®rst time, they did not appear the

second time, and if they appeared the ®rst time, in general they

also appeared the second time (P. J. Folgarait, pers. obs.).

These patterns suggest both low density and patchiness of

Pseudacteon populations, possibly resulting from the high rate

of disturbance evident at the study site or from undetected

habitat heterogeneity.

Great variability has also been found in the effect of

Pseudacteon on S. richteri behaviour, possibly due to the mode

of attack by the species of Pseudacteon. Some Pseudacteon

spp. seem to prefer to attack ants at the bait, others at the trail,

and still others at mound entrances; some behave like a sit and

wait predator (Feener & Brown, 1993, on leaf-cutter

parasitoids), resting on a piece of litter or rock until an ant

passes by before deciding to attack. Others behave like pursuit

predators, following ants intermittently or continuously along

foraging trails. It appears that these different modes of attack

lead to different behavioural responses by the ants, and

probably explain why, in response to Pseudacteon attack, some

ants remain frozen on the trails, others run towards the mound

or hide under available shelter, others remain at the bait for a

while, heads down and gasters elevated, and in 35% of the

cases when only one phorid attacked, ant activity resumed to

pre-phorid levels within 30 min, while in 41% of cases the ant

activity had not returned to normal after 1.5 h.

From eleven cases in which other ant species arrived at the

bait, in 73% of the cases they arrived only after phorids had

attacked, and in 18% of the latter cases, S. richteri lost all the

bait to competing ants. These preliminary observations suggest

that phorids may affect the outcome of interspeci®c competi-

tion among ants in this Argentinian community (as ®rst

documented by Feener, 1981, in Texas), and that harassment of

R

Fig. 4. Head width distribution for ants collected randomly in the

absence and presence of phorids from experiment 1. The median

size of foraging ants in the presence of phorids (head width: 0.72)

was signi®cantly smaller than in the absence of phorids (head width:

0.76).



®re ants may actually promote the coexistence of many

different ant species. Baits were mainly occupied by small

workers in the presence of phorids (although there was always

a ring of large individuals around the cut tip of the syringe),

but also by many larger workers in their absence. Such a shift

in worker sizes (Fig. 4) could be a compromise allowing

defence of the bait against competing ants but mainly with

workers that are not preferred by phorids. Both phorids and

competing ants are important aspects to consider with regard to

potential introduction of Pseudacteon into an exotic environ-

ment with the goal of controlling imported ®re ants. In Texas,

the Pseudacteon spp. speci®c to native S. geminata have been

shown to reduce their competitive parity with S. wagneri

(L. W. Morrison, unpublished data). Thus freedom from co-

evolved phorids is perhaps suf®cient to account for the

invading species' exclusion of the native Solenopsis over most

of its former range (e.g. see Porter et al., 1988).

The fact that only one food-related variable in experiment 2

(the number of ants present at the bait) proved to be signi®cant

was probably an outcome of the fact that these ants must be

concerned with protecting food from competing ants while

facing parasitoid threat. Only after ants cover the bait surfaces

do numbers moving on the trails reach equilibrium. Thus, after

workers found the baits, ants approaching the baits increased

continuously until saturation (e.g. Wilson, 1962a), the larger

bait being saturated by a greater number of ants than the

smaller bait. When saturation was reached, the bait was

protected (saturated baits were never taken by competing ants;

P. J. Folgarait, pers. obs.) and the numbers of ants moving on

the foraging trail stabilized (note that data collection began at

this point or one measurement period later). Remarkably, no

signi®cant differences were found between the number of ants

approaching and leaving saturated baits that differed in surface

area by a factor of two. Roughly twice the number of ants

going to and/or from the larger bait was expected had all the

ants simply arrived, collected food and returned to the colony.

