Attack pheromone response among colonies and castes in the leaf cutting ant Atta cephalotes Matthew M. Ward Department of Biological Sciences, University of Iowa ABSTRACT Attack pheromones in Atta cephalotes are important in coordinating defense respon ses to protect the colony. Little is known about the dynamics of A. cephalotes defenses and their underlying mechanisms Whitehouse and Jaffe 1996. The objective of this study was to test for differences in response between four colonies and between the three classes of ant; minor, worker, and soldiers. Eight treatments were performed to carry out the two experiments. Toothpicks were coated with the attack pheromone from each of the four colonies and each of the three castes and one toothpick was teste d without any pheromone as a control. Using a Kruskal Wallis and Post Hoc test, eight relationships were found to be significant. It was determined that the minor class is the primary attacking class for attack pheromones Kruskal Wallis p value < .000 1, df = 2. These data indicate that A. cephalotes follows the square law proposed by Franks and Partridge 1993. Initially, these minors responded more intensely to the pheromone of their own pheromones Kruskal Wallis p value = .0036, df = 3. Over t he full 20 minute duration of the trial they responded more greatly to the worker pheromone Kruskal Wallis p value = 0.0042, df = 3. The response of the worker class during the trials was greater for the minor pheromone Kruskal Wallis p value = 0.0099, df = 3. The overall response by A. Cephalotes was greater for the toothpicks with the attack pheromone than it was for the control Kruskal Wallis P Value = .0260, df= 4. An interesting relationship was uncovered between two of the colonies studied. When the attack pheromone from colony C was placed in the trails of colony D, the response from the colony was not significantly different from that of the control for both the initial and total responses respectively, Kruskal Wallis p value = .0030, df = 4 and Kruskal Wallis p value = .0269, df = 4. These data may have indicated that when colonies are close to one another, as colony C and colony D are, then interactions may cause the dynamic between them to change. RESUMEN Las feromonas de ataque de la hormiga Atta cephalotes son importantes en la coordinaciÃ³n de las respuestas de defensa para proteger a la colonia. Se sabe muy poco acerca de la dinÃ¡mica de las defensas de A. cephalotes y sus mecanismos fundamentales Whitehouse y Jaffe 1996. El o bjetivo de este estudio fue probar las diferencias en la respuesta entre cuatro colonias y entre las tres categorÃas de hormigas; la menor, la trabajadora y el soldado. Ocho tratamientos se realizaron para llevar a cabo los dos experimentos. Se revistie ron palillos de dientes con la feromona de ataque de cada una de las cuatro colonias y cada una de las tres castas y un palillo de di entes sin ninguna feromona se u tilizÃ³ como control. Se aplicaron las pruebas de Kruskal Wallis y de post hoc para deter minar si las ocho relaciones eran significativas. Se determinÃ³ que la clase menor es la clase atacante primaria para las feromonas de ataque Kruskal Wallis p < 0.0001, df = 2. Esto indica que A. cephalotes sigue la ley cuadrada propuesta por Franks y P artridge 1993, donde el nÃºmero de individuos en la batalla es mÃ¡s importante que la calidad. Inicialmente, estas menores respondieron mÃ¡s intensamente a su propia feromona Kruskal Wallis p = 0.0036, df = 3. Las menores respondieron mÃ¡s a la feromona de las trabajadoras durante los 20 minutos totales que durÃ³ la prueba Kruskal Wallis p = 0.0042, df = 3. La respuesta de la clase trabajadora durante los ensayos fue mayor para la feromona de las menores Kruskal Wallis p = 0.0099, df = 3. La
respues ta general de A. cephalotes fue mayor para los palillos de dientes con la feromona de ataque que la que fue para el control Kruskal Wallis p = 0.0260, df= 4. Se descubriÃ³ una relaciÃ³n interesante entre dos de las colonias estudiadas: cuando la feromona d e ataque de la colonia C se colocÃ³ en los senderos de la colonia D, la respuesta de la colonia no fue significativamente diferente de la del control para las respuestas iniciales y totales Kruskal Wallis p = 0.0030, df = 4 y Kruskal Wallis p = 0.0269, df = 4, respectivamente. Estos datos pueden indicar que cuando las colonias se localizan cerca la una de la otra, como en el caso de las colonias C y D, entonces sus interacciones pueden causar que la dinÃ¡mica entre ellas cambie. INTRODUCTION Leaf cutting ants in the genus Atta , are influential organisms in the Neotropics. They live in colonies numbering in the millions and cause between 12 and 17 percent of herbivory in the Neotropics Holldobler and Wilson 1990. The species in the Atta genus such as A tta cephalotes can be highly polymorphic; ranging from the miniscule minors to the over sized soldiers with menacing jaws. This high level of polymorphism has been implicated in increasing foraging efficiency as well as for defense mechanisms. Many oppor tunistic creatures would gladly make a meal of A. cephalotes. Paraponera clavata has been observed preying upon individuals from A. Cephalotes foraging trails Wetterer 1994. There is a species of army ant, Namamyrmex esenbeckii , that preys almost exclu sively on a leaf cutter colonies and there have been cases where an A. cephalotes has been destroyed by these ants Swartz 1998. Phorid flies will attack medium sized workers as they carry leaves and lay an egg on the back of the ant, where it will hatch and kill the individual Miller 1992. The responses to these attacks varies greatly; from simple avoidance in the case of P. clavata, to the active defense of the colony from N. esenbeckii and plugging of entrances by the solider class, to the presence of