Experimental analysis of nestmate recognition in the imported fire ant Solenopsis invicta Buren (Hymenoptera


Material Information

Experimental analysis of nestmate recognition in the imported fire ant Solenopsis invicta Buren (Hymenoptera formicidae)
Physical Description:
ix, 110 leaves : ill. ; 28 cm.
Obin, Martin S., 1949-
Publication Date:


Subjects / Keywords:
Solenopsis invicta   ( lcsh )
Kin recognition in animals   ( lcsh )
Fire ants   ( lcsh )
bibliography   ( marcgt )
theses   ( marcgt )
non-fiction   ( marcgt )


Thesis (Ph. D.)--University of Florida, 1987.
Includes bibliographical references.
Statement of Responsibility:
by Martin S. Obin.
General Note:
General Note:

Record Information

Source Institution:
University of Florida
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
aleph - 001088017
notis - AFH3390
oclc - 19299314
System ID:

Full Text








The research presented here is the result of

discussions (some heated) with Drs. Robert Vander Meer,

Laurence Morel, Norm Carlin, James Trager, Jane Brockmann,

Ken Ross, Barry Lavine, Les Greenberg, Klaus Jaffe, Alex

Mintzer, Brad Vinson, Sanford Porter, John Sivinski, James

Lloyd and with Bob Weeks, Genie Avery and Bruce Sutton.

Financial support was provided by teaching assistantships

from the Department of Zoology, University of Florida, and

by a research assistantship from the Imported Fire Ant

Project of the Agricultural Research Service, United States

Department of Agriculture. Technical support during all

phases of the research was graciously provided by USDA

research scientists and their staffs, and especially by Drs.

C. S. Lofgren and R. K. Vander Meer. Additional thanks in

this regard are extended to Drs. D. P. Wojcik and B. M.

Glancey for collecting fire ant colonies in Mississippi, and

to Drs. D. F. Williams and D. P. Jouvenaz for equipment and

space. Equally important to the successful completion of

this research was the tireless assistance and tolerance of

Terry Krueger, Bob Weeks and Lloyd Davis.

Moral support, friendship and academic guidance above

and beyond the call of duty were unstintingly offered by

Drs. Frank Nordlie, Bob Vander Meer, Jim Lloyd and Jon

Reiskind, and above all, by my wife Susan. It can be stated

with certainty that none of them knew what they were getting




ACKNOWLEDGMENTS .......................... ..................ii

LIST OF TABLES............................................vi

LIST OF FIGURES.......................................... vii

ABSTRACT................ ...... ....... .... ... ........ viii


I INTRODUCTION......... ................................. 1


Introduction ........................................... 6
Materials and Methods.............................. ... 7
Results...................... .............. ........... .11
Discussion........ ................... ............... 14


Introduction........................ ..... ........... 26
Materials and Methods................................ 28
Results................. ............... .......... .... 34
Discussion.................... .................... ..37


Introduction......................... .... ... ......... 50
Materials and Methods......................... ...... 51
Results................................. ............. 55
Discussion.................................... ...... 57


Introduction.................. ....................... 67
Materials and Methods................................ 69
Results.................... .................... ...... 73
Discussion..................... .............. .. ....... 74

TO HYBRID ZONE DYNAMICS .............................. 80

Introduction.......................................... 80
Materials and Methods ................................ 82
Results .............................................. 85
Discussion ........................................... 85

VII CONCLUSION ............................................. 90

LITERATURE CITED .......................................... 95

BIOGRAPHICAL SKETCH ...................................... 110



2.1 Five-unit behavioral scale used in preliminary
nestmate recognition experiments.................. 22

2.2 S. invicta nestmate recognition
bioassay summary...................................23

2.3 Resident response when intruder and resident
colonies are from the same population or from
different populations ............................. 24

2.4 Peak areas of S. invicta cuticular hydrocarbons
in lab and field colonies ......................... 25

3.1 Nine-behavior recognition assay used
in chapters I V................................. 48

3.2 Frequency of least aggressive behavioral response
during four series of nestmate introductions......49

4.1 Response of laboratory colonies to queenless
and queenright kin and non-kin .................... 65

4.2 Effect of queen-tending on worker labels..........66

5.1 Aggression elicited from workers in queenright
colonies by kin intruders maintained on diets
containing identical, common or foreign
components ........................................ 79

6.1 Recognition response of S. invicta, S. richteri
and S. invicta/S. richteri hybrids................ 89



2.1 Chromatographic pattern of _. invicta worker
cuticular extracts.................................21

3.1 Effect of laboratory vs field rearing on
nestmate recognition between split colonies
of S. invicta..................................... 45

3.2 Effect of diet on nestmate recognition
in S. invicta ..................................... 46

3.3 Effect of the presence (28 days) of an alien
queen or alien worker on recognition between
former nestmates.......................... ....... 47


Abstract of Dissertation Presented to the Graduate School
of the University of Florida in Partial Fulfillment of the
Requirements for the Degree of Doctor of Philosophy



Martin S. Obin

December, 1987

Chairman: Dr. Frank Nordlie
CoChairman: Dr. Robert Vander Meer
Major Department: Zoology

This study examines the mechanism and adaptive

significance of nestmate recognition in monogyne colonies of

the imported fire ant, Solenopsis invicta. Discrimination

of nonnestmates was quantified in a bioassay that measured

agonism in the context of nest defense directed at

"intruder" ants placed near (but not in) the nest. The

following results were obtained: (1) Environmentally

correlated cues dominated the recognition cue hierarchy of

laboratory and field colonies, and diet alone significantly

altered recognition "labels" and "templates" of laboratory-

reared workers. (2) Heritable recognition cues

("discriminators") of workers constituted some portion of

the recognition-cue hierarchy of this species. (3) Gas-

liquid chromatographic and multivariate analyses indicated

that cuticular hydrocarbon pattern was a poor predictor of


recognition response when colonies were reared under

different environmental regimes (laboratory vs field). (4)

Queen-derived cues ("queen discriminators") did not

significantly alter labels or templates of workers in nests

requeened with an alien field queen for 1 month, or reared

from the larval stage for 5 months in the absence of a

queen. (5) However, workers that tended mother queens in

small cups were attacked more often by kin than non-tending

workers. Aggression directed at these workers was not

significantly associated with ovarian development of the

tended queen. (6) Template-label matching of dietary odors

was based on overall "similarity" rather than on a "discrete

odor" mechanism. (7) The recognition bioassay was extended

to within and between species tests among laboratory-reared

(1 year) colonies of S. invicta, S. richteri and their

Mississippi hybrid. Within-group recognition scores were

highest for hybrids and lowest for S. invicta, suggesting

greater genetic variability for heritable recognition cues

in hybrid populations and least variability in S. invicta

populations. Results are applied to the dynamics of the

Mississippi fire ant hybrid zone and contrasted with

electrophoretic studies of "neutral," Mendelian markers in

fire ant populations.


"Recognition" in animals can be defined as the ability

to distinguish classificatory differences among organisms

and to base subsequent behavior on these differences. For

example, many species ensure reproductive isolation by

distinguishing conspecific mates from among a pool of

sympatrically occurring congeners (Ehrman 1964; Hoy et al.

1977; Ratcliffe and Grant 1983). Animals commonly make

intraspecific distinctions as well, responding

differentially to population members of different sexes,

ages and physiologies (Muller-Schwarze 1974; Barrows et al.

1975; review in Holldobler & Carlin, in press). Much

current interest surrounds the demonstrated capacity of many

species to assess relatedness ("kin recognition") and

individual differences ("individual recognition") among

conspecifics (for reviews see Barrows et al. 1975;

Holldobler & Michener 1980; Beckoff 1981; Lloyd 1981;

Beecher 1982; Holmes & Sherman 1982; Gadagkar 1985;

Holldobler & Carlin, in press; Fletcher & Michener 1987).

This highly developed recognition capability directs animal

decisions associated with three categories of behavior

(discussed below). These are (1) nepotism (Holmes & Sherman

1983) and "reciprocal altruism" (Trivers 1971), (2) mate


choice and pair-bonding, and (3) the formation and

maintenance of dominance and territorial relationships.

Perhaps the greatest attention has been focused on how

recognition systems direct nepotism and reciprocal altruism

(Petrinovich 1974; Ross & Gamboa 1981; Blaustein & O'Hara

1982; Davis 1982; Hamilton et al. 1981; Breed 1983, 1986;

reviews in Holmes & Sherman 1983; Sherman & Holmes 1985;

Hepper 1986; Fletcher & Michener 1987). This is due in

large part to the fact that Hamilton's kin selection

hypothesis (1964) argued strongly for the existence of a kin

assessment mechanisms) ("kin recognition"), especially

among the social insects (see also Wilson 1971; Alexander

1974, 1979; review in Gadagkar 1985). Recognition systems

also figure prominently in mate choice and pair-bonding,

during which prospective mates can be assessed with respect

to species, outbreeding and mate quality criteria (Ehrman

1966; Averhoff & Richardson 1974; Muller-Schwarze 1974;

Yamazaki et al. 1976, 1982; Johnson 1977; Bateson 1979,

1980, 1982, 1983; Wickler & Seibt 1980, 1983; Lloyd 1981;

Hoogland 1982; Lenington 1983; Spiess 1987). It should be

emphasized that kin recognition mediates mate choice in

those species where the assessed degree of relatedness

directs optimal outbreeding (review in Bateson 1983).

Population geneticists and evolutionary biologists have

debated the existence of mate choice based on mate quality

(see discussion in Lenington 1983). Recognition studies

indicate that selection favoring mate quality assessment may


be widespread and suggest the probable nature of those

selective forces (Bateson 1983; Hamilton & Zuk 1982).

Finally, recognition has been shown to be an important

factor in the formation of dominance and territorial

relationships (Weedon & Falls 1959; Emlen 1971; Barnard &

Burk 1979, 1981; Caldwell 1979, 1982; Searcey et al. 1981;

Beletsky 1983; Holldobler & Carlin, in press). As pointed

out by Barnard and Burk (1979) the role of (individual)

recognition in the maintenance of dominance hierarchies has

been debated since the concept of social dominance was

introduced in 1922.

The following study addresses discrimination of

nestmates from nonnestmates ("nestmate recognition") in

monogyne (single-queen) colonies of imported fire ant

Solenopsis invicta Buren (Hymenoptera: Formicidae). This

neotropical, myrmicine ant was introduced to the Southeast

United States from South America during the 1930s and

currently dominates the ant fauna of disturbed habitats

throughout the region (for reviews of systematics and

biology see Buren 1972; Lofgren et al. 1975; volume edited

by Lofgren & Vander Meer 1986). Fire ants are highly

eusocial--i.e., a queen(s) and her worker daughters

("nestmates") live in a colony in which reproductive

division of labor and distinct (temporal) "castes" are

observed (for review of fire ant biology see Wilson 1971,

1978 and citations therein; Mirenda & Vinson 1976; Vander

Meer 1983; volume edited by Lofgren & Vander Meer 1986).


Nonnestmate conspecifics are immediately recognized and

attacked in and around the nest, and this discrimination is

mediated by chemoreception of cuticular-borne odorants.

Although the ability of ants and bees to discriminate colony

members from nonmembers (nestmate recognition) has been

known for over a century (citations in Wilson 1971), the

intense interest in animals' abilities to distinguish

degrees of kinship within a colony ("kin recognition") is a

more recent outgrowth of challenges to "inclusive fitness"

arguments (Hamilton 1964) posed by the existence of polygyne

and multiply-mating social insects (review by Gadagkar

1985). However, because fire ant queens mate with only one

male monandryy), all nestmates in monogyne colonies are

equally related to one another (on average). As a result,

nestmate (colony-level) recognition in monogyne fire ant

colonies is theoretically and functionally indistinguishable

from "kin recognition."

In general, kin and nestmate recognition studies of the

last decade have addressed three issues. These are (1) the

nature and origin of recognition cues (Gadagkar 1985, Carlin

& H611dobler 1986, 1987; Hepper 1986; Breed & Bennett 1987),

(2) the ontogeny of recognition cues and recognition

capability (Porter & Wyrick 1979; Porter et al. 1981; Holmes

& Sherman 1982; Morel 1983; Morel et al., in press;

Isingrini et al. 1985; Brian 1986; Gamboa et al. 1986b;

Stuart 1987), and (3) the nature of decision rules governing

discrimination (reviews by Holmes & Sherman 1983; Gadagkar


1985; Breed & Bennett 1987). This study uses a recognition

bioassay to experimentally analyze cues and decision rules

mediating fire ant nestmate recognition. Issues addressed

include whether cues mediating worker/worker recognition are

heritable or environmentally acquired (chapter II), whether

cuticular hydrocarbons are nestmate recognition cues in this

species (chapter II), the relative contributions of

environment (diet), queens and workers to recognition cues

(chapters III and IV), and how odor phenotypes of

nonnestmates are recognized as "foreign" (chapter V). In

chapter VI, I evaluate the species-level discrimination

capabilities of S. invicta, S. richteri and their hybrid and

extend the results to an explanation of hybrid zone dynamics

among the three populations. Concluding remarks discuss the

more global implications of results presented here, and

suggest future directions for recognition studies of fire




The ability to discriminate colony members or nestmates

from nonnestmates (i.e., "nestmate recognition") is

essential for the stability of insect societies in both

ecological and evolutionary time (Wilson 1971). As in

recognition systems of many invertebrate and vertebrate

species (reviewed in Hepper 1986; Fletcher & Michener 1987),

nestmate recognition in social insects is mediated by

olfactory cues (H611dobler & Michener 1980; Bradshaw & Howse

1984). Elements of an individual's odor profile determine

to a large extent whether it is fed, tolerated or attacked

in or near a foreign colony. Chemical recognition cues can

be genetically correlated as well as environmentally

determined or acquired (reviews in Hlldobler & Michener

1980; Gadagkar 1985; Fletcher & Michener 1987). As a

consequence, recognition systems can be complex and dynamic,

rendering chemical elucidation difficult.

This chapter reports a preliminary assessment of

genetic and environmental inputs to recognition cues of the

imported fire ant Solenopsis invicta Buren (Myrmicinae).

