2. Tierpsychol., 39, 24-32 (1975) @ 1975 Verlag Paul Parey, Berlin und Hamburg ISSN 0044-3573/ASTM-Coden ZETIAG

The University of Kansas, Department of Entomology and Department of Physiology and Cell Biology, Lawrence, Kansas

Agonistic Behavior in the German Cockroach, Blattella germanicu

BY

MICHAEL D. BREED, CAPRICE M. HINKLE and WILLIAM J. BELL

With 3 figures

Received: 25. 11. 1974

Introduction Agonistic behavior in the Blattaria has previously been investigated by

BELL and SAMS (1973), EWING (1967, 1972), SIMON and BARTH (in press) and RITTER (1964). ALEXANDER (1961) investigated aggressive behavior in another Orthopteriod group, the field crickets (Acheta sp.). Social interactions have been observed in each of these species in laboratory populations, but the pur- poses of aggressive, hierarchal or territorial behavior in some of these species is unclear. Previous work has focused on components and releasers of aggressive behavior and on the degree of territoriality found in the laboratory in a given species. The purpose of this paper is to present data which relates agonistic behavior to the ways in which it enhances the overall fitness of the individuals in the population.

Material and Methods Observations were made on groups of cockroaches in covered 30 X 19 x 12.5 cm

plastic cages. All observations, except where noted, were made on groups of 20 cockroaches during the first three hours of the dark period of the photocycle; GE photography lamps (590-640 nm.) were used to simulate darkness (BELL et al., 1972). Animals for experimenta- tion were caught by live-trapping and quarantined in glass aquaria for at least 19 days prior to use in experiments. Cockroaches were maintained under a 12 : 12 hr. photoperiod prior to and during experimentation. Means were compared using Students t-tests or analysis of variance. Significance of regression coefficients was tested following the method of SOKAL and ROHLF (1969).

Results a) Categories of interactions and fighting techniques

The German cockroach displays agonistic action patterns which resemble those of other species of Blattaria, for example, Periplaneta americana (BELL and SAMS 1973; SIMON and BARTH in press) and Nauphoeta cinerea (EWING 1972). Stilt-walking and body-jerking are threat or intention movements. In stilt-walking the animal raises off the substratum by extending the legs ver-

Agonistic Behaviour in the German Codcroach, Bfattella gerrnanica 25

tically. In B. germanica body jerking, which occurs when the animal has reached the stilt-walking position, consists of a rapid series of vibrations of the body, usually directed anteriorly. More aggressive actions include butting with the head (a follow-through of the body-jerk), biting, and kicking. Non- aggressive agonistic actions are truce, retreat, and aatennation. In retreat one animal stays a t the site of the encounter and the other one leaves, while in truce, both animals remain. Antennation occurs when one animal contacts another with the flagellum of its antennae.

b) Sequences Observed in Interaction Patterns During Agonistic Encounters In order to interpret the observations, encounters were classed into five

levels of intensity based on the categories discussed above. Level one encoun- ters are characterized by one animal antennating the other. In level two en- counters, mutual antennation occurs. Neither of these levels may be considered true aggressive encounters, since no overt aggressive act occurred. Level three encounters involve individual or mutual intention movements, such as stilt- walking and body-jerking, but no contact between animals occurs at this level. Higher level encounters involve actual contact. In level four encounters one-sided aggressive contacts occur, while in level five, mutual kicking or biting is observed. Interactions may result in either increased aggression (pro- gression to the next level) or termination. N o interaction began a t any level except one, nor did any interaction decrease in level (i. e. stilt-walking never followed biting or kicking). Thus the German cockroach appears to have a fairly rigid pattern of actions which comprises its agonistic behavior. Level one and two encounters always resulted in truce, and higher level encounters resulted in retreat 37 O/o of the time and truce in the remainder. In P. ameri- cana encounters where stilt-walking or body-jerking takes place truce some- times occurs (18 O/o), but in higher level encounters truce was never observed and retreat always followed (BELL and SAMS 1973).