It should be noted that Solenopsis ants have important defence

strategies, the effects of which were not quanti®ed in this

study. These include the use of subterranean access to food to

shorten above-ground foraging trails, the building of trenches,

and covering food with leaves or gravel. While all these

behaviours should reduce the hazard of phorid attacks while

foraging, they must also be viewed in the context of countering

competition with other ants. Pseudacteon±Solenopsis inter-

actions in Argentina suggest that activities such as covering

food items would be inhibited during phorid attack, thus

increasing the likelihood of losing competitive encounters.

Data from the second experiment demonstrate that (a) S.

richteri workers respond differently to varying quantities of

food (Taylor, 1977; Davidson, 1978; Nonacs & Dill, 1990),

(b) S. richteri perceive predation hazards that affect their

foraging decisions (Whitford & Bryant, 1979; MacKay, 1982;

Nonacs & Dill, 1988), and (c) the negative effects of

Pseudacteon species on trail-level foraging activity persist

despite changes in the size of food items found by S. richteri

and are proportional to the parasitoid hazard (risk-adjusting).

The foraging response of S. richteri to phorids corresponds to

Fraser and Huntingford's concept of risk-adjusting for four

reasons: (1) it is not risk-reckless because the presence of

phorids always had a signi®cant effect, reducing ant activity,

L

Table 1. ANOVA summaries for S. richteri responses in a 2 3 2 factorial experiment.

Source SS d.f. MS F P

Number of ants at the bait

Food 0.505 1 0.5050 21.600 0.0001

Phorids 0.522 1 0.5220 22.330 0.0001

Food 3 Phorids 0.001 1 0.0001 0.003 0.9550

Error 1.122 48 0.0230

Total 2.148 51

Number of ants moving

Food 0.165 1 0.1651 3.200 0.0803

Phorids 1.937 1 1.9373 37.590 0.0001

Food 3 Phorids 0.008 1 0.0082 0.160 0.6901

Error 2.267 44 0.0510

Total 4.378 47

Rate of ants approaching the bait

Food 0.329 1 0.3290 3.975 0.0525

Phorids 4.550 1 4.5500 54.890 0.0001

Food 3 Phorids 0.079 1 0.0790 0.963 0.3320

Error 3.564 43 0.0830

Total 8.587 46

Rate of ants leaving the bait

Food 0.360 1 0.360 1.910 0.1740

Phorids 9.670 1 9.670 51.150 0.0001

Food 3 Phorids 0.001 1 0.002 0.008 0.9270

Error 8.510 45 0.180

Total 18.460 48

170 Patricia J. Folgarait and Lawrence E. Gilbert170 Patricia J. Folgarait and Lawrence E. Gilbert


(2) it is not risk-balancing because ants did not risk

proportionally more at greater food levels and none of the

interaction terms was signi®cant, (3) it is not risk-avoiding

because the reduction of foraging activity in the presence of

phorids was never shut down completely, and (4) it is risk-

adjusting because for two ant activity dependent variables

there was a response of this kind (terms for food and phorids

were both signi®cant at P < 0.05, while the interaction term

was not signi®cant; see also next paragraph).

Nonacs and Dill (1990) tested the Fraser and Huntingford

model in the laboratory, by relating food quality (different

concentrations of sugar solutions) and predation risk in terms

of colony growth. They found a proportionally greater bene®t

in terms of colony growth rate for the ant Lasius pallitarsis

when foraging in patches made more risky by the presence of

predaceous ants, Formica subnuda. In contrast to results from

this study, their results support the risk-balance decision sensu

Fraser and Huntingford (1986). These differences could be

explained by considering that Lasius ants are monomorphic or

weakly polymorphic (HoÈlldobler & Wilson, 1990), and

because in Nonacs and Dill's experimental arena each Lasius

ant had the same probability of being killed by a Formica ant

at different concentrations of sugar (food treatment). In

contrast, mass recruiting polymorphic Solenopsis respond

differently, because the evolution of their foraging decisions

has been shaped by the hazard imposed by parasitoids.

Solenopsis colonies in response to a higher quantity resource

have the option to send larger (more valuable) workers, which

are also preferred by the parasitoid (Morrison et al., 1997).