Workers of this neotropical species aggressively defend



their nests against intruders, including conspecific

neighbors, and thus agonism in the context of nest defense

is a reliable index of discrimination. As in other

myrmicine ants studied (Jutsum 1979; Vilela & Howse 1986),

attack is released by close-range or contact chemoreception

of surface compounds (see also Wilson 1971; Bradshaw & Howse

1984). By rearing fire ant colonies under controlled,

laboratory conditions, it was possible to assess the

potential contribution of environmental and genetic inputs

to nestmate recognition cues of workers. It was also

possible to indirectly test the suggestion of Howard and

Blomquist (1982) and Vander Meer and Wojcik (1982) that

these cues might be provided in part by cuticular


Materials and Methods

Ant Rearing

Monogyne S. invicta colonies were established from

newly-mated queens collected near Gulfport, Mississippi, and

Gainesville, Florida, and established at the USDA/ARS

Insects Affecting Man and Animals Research Laboratory,

Gainesville, Florida. For details of collection and

laboratory rearing see Banks et al. (1981). Colonies were

maintained in plastic petri dish cells (diameter = 14.0 cm)

with moist Castone floors at 26-27 degrees centigrade on a

diet of honey-water, fly pupae, roaches and hard-boiled egg.

Illumination was provided by fluorescent lamps, and the


light/dark cycle was variable. The petri dish cells were

placed in plastic trays (52.0 x 39.0 x 7.5 cm) that served

as foraging arenas. The sides of the tray were Fluon-


Nestmate Recognition Bioassay

The recognition bioassays developed for this study are

based on the belief that the magnitude of aggression

elicited from a fire ant worker by a conspecific is

positively correlated with the extent of perceived

differences in nestmate recognition cues. Eighteen months

after establishment in the laboratory, six of these colonies

were tested for their ability to discriminate non-nestmates

introduced from (a) each of the five other lab-reared

colonies and from (b) S. invicta field colonies (N=6)

situated in a 24.0 x 55.0 meter section of lawn adjacent to

the USDA/ARS Fire Ant Project laboratories. Ants introduced

into foreign colonies (i.e., "intruders") were collected in

groups of 12-15 from either the mound perimeter of field

colonies or from the foraging tray of laboratory colonies.

Medias--i.e., ants of the intermediate size caste (Wilson

1978) were selected by visual inspection, and only these

were introduced into foreign colonies. Collected ants were

maintained in small, glass vials and then introduced

individually on the end of forceps into the foraging arena

of a laboratory colony. Only ants that walked undisturbed

from the forceps into the "resident" colony tray were

tested. Intruders were positioned on the tray floor so as


to maximize both the initial distance (5-10 cm) between the

site of introduction and resident ants as well as the

distance from any previous introduction. The forceps were

rinsed with acetone and air-dried after each introduction.

Controls consisted of lab-reared ants that were removed and

subsequently introduced back into their colony of origin (as

above). Tests were conducted on two consecutive days. On

day one, control introductions were conducted first,

followed by tests of lab-reared intruders and finally by

tests of field intruders. This sequence was reversed on the

second day. A 90 minute period was observed between the

three series of introductions. Within each series (other

than controls) replicates were randomized with respect to

intruder origin and resident colony, and no resident colony

was re-tested before all resident colonies had been

replicated equally. When possible, introduced ants

(including at times those residents clinging to them) were

removed after each introduction.

The recognition bioassay consisted of five behaviors

(Table 2.1). The two most aggressive, "hold/sting" and

"hold," always resulted in intruder death, whereas the three

least aggressive responses--"hold/release," "antennate/

follow" and "no follow"--resulted in continued intruder

mobility throughout the foraging tray. An introduction was

considered completed either when "hold/ sting" was observed,

or after five resident ants had contacted the intruder.


Behavioral data were analyzed according to statistical

methods described in Sokal and Rohlf (1981).

Chemical and Statistical Analysis of Cuticular Hydrocarbons

Samples (N = 3) consisting of three media workers were

collected (as above) from each of the lab and field colonies

on the second day of tests. Samples were extracted for 7

minutes in 150 microliters of hexane. Previous studies

(Vander Meer, unpublished data) indicate that hydrocarbons

extracted by this method are almost exclusively of cuticular

origin. Solvent was removed, reduced under a nitrogen

stream to approximately 25 microliters and analyzed by

capillary gas chromatography: 15 m x 0.33 mm DB-1 column

(J. and W. Scientific, Inc.), splitless; Varian 3700, FID,

temperature program = 150 degrees centigrade (1 min hold),

then 4 degrees/min to 285 degrees (Fig 1). Normalized and

autoscaled areas (Luc Massart and Kaufman 1983) of the five

major S. invicta cuticle hydrocarbons were obtained. These

peaks have been chemically identified (Lok et al. 1975;

Nelson et al. 1980) and are qualitatively invariant in this

species. Analysis included t-tests of individual peak areas

in lab and field colonies as well as canonical discriminant

function and nearest neighbor discriminant analyses (SAS

1982) of cuticular hydrocarbon patterns in the two types of





Most introduced ants moved immediately about the tray,

although some groomed momentarily before doing so. Since

the foraging trays contained a minimum of several hundred

workers, intruder-resident contact (antennation) occurred

within 1 minute of introduction in all cases. No

introduction took more than 3 minutes to complete. Forty-

eight "control," 120 lab-reared and 96 field intruders were

tested. Eight lab-reared and 14 field intruder replicates

were excluded from analysis as a result of either

experimenter confusion as to ant identity (N = 2),

trophallactic appeasement (Bhatkar 1979) of resident ants by

an intruder (N = 2), initiation of aggressive behavior by

the intruder (N = 7), or rapid, erratic escape attempts

following introduction (N = 11). Intruders from field

colonies (5 of 96 introductions) were no more likely to

exhibit initial aggressive responses than were intruders

from laboratory colonies (2 of 120 introductions) (chi-

square, 1 df = 2.15, P = 0.14), but field intruders were

more likely to attempt escape by moving rapidly to one side

of the test arena and attempting to climb out (9 of 96 vs 2

of 120; chi-square, 1 df = 6.56, P = 0.01).

Data (Table 2.2) clearly demonstrate that intruders

introduced from either lab-reared or field colonies elicit

more aggression from resident ants than "control" intruders.

No control ants were killed, whereas 18.8% (21 of 112) of


the laboratory intruders were dispatched (Test of

Percentages; t, 158 df = 5.17, P < 0.001). In addition,

only 10.4% (5 of 48) of controls were physically detained by

residents (i.e., elicited "hold/release," "hold," or "hold/

sting"), compared with 54.5% (61 of 112) of lab-reared

intruders (chi-square, 1 df = 26.90, P < 0.001). Resident

ants responded more aggressively to introductions of field

ants than to introductions of lab-reared ants. Both the

proportion of intruders killed (61 of 82 vs 21 of 112; chi-

square, 1 df = 60.06, P < 0.001) and the proportion of

intruders physically detained (76 of 82 vs 61 of 112; chi-

square, 1 df = 5.44, = 0.02) were significantly greater

among ants introduced from field colonies. Data are

homogeneous with respect to the sequence of trials (i.e.,

day one vs day two) for both the proportion of intruders

killed (chi-square, 2 df = 3.83, P = 0.15) and the

proportion detained (chi-square, 2 df = 2.34, P = 0.31).

Finally, not significantly different proportions of

intruders were killed when intruders and residents were from

the same population (i.e., Florida or Mississippi) or from

different populations (Table 2.3).

Analysis of Cuticular Hydrocarbons

A total of 36 samples was analyzed by gas

chromatography. Hydrocarbon patterns gave no evidence of

hybridization events involving S. richteri (Vander Meer et

al. 1985), and all colonies tested were therefore included

in the analysis. The five major hydrocarbons (Fig 2.1)


constituted 72-77% of total cuticle hydrocarbon. Univariate

analysis revealed significant differences between laboratory

and field colonies in the means of normalized peak areas of

peaks A, B and E (Table 2.4). (Variances associated with

non-normalized peak means were not significantly different

in the two types of colonies.) When particular laboratory

vs field colony comparisons were made, individual areas of

peaks A, B and E were not always significantly different.

However, absence of a significant univariate peak area

difference between individual field and laboratory colonies

could not be statistically correlated with significantly

reduced resident aggression in the recognition bioassay.

For example, it was determined (Neuman-Keuls' test;

experiment-wise Type 1 error rate = 0.05) that 19 of 36

field vs laboratory bioassay combinations were between

colonies with not significantly different normalized, mean

peak areas for peak A. Yet, the proportion of intruders

killed (36 of 45) was not significantly different from the

proportion killed when intruder and resident colonies

exhibited significantly different peak A areas (25 of 37

killed; chi-square, 1 df = 1.65, P = 0.20). Similarly, not

significantly different proportions of intruders were

detained when colony pairs with significant peak B

differences were compared with those exhibiting no

significant differences (34 of 37 vs 43 of 45 detained; chi-

square, 1 df = 0.43, P = 0.49).


Both parametric discriminantt function) and

nonparametric (nearest neighbor) analysis generated

classifications in which colony "blends" of the five major

hydrocarbons were maximally correlated with colony class

(i.e., field or lab origin). In the former analysis, a

classification criterion based upon the pooled covariance

matrix resulted in nine misclassifiedd" field colony

samples, i.e., samples from three colonies that, based upon

cuticular hydrocarbon pattern, were more appropriately

classified as originating from laboratory-reared colonies.

The nearest neighbor discriminant analysis produced

identical results using both Euclidean and Mahalanobis

distances. In summary, the classification rules developed

from the five chemical descriptors (peaks A-E) accurately

classified only 50% of the samples. More importantly, of

the 21 field intruders not killed by residents (Table 2.1),

only nine originated from these three misclassified ("lab-

like") colonies (chi-square, 1 df = 0.43, P = 0.51).


Workers in monogyne S. invicta field colonies

vigorously attack intruders from neighboring, conspecific

colonies. This study assumes that the magnitude of

aggression is positively correlated with the extent of

perceived differences in recognition labels between two

colonies. Partial justification for this assumption is

provided by the fact that S. invicta workers are more


aggressive toward members of other Solensopsis species than

they are toward conspecifics (chapter VI; Obin, unpublished

data for S. geminata). Aggressivity between conspecific

colonies may also be affected by ecological and motivational

factors in addition to differences in recognition cues

(Wallis 1962; Wilson 1971; Holldobler 1976; Davies & Houston

1984; Jaffe & Puche 1984), and olfactory discrimination

itself may vary with time of day (Hangartner et al. 1970).

Recognizing the potential problems presented by variation in

aggression thresholds and fighting capabilities of workers,

I restricted the choice of "intruders" to the "media"

physical subcaste (Wilson 1978).

Effect of Environment on Nestmate Recognition Cues

Fire ant workers reared under controlled, laboratory

conditions respond less aggressively to nonnestmate

conspecifics reared under the same conditions. This

phenomenon has been observed in other laboratories as well

(Les Greenberg, pers. comm., 1983). That this effect is not

due to reduced aggressiveness overall is indicated by the

almost total recognition capability observed when intruders

were selected from field colonies (76 of 83 intruders

detained). Observed behavioral differences between lab-

reared and field intruders do not account for these data,

although the bioassay is admittedly insensitive to more

subtle, social interactions such as alarm and recruitment

(see chapter III). However, tests involving introductions

of non-nestmates to the mound exterior of field colonies


(chapter III) indicate that the recognition bioassay used in

these experiments is ethologically realistic. It can

therefore be concluded that the nestmate recognition cues of

laboratory- reared S. invicta colonies were less distinctive

than those of field colonies. These recognition factors or

other colony-specific cues may be transferred to the

substrate. This is suggested by the significantly greater

pre-contact escape response of field intruders, although an

alternative explanation (J. Sivinski, pers. comm., 1984) may

be that lab-reared ants have learned that escape is


The effect of laboratory rearing on detection of non-

nestmates supports the hypothesis that colony odor in S.

invicta is subject to environmental modification.

Environmental factors known to affect nestmate recognition

include diet (Kalmus and Ribbands 1952; Ribbands et al.

1952; Lange 1960), ambient odors (Renner 1960; Free 1961,

cited in Wilson 1971) and nest material (Lange 1960; Gamboa

et al. 1986a). With respect to the latter, Hubbard (1974)

has suggested that nest soil may contribute to colony odor

in S. invicta. The lack of nest soil in lab-reared colonies

used in this study removed potential odor cues that could

conceivably be transferred to the ants' cuticle. In

addition, absence of nest soil exposed workers to rearing

room odors that could have become incorporated into the

cuticle (see also Shellman-Reeve and Gamboa 1985). However,

unless "trapped" by cuticular lipids, olfactory cues


obtained in this manner would tend to be more volatile than

the contact cues addressed in this study. Additional

features of environment could conceivably contribute to

distinctive "bouquets" of workers from particular colonies.

For example, it is known (Toolson 1982, and references

therein) that the cuticular lipid composition of a number of

arthropod species examined is temperature- and humidity-

labile. Without knowledge of field colony response to lab-

reared intruders (see chapter III), one cannot determine to

what extent laboratory rearing homogenizes environmentally

determined recognition cues among laboratory colonies or

precludes their full expression.

It should be noted that endogenous, less

environmentally affected cuticular cues contribute somewhat

to colony odor in S. invicta. Although clearly reduced,

resident response to lab-reared intruders was significantly

greater than response to controls (Table 2.1), indicating

that detectable differences remained despite 18 months of

controlled rearing. That colony collection locale had no

measurable effect on the bioassay response of resident ants

to these cues (Table 2.3) suggests that variability in these

less labile contact recognition cues is minimal in North

American populations. Whether this is a consequence of

limited introduction events and genetic drift particular to

a. invicta (Tschinkel & Nierenberg 1983) or a more general

phenomenon in ants is unclear.