In the German cockroach, aggressive actions have a lower level of inten- sity than in the other species previously reported; intense mutual grappling

LEVEL

1

2

3

4

EITHER

ANTENNATES i l l 2

TRUCE I2O1 82f A . . \

T T T MUTUAL / ANTE N N AT I0 N

BODY-JERK I

1"" RETREAT

K 3 9 15 \ONE-SIDED BITING ,/

KICKING, BUTTING I

5 RECIPROCAL / BITING, KICKING

occurs in P . americana which may result in the loss of appendages by the combatants, but in B. germanica even the most intense encounters lasted only a few se- conds and were never observed to result in dismemberment.

F i g . I : Summary of tech- niques and sequences used

in agonistic interactions

26 M. D. BREED, C. M. HINKLE and W. J. BELL

Table I : Percent incidence of each level of aggression among and between sexes (see text for

explanation of levels).

I Level 1 1 1 2 1 3 1 L I 5 1 Mean Level 1 2.36 2.26 2.16

The progression of movements in interactions is outlined in Fig. 1. Each encounter begins with approach by one individual. In order for the interaction to proceed to aggressive levels, antennation must be reciprocated. Stilt-walk- ing and body-jerking follow; the manner in which the individual against whom these actions are aimed perceives them in the dark is unclear. How- ever, antenna1 contact may establish the position and action of the combatant and jerks may be transmitted as vibrations in the substratum.

In the following sections, encounters are referred to by the maximum level of aggression observed; for example, an encounter terminating after an animal stilt-walks is a level three encounter.

1,51 / Y=.O3X+1.51

1 .o / 10 15 20 25 30

POPULATION SIZE

Fig . 2: Relationship between population density (number of animals per container) and mean level of aggression. Points represent means of five replicates

I I I I I I 0 5 10 15 20 25 30

POPULATION SIZE

Fig. 3: Relationship between population density (number of animals per container) and frequency of agonistic encounters expressed as encounters per animal per minute. Points

represent means of five replicates

Agonistic Behaviour in the German Cockroach, Blattella germanica 27

Level 1 2 3 4

Number of animals

2 38 50 12 0 5 31 42 21 0

10 66 38 14 3 15 33 32 16 17 20 28 42 25 6 30 21 38 26 13

5

0 0 0 1 0 1

28 M. D. BREED, C. M. HINKLE and W. J. BELL

Level

? ? with oothecae $ 7 without oothecae

Table 3: Comparison of percent incidences of aggressive levels for pregnant and non-pregnant ?E

(see text for explanation of levels).

1 2 3 4 5

30 41 18 12 0 27 54 19 0 0

Level

Starved Fed

(p < 0.01); the mean number of encounters per minute was 0.70 in the group with oothecae and 0.26 in the group without. 99 without oothecae did not participate in level four encounters (Table 3), while level four encounters com- prised 120/0 of the encounters for 99 carrying oothecae. This increase in aggressive behavior in 99 which are carrying egg cases externally probably serves a protective function. Egg cannibalism is common in cultures of P. umericana, where oothecae are simply deposited on the substratum about a month before the young hatch. While the reason for increased aggressive activity is not clear, it may result in dispersal of 99 carrying oothecae to areas with low population density, or i t may prevent other cockroaches from feeding on the ootheca while it is being carried by the Q.