minors Â€hitchhikingÂ on the leaf fragment fending phorid flies Miller 1992. The aforementioned examples are only cases in which the threat is interspecific. A. cephalotes also must be prepared to encounter conspecifics of a different colony, which a t times may result in battling over food resources Whitehouse and Jaffe 1996. How is A. cephalotes able to orchestrate these defenses and what factors govern the measure of their response? Franks and Partridge 1993 proposed that two models used in wa rfare can be applied to ants. The first, the square law, states that in battles where all individuals are endangered, then sheer numbers of individuals will win the battle. The second, the linear law, applies when the battle is a series of one to one sk irmishes with an excess number of individuals waiting to fight. These conditions favor a few good fighters soldiers over many poorly equipped fighters minors. Little is known about the dynamics of A. cephalotes defenses and their underlying mechanism s Whitehouse and Jaffe 1996. Attempting to gain a better understanding of the attack responses, data were taken for two experiments using eight different treatments. The first experiment involved three treatments using the attack pheromone from each of the three different castes, minor, worker, and soldier, as well as a control. The second experiment was to see if the attack pheromones from different colonies would elicit a different response then the colonyÂ‚s own attack pheromone. These data were als o intended to determine whether intercolony battles would be governed by square law lots of ants or the linear law fewer ants but better equipped to do battle.
MATERIALS AND METHODS This pro ject was conducted in the Santu ario EcolÃ³gico in Cerro P lano, Puntarenas, Costa Rica Fig. 1. Using eight treatments, 4 with the attack pheromone from different colonies, 3 with the attack pheromone from the different classes, and a control, the data were analyzed to determine the associations between colonie s and between classes. Each of the eight treatments was performed 1m from an entrance close to the colony. A total of 4 colonies were used in data collection, each of which was utilized 3 times, yielding twelve data points for each treatment. Data were c ollected between 0800 and 1700 hours from April 25 to May 8, 2006. The 4 colonies were labeled A, B, C, and D; the colony closest to the entrance of the sanctuary was labeled A and the farthest, D Fig.2. Prior to data collection for the day, all colon ies were visited so that ants from each colony could be collected using tweezers. While collecting, each ant was assigned to one of the three classes: minor, worker, and soldier. A caliper was used to classify each ant. Ants with a the widest part of th e head measuring less than 1 mm was a minor, ants with heads wider than 3 mm were soldiers, and any ant between 1 and 3 mm was a worker. Ants were measured until the experimenter was able to categorize them without the use of a caliper. Each colony was re presented by 3 5 soldiers, 9 13 workers, and 15 20 minors. These samples from each class were approximately equal at 0.2 grams. Additional ants from the colony that was tested were collected and the three classes were kept separate. The abdomens from the ants collected were crushed using tweezers and the excretion was applied along the length of a toothpick, being careful not to touch the toothpick or ants to avoid contamination. A toothpick for each colony was prepared using the scents from all three classes. Toothpicks with only the scent of one of the three classes were prepared using ants from the colony being observed. One toothpick without any scent was used in a trial to serve as a control. Toothpicks were placed parallel to the foraging tra il, to minimize the trail clearing response often observed in A. cephalotes , at the approximate center Holldobler and Wilson 1990. The Â€attack responsesÂ were recorded at 1, 2, 3, 5, 10, 15, and 20 minutes after the start of the trial. The cataloging o f an ant as Â€attackingÂ required that its mandibles were attached to the toothpick at the time of observation. The number of ants attacking from each class was recorded at the aforementioned intervals. The procedure was repeated 96 times, allowing for 3 trials from all 4 colonies. To test for significant differences between responses to attack pheromones of the 4 different colonies, Kruskal Wallis tests were performed and post hoc tests were used to determine where to assign significance. Kruskal Wall is and post hoc tests were also used to determine if there were any significant differences among the attack pheromones of the three classes. RESULTS The relationships between the different colonies and the different castes were analyzed using data fro m the initial response number of ants that attacked in the first minute and the total response number of ants that attacked during the 20 minute trial; when the number of ants, in each class displaying an attack response in the first minute was tallied ,
a significant difference was found between the three classes Kruskal Wallis p value < 0.0001, df= 2. The minor class displayed a significantly higher initial response than the worker and soldier classes and the worker class response was significantly higher than that of the soldier class see Figure 3 Kruskal Wallis p value < 0.0001, df= 2. This trend, soldier response< worker response< minor response, was also seen in the total response data Fig. 4. An analysis of only the data in which minor s were attacking, yielded two significant relationships; when the initial response of the minors was compared for the three caste pheromones and the control, the minor showed preference to their own pheromone Fig. 5 Kruskal Wallis p value = 0.0036, df = 3. Total response