The effect of environment on nestmate recognition in S.

invicta strongly suggests that colony odor in this species

is a dynamic phenomenon. If so, some form of learning must

be involved in worker/worker discrimination (Fields 1903;

Soulie 1960; Kukuk et al. 1977; Breed 1981; Carlin &

Holldobler 1983; Morel 1983; Isingrini et al. 1985; Morel et

al. in press). There are both costs and benefits to a

flexible recognition system based on learning. While such a

system may increase colony susceptibility to slave-making

(i.e., dulotic) species, it may facilitate efficient colony

functioning when queens are multiply-inseminated or when, as

is sometimes the case for S. invicta in North America

(Glancey et al. 1973), colonies are polygynous (contain more

than one functional queen). Dulosis involving S. invicta is


Role of Cuticular Hydrocarbons

Cuticular hydrocarbons were of interest to this study

for several reasons. To begin with, the experiments of

Howard et al. (1980, 1982) argued strongly in support of

cuticular hydrocarbons as interspecific recognition cues in

the termite genus Reticulitermes. Secondly, fire ant

cuticle hydrocarbons are species-specific (Howard and

Blomquist 1982). In addition, Vander Meer and Wojcik (1982)

observed that myrmecophilous beetles inquilinous in the

nests of S. invicta, S. richteri and S. xyloni acquired the

species-specific hydrocarbon pattern of their host. They

suggested that the predacious beetles escaped detection by


chemically mimicking the host species (see also Howard et

al. 1980). Finally, hydro-carbons represent 70-75% of all

S. invicta cuticular lipids (Lok et al. 1975).

However, although differences between the cuticular

hydrocarbon patterns of laboratory and field colonies were

observed, these differences were inadequate predictors of

resident ant response to nestmates in the recognition

bioassay presented here. Rather, intruders from

misclassifiedd" colonies (i.e., those with hydrocarbon

patterns statistically not unlike residents') elicited as

much aggression as did intruders from correctly classified

field colonies. The same conclusion resulted when

individual hydrocarbon peaks were considered separately.

Possible explanations for these results include the

following: (1) Cuticular hydrocarbons are in fact not

nestmate recognition cues in S. invicta. (2) Cuticular

hydrocarbons other than or in addition to those addressed in

this study provide intraspecific nestmate recognition cues.

(3) Cuticular hydrocarbons do provide nestmate recognition

cues, but environmentally influenced or acquired surface

compounds (perhaps in concert with cuticular hydrocarbons)

permit discrimination when hydrocarbons alone do not.

Hypotheses 2 and 3 are not mutually exclusive, and each

could explain the high levels of aggression directed at

intruders from misclassifiedd" field colonies. It is

possible that the major cuticular hydrocarbons represent a

part of the more genetically controlled recognition cue


profile. If so, it is not surprising that hydrocarbon

pattern was such a poor predictor of environmentally based

behavioral differences in this study.


Both innate (heritable) and environmentally based

chemical cues affect nestmate recognition in Solenopsis

invicta. This conclusion supports findings for other

Hymenoptera (Wilson 1971; Jutsum et al. 1979; Holldobler &

Michener 1980; Breed 1983; Gamboa et al. 1986a,b; Holldobler

& Carlin, in press; Fletcher & Michener 1987). It should be

possible with the bioassay presented here to elucidate the

environmental factors affecting recognition chemistry in

this species. In addition, it has been demonstrated that

the major S. invicta cuticular hydrocarbons cannot account

for the dramatically different response of laboratory-reared

colonies to field-collected and laboratory-reared workers,

respectively. However, it cannot be concluded at this time

that these compounds do not constitute a portion of the

recognition profile mediating intraspecific interactions.








Fig 2.1. Chromatographic pattern of S. invicta worker
cuticular extracts (see Materials and Methods). Structure
of the five major hydrocarbons (A-E) are included (after
Nelson et al. 1980).


Table 2.1. Five-unit behavioral scale used in preliminary
nestmate recognition experiments.






No Follow


Resident holds intruder with forelegs
and mandibles, curls abdomen and
attempts vigorous stinging during
first minute of interaction.

Resident holds intruder (usually by
the petiole), but does not attempt
stinging; intruder is not released,
but eventually dismembered by other

Resident releases intruder after
initial holding (as above); residents
may repeatedly hold, release and
follow intruder.

Resident antennates (> 5s), follows,
but does not hold intruder.

Resident antennates intruder, but does
not follow.


Table 2.2 S. invicta nestmate recognition bioassay summary.






No Follow



0 (0.0) 1 (0.9) 21 (25.6)

0 (0.0) 20 (17.9) 40 (48.8)

5 (10.4) 40 (35.7) 15 (18.3)

18 (37.5) 28 (25.0) 6 (1.3)

25 (52.1) 23 (20.5) 0 (0.0)

112 (100)

82 (100)

Note: Scores are frequencies of resident ant responses to
(1) nestmates (control), (2) non-nestmates from other lab-
reared colonies (lab/lab) and (3) non-nestmates from field
colonies (field/lab). Percent total response is included in


48 (100)


Table 2.3. Resident response (No. intruders killed and No.
intruders released) when intruder and resident colonies are
from the same populations (i.e., Alachua County, Florida, vs
Gulfport, Mississippi).

Resident and Intruder
From Same Locale From Different Locale

Ttrl*tvnvA -

1r 1 1 ,A TD^I ^.-.-.A

a.,~~~..V IA~ IS A 10 "1 A t. .CA S O



X2. ldf



Note: Chi-square values were generated by contingency table
analyses of resident response to (a) lab-reared intruders
and (b) intruders from field colonies. Chi-square (0.05) =

Vill 1 l Dal1aacA^


Table 2.4. Mean ( SEM), normalized peak areas of the five
major S. invicta cuticle hydrocarbons in laboratory (N=18)
and field (N=18) colonies.


0.209 (0.009)

0.268 (0.007)

0.170 (0.008)

0.194 (0.010)

0.164 (0.008)


0.165 (0.012)

0.301 (0.009)

0.183 (0.008)

0.166 (0.011)

0.203 (0.011)

T. 34 df






Note: Univariate test statistics and attained significance
levels are also presented.















Data presented in chapter II indicate that cues

mediating worker-worker recognition in S. invicta can be

genetically or environmentally determined. In theory

genetically correlated recognition cues or "discriminators"

can be produced by both workers and queens of social

Hymenoptera (Crozier & Dix 1979; Holldobler & Michener 1980;

Gadagkar 1985; Breed & Bennett 1987). To function as

nestmate recognition cues, queen discriminators must be

transferred to workers. These transferred discriminators

are thus analogous to the maternal odors used for sibling

recognition by certain vertebrates (Porter et al. 1981;

Waldman 1981). To date, transferable queen discriminators

have been demonstrated for small laboratory colonies of the

ants Camponotus spp. (Carlin & Holldobler 1983, 1986, 1987;

Carlin et al. 1987), Myrmica spp. (Brian 1986) and

Leptothorax lichtensteini (Provost 1986; but see Stuart

1985), and suggested for honey bees (Breed 1981, 1986).

Environmental sources of recognition cues are potentially

innumerable, with diet, nest substrate and ambient odors



demonstrated to modify nestmate recognition cues of social

insects (see Discussion, chapter II).

Here I continue to explore the sources) of cues

mediating worker-worker recognition in monogyne colonies of

the imported fire ant. Using an expanded bioassay that

includes nine (rather than five) categories of aggressive

behavior, I first measure the recognition response between

former nestmates reared in different environments (lab or

field). I subsequently evaluate the effect of diet alone on

nestmate recognition in small (< 200 workers), queenless

colonies and in larger (> 10,000 workers), queenright

colonies. I then investigate the separate effects of alien

queen and alien worker presence on recognition between

nestmates and nonnestmates in small, laboratory colonies.

Chapter II and evidence obtained for other social

insects (Gadagkar 1985; Breed & Bennett 1987) suggest that

S. invicta workers discriminate nestmates from nonnestmates

by matching phenotypic "recognition labels" expressed by

encountered individuals with a "template" (Alexander 1979)

of learned, nestmate-borne or self-borne cues. This

mechanism is commonly referred to as "phenotype matching"

(Holmes & Sherman 1982; see also Beecher 1982; Blaustein

1983). By performing reciprocal tests between treatment

groups in each experiment, I was able to assess the effects

of treatments on both worker recognition labels and

recognition templates.


Materials and Methods

Behavioral Assay and Data Analysis

The recognition bioassay quantitated the behavioral

response of workers to individually introduced "intruder"

conspecifics (see Materials and Methods, chapter II).

Introduced ants were intermediate-sized (media) workers of

the "reserve" temporal (age) subcaste (Mirenda & Vinson

1981), and were selected from the area of rearing tray

immediately surrounding the brood cell (lab colonies,

experiments 3.1 and 3.2) or from immediately beneath the

mound tumulus (field colonies, experiment 3.1; lab colonies,

experiment 3.3). Intruders were restricted to one age

class, because recent evidence (Sorensen & Fletcher 1985;

Vander Meer, unpublished) suggests that workers of different

temporal subcastes may have different olfactory

sensitivities. Ant behavior was observed with 10X

magnifying glasses. In addition, the observer (M. 0.) wore

a particle mask to minimize the agitation-inducing effects

of exhalation on the ants. Some tests required more than

one day to complete. To control for possible circadian

differences in olfactory response (Hangartner et al. 1970)

or motivation, colonies were tested at the same time of day

( 0.05 h) throughout the experiment.

Colony response to each intruder ant was scored as the

numerical rank (1-9) of the most aggressive of nine

behavioral responses (Table 3.1) elicited from 20 resident

ants during each introduction. These behaviors can be


grouped into three, distinct levels of increasing

aggression: Level I (behaviors 1-3), Level II (behaviors 4-

7) and Level III (behaviors 8-9). Introduced ants eliciting

behaviors 4-7 are challenged or attacked, but they continue

to move about the colony; intruders eliciting behaviors

eight or nine are always killed. The null hypothesis of

independence between treatments and the distributions of

Levels I, II and III behavior was tested by 3 x 2

contingency tables using the G statistic (Sokal & Rohlf

1981). When three hypotheses were tested with the same set

of data (e.g., data for "control" introductions in

experiment 3.2), the comparison-wise attained significance

level for rejection of the null hypothesis was set at 0.02.

Experiment 3.1. Split Colonies: Lab vs Field

This experiment tested reciprocal recognition between

workers removed from field colonies and maintained in the

laboratory and workers from the field colony of origin.

Worker ants (N = 400-600) and brood were collected from 15

individual monogyne field colonies and reestablished with

original nest soil in the laboratory. Ants were maintained

at 26-27 degrees Centigrade under variable, fluorescent

illumination in ceramic pans (5.5 x 17.5 x 29.0 cm) equipped

with a petri dish brood cell, and a cotton-stoppered water

tube. A honey-water and roach (Periplaneta americana) diet

was provided on the third day after collection and

subsequently every third day until the completion of the



The aggressive response of lab-maintained subcolonies

toward individual workers introduced from the field colony

of origin (N = 2) and from their own subcolony (N = 2) was

measured at 2 days and 27 days post-collection. Comparisons

were therefore based on 30 replicates per treatment.

"Intruder" ants were carefully introduced on the end of

forceps and removed after each trial (see chapter II for

details). The sequence of intruder types was alternated.

We also measured the aggressive response of each field

colony to former nestmates maintained in the laboratory and

to field colony nestmates. In this procedure, a Castone-

filled petri dish was baited with a roach and placed ca.

35cm from the nest mound. Following discovery and

recruitment to the arena by workers, alarm was generated by

placing the crushed head of a worker (source of alarm

pheromone; see Wilson, 1962) in the arena. After 1 min, an

"intruder" was introduced into the arena and behavioral data

were collected. The sequence of intruder types was

alternated among colonies, and a 5 min interval separated

introductions into any one colony. As the exact location of

three field colonies could not be determined on day 27, only

12 colonies were tested at that time (N = 24 introductions

per treatment group).

Experiment 3.2. Effect of Diet on Intercolony Recognition

These experiments evaluated whether nonnestmates

maintained on similar diets are less aggressive toward one

another than nonnestmates maintained on dissimilar diets.


Twenty-four monogyne field colonies were excavated. Queens

were placed in vials with 10-30 workers to insure queen

safety and queen tending during transport to the laboratory.

The remainder of each excavated colony (> 10,000 workers)

was reestablished in the laboratory with original nest soil

in large rearing trays (64.0 x 78.5 x 9.5 cm). Queens were

reintroduced at this time. Colonies were provided with

water tubes on the soil mound but were supplied no food for

two weeks. During this time, workers foraged on arthropods

in the nest soil. After two weeks, ceramic foraging arenas

were connected to each colony by a Tygon tube bridge.

Twelve colonies received (ad libitum) honey-water and

roaches (group I), while twelve received 5% sucrose solution

and moth (Anticarsia gemmatalis) pupae (group II). Nest

soil was watered throughout the experiment when necessary.

Aggression bioassays were conducted at two weeks (prior

to presentation of diet) and at five months post-collection.

In the pre-diet tests, colonies were tested against four

within-group intruders (two each from two colonies, N = 48),

four between-group intruders (two each from two colonies, N

= 48), and against two "control" intruders from the resident

colony being tested (N = 24). Intruders were introduced

onto the nest soil at least 10 cm from a nest opening. The

sequence of intruder types was alternated among colonies on

both days.

Based on brood production and the intensity of

foraging, I concluded that only 14 of the original 24


colonies retained healthy, productive queens after 5 months

in the laboratory. Consequently each of six colonies in

either diet group was tested against the other 11 for within

and between group effects. To test recognition within

diets, one intruder was introduced between each of the 30 (6

x 5) colony pairs in each of the two diet groups (N = 60

introductions). Between diet response was measured with two

(reciprocal) introductions between each of 30 (6 x 5) paired

colonies, one from each diet group (n = 60 introductions).

Tests were conducted on days 1 and 2 following the most

recent feeding.