1 2 3 4 5

24 18 27 29 2 29 51 14 6 0

Table 4 : The effect of starvation on levels of aggression (percent incidence) in B . germanica.

f ) Effects of Starvation on Aggressiveness Starvation induced the most marked change in level of aggressiveness of

any variable studied (Table 4). The average level of aggressiveness in control groups of 10 cockroaches was 1.97, close to the value predicted from Fig. 2. In populations which had been starved from four to eleven days, the average level was 2.69; this is significantly greater (p < 0.05) than fed populations. Although starved animals had a greater number of encounters per min (8.5 versus 3.8), because of a large variance, the difference was not statistically significant. The length of the starvation period was not found to have a significant effect after the fourth day.

g) Relationship Between Photoperiod and Intensity of Aggressiveness The observations described above were all conducted during the first

three hours of the dark part of the photocycle. This corresponds to the period of time when most activity occurs ( B . germanica has a diurnal rhythm of acti- vity similar to that of P. americana reported by HARKER 1960). Most aggressive encounters were observed away from nest shelters, food, or water; they were the result of meetings between roaches actively moving on the sub- strate. During the portion of the photocycle when the cockroaches are inact- ive, few aggressive encounters were observed, since nearly all cockroaches remained in the shelters. While few animals actually touched neighboring individuals, except for antenna1 contact. no interactions which could be interpreted as agonistic were observed in the shelters. Spacing appeared to be even (non-random) but did not appear to be the result of aggression.

Agonistic Behaviour in the German Codtroach, Blurtella gerrnanica 29

Discussion MOYER (1 968) distinguishes six types of aggression; inter-male, fear-

induced, irritable, territorial, maternal and instrumental. Of these, inter- male, territorial and maternal are most easily distinguished in cockroaches. In the cockroach, “fear” usually induces fleeing responses, while the other two types involve a complexity of motivation which is probably beyond the capabilities of a cockroach. KING (1973) put aggression into an ecological framework, grouping the effects of aggression as: 1. enhanced recruitment to the population (e. g. maternal behavior), 2 . reduction of population by morta- lity or dispersal, and 3. spacing of individuals within the population.

Each of MOYER’S types of aggression will probably affect more than one of KING’S categories. For example, inter-male aggression may serve both to reduce the population by mortality or dispersal and to space the remaining individuals in the population into territories. Similarly maternal aggression will enhance recruitment by protecting the young and will also result in spac- ing of individuals. Since MOYER’S categories provide a more precise frame- work they will be used in the following discussion.

a) Inter-male Aggression Fighting among d‘d may serve to improve mating chances for the winner;

losing dd may leave the area, reducing the 0“ : 9 ratio, or they may be dismembered and suffer a disadvantage in finding mates. In Peripfaneta americana (BELL and SAMS 1973), fighting among dd seems to serve this purpose, since dismemberment often occurred in laboratory cages. Studies on a wild population of P. americana (BREED, COLE and WETHERELL, in press) revealed many more 99 (70 to 90 “ / o of the population) than d‘d, lead- ing to the conclusion that dd were more highly dispersed than Q?, probably due to a process mediated by agonistic behavior. EWING (1972) found that low density populations of N . cinerea formed dominance hierarchies and that the top ranking dd possessed territories. The rank of the cr* did not confer any special mating status, although sometimes a territory holder would have sole possession of the food and water. A defect, however, in studies of the type made by EWING is that the dd did not have open routes for dis ersal - in natural circumstances low-ranking or non-territory-holding 6 c? may leave the area and territory-holding Crd may gain reproductive advantage simply by being in an area where greater numbers of Q? are present.

Dismemberment in B. germanica is practically absent and aggression appears to be independent of the sex of those involved in the encounter; inter-male fighting does not seem to be important in determining mating fitness in this species.

b) Maternal Aggression Maternal behavior expressed as aggression increases the probability of

survival of that individual’s young, and an increase in aggressiveness in ootheca-carrying 99 would be highly advantageous from a selective point of view. Other types of maternal behavior are known in cockroaches. For example, LIECHTI and BELL (in press) found that Q? of Byrsotria fumigata exhibit brooding behavior after the young hatch, and many species (e. g. P. americana, Supella longipalpus, personal observations) exhibit care in select- ion of oviposition sites. Maternal aggression, however, is rare in insects out- side of the eusocial groups, largely because reproductive strategies are oriented to the production of a large number of eggs in a short amount of time and because the Q often does not live until the next generation hatches.