Small ants, which are less suitable hosts for parasitoids, can

harvest food of higher quality without incurring a greater risk

of mortality than when harvesting food of low quality (no

trade-off involved). Colony response to increased food

quantity, however, involves the recruitment of ants of greater

size (in order to handle bigger solid food particles that cannot

be teased apart or in order to defend the food from competing

ants). These larger workers are, in turn, subjected to increased

parasitoid hazards (trade-off involved). In the system studied

here, then, the effect of polymorphism can explain why the

number of ants at the bait was the only de®nitive signi®cant

variable for the food term in the ANOVA. Up to the point of

saturation, large baits accumulate a greater number of large

workers (P. J. Folgarait, pers. obs.), but after the bait surface is

covered, and in the presence of phorids, the median size of

workers walking on the trail shifts towards smaller ants. The

activity of these ants, which are at or below the threshold of

R

Fig. 5. Evidence for a risk-adjusting response. Means and standard errors for the number of ants at the bait, moving, approaching, and leaving

the bait, in the absence (thick line) and presence (thin line) of phorids at the small (smaller food availability) and large (greater food availability)

bait in experiment 2. Because the presence of phorids always reduced, but did not eliminate ant foraging activity, and phorid effects on ant

activity did not change for different levels of food availability (see Table 1), the ants were responding to the applied treatments in a risk-

adjusting manner (see Fig. 1).



vulnerability to the parasitoid, remained indistinguishable

between food treatments. The risk of losing the resource to

competing ants is another pressure shaping the trade-off as

de®ned in terms of food and parasitoids. It might be

worthwhile for the colony to risk a few big workers at the

bait to assure its defence against other ants, but it might not be

worthwhile risking many large-size workers along the foraging

trail where phorids are also active and effective. The main

consequence of using ants smaller than the sizes preferred by

phorids for most food retrieval along the trail would be that it

would take longer to gather all the food from the high-quantity

food treatment. It is hypothesized that in this system, slower

food retrieval is less costly than losing large workers to

parasitoids along the trail, but only measurements of colony-

level consequences of such alternatives will provide a

de®nitive test of this view.

For the range of food quantities provided and natural

risks observed in the ®eld, the data (Fig. 5) show that S.

richteri workers respond to the phorid threat in a risk-

adjusting manner (Fig. 1). This type of response would help

to ensure the effectiveness of these parasitoids as biological

control agents across environments or seasons in which

distributions of food particle sizes might be likely to vary.

It is possible that larger or higher quality resources than

those used in these experiments, or a situation of

interference competition that requires greater ant recruit-

ment to defend the food, could induce the ants to incur a

greater risk from phorid attack, resulting in a risk-balancing

response (Fig. 1, Risk-balancing panel, top graph). If this is

the case, an upward in¯ection of the ant activity curve

would have been expected at some threshold state of the

food resource. This possibility (which the experiment in

this study cannot exclude) holds more than theoretical

interest. To increase the chances of establishment of an

exotic phorid introduced for biological control, maximizing

potential attack rates for these short-lived (Morrison et al.,

1997) parasitoids may be crucial. Risk-balancing responses

include cases in which the rate of increase of ant activity

actually decreases as food item quantity/quality increases

beyond some value (Fig. 1, Risk-balancing panel, bottom

graph). This could occur if phorid recruitment or attack rate

increase is nonlinear after a particular density of ants on or

near bait is reached. Again, this possibility is of practical

interest if, for example, food-bait application is to be used

in phorid introduction attempts.

Acknowledgements

We are indebted to S. Recio, Director of the Costanera Sur

Reserve, who aided this research by providing permits and

logistical support. M. Muia and S. Mariani helped collect data

for experiment 1 in the ®eld and in the laboratory, respectively.