Experiment 3.3. Split Colonies:
Queen and Nonnestmate Additions

This experiment determined whether the addition of an

alien queen or nonnestmate worker for 28 days could

detectably modify recognition labels and templates of

recipient subcolonies. Minor and media workers and brood

from monogyne field colonies (N = 14) were established

without queens in the laboratory for 10 weeks and then

divided into 4 subcolonies of equal numbers of workers

(range = 20 200 workers). One subcolony (Q+) was

requeened with a physogastric (i.e., highly fecund) field

queen from a different site. This was accomplished by

gently cooling the workers and introducing the field queen

into the brood cell (orphaned S. invicta colonies readily

accept replacement queens by this method). A major worker

control from the field queen's colony of origin was added to

another subcolony (W+) by repeating cooling and


introduction. Two subcolonies (AQ- and BQ-) were cooled,

but received neither queen nor worker nonnestmates.

Subcolonies were maintained on a honey-water and roach diet,

and were covered with tight-sealing plastic tops during the

course of the experiment. Pharate pupae (distinguishable by

pigmentation) were removed from requeened subcolonies

starting at day 20, thereby insuring that no subcolonies

contained adult offspring of added queens. Aggression

bioassays were conducted (as above) between days 25-28 to

assess the effect of queen presence on worker-worker

recognition. The following series of reciprocal

introductions were performed within colonies: (1) control

introductions (AQ- vs AQ- and BQ- vs BQ-), (2) queenless vs

queenless (AQ- vs BQ-), (3) queenless vs queen-added (AQ-

and BQ- vs Q+), and (4) queenless vs worker-added (AQ- and

BQ- vs W+). Scores for each series of introductions

represent the sum of four introductions per colony (i.e.,

reciprocal sets of 2 introductions). Thus, 56 introductions

were conducted for each series of comparisons. Tests of AQ-

vs BQ- and AQ- and BQ- vs Q+ were also conducted between

colonies (N = 2 nonnestmates introduced per colony). One

series of introductions was conducted each day. In all

tests, introductions into any one subcolony were spaced five

minutes apart, and at least 90 min elapsed between the time

a colony received intruders and the time it provided




Experiment 3.1. Split Colonies: Lab vs Field

After two days in the laboratory, subcolonies exhibited

no greater aggression toward field nestmates than they did

toward lab subcolony nestmates (Fig 3.1a). No intruders in

either series of tests were killed. However, after 27 days

in the laboratory, subcolonies were significantly more

aggressive toward field nestmates than toward lab subcolony

nestmates (Fig 3.1b). In addition, although the magnitudes

of aggression directed at lab-maintained nestmates on day 2

and day 27 were not significantly different (G, 2 df = 0.02,

P = 0.99; Fig. 3.1a,b), lab colonies were more aggressive

toward field nestmates on day 27 than on day 2 (G, 2 df =

22.93, P < 0.001; Fig 3.1a,b). Twenty-one percent (5/24) of

field nestmate intruders were killed (i.e., elicited Level

III response).

Similar results were generated by the reciprocal

presentation of lab-maintained workers to former nestmates

in the field. On day 27 field colonies directed

significantly more aggression at lab-maintained workers than

at field nestmates (Fig. 3.1d). While no field nestmate

elicited Level III response, 37.5% (9/24) of the lab-

maintained nestmates did so. Aggression directed at lab-

maintained nestmates on day 27 was also greater than that

directed at lab-maintained nestmates on day 2 (G, 2 df =

19.84, P < 0.001, Fig. 3.1c,d).


Experiment 3.2. Effect of Diet

After two weeks in the laboratory, workers introduced

into their own colony (controls) elicited only Level I

response (N = 24 trials). Before presentation of the

experimental diet, introductions between and within test

groups did not produce significantly different distributions

of behavioral response (Fig. 3.2a). However after five

months of diet treatments, within-group introductions

elicited significantly less aggression than between-group

introductions (Fig. 3.2b). In addition, within-group

aggression was significantly reduced with respect to pre-

feeding levels (G, 2 df = 11.76, P = 0.003; Fig. 3.2a,b).

In contrast aggression among ants reared on different diets

was not significantly different from pre-diet levels (G, 2

df = 2.58, P = 0.28; Fig. 3.2a,b).

Experiment 3.3 Split Colonies:
Queen and Nonnestmate Additions

All queens added to recipient subcolonies were

accepted, and all produced viable brood during the tests.

No major worker controls were found dead in any subcolony

for up to 48 hours post-introduction, and we assumed that

all remained alive and integrated into their new colonies

during the course of this experiment. Aggression scores for

between-treatment introductions were not significantly

different, with almost all introductions (162/168) between

nestmate subcolonies eliciting only Level I

("investigative") behaviors (Fig. 3.3a,b). No significant

difference in recognition response was detected when


reciprocal introductions between queenless and queen-added

subcolonies were compared (G, 2 df = 0.99, P =0.61), or when

reciprocal introductions between queenless and worker-added

subcolonies were compared (G, 2 df = 0.00, P = 0.99).

Because aggression was so muted, we compared the

frequency of the lowest behavioral response (rank #1) among

the four series of introductions (Table 3.2). The frequency

of rank #1 response was not significantly different when

queenless vs queenless introductions (32/56) were compared

with queenless vs queen-added introductions (27/56) (G, 1 df

= 0.88, P = 0.35). Similarly, no significant differences in

the frequency of rank #1 response were detected when control

introductions of queenless workers back into their own

subcolony were compared with any series of between-treatment

introductions, although data for alien queen treatments are

suggestive (Table 3.2).

As with tests between nestmates, presence of an alien

queen did not significantly affect introductions between

nonnestmate subcolonies (Fig. 3.3c). However, nonnestmates

clearly elicited greater aggression than nestmates when

introduced between queenless subcolonies (G, 2 df = 21.53, P

< 0.001; Fig. 3.3a,c) and between queenless and queenright

subcolonies (G, 2 df = 10.34, P =0.006; Fig. 3.3a,c).



Environmentally Correlated Cues

Laboratory and field bioassays indicate that colony

level recognition in S. invicta is mediated in large part by

environmentally correlated cues. In experiment 3.1,

aggression between laboratory-reared and field-reared

nestmates was significantly increased within 27 days of

laboratory rearing. As in honeybees (Kalmus & Ribbands

1952), other ants (Wallis 1962; Jutsum et al. 1979), and as

suggested for vertebrates (Galef 1981; Hepper 1986), group

recognition cues used by S. invicta workers can be derived

(either directly or indirectly) from diet. In the present

study, diet significantly modified worker labels in large,

productive queenright colonies in original nest soil

(experiment 3.2). Aggression between nonnestmates was

reduced by maintaining colonies on the same diet, and

aggression between colonies maintained on different diets

was not significantly different from that observed when

colonies were first collected.

The strong reaction of laboratory colonies to field

colony nestmates on different diets (experiment 3.2)

additionally argues against artifactually "impoverished"

worker labels or templates (see Discussion, chapter II) or

reduced motivation among laboratory colonies (N. Carlin,

pers. comm.) as underlying causes of the reduced aggression

observed among large, queenright laboratory colonies of S.

invicta. Rather, low levels of aggression between


laboratory colonies are most likely due to the effects of

homogeneous laboratory conditions on recognition labels and


We are particularly intrigued by several implications

of environmentally derived recognition cues in S. invicta.

One is the possible relationship between such cues and

polygyny. Polygyne S. invicta colonies are known from North

America (Glancey et al. 1973; Fletcher et al. 1980). These

colonies are typically polydomous (i.e., include a number of

mounds), and frequently extend over several hundred square

meters. As in other species examined (Breed & Bennett

1987), polygyne S. invicta workers appear to be less

aggressive than monogyne workers (Mirenda & Vinson 1982).

Although increased genetic (and thus discriminator)

variability in polygyne societies may explain the reduced

recognition performance of polygyne workers (H611dobler &

Michener 1980; Carlin & Holldobler 1983; Breed & Bennett

1987), we suggest that inter-mound variation in

environmentally correlated recognition cues could function

similarly to make workers in polygyne colonies more


Moreover, food exchange (trophallaxis) between

nonnestmate workers could, by fostering intercolonial worker

adoption, actually promote the formation of polydomous,

polygyne colonies (Bhatkar 1979). "Appeasement"

trophallaxis involving agonistic, nonnestmate S. invicta

workers has been observed by several investigators (Bhatkar


1979; Obin, unpublished). Intruding workers appear to

behave a "satellites" (Davies & Houston 1984), offering

payment (food) in exchange for access to a foreign territory

or nest ("appeasement"). Appeasement trophallaxis and the

consequent sharing of recognition cues may, under a model of

"habituated label acceptance" (Getz 1982) or "cue

similarity" (Gamboa et al., 1986a,b; Getz & Chapman 1987;

see chapter V), promote the functional integration of

nonnestmate worker forces. A similar mechanism may also

facilitate colony desertion and intercolonial adoption

observed in the ant genus Myrmecosystus (Bartz & H6lldobler

1982) and strongly suggested for incipient S. invicta

colonies (Tschinkel 1986b).

Second, environmentally derived cues may be significant

in the evolution of recognition systems (Gamboa et al.

1986b). Discrimination of nestmates from nonnestmates could

evolve from the ability to use substrate odors as nest

recognition cues. It will be of special interest,

therefore, to determine if S. invicta workers can use the

same environmentally derived chemical cues for both nest

recognition and nestmate recognition (cf Hangartner et al.

1970; Hubbard 1974).

A changing environment and the importance of

environmentally correlated recognition cues in S. invicta

additionally suggests that worker labels are dynamic, and

that updated recognition templates must be learned by

workers throughout their lifetime (see also Wallis 1963;


Vander Meer, in press). The cuticular hydrocarbon blend, a

model for potential genetic components of the S. invicta

colony label, also changes through time (Vander Meer et al.,

unpublished). If labels can change more rapidly than

templates, the possibility exists for false-negative

recognition of nestmates as intruders. Such recognition

errors may be evident in the high levels of aggression

directed at S. invicta foragers returning to their colony

after contacting novel food (Obin, unpublished; see also

Wallis 1963).

Genotype-Correlated Cues (Discriminators)

There were good reasons to suspect that fire ant queens

transferred heritable, queen-specific odorants to workers.

S. invicta queens are much larger than workers, and they

release several different queen pheromones within the colony

(reviewed in Fletcher 1986). Moreover, the individual

specificity of one of these queen pheromones had been

previously suggested (Jouvenaz et al. 1974). If fire ant

queen discriminators are transferred to workers and

constitute some portion of the colony label, reciprocal

introductions among recipient (queen-added) and nonrecipient

queenlesss) colonies should elicit more pronounced

aggression than introductions among nonrecipient colonies.

If labels alone were affected by queen addition, we might

expect an asymmetry in the direction of aggression directed

at former nestmates, with queenless colonies more aggressive

toward queenright intruders than vice versa. Alternatively,


if queen presence modified worker templates, but queen

discriminators were not transferred to workers, queenright

subcolonies might be more aggressive toward queenless

intruders than vice versa. The same rationale formed the

basis of tests between queenless and worker-added

subcolonies. In addition, comparison of queen-added and

worker-added trials might indicate whether or not effects

observed with queen-added groups were due specifically to

the presence of queens.

Results of experiment 3.3 indicate that addition of an

alien queen or an alien worker of comparable size to small,

queenless laboratory subcolonies did not significantly

increase aggression between recipient and nonrecipient

colonies. These results argue against the possible effects

of "queen discriminators" (Carlin & H611dobler 1983, 1986,

1987; Carlin et al. 1987) in experiment 1, and suggest that

under the experimental protocol employed, adult S. invicta

workers do not incorporate alien queen discriminators into

the recognition template mediating colony defense (see also

Mintzer 1982; Stuart 1985). Subtle differences in the

frequency of "investigative" behaviors (Table 3.2) are

probably inconsequential for nest defense in monogyne

colonies, but could conceivably function in close-kin (i.e.,

matriline) recognition within polygyne colonies of this

species (cf Carlin et al. 1987).

It must be noted that our data do not exclude the

possibility that queen discriminators are learned during a


critical period early in worker life (Brian 1986) or that

conditioning by adult workers to compounds of newly-

introduced queens requires more than the 28 days that

experiment 3.2 lasted. If either is found to be the case,

it would suggest that queen discriminators and

environmentally derived cues are incorporated into worker

templates by different neural processing. In addition, S.

invicta is an introduced species in the Southeastern United

States, with genetic variability presumably reduced as a

result (Tschinkel & Nierenberg 1983). It is possible that

additive genetic variability for queen discriminator

production is insufficient to mediate colony-level

discrimination in S. invicta (see also Tschinkel &

Nierenberg 1983). Production of worker discriminators (see

below) could be similarly affected. Clearly, the relative

roles of environmental and genetic recognition cues must be

assessed in endemic, South American S. invicta populations.

Although similar environmental cues can potentially

eliminate recognition of nonnestmates in the leaf-cutting

ant Acromyrmex octospinosus (Jutsum et al. 1979) and the

social wasp Polistes fuscatus (Gamboa et al. 1986a),

aggression persists among nonnestmate S. invicta reared

under similar laboratory conditions (chapter II; present

study). This fact suggests that genotypically correlated

cues constitute some portion of worker recognition labels

and templates. In this study, these discriminators were

most probably worker (rather than queen) derived. With


respect to label acquisition, data for worker-added trials

(experiment 3.3) do not appear to support a model in which

an individual's discriminators blend with those acquired

from nestmates, thereby producing a unique colony blend or

"gestalt" (Crozier & Dix 1979; Stuart 1986). However,

strong evidence for mixing of cuticular compounds among

colony members has been presented for a. invicta (Vander

Meer & Wojcik 1982), and it is likely that the transferred

discriminators of a single, alien conspecific were

insufficient to detectably modify the "gestalt" label of a

recipient subcolony.