30 M. D. BREED, C. M. HINKLE and W. J. BELL

c) Territoriality Territoriality has been observed in several cockroach species: Cryptocer-

cus punctulatus (RITTER 1964), N . cinerea (EWING 1967, 1972) and P. ameri- cana (BELL and SAMS 1973). Degree of territoriality is variable, with P. americana being only vaguely territorial and lacking a fixed dominance hierarchy and Cryptocercus being highly territorial and possessing mating chambers. Loosely defined, fighting implies territoriality (NOBLE 1939; see discussion in ALEXANDER 1961) even though an individual may not be con- stant to a specific location or feature of the environment. In B . geumanicu, however, aggressive encounters appear to be the result of meetings between animals whose occurrence is more dependent upon population factors (such as density) than on the spatial relationship of animals with resting places, food, water, or mating sites. While these encounters may be territorial in the senlse that they result in the spacing of animals in the larger environment (dispersal), they do not establish fixed spatial relationships among the ani- mals and their immediate environment. Behavior of this type may f i t a broad definition of territoriality, but it is probably best to label this aggression “dis- persive”.

B . germanica is a highly gregarious species; animals cluster together in dense aggregations during the light portion of the photocycle and an aggre- gation pheromone has been demonstrated in the feces (ISHII 1970). Agonistic behavior serves to balance gregariousness in a system of population regulat- ion similar to that described for crickets by ALEXANDER (1961). In times of low population density, factors increasing gregariousness have a major im- portance, while in times of high density, factors increasing dispersal, such as agonistic behavior, are more important.

Aggressive interactions in the German cockroach appear to be motivated by many factors relating to individuals and the population. 99 bearing oothe- cae show increased aggressiveness which probably repels other individuals who are potential egg cannibals. This behavior may also lead to greater spac- ing of individuals in the environment, providing more space for the newly emerged young. Starvation and high population density are two other situa- tions in which increased aggression may lead to increased dispersion; this pro- vides a method of population size regulation which is sensitive to sudden changes in ecological conditions. While it is difficult to link aggression and dispersion experimentally, i t is reasonable to hypothesize that the aggressive interactions in a population constitute communication among members of the population concerning various factors affecting the population, and that the population will respond to this communication.

d) Comparative ethology of the Blattaria While agonistic techniques are relatively constant throughout the Blatta-

rian groups so far studied, there is a high degree of variability in the purpose of the agonistic encounters. Cryptocercus (RITTER 1964) is highly territorial; both territorial and inter-male aspects of fighting are emphasized. Nauphoeta (EWING 1972) and Periplanetu (BELL and SAMS 1973), which are somewhat territorial, generally restrict their fighting to inter-male conflicts, but also use fighting as a dispersive mechanism. Blurtella completely lacks territoriality, has de-emphasized inter-male aspects of fighting, but uses fighting as a mater- nal and dispersive mechanism. While data is not available on sufficient spe- cies to draw firm phylogenetic conclusions it is interesting to note that Nauphoeta and Periplaneta, with similar behavioral patterns, are quite distantly related morphologically (MCKITTRICK 1964). A tentative conclusion

Agonistic Behaviour in the German Cockroach, Blattella germanica 31

is that agonistic behavior in the Blattaria is probably more related to the ecology of each species than to a phylogenetic progression in the order. This is substantiated by the fact that primitive Blattaria, represented by Cryptoccr- LUS, display the full repertoire of agonistic actions. The main direction of evo- lution may be a ritualization of what primitively was inter-male and territo- rial aggression into dispersive and maternal aggression accompanied by an attenuation of the intensity of encounters.

Summary Agonistic behavior in Blattella germanica is delineated with an empha-

sis on fighting techniques and population factors affecting aggressiveness. 6-6, 6-9, and 9-9 encounters are not significantly different in either level of aggressiveness or frequency. 99 carrying oothecae are, however, more aggressive than other 99. The mean intensity of aggression increases as popu- lation density is increased, but number of contacts per individual per unit time remains constant. Defense of specific territories does not occur and aggressive interactions are most common during the dark portion of the photo- cycle, when many of the cockroaches are observed to be foraging.