We thank M. R. Orr for his continuous encouragement,

suggestions on experimental methods, and comments on the

manuscript. S. Porter identi®ed the phorids and J. Trager

identi®ed the Solenopsis ants. L. W. Morrison, D. H. Feener,

S. Porter, B. A. Hawkins, R. H. Disney and one anonymous

reviewer also made suggestions for improving the manuscript.

C. Wuellner provided some key references. This research was

supported by grants from the R. J. Kleberg and H. C. Kleberg

Foundation to L.E.G. and NSF grant DEB-9528120 (Donald H.

Feener and Lawrence E. Gilbert co-P.I.s).

References

Campiolo, S., Pesquero, S.A. & Fowler, H.G. (1994) Size selective

oviposition by phorid parasitoids on workers of the ®re ants,

Solenopsis saevissima. Etologia, 4, 85±86.

Davidson, D.W. (1978) Experimental tests of the optimal diet in two

social insects. Behavioral Ecology and Sociobiology, 4, 35±41.

Detrain, C. & Deneubourg, J.L. (1997) Scavenging by Pheidole

pallida: a key for understanding decision making systems in ants.

Animal Behaviour, 53, 537±547.

Disney, R.H. (1994) Scuttle Flies: the Phoridae. Chapman & Hall,

London.

Feener, D.H. (1981) Competition between ant species: outcome

controlled by parasitic ¯ies. Science, 214, 815±817.

Feener, D.H. (1987) Size selective oviposition in Pseudacteon

crawfordi (Diptera: Phoridae), a parasite of ®re ants. Annals of

the Entomological Society of America, 80, 148±151.

Feener, D.H. & Brown, B.V. (1992) Reduced foraging of Solenopsis

geminata in the presence of parasitic Pseudacteon spp. Annals of the

Entomological Society of America, 85, 80±84.

Feener, D.H. & Brown, B.V. (1993) Oviposition behavior of an ant

parasitizing ¯y, Neodohriphora curvinervis (Diptera: Phoridae), and

defense behavior by its leaf cutting ant host Atta cephalotes

(Hymenoptera: Formicidae). Journal of Insect Behavior, 6, 675±

688.

Fraser, D.F. & Huntingford, F.A. (1986) Feeding and avoiding

predation hazard: the behavioral response of prey. Ethology, 73,

56±68.

Gilbert, L.E. & Morrison, L.W. (1997) Patterns of host speci®city in

Pseudacteon parasitoid ¯ies (Diptera, Phoridae) that attack

Solenopsis ®re ants (Hymenoptera, Formicidae). Environmental

Entomology, 26, 1150±1154.

Hangartner, W. (1969) Structure and variability of the individual odor

trail in Solenopsis geminata Fabr. (Hymenoptera, Formicidae).

Zeitschrift fuÈr Vergleichende Physiologie, 62, 111±120.

Hawkins, B.A. & Sheehan, W. (1994) Parasitoid Community Ecology.

Oxford University Press, Oxford.

HoÈlldobler, B. & Wilson, E.O. (1990) The Ants. Belknap Press of

Harvard University Press, Cambridge, Massachusetts.

La Salle, J. & Gauld, I.D. (1993) Hymenoptera and Biodiversity. CAB

International, Wallingford, U.K.

Mackauer, M., Ehler, L.E. & Roland, J. (1990) Critical Issues in

Biological Control. Intercept, Andover, U.K.

MacKay, W.P. (1982) The effect of predation of western widow

spiders (Aranae: Therididae) on harvester ants (Hymenoptera:

Formicidae). Oecologia, 53, 406±411.

Markin, G.P., O'Neal, J. & Diller, J.H. (1975) Foraging tunnels of the

red imported ®re ant, Solenopsis invicta (Hymenoptera: Formici-

dae). Journal of Kansas Entomological Society, 48, 83±89.

Morrison, L.W., Dall'Aglio-Holvorcem, C.G. & Gilbert, L.E. (1997)

Oviposition behavior and development of Pseudacteon ¯ies

(Diptera: Phoridae), parasitoids of Solenopsis ®re ants (Hymenop-

tera: Formicidae). Environmental Entomology, 26, 716±724.