In summary, data for S. invicta support the current

view (H6lldobler & Carlin, in press; Breed and Bennett, in

press) that ants, not unlike other social Hymenoptera

studied (sweat bees, honey bees, wasps), are capable of

using both genetic and environmentally derived nestmate

recognition cues. These cues form a functional "cue

hierarchy" (Carlin & H6lldobler 1986) for a particular

species. The cue hierarchy suggested for S. invicta by this

study is environmental > worker discriminators > queen

discriminators (if in fact the latter exist). These results

conform to our emerging expectation that "unifying cues"

derived either from the environment or from the queen and

shared by all workers should be important in species with

large colonies (i.e., S. invicta), and particularly in the

context of colony-level activities such as nest defense

(Holldobler & Carlin, in press; Breed & Bennett 1987). But


why do S. invicta workers rely on unifying cues derived from

the environment rather than from the queen? Although

Holldobler and Michener (1980) originally suggested the

queen as the logical source for the colony odor in monogyne

societies, more recent arguments (Hepper 1986) suggest that

predominantly environmental factors are reliable

determinants of group and kin recognition cues in species

where group membership and kinship are highly correlated. A

high correlation of this type is expected in monogyne

colonies of S. invicta, since females are monoandrous (Ross

& Fletcher 1985). The negligible contribution of queen

discriminators to the fire ant colony label might also

reflect the workers' obligate sterility (Holldobler &

Carlin, in press). More specifically, Solenopsis queens may

represent less of a chemical (olfactory) presence within the

colony than queens of species exhibiting strong queen

suppression of worker oviposition (e.g., Camponotus).

Clearly, comparative studies of recognition-cue hierarchies

remain challenging and rewarding tasks for the future.



(0) DAY 2


G, 2df=0.07, P=0.97



II n Ur


(b) DAY 27
60 <0.001 1
60 .- .

I ll I


(d) DAY 27
G, 2df = 25.14, P<0.001
60- 13
M -12

JI n


0 lab nestmates

0 field nestmates

Figure 3.1. Experiment 3.1. Effect of laboratory vs field
rearing on nestmate recognition between split colonies of S.
invicta. Fifteen queenless subcolonies were established in
the laboratory. The parental field colonies were marked.
At two days and 27 days post-collection, two reciprocal
introductions of "intruder" nestmates were presented between
each field colony and its laboratory subcolony
("residents"). Only 12 of 15 field colonies tested on Day 2
were located for tests on day 27. (a) Day 2 responses of
laboratory subcolonies to laboratory subcolony nestmates (N
= 30) and to field colony nestmates (N = 30), and (b) Day 27
responses of laboratory subcolonies to laboratory (N = 24)
and field (N = 24) nestmates. (c) Day 2 responses of field
colonies to laboratory (N = 30) and field nestmates (N =
30), and (d) Day 27 responses of field colonies to
laboratory (N = 24) and field (N = 24) nestmates.
Comparison of (a) with (b) and (c) with (d) demonstrates
increased aggression between nestmates maintained under
different environmental conditions.

(C) DAY 2






G,2df =0.17, P=0.92



10 9







G, 2df= 24.41, P<0.O01

I ~EA~


- within groups

0 between groups

Figure 3.2. Experiment 3.2. Effect of diet on nestmate
recognition in S. invicta. Twenty-four monogyne, queenright
field colonies were reestablished in the laboratory in
original nest soil and randomly assigned to one of two
groups. (a) Results of reciprocal recognition tests
conducted after two weeks, but before presentation of diet
(4 introductions per colony, N = 96). (b) Results of
recognition tests conducted after 5 months among 6 colonies
in each of the two diet groups (N = 120, see Materials and
Methods). Aggression within groups receiving the same diet
is significantly reduced (P < 0.001, G-test), whereas
aggression between groups receiving different diets is not
significantly different from levels obtained in pre-feeding
trials (P = 0.28, G-test).










55 53


O queenless x
1 queenless x

G, 2df= 0.99
P =0.61


[ queenless x
0 queenless x

P >0.99


O queenless x queenless
80 M queenless x queen-added
S 19





Figure 3.3. Experiment 3.3. Effect of the presence (28
days) of an alien queen or alien worker on recognition
between former nestmates established in small, laboratory
subcolonies. Results of reciprocal aggression tests between
(a) queenless and queen-added nestmate subcolonies (N = 112
introductions), and (b) queenless and worker-added nestmate
subcolonies (N = 112 introductions). (c) Aggression between
nonnestmate subcolonies of queenless and queen-added workers
(N = 56 introductions), indicating no significant effect of
queen presence, but significantly greater aggression between
nonnestmates than between nestmates (Fig. 3.3a) in queenless
subcolonies (P = 0.006, G-test).

I 11 IE


I I I JJJf [




Table 3.1. Nine-behavior assay used in nestmate recognition
experiments in chapters III-VI.

Rank Behaviors

9 Immediate lunge, grab and stinging

8 Intruder surrounded and "held" in mandibles by
petiole and appendages; appendages pulled/bitten
off; eventual stinging.

7 Intruder "held" (as in #8), but released;
abdomen-curling (stinging posture) by residents,
but no stinging; biting.

6 Intruder "held," but released; biting; no

5 Alarm (running, abdomen elevation and vibration)
and recruitment.

4 Mandible gaping; rapid antennation; "sidling"
(maintaining a lateral orientation to and slowly
circling intruder).

3 Rapid antennation of intruder, antennae extended
for > 2 s.

2 Intruder antennated for less than 2 s; if mobile,
intruder is followed slowly for several cm; if
intruder is stationary, resident stops.

1 Intruder antennated (as in #2), but if mobile, is
not followed; if intruder is stationary, resident
ant does not stop.

Note: Bioassays tested post-contact colony ("resident")
response to individual, introduced workers ("intruders").
These behaviors can be grouped into three levels of
increasing aggression (see Materials and Methods):
behaviors 1-3 (Level I), behaviors 4-7 (Level II), and
behaviors 8-9 (Level III).


Table 3.2. Experiment 3.3. Frequency of the least
aggressive behavioral response (rank #1) elicited by four
series of nestmate introductions.

Response Frequency (%)


Controls (28)

Q- x Q- (56)

Q- x Q+ (56)

Q- x W+ (56)

Rank #1

19 (67.9)

32 (57.1)

27 (48.2)

33 (58.9)

Ranks > #1

9 (32.1)

24 (42.9)

29 (51.8)

23 (41.1)

G. ldf ALS







Note: Greater than 95% of all responses were rank 1-3
(Level I) behaviors. "G" statistics and attained
significance levels (ALS) are computed for comparisons of
each between-treatment series with the control series (i.e.,
workers introduced into their own subcolony). Queenless
workers (Q-); workers requeened with alien, field queen
(Q+); workers housed with alien, field worker (W+).



It has been suggested that the queen should contribute

most to nestmate recognition cues (colony odor) in monogyne,

eusocial Hymenoptera (Holldobler & Michener 1980),

especially in those species with pronounced queen-worker

dimorphism and queen suppression of worker reproduction

(Hol6dobler & Carlin, in press). This prediction was not

supported by experiments (chapter III) with the imported

fire ant Solenopsis invicta, a species evidencing pronounced

queen-worker dimorphism and queen suppression of gyne

reproduction (reviewed in Fletcher & Ross 1985). Rather,

environmentally derived odors and heritable worker

"discriminators" were the only nestmate recognition cues

experimentally verified. This result was especially

surprising, since fire ant queens release potent pheromones

within the colony, and worker cuticles were likely to

contain these compounds (Fletcher & Blum 1983a,b; reviews in

Vander Meer 1983; Fletcher 1986). Moreover, there is some

evidence that fire ant workers recognize pheromone blends of

individual queens (Jouvenaz et al. 1974; Fletcher & Ross



1985). If so, transferred queen pheromones could readily

function as nest-specific worker signatures.

The experimental analysis of queen inputs to worker

labels presented in chapter III used small, queenless worker

groups (n = 20-200) into which alien field queens were

placed for several weeks. The rationale behind this

approach was that the artificially small worker force would

enhance the probability of queen odors being transferred to

workers. Unfortunately, queens maintained in such small

colonies exhibit pronounced decrements in reproduction

(explained in Tschinkel 1986), and recent theoretical and

empirical arguments (Carlin & Holldobler 1987) suggest that

a queen's ability to "label" workers and dominate the colony

odor may be positively associated with her fertility.

Positive associations of queen pheromone release and queen

reproductive capability are recognized for a number of

insects, including _. invicta (Fletcher & Ross 1985). With

this issue in mind, I undertook additional experiments on

the role of queen derived cues in fire ant nestmate

recognition. These experiments, involving robust and fecund

queens, are reported below.

Materials and Methods

Ant Rearing

Monogyne S. invicta colonies were established in the

laboratory from newly-mated queens collected near

Gainesville, Florida (see chapter II or Banks et al. 1981


for details of collecting and initial queen maintenance).

When tested at maturity, these "parent" colonies each

contained 90-120,000 workers and all brood stages, and were

housed in trays as described in chapter II. Two queenless

subcolonies containing 30 worker ants of the forager and

reserve temporal subcastes (Mirenda & Vinson 1981) and 500-

600 larvae were established from each mature colony. These

queenless sister groups were maintained in individual metal

pans (5.5 x 17.5 x 29.0 cm) equipped with a petri dish nest

(with Castone floor) and a cotton-stoppered water tube. All

ants were fed (ad libitum) honey-water, thawed roach

(Periplaneta americana), fly pupae (Musca domestic) and

moth pupae (Antecarsia gemmatalis). Rearing room

temperatures ranged from 21-24 degrees Centigrade, and the

light-dark cycle was variable.

Nestmate Recognition Bioassay and Data Analysis

Behavioral tests were conducted 26 weeks after

queenless subcolonies were established. Subcolonies were

broodless at this time, and contained from 200-375 workers.

Colony response to each introduced "intruder" ant was scored

according to the nine-point behavioral scale presented in

chapter III (Table 3.1). Resident colony responses were

grouped into Level I, II and III behaviors (chapter III).

The null hypothesis of independence between treatments and

the distributions of Levels I, II and III behavior was

tested by 3x2 contingency tables using the G statistic

(Sokal and Rohlf 1981). In addition, means ( sd) of


individual trial scores were compared between treatments by

t-tests (unpaired data) (Steel and Torrie 1960). When tests

required more than one day to complete, each colony was

tested at the same time of day ( 0.5 h) throughout the


Queen Weight and Fertility

Twenty-four hours prior to bioassays, parent colony

queens were observed for qualitative evidence of

reproductive vigor. Evidence included obvious physogastry

(swollen abdomen), copious brood production, non-dealated

gynes (Fletcher & Blum 1983a,b; Willer & Fletcher 1986), and

pronounced, pheromonal attraction of queen-tending workers.

As part of the last experiment (Experiment 4.2C below)

parent colony queens were removed from their nests and

individually weighed ( 0.005 mg) on a Mettler balance.

Each queen together with 20 daughters reared queenless for

26 weeks was housed in a plastic cup (height = 3.0 cm,

diameter at base = 1.8 cm) containing a damp, Castone floor.

After 24 hours at 22 degrees Centigrade, queens and workers

were removed, and the number of eggs oviposited by queens

counted with the aid of a dissecting microscope. Hourly

oviposition rates were determined for each queen by dividing

the number of eggs counted by the duration of the test.

Experiment 4.1. Effect of Long-Term Exposure to Physogastric
Queens in Large Laboratory Colonies

In Experiment 4.1A, nine parent colonies (residents)

were each tested against five queenless (Q-), intruder kin


introduced from queenless subcolonies, and against six

queenright (Q+) kin removed from the parent colony and

immediately reintroduced to another position in the foraging

tray (nl = 45; n2 = 54). In Experiment 4.2B, eight

queenless (Q-) and eight queenright (Q+) non-kin workers

(one from each colony) were introduced into each of the nine

parent colonies (nl = n2 = 72). In both series of tests,

the sequence of intruder types was alternated for each

parent colony, and tests were replicated an equal number of

times in all colonies before any colony was tested again.

All intruders were selected from the outer lid or sides of

nest cells, and were therefore assumed to be members of the

"reserve" temporal subcaste (Sorensen et al. 1981; Mirenda

and Vinson 1981; see rationale in Materials and Methods,

chapter III and Discussion, present study).

Experiment 4.2. Effect of Direct Exposure to Queens

In Experiment 4.2A, 40 workers (20 from the foraging

arena and 20 from inside the nest cell) were removed from

their queenless subcolony (n = 5) and divided equally

between two, metal dissecting pans, each containing a foil-

covered water tube (nest) and a dish of honey-water. After

30 minutes, a freshly-dug, physogastric field queen (> 20.0

mg) was placed in one pan of each of the paired sub-

colonies. Twenty-four hours later, alternate introductions

of workers tending alien queens (Q+T), those moving about

the foraging area of pans containing alien queens (Q+F) and

non-exposed, control workers (Q-) were made to queenless kin


in the original subcolony (nl = n2 = n3 = 20).

In Experiment 4.2B, workers from six additional

queenless subcolonies were exposed to alien queens, and

those tending queens (Q+) in each pan were introduced into

non-kin queenless groups. Two groups of alien queens were

used: "high fertility" queens (weight = 21.8-27.4 mg, n =

3) and "low fertility" queens (weight = 13.7-16.7 mg, n =

3). Five workers from each pan containing queenless workers

but no queen were also tested (nl = n2 = 30). To assess

whether queen fertility was positively associated with

aggression directed at exposed workers, the differences

between scores of matched Q+ and Q- introductions were

calculated. These differences were compared between "high"

and "low" fertility groups (Wilcoxon's two-sample test).

In Experiment 4.2C, 20 workers from queenless

subcolonies were housed with their mother queen (see Queen

Weight and Fertility, above). Three of these and three

queenless kin treated identically with the exception of

queen exposure were subsequently introduced to their

original parent colony (nl = n2 = 27). Mother queens were

not present in parent colonies for 24 hours prior to and

during these tests.


Queen Weights and Fertility

Mean weight ( sd) of queens in the nine parent

colonies was 21.56 mg ( 1.26; range = 19.72-23.74 mg; n =


9). The mean ( sd) hourly oviposition rate obtained for

these queens was 42.4 eggs ( 3.3 eggs; range = 36.75-47.39;

n = 9).

Experiment 4.1

Parent colony response to intruders from queenless

subcolonies was not significantly different from parent

colony response to queenright nestmates (Table 4.1A). As

expected, parent colonies responded more aggressively to

non-kin intruders, but these responses were not

significantly affected by whether or not intruders were

reared in queenright or queenless groups (Table 4.1B).