Acknowledgements This work was supported by National Science Foundation-Student Originated Studies

grant GY-11465 to the University of Kanszs and a University of Kansas General Rcsearcli Grant . The authors wish to thank RUDOLF JANDER for critically reading the manuscript. Contribution 1575 from the Department of Entomology, the University of Kansas.

Zusammenfassung Das agonistische Verhalten der Deutschen Schabe, Blattella germanica,

wird beschrieben und in fiinf Intensitatsstufen gegliedert. d-d, d-9- und 0-9- Begegnungen unterscheiden sich nicht in ihrer durchschnittlichen Hau- figkeit und Aggressivitat. 99 mit Ootheken sind aggressiver als 99 ohne solche. Die mittlere Aggressionsintensitat steigt mit der Populationsdichte, jedoch bleibt dabei die mittlere Haufigkeit der aggressiven Kontakte pro Individuum lronstant. Territorialverhalten fehlt, und die Aggressionshaufigkeit steigt in der Dunkelphase der Photoperiode in Zusammenhang mit der Nahrungsauf- n a h me.

Literature cited AI.I:XANL)ER, R. D. (1961): Aggressiveness, territoriality, and sexual bchaviour in field

crickets (Orthoptera: Gryllidae). Behaviour 17, 130-223 BELL, W. J., C. PARSONS and I,. hlARTiNtio (1972) : Cockroach aggregation pheromones: analysis of aggregation tendency a n d species specificity. J. Kansas Entomol. Soc. 45, 414-421 BELL, W. J., and R. SAMS (1973): Aggressiveness in the cockroach Periplaneta americana (Orthoptera, Blattidae). Uchavioral Biology 9, 581-593 BREED, M. D., B. COLE and G. WETHERELL (in press): Studies on the population biology of the American cockroach, Periplaneta americana EWING, 1.. S. (1967): Fighting and death from stress in a cockroach. Science 155, 1035-1036 EWING, L.. S. (1972): Hierarchy and its relation to territory in the cockroach Nauphoeta cinerea. Bchaviour 42, 152-174 HARRER, J. D. (1960): The effect of perturbations in the environ- mental cycle on the diurnal rhythm of activity of Periplaneta americana L. J. Exp. Biol. 37, 154-163 ISHII, S. (1970): An aggregation pheromone of the German cockroach, Blattella germanica (L.) 2. Species specificity of the pheromone. Appl. Ent. Zool.5,33-4l KING, J. A. (1973): The ecology of aggressive behavior. Ann. Rev. Ecol. Syst. 4, 117-138 LIECHTI, 1'. M., and W. J. BELL (in press): Brooding behavior in the Cuban burrowing cockroach Byrsotria frrmigata. Insectes Sociaux MCKITTRICK, F. A. (1964) : Evolutionary Studies on cockroaches. Cornell Univ. Agric. Expt. Sta. Mem. 389, 1-197 MOYER, K. E. (1968): Kinds of aggression and their physiological basis. Commun. Behav. Biol. 2, 65-87 NOBLE, G. K.

32 M. D. BREED, C. M. HINKLE and W. J. BELL

(1939): The experimental animal from the naturalist’s point of view. Amer. Natur. 73, 113-126 RITTER H., Jr. (1964): Defense of mate and mating chamber in a wood roach. Science 143, 1459-1460 0 SIMON, D., and R. H. BARTH (in press): A comparative study of sexual behavior in the genera Periplaneta and Blatta . SOKAL, R. R., and F. J. ROHLF (1969): Biometry. S. H. Freeman and Company, San Francisco. 776 p.

Author’s address: MICHAEL D. BREED, Department of Entomology, University of Kansas, Lawrence, Kansas 66045, U. S. A.