Nonacs, P. (1990) Death in the distance: mortality risk as information

for foraging ants. Behavior, 112, 23±35.



Nonacs, P. & Dill, L.M. (1988) Foraging response of the ant Lasius

pallitarsis to food sources with associated mortality risk. Insectes

Sociaux, 35, 293±303.

Nonacs, P. & Dill, L.M. (1990) Mortality risk vs. food quality trade

offs in a common currency: ant patch preferences. Ecology, 71,

1886±1892.

Obin, M.S. & Vander Meer, R.K. (1985) Gaster ¯agging by the ®re

ants (Solenopsis spp.): functional signi®cance of venom dispersal

behavior. Journal of Chemical Ecology, 11, 1757±1768.

Orr, M.R., Seike, S.H., Benson, W.W. & Gilbert, L.E. (1995) Flies

suppress ®re ants. Nature, 373, 292±293.

Orr, M.R., Seike, S.H. & Gilbert, L.E. (1997) Foraging ecology and

patterns of diversi®cation in dipteran parasitoids of ®re ants in South

Brazil, genus Pseudacteon. Ecological Entomology, 22, 305±314.

Porter, S.D., Eimeren, B.V. & Gilbert, L.E. (1988) Invasion of red

imported ®re ants (Hymenoptera: Formicidae): microgeography of

competitive replacement. Annals of the Entomological Society of

America, 81, 913±918.

Porter, S.D., Fowler, H.G., Campiolo, S. & Pesquero, M. (1995a) Host

speci®city of several Pseudacteon (Diptera: Phoridae) parasites of

®re ants (Hymenoptera: Formicidae) in South America. Florida

Entomology, 78, 70±75.

Porter, S.D., Pesquero, M.A., Campiolo, S. & Fowler, H.G. (1995b)

Growth and development of Pseudacteon phorid ¯y maggots

(Diptera: Phoridae) in the heads of Solenopsis ®re ant workers

(Hymenoptera: Formicidae). Environmental Entomology, 24, 475±

479.

Porter, S.D., Vander Meer, R.K., Pesquero, M.A., Campiolo, S. &

Fowler, H.G. (1995c) Solenopsis (Hymenoptera: Formicidae) ®re

ant reactions to attacks of Pseudacteon ¯ies (Diptera: Phoridae) in

Southeastern Brazil. Annals of the Entomological Society of

America, 88, 571±575.

Porter, S.D., Williams, D.F., Patterson, R.S. & Fowler, H.G. (1997)

Intercontinental differences in the abundance of Solenopsis ®re ants

(Hymenoptera: Formicidae): an escape from natural enemies?

Environmental Entomology, 26, 373±384.

Price, P.W. (ed.) (1975) Evolutionary Strategies of Parasitic Insects

and Mites. Plenum, New York.

Taylor, F. (1977) Foraging behavior of ants: experiments with two

species of myrmecine ants. Behavioral Ecology and Sociobiology,

2, 147±167.

Waage, J. & Greathead, D. (eds) (1986) Insect Parasitoids. Academic

Press, London.

Whitford, W.G. & Bryant, M. (1979) Behavior of a predator and its

prey: the horned lizard (Phrynosoma cornutum) and harvester ants

(Pogonomyrmex sp.). Ecology, 60, 686±694.

Wilson, E.O. (1962a) Chemical communication among workers of the

®re ant Solenopsis saevissima. 1: The organization of mass-

foraging. Animal Behavior, 10, 134±147.

Wilson, E.O. (1962b) Chemical communication among workers of the

®re ant Solenopsis saevissima. 2: An information analysis of the

odour trail. Animal Behavior, 10, 148±158.

Accepted 11 September 1998

RParasitoid suppression of ®re ant activity 173Parasitoid suppression of ®re ant activity 173


Documents

Phorid parasitoids affect foraging activity of Solenopsis richteri under different availability of food in Argentina