Experiment 4.2

In all tests, queens elicited the stereotyped worker

attraction, "rescue" and tending behaviors (Glancey et al.

1983, 1984) indicative of attractant/recognition pheromone

release by queens. Each queen (with her coterie of grooming

and licking attendant workers) was "settled" and ovipositing

within 15 minutes of queen introduction. However, not all

workers tended queens. Some moved about the periphery of

the metal pan, often huddling together in groups of 2-5 for

extended periods. Results of Experiment 4.2A indicate that

those intruders selected while tending alien queens elicited

significantly more aggression from kin than non-exposed

workers (Table 4.2A). However, workers selected from the

periphery of the metal pan elicited no more aggression than

non-exposed workers (Table 4.2A). Results of Experiment

4.2B (Table 4.2B) indicate that workers tending alien queens


for only 15 minutes elicited significantly more aggression

from non-kin than non-exposed workers. The magnitude of

this aggression was not significantly associated with alien

queen fertility (weight) (T' = 213, nl = n2 = 15; P > 0.20).

Finally, workers housed for 24 hours with their mother queen

in Experiment 4.2C elicited significantly more aggression

from kin than non-exposed workers (Table 4.2C). Resident

workers directed less intense aggression at kin exposed to

the mother queen than they did at kin exposed to alien

queens (G, 2df = 11.96, P < 0.002; Table 4.2A,C). In all

three test series, intruders selected while tending queens

were attractive to resident workers over 1-2 centimeters.


Environmental and genetic worker inputs to fire ant

labels were verified in chapters II and III (see also Obin

1986). This study pursued the issue of inputs to the worker

label, and in particular, the issue of cue acquisition from

queens. Genetically correlated queen cues or "queen

discriminators" (Holldobler & Michener 1980) have been shown

to constitute a portion of the worker label in several ant

species studied (reviewed in Holldobler & Carlin, in press),

and are suggested for honey bees (Breed 1981, 1986). In a

broader sense, these transferred cues are analogous to the

maternal odors used for sibling recognition by certain

vertebrates (Porter et al. 1981; Waldman, 1981).

Ideally, queen-derived recognition cues are best


studied by apportioning sister brood among unrelated queens

(cross-fostering) and then testing the adults against each

other for recognition (for definitive examples, see Carlin &

Holldobler 1983, 1986, 1987). This method demands that the

queen's own brood be removed before eclosion for the

duration of the experiment. The issue of queen fecundity

(see Introduction, present study) renders this approach

logistically untenable in fire ants, since queen

reproductive robustness in this species requires tens of

thousands of workers and late-instar larvae (Tschinkel

1986a). The approach taken in this study was to compare the

recognition response elicited by workers reared in the

absence of queens to the response elicited by either workers

reared in queenright colonies or workers tending

artificially introduced queens.

Two assumptions are inherent in all experiments

presented: (A) Adult workers reared from the larval stage

in the absence of a queen and maintained for 26 weeks

possessed no queen-derived recognition cues (cf Breed 1983).

(B) Potential queen cue contamination of newly-eclosed

workers by adult workers introduced with the brood (to aid

in eclosion) would be minimal and would not persist over the

6 months of queenless rearing. Moreover, most of the

original workers would be either dead or visibly senescent

by the time bioassays were conducted, thereby minimizing the

possibility that a worker, previously exposed to a queen in

the parent colony, would be tested.


Experiment 4.1 addressed whether or not queen-derived

cues contribute to the nestmate recognition response of S_.

invicta in large laboratory colonies. Data indicate that

given similar environmental histories (chapters II and III;

Obin 1986), kin reared from the larval stage in the absence

of the mother queen (and thus lacking potential queen

labels) were recognized as nestmates by kin reared in the

presence of the mother queen. Homogeneous, environmental

label components and genetically correlated worker

discriminators were sufficient for acceptance. Similarly,

although aggression directed at non-kin was predictably more

pronounced than aggression directed at kin, no enhanced

discrimination of non-kin resulting from queen-derived cues

was detected by the bioassay employed. Genotypic cues

expressed by workers probably mediated the full recognition

response observed when non-kin intruders were tested.

Because all ants were reared uniformly, possible interaction

between queen-derived and environmentally derived cues was

not assessed.

Experiment 4.2 examined whether or not queen-derived

cues were sufficient to mediate nestmate recognition. I

presented artificially small groups of workers with queens,

thereby maximizing worker exposure to queen-borne odorants.

Data (Table 4.2) indicate that tending of queens in this

artificial situation renders workers more likely to be

strongly aggressed by both kin and non-kin. Queen fertility

was not associated with levels of aggression directed at


workers, a result similar to that obtained by Sorensen and

Fletcher (1985) in a study of queen execution. They

determined that fire ant workers rejected (killed) foreign

queens independent of queen fertility (weight), and

concluded that workers reacted to unfamiliar, qualitative

blends of queen pheromones on foreign queens, not to the

quantity of queen pheromone released. They did not address

the possibility that workers were responding to

environmentally acquired cues and/or worker discriminators

present on alien queens (see below), and that these cues

effectively swamped any effect of physogastry.

The pronounced attraction of resident ants to intruders

selected while tending queens strongly suggested that the

volatile queen attractant/recognition pheromone (Vander Meer

et al. 1980; Rocca et al. 1983) was transferred to

individuals that tended queens. This was confirmed by gas

liquid chromatography (Obin, unpublished), using queen-

specific venom components as indicators of queen pheromone

presence (both are released from the venom reservoir).

However, other exocrine queen substances (e.g., cuticular

hydrocarbons) may also have been transferred from queens to

workers. In addition, because alien queens were freshly-

removed from field colonies, the transfer of environmentally

derived cues and alien worker discriminators cannot be ruled

out in Experiments 4.2A and 4.2B. This may partly explain

why significantly more aggression was directed at kin


tending alien queens than at kin tending mother queens

(Table 4.2A,C).

Results of Experiment 4.2C demonstrate that substances

specific to queens are sufficient to alter worker labels.

Residents and intruders possessed the same environmental

cues and worker discriminators, yet intruders exposed to

queens were attacked more vigorously. Furthermore, the fact

that exposure to mother queens significantly increased

aggression from queenright kin strongly suggests that the

quantity of queen-derived material may be important in

worker-worker recognition. Since residents and intruders

were exposed to the same queen, their labels probably shared

the same blend of queen-derived components. However, labels

of those individuals confined with queens may have been

abnormally high in total queen components, and as such, did

not quantitatively "match" the templates of resident kin.

The quantitative disparity between intruder and resident

labels may also have been enhanced by the 24 hour absence of

the mother queen from the parent colony, although similar

results were obtained when queens were housed with intruder

groups for only 15 minutes (Obin, unpublished data). The

notion of worker discrimination based on quantitative

differences in queen input to worker labels is consistent

with the demonstrated ability of fire ant workers to detect

differences in the quantity of queen pheromones produced by

different queens within the colony (Fletcher & Blum 1983b).


In addition to the likelihood of an early sensitive

period for conditioning to queen and other nest odors (Morel

1983; Evesham 1984; Isingrini et al. 1985; Brian 1986),

templates of adult workers may also undergo age-related

changes associated with temporal polyethism (division of

labor). As they age, fire ant workers switch from nurse

roles to forager roles (Mirenda & Vinson 1981). This

alternation in subcaste affiliation is marked by movement

further from the queen and brood cells (see also Sorensen et

al. 1981a; Vander Meer, in press), and although queen-

derived compounds may be transferred throughout the colony

by mutual grooming and trophallaxis (Fletcher & Blum 1983b;

see also Sorensen et al. 1981b), the relative concentration

of these odors outside the immediate vicinity of the queen

may be relatively low. (Recall that workers not observed

actually tending alien queens in Experiment 4.2A elicited

levels of aggression not significantly different from levels

elicited by queenless kin.) Subcaste-specific templates,

reflecting the relative strength of ambient queen odors

and/or age-related changes in sensitivity to these odors are

suggested by the observation of Sorensen & Fletcher (1985)

that fire ant foragers are more apt to execute foreign

queens than are nurses or reserves. Since we introduced

intruder ants into the foraging arenas of colonies, data

reflect the recognition responses of foragers almost

exclusively (Sorensen and Fletcher 1985), and may not


accurately reflect worker labels and templates near the

queen and brood.

In conclusion, fire ants from mature, queenright

laboratory colonies exhibited no effect of queen-derived

cues on nestmate recognition response. This result

corroborates results obtained in chapter III suggesting that

queen-derived cues contribute minimally, if at all, to the

nestmate recognition cue hierarchy of S. invicta. Exposing

a small number of fire ant workers to physogastric queens

under manifestly unnatural conditions was, however

sufficient to alter workers' nestmate recognition labels.

Queen attractant/recognition pheromone was transferred to

workers tending queens, although a variety of queen-derived

cues (both pheromonal and otherwise) may have also been

transferred. Data are consistent with the putative ability

of fire ant workers to detect both qualitative and

quantitative differences in queen pheromones. Although

weight and fertility of queens in our large, laboratory

colonies were within the ranges reported for "physogastrous"

queens in other fire ant studies (Fletcher & Blum 1983b),

these measures were not maximal. During the summer months,

queens in large, monogyne fire ant colonies can attain 28 mg

and lay over 60 eggs/hour (Fletcher et al. 1980; see also

Willer and Fletcher 1986, Table 1). At such times, queen

inputs to nestmate recognition labels and templates may be

functionally detectable, although the recognition cue

hierarchy will probably remain dominated by environmental


cues and worker discriminators. Finally, the absence of a

significant role of queen discriminators in fire ant

nestmate recognition strongly suggests that the reduced

recognition response of polygynous S. invicta colonies

(Mirenda & Vinson 1982; Obin, unpublished) does not result

from workers learning the discriminators of a number of

queens (cf Carlin & Holldobler 1983).


Table 4.1. Experiment 4.1 (A & B). Responses of S. invicta
laboratory colonies to (A) queenless (Q-) and queenright
(Q+) kin, and (B) queenless (Q-) and queenright (Q+) non-

(A) Queenright and Queenless Kin

Mean (sd)
Intruder Response
Q- (45) 2.86(1.42)
Q+ (54) 2.75(1.37)

Response Level (%/N)
73.3/33 26.7/12 0.0/0
0.39 ns 81.4/44 18.6/10 0.0/0

(B) Oueenright and Queenless Non-Kin

Q+ (72)
Q- (72)

Mean (sd)


Response Level

27.8/20 69.4/50
0.33 ns 29.2/21 69.4/50



Note: Colonies contained 90,000-120,000 workers and a
physogastric queen. Mean ( sd) aggression scores elicited
by different classes of intruders were compared by t-tests
(unpaired data), and the distributions of Level I, II and
III behavior were analyzed by contingency tables, using the
"G" statistic. ns, P > 0.05.


0.94 ns


0.35 ns


Table 4.2. Experiment 4.2 (A-C). Effect of queen-tending
on worker nestmate recognition labels.

(A) Exposed to Alien Queen (24 hours), Tested Against Kin

Q- (20)
Q+T (20)
Q+F (20)

Mean (sd)


0.37 ns

Response Level (%/N)
55.0/11 45.0/9 0.0/0
10.0/2 60.0/12 30.0/6
45.0/9 55.0/11 0.0/0


0.41 ns

em nr~aaj +r~ hltan Aii~n f1~

mini m~4-L~A ~inc~t ?~bm-Vh,

Q- (30)
Q+ (30)

Mean (sd)



Response Level (%/N)
30.0/9 67.7/20 3.3/1
10.0/3 67.7/20 23.3/7 8.17*

(C) Exposed to Mother Queen (24 hours), Tested Against Kin

Mean (sd)
Intruder Response
Q- (27) 4.48(1.95)
Q+ (27) 5.48(1.34)

Response Level (%/N)
33.3/9 66.7/18 0.0/i
3.13*** 7.4/3 92.6/25 0.0/1

0 5.98*

Note: (A) Workers were removed from their queenless parent
colonies, housed with physogastric alien queens for 24 hours
and then introduced into the foraging arena of their parent
colony. Workers tending the alien queen (Q+T) and those
foraging (Q+F) were tested. (B) Workers were exposed to
alien queens for 15 min, and those tending the queen were
introduced into foreign, queenless colonies. (C) Workers
from queenless subcolonies were housed with their mother
queen for 24 hours, and those tending the queen were
reintroduced back into the foraging arena of their parent
colony. In all trials, non-exposed workers (Q-) were
treated exactly as exposed workers, but queens were not
provided. Mean ( sd) aggression scores elicited by
different classes of intruders were compared by t-tests
(unpaired data), and the distributions of Level I, II and
III behavior were analyzed by contingency tables, using the
"G" statistic. *, P < 0.05; **, P < 0.01; ***, P < 0.001;
ns, P > 0.05.

f, UN f--*J. ^^--=> ---^ -- I =" --Wn=" 1 9 -- i I^y- '" M---J. -*>MA A Min^J a.^-* '*"- -



Evidence of learning in fire ant nestmate recognition

(chapters II-IV) implicates "phenotype matching" (Holmes &

Sherman 1983) in the recognition process of this species

(for other examples of phenotype matching by social

Hymenoptera, see Buckle & Greenberg 1981; Pfennig et al.

1983; Carlin & Holldobler 1986). During phenotype matching,

colony members compare olfactory attributes or "labels" of

encountered individuals with a neural "template(s)" of odors

possessed by nestmates or by self (see also Alexander 1979;

Lacy & Sherman 1983; Sherman & Holmes 1985). Decision rules

by which animals compare labels and templates are for the

most part, unaddressed by empirical studies, although

several discrimination mechanisms have been proposed

(Crozier & Dix 1979; Lubbock 1980; Getz 1982; Scofield et

al. 1982; Neigel & Avise 1983a; Getz & Chapman 1987). Getz

(1982) suggested three mechanisms of template-label matching

with respect to heritable odor cues mediating hymenopteran

kin recognition. These are (1) Accept an encountered

individual if all cues in that individual's label are

present in the template ("genotype matching" of Getz'



original model, to be referred to hereafter as "all or none

acceptance"), (2) Reject an individual if its label contains

any cue not present in the template ("foreign label

rejection"), and (3) Accept an individual if its label

contains any cue present in the template ("habituated" or

"common label acceptance"). Habituated label acceptance has

been proposed (chapter III) as a proximal mechanism

maintaining polygyne invicta societies and promoting

intercolonial desertion and adoption observed in several ant

species (Bartz & Holldobler 1982), including S. invicta

(Tschinkel 1986b).

Getz' original models assume that labels and templates

comprise a limited number of discrete and individually

recognizable odors. This assumption is not required in more

recently proposed "cue similarity" models (Gamboa et al.

1986a,b; Getz & Chapman 1987). In these models, phenotype

matching is based upon an individual perceiving that the

incoming odor label is sufficiently "similar" to its

template of "odor images" (Getz & Chapman 1987) to be

accepted. While both models postulate a "threshold" (i.e.,

discontinuous) response to similarity, the model of Gamboa

et al. (1988a,b) is vague about how matching actually

occurs. Getz & Chapman (1987) posit several neural

processing attributes of insects and develop a quantitative

model in which the similarity between the perceived odor and

the odor template is represented by a scalar quantity.


Here, I investigate template matching decision rules in

monogyne, laboratory colonies of the imported fire ant

Solenopsis invicta Buren (Myrmicinae). Under laboratory

conditions, cues associated with diet can dominate the

nestmate recognition-cue hierarchy of S. invicta workers

(chapters II and III). I therefore tested recognition

between kin groups reared on diets that either contained

identical components, contained "foreign" components, or

"common" components. Results would suggest how olfactory

cues were organized in templates and how these templates

were matched to incoming odor labels of encountered


Materials and Methods

Ant Rearing

Monogyne S. invicta colonies were established in the

laboratory from newly-mated queens collected near

Gainesville, Florida (see Banks et al. 1981, for details of

collecting and initial queen maintenance). At maturity,

these mature "resident" colonies each contained 60-100,000

workers, a physogastric queen, and all brood stages.

Colonies were housed as described in chapter II. Queenless

subcolonies containing 200 workers of each of the three

temporal subcastes (i.e., nurses, reserves and foragers;

Mirenda & Vinson 1981) and 500-600 larvae were established

from each mature colony (see below). These queenless

"intruder" groups were maintained in individual metal pans


(5.5 x 17.5 x 29.0 cm) equipped with a petri dish nest (with

Castone floor) and a cotton-stoppered water tube. Rearing

room temperatures ranged from 21-24 degrees Centigrade, and

the light-dark cycle was variable.

Fire ants require a diet containing carbohydrate, lipid

and protein. Accordingly, diets included some or all of the

following material (see below): 50% (v:v) clover honey in

water ("H"), 50% (v:v) dark cane syrup in water ("Sy"),

equal parts honey-water and syrup solutions ("H/Sy"), 5.0%

(w:v) granulated sugar solution ("S"), thawed moth pupae

(Anticarsia gemmatalis) ("P"), thawed roaches (Periplaneta

americana) ("R"), and live mealworms ("W"). Solutions were

stored at 4 degrees Centigrade, and moth pupae and roaches

were stored at -16 degrees Centigrade. Colonies were fed ad

libitum, and unconsumed diet was removed and replaced every

24-38 hours.

Dietary odors may be adsorbed directly into the

integument and/or modified metabolically after ingestion.

This study assumes that (1) metabolically produced odor cues

of workers are identical unless colony diets contain

unshared elements, and that (2) no direct or metabolically

correlated, unique odor cues result from sucrose solution.

Both assumptions preclude potential, asymmetric effects of


Experiment 5.1

Three subcolonies were established from each of 12

queenright, resident colonies, and diet treatments were


begun. Queenright resident colonies received honey-water

and moth pupae (H + P). Subcolonies received either sugar

solution and roaches (S + R), honey-water and roach (H + R),

or the resident colony diet (H + P). After 1 month, three

intruders from each queenless subcolony were introduced into

the foraging arena of their parent resident colony, and

subsequent aggression was quantified (see Bioassay). The

sequence of intruder types presented to any colony was

alternated between replications, with no colony receiving an

additional intruder until all colonies had been tested an

equal number of times. A replicate series of tests was

conducted one week later. Thus, 216 (3 x 72) introductions

were performed.

Experiment 5.2

Four subcolonies were established from each of 11

resident colonies, and diet treatments were initiated.

Resident colonies received honey-water and equal parts by

weight moth pupae and roaches (H + P + R). Subcolony diets

included sugar solution and roaches (S + R), honey-water and

roaches (H + R), honey-water and moth pupae (H + P) and the

resident colony diet (H + P + R). After six weeks, four

intruders from each subcolony were introduced in alternating

sequence (as above) into the resident colony (N = 176).

Experiment 5.3

Nine colonies tested in Experiment 5.2 and three of

each of their subcolonies were used in this experiment.

Mealworms were added to resident colony diets, and honey-


water was replaced with the honey-water/cane syrup mixture

(H/Sy + W + P + R). Subcolonies were fed identical insect

material as parent colonies, but received either 50% honey-

water (H + W + P + R), 50% cane syrup (Sy + W + P + R) or

the honey/syrup mixture (H/Sy + W + P + R). Four months

after diets were established, three workers from each

subcolony were introduced in alternating sequence into their

parent colony. A replicate series of tests was conducted

one week later (N = 182).

Nestmate Recognition Bioassay And Data Analysis

Workers from subcolonies (intruders) were individually

introduced into queenright colonies of sisters (residents)

(see chapters I and II for introduction technique).

Statistically significant treatment effects were inferred

from a consensus between parametric and nonparametric

methods (experiment-wise error rate = 0.05). First,

variances associated with treatment means of introduction

scores (Table 5.1) were tested for homogeneity with the F-

max test (Sokal & Rohlf 1981; error rate = 0.05), and found

not significantly different. Following analysis of

variance, multiple comparisons of treatment means were

performed by the "honestly significant difference" method of

Tukey (Steel & Torrie 1960). In addition, the null

hypothesis of independence between treatments and the

distributions of Levels I, II and III behavior was tested

for sets of treatments by contingency tables using the G

statistic (Sokal & Rohlf 1981). Total degrees of freedom


were used when determining attained significance levels for

any comparison. When no Level III behaviors were recorded

for a treatment (Experiments 4.2 and 4.3), frequencies of

Level II and III behaviors were "lumped." Because of

circadian and seasonal variation in recognition response, we

avoided statistical comparisons involving results of

different experiments.


Experiment 5.1

Data (Table 5.2A) indicate that intruders fed either

S + R diet or H + R diet elicited significantly more

aggression from resident colonies than intruders maintained

on the H + P diet shared with resident colonies. Aggression

directed at intruders fed either S + R or H + R diets was

not significantly different.

Experiment 5.2

Data (Table 5.2B) demonstrate that significantly more

aggression was directed at intruders maintained on the S + R

diet than at all other treatment groups. The magnitudes of

aggression directed at intruders from the three other diet

groups were not significantly different.

Experiment 5.3

Data (Table 5.2C) indicate that significantly more

aggression was directed at intruders fed either diet lacking

a component of the resident colony diet. The magnitudes of


aggression directed at intruders from these two diet groups

were not significantly different.


This study addressed how fire ant workers compare

memory-based odor templates to the incoming odor labels of

conspecifics. Since only kin were tested against each

other, and since transferred queen discriminators appear to

play no role in fire ant nestmate discrimination (chapters

II and III), discordance between resident templates and

intruder labels was due solely to differences in

environmentally (diet)-derived odors. Labels and templates

based on these odors are temporally labile in adult fire ant

workers, and are thus amenable to manipulation. In reality,

labels of encountered nonnestmates differ with respect to

heritable worker discriminators (chapters II and III) as

well as in cues acquired from environmental sources other

than diet. Accordingly, the results of this study may not

accurately reflect events as they occur in the field.

Results suggest that S. invicta workers match templates

and labels of environmentally derived recognition cues by a

mechanism other than those proposed by Getz (1982) for

genetically determined cues. Acceptance under an "all or

none" model of template matching requires that all diet

components be represented in the intruder label. Although

results of experiments 5.1 and 5.2 appear to support this

mechanism, intruders from the H + P and H + R diet groups in


experiment 5.2 were treated no differently by resident

colonies than intruders maintained on the complete resident

colony diet (H + P + R). A mechanism involving "foreign-

label rejection" requires that any component not represented

in the resident colony diet elicit aggression. While the

results of experiment 5.1 cannot rule out this mechanism

(roach is a "foreign" component), the results of experiments

5.2 and 5.3 conclusively exclude it. Under our assumption

that sucrose solution provides no olfactory recognition

cues, the S + R diet that elicited strong aggression in

experiment 5.2 has no "foreign" cues. More convincing

perhaps are the results of experiment 5.3, in which

"foreign" cues were unequivocally absent from any subcolony

diet, yet two of the diets elicited increased aggression.

Finally, intruders were attacked despite the fact that their

diets shared common elements with resident diets

(experiments 5.1, 5.2 and 5.3). This result argues against

a recognition mechanism involving "habituated label


Taken together, data from the three experiments

presented are consistent only with an "overall similarity"

model (Getz & Chapman 1987; see also Gamboa et al. 1986a,b).

Our results suggest that "similarity" may depend on the

total number and relative contributions of colony odor cues

to labels and templates. When differences in labels were

derived from two-component diets (experiment 5.1), odor cues

from both components of the resident diet were necessary for


complete concordance between labels and templates. With a

three component diet (experiment 5.2), only two shared

components were necessary for a functional match between

label and template. In experiment 5.3, odor cues derived

from honey and syrup appeared to dominate the template such

that, despite four shared diet components, intruder labels

required both honey and syrup in order to match worker

templates. Although no experiments were designed expressly

to evaluate whether similarity was a threshold or

nonthreshold phenomenon, multicomponent diet experiments

such as these seem particularly suited to such studies.

Data supporting "cue similarity" rather than discrete

odor recognition are not surprising when only

environmentally derived odors differ between individuals, as

these odors are expected to be especially variable. Models

involving discrete odor cues often assume a relatively

simple (i.e., one or two locus) genetic basis for the cues

(see for example, Crozier 1986), and such models are

probably more appropriate for recognition systems

characterized by potentially less complex and less variable

cues, such as queen discriminators (review in Holldobler and

Carlin, in press). Of significance for fire ants is the

fact that "cue similarity" matching is consistent with the

hypothesis (chapter III; see also Bhatkar 1979) that food

sharing (trophallaxis) between nonnestmates promotes both

intercolonial adoption (Tschinkel 1986b) and polygyny

observed in this species. A mechanism involving "all or


none" matching or foreign label rejection would effectively

preclude such an hypothesis.

Finally, it is proposed here that "similarity" provides

a more coherent framework than discrete odor models from

which to attack the recognition process (as discussed in

Getz & Chapman 1987). To begin with, empirical distinctions

between any of the discrete models may in fact, be

epiphenomena. For example, recognition studies of the

carpenter ant (Camponotus spp.) have been interpreted as

evidence for both "genotype matching" (Morel et al., in

press) and "foreign label rejection" (Carlin & Holldobler

1987; Morel, unpublished data). While it is possible that

animals use different matching mechanisms in different

behavioral contexts and under different experimental

conditions, the parsimonious explanation consistent with all

the data is one involving "cue similarity." In addition,

discrete odor models of template-label matching may blur

important distinctions with respect to mechanisms of label

acquisition. Consider the oft-invoked "gestalt" model

(Crozier & Dix 1979) in which odor cues are transferred

among workers, thereby producing a colony odor "blend"

shared by all nestmates. One empirical demonstration of a

colony odor "gestalt" involves the generation of a graded

recognition response as a function of incremental changes in

the proportion of kin in mixed groups (see for example,

Carlin & Holldobler 1986). Because they assume an "all or

none" recognition response, none of the discrete odor models


are germane to this approach, whereas a nonthreshold "cue

similarity" model is entirely appropriate. Failure to

obtain a graded response may in reality be evidence in favor

of a threshold matching mechanism, not evidence against a

colony odor blend transferred among workers. Certainly

coherent models of phenotype matching require consistency

between mechanisms of label acquisition and mechanisms of

template-label matching. Although elucidation of the latter

may require the combined efforts of chemists,

neurophysiologists and ethologists, a more unified

understanding of recognition may be the welcome outcome (cf

Getz & Chapman 1987).


Table 5.1. Experiments 5.1-5.3. Response of queenright
colonies to queenless kin maintained on diets containing
identical "common" or "foreign" components (see Materials
and Methods).

(A) Experiment 5.1. Resident Colony Diet = H + P.

Intruder Diet
H + P (72)
H + R (72)
S + R (72)

Mean (sd)
3.25 (2.12)a
4.07 (2.42)b
4.50 (2.38)b


% Response Level
1(49) 27.8(20)
6(35) 41.7(30)
4(32) 45.8(33)

(B) Experiment 5.2. Resident Colony Diet = H + P + R.

Intruder Diet
H + P + R (44)
H + P (44)
H + R (44)
S + R (44)

Mean (sd)
3.30 (2.08)a
3.55 (1.98)a
3.41 (2.28)a
5.11 (1.87)b

% Response Level




(C) Experiment 5.3. Resident Colony Diet = H/Sy + W + P + R.

Intruder Diet
H/Sy + W + P +
H + W + P +
Sy + W + P +


Mean (sd)
2.76 (1.37)a
3.54 (1.65)b
3.83 (1.67)b

% Response Level (N)




Note: Individual scores represented the highest rank of
aggressive acts directed at each intruder by 20 resident
ants. Mean ( sd) aggression scores were compared among
treatments (diets) by the "hsd" method of Tukey (Steel and
Torrie 1960). Individual scores were also assigned to one
of three, progressive Levels of aggressive behavior. The
distributions of Level I, II and III behavior were tested
for homogeneity by contingency tables, using the "G"
statistic. Means or distributions of behavior followed by
the same letter are considered not significantly different
(experiment-wise error rates = 0.05).




Studies of fire ants in the Solenopsis saevissima

species complex have contributed significantly to our

understanding of social insect biology (see for example,

Wilson 1971; Fletcher & Ross 1985a; volume edited by Lofgren

& Vander Meer 1986; Vander Meer, in press). With the recent

documentation of introgression ("hybridization") between

populations of the two parental forms, S. invicta and S.

richteri (Vander Meer et al. 1985), the imported fire ant

has emerged as an important model system with which to

address issues in systematics and biogeography (Ross et al.

1987; see also Buren et al. 1974). Central among these

issues is the question of hybrid zone dynamics, and in

particular, postulated mechanisms that explain the temporal

stability (ca 45 years) and northward movement (from

southern Alabama to east-central Mississippi) of the fire

ant hybrid zone. Of a number of models advanced as possible

explanations (Ross et al. 1987), the most plausible invoked

either hybrid superiority resulting from increased genetic

diversity in recombinants, the competitive superiority of S.



invicta, and/or selection favoring or disfavoring hybrids.

Little is known in regards to these models.

In this study, I use the phenomenon of "recognition" as

a measure of both relative competitive ability and genetic

diversity among the imported fire ants S. invicta and S.

richteri (sensu Buren 1972) and their Mississippi hybrid(s).

Intra- and interspecific competitive ability is an important

determinant of colony fitness and ant community structure

(Holldobler & Lumsden 1980; Levings & Traniello 1981;

Holldobler 1983). This competition is mediated to a

significant degree by discrimination of chemical cues

mediating territory, trail and competitor/enemy

"recognition" (reviewed in Holldobler & Carlin, in press.

Fire ants vigorously defend colony-specific territories and

exhibit a complex alarm/recruitment defense behavior when

intruders are encountered in or near the nest (Wilson 1962;

Obin & Vander Meer 1985; Obin 1986; Table 3.1). In S.

invicta, recognition of territory (Hubbard 1974; Obin,

unpublished), conspecifics (chapters I-IV; Obin 1986), and

congeners (Vander Meer & Wojcik 1982) is mediated either

wholly or in part by chemoreception, and this appears to be

the case throughout the genus (Jaffe & Puche, 1984 for S.

geminata; Obin, unpublished data for S. richteri). Both

heritable odor cues ("discriminators") and nonheritable

(environmentally acquired) cues are implicated in fire ant

nestmate recognition (chapters II and III). Here I assess

the ability of the two parental fire ant species and their


hybrid to recognize and repulse conspecific and congeneric

"intruders." By controlling for differences in

environmentally derived recognition cues, I was also able to

indirectly evaluate the relative diversity of heritable

recognition cues in the three populations studied. These

data are compared with heterozygosity values obtained from

isozyme data (Ross et al., in press) and interpreted with

respect to temporal and geographic dynamics of the

Mississippi fire ant hybrid zone.

Materials and Methods

Collection and Rearing of Stock Colonies

Solenopsis invicta were reared from newly-mated queens

collected 17-23 months prior to tests in Alachua County,

Florida. Solenopsis richteri and hybrids were collected 12

months prior to tests from single, Mississippi locales

(populations). Solenopsis richteri were collected at mile

279 on the Natchez Trace in Lee County. Hybrids were

collected in Loundes County, 14 miles west of Columbus on

Interstate 82. The queen and several thousand workers and

brood were collected from each colony. Characterization of

all colonies as either S. invicta, a. richteri or hybrid was

obtained in the laboratory by gas-liquid chromatographic

(GC) analysis of worker cuticular hydrocarbons and venom

alkaloids (Vander Meer et al. 1985; Ross et al. 1987). This

method is highly concordant with population designations

based on enzyme polymorphism. A numerical index was


developed with GC data to identify hybrid and parental

populations (Ross et al. 1987). Hybrid colonies with

"hybrid indices" closest to 0.5 (the midrange between values

for the parental species) were selected for the present

study. Colonies were maintained in large trays as described

in chapter II. Ants were fed fly pupae (M. domestica,

hard-boiled egg and honey-water, and were maintained at 24-

26 degrees Centigrade on a variable light-dark cycle.

Recognition Bioassay

Ten colonies each of S. invicta and S. richteri and

five hybrid colonies were used during the experiments. Nine

different combinations of between and within-group

(genotype) introductions were conducted. Interspecific

tests were performed with five, randomly selected colonies

of one parental form ("residents") and individual intruders

from ten colonies of the other parental form (nl = n2 = 50).

This was done so that half of the colonies of each parental

species would be naive with respect to chemical recognition

cues possessed by heterospecific intruders and potentially

shared with the hybrids. The "naive" colonies of each

parental species were tested against intruders from the five

hybrid colonies (n3 = n4 = 25). Hybrid colonies were tested

against intruders from five randomly selected colonies of

each of the parental forms (n5 = n6 = 25). Intraspecific

tests were performed with nine colonies of each of the

parental forms ("residents") tested against individual

intruders from the other eight, conspecific colonies (n7 =


n8 = 72). Each of the five hybrid colonies were tested

against intruders from the other four (n9 = 20). Bioassays

were conducted at 23-24 degrees Centigrade.

Data Analysis

Statistically significant effects were inferred from a

consensus between parametric and nonparametric methods.

First, due to significantly heteroscedastic data (F-max

test), treatment means were compared by sequential t-tests

(rather than by simultaneous test procedures appropriate to

analysis of variance). Inspection of the data determined

which comparisons were required to construct maximal,

nonsignificant sets of means. As six pairs of means were

tested, I set the per-comparison Type I error rate for

rejection of the null hypothesis of equality of means at

0.01. The null hypothesis of independence between genotypes

confronting each other in the bioassay and the distributions

of Levels I, II and III behavior was tested by sequential G-

tests (comparison-wise error rate = 0.01), (Sokal and Rohlf

1981). When no Level I behaviors were recorded for an

intruder x resident type, frequencies of Level I and II

behaviors were "lumped" and degrees of freedom adjusted

accordingly. Fisher Exact Tests were performed when only

Level III behavior was recorded for an intruder x resident




Data for between-genotype tests (Table 6.1) indicate

that hybrid colony response to S. invicta and S. richteri

was not significantly different from the response of either

parental species to hybrids or heterospecifics. Similarly,

hybrid response to hybrid intruders was not significantly

different from hybrid response to either of the parental

species. In contrast, both parental species were less

aggressive toward conspecific intruders than toward

heterospecific or hybrid intruders, with intraspecific

recognition scores for S. invicta significantly lower than

scores for S. richteri (t-test, P < 0.01).


Although numerous factors ultimately determine colony-

level fitness, behavioral-ecological studies of species and

colony interactions among ants (Holldobler & Carlin, in

press) and the potentially broad niche overlap between

hybrids and parental species suggest that detection and

repulsion of congeners is an important measure of a fire ant

colony's competitive ability and relative fitness. This

study assumes that the laboratory bioassay used to quantify

this ability reflects competitive vigor as manifest in the

field. Results indicate no significant differences in

aggressive response when all six between-genotype

introduction sequences are considered. These data argue

against either superior or inferior species-level


recognition capabilities of hybrids. To the extent that

this capability contributes to colony fitness, results of

this study fail to support the "dynamic equilibrium" model

(Barton & Hewitt 1985) as an explanation for the temporal

stability of the Mississippi fire ant hybrid zone (see Ross

et al. 1987). In the "dynamic equilibrium" model, the

hybrid zone persists despite selection against hybrids as a

consequence of gene flow from the parental populations.

Data do, however, suggest that genetic diversity is

greater in the hybrid population tested than in populations

of either of the parental species. Intraspecific (nestmate)

recognition in imported fire ants is mediated by both

environmentally acquired cues and heritable, worker

discriminators (chapters II and III; Obin, unpublished data

for S. richteri). Under homogeneous laboratory conditions,

recognition is theoretically based solely on heritable odor

differences. In this study, the intraspecific recognition

response within each of the parental populations was

significantly less than their interspecific (species)

recognition response. In contrast, recognition response

among hybrid colonies was not significantly different from

hybrid response to either of the parental genotypes. This

implies a relatively greater ability of hybrids to make

within-population discrimination based upon heritable odor

cues (discriminators), and is consistent with the hypothesis

that genetic variability underlying the expression of these

cues is greater in hybrid populations than in populations of


either parental species. This is to my knowledge the first

extension of a recognition bioassay to the population

genetics of insects, although histocompatability bioassays

have been used to analyze the population structure of clonal

marine organisms (Neigel & Avise 1983a,b).

If, as reviewed elsewhere (Hepper 1986; Crozier 1986,

1987), discriminator expression reflects the contributions

of polymorphic loci under strong selection by pathogens and

parasites, population differences in discriminator

variability may indicate important, fitness-correlated

differences in overall genetic variability. Reductions in

genetic variability among the parental fire ant populations

may be a consequence of limited introduction events and

genetic drift (Obin 1986; Ross et al. 1987), and are

demonstrated to have severe consequences for colony fitness

(Ross & Fletcher 1985, 1986). Data therefore support the

contention of Ross et al. (1987) that hybrid zones of the

imported fire ant may persevere as a consequence of

relatively high levels of genetic diversity and resulting

hybrid superiority.

We can also extend our interpretation of the within-

genotype bioassay results to relative differences in genetic

diversity in the three populations studied. Using

presumably "neutral" Mendelian markers, Ross et al. (in

press) determined that heterozygosity values for S. richteri

colonies were significantly lower than values obtained for

S. invicta. Their results are intuitively satisfying given


the extremely small effective population size (N) in S.

richteri (Ross et al., in press; Ross, pers. comm.) and the

positive association between effective population size and

the extent of neutral variation (Kimura & Ohta 1971). In

contrast, our behavioral data suggest the following

hierarchy of heterozygosity for loci on which selection is

presumably acting: S. invicta < S. richteri < hybrid(s).

To the extent that genetic diversity is reflected in colony

fitness, these data do not support the hypothesis (Ross et

al. 1987, in press) that the northward movement of the

hybrid zone toward S. richteri populations has been in the

direction of the competitively inferior parental species.

The idea of genetic variability being enhanced by

hybridization between bottlenecked species may be generally

applicable to social Hymenoptera, a group distinguished by

low levels of genetic variability (reviews in Graur 1985;

Sheppard & Heydon 1986). Such a mechanism may play a

particularly significant role in speciation within the S.

saevissima complex, both because hybridization in contact

zones may be widespread in South America (Wilson 1958), as

well as because the "weedy," colonizing life-history of the

ant (Tschinkel 1986b) renders populations especially

susceptible to founder effects (Mayr 1942) and allele

fixation by genetic drift (Wright 1951). This intriguing

possibility should be addressed by extensive, long-term

studies of natural fire ant populations here and in South



Table 6.1. Recognition response of S. invicta (Si), S.
richteri (Sr) and S. invicta/S. richteri hybrids (HY)
maintained in the laboratory.

Mean (sd)

Int x Res

Aggression Level



(0.37) ab








100. 0/50a


Note: Within-genotype tests measure intraspecific
(nestmate) recognition, whereas between group tests measure
interspecific (species) recognition. Individual datum
represented the highest rank (1-9) of aggressive acts
directed at each intruder (Int) by 20 resident ants (Res).
Mean ( sd) aggression scores were compared by t-tests (see
Materials and Methods). Individual scores were also
assigned to one of three, progressive Levels of aggressive
behavior. The frequency distribution of Level I, II and III
behaviors were tested for homogeneity by contingency tables,
using the "G" statistic or by Fisher Exact Tests. Means or
distributions of behavior followed by the same letter are
considered not significantly different (comparison-wise
error rate = 0.05). Means or distributions of behavior
followed by different letters are considered significantly
different (comparison-wise error rate = 0.01).


Sr x Si
HY x Si
Si x Si

Si x Sr
HY x Sr
Sr x Sr

Si x HY
Sr x HY








This study presents (1) a detailed and replicated

inquiry into the recognition-cue hierarchy of the imported

fire ant (chapters II-IV), (2) an experimental analysis of

the mechanism by which fire ant workers match odors of

encountered individuals with a memory template of nestmate

odors (chapter V), and (3) an extension of the recognition

concept to the population genetics and ecological

biogeography of parental and hybrid fire ant populations

(chapter VI).

Chapter II suggested the importance of both

environmentally acquired and heritable components of worker

labels. This was a somewhat unexpected (unwelcome?) result

in light of the recent emphasis on "genetic" models of kin

recognition (see Gadagkar 1985), as well as because it made

chemical elucidation of recognition cues more difficult (see

below). The preliminary results with environmental

differences were verified with more extensive laboratory and

field experiments in chapter III. Data confirmed the fact

that diet was a potentially important determinant of worker

labels, thereby establishing the methodological framework

for chapter IV. Chapter III also presented the first

negative results with queen discriminators. The issue of



queen inputs to worker labels was pursued in chapter IV,

where it was concluded that such inputs contribute minimally

if at all to worker labels in a nest defense context.

Negative evidence for the role of queen discriminators in

nestmate recognition makes the fire ant the exception to the

"theoretical" rule suggested by H611dobler and Michener

(1980). Clearly, the fire ant system must be addressed in

any meaningful, comparative studies of queen pheromone and

queen discriminator evolution. The aggressive response of

foragers to queen-tending kin argued for the existence of

subcaste-specific labels and templates that reflect

workers's physiological age and olfactory surroundings.

These labels and templates may mediate integrated, within-

nest worker activities beyond the scope of this study. The

existence of sub-caste specific labels is further suggested

by the results of Vander Meer et al. (unpublished), who

distinguished nurses, reserves and foragers based upon

cuticular hydrocarbon pattern. In chapter V, diet-derived

cues were manipulated to investigate the mechanism of

template-label matching in fire ants. This is to my

knowledge, the first empirical study that specifically tests

proposed odor-matching mechanisms) underlying phenotype

matching. Results argue against the discrete odor models of

Getz (1982) and others, and support the "overall similarity"

concept of Getz and Chapman (1987) and Gamboa et al.

(1986b). In chapter VI, recognition response among S.

invicta, S. richteri and their Mississippi hybrid(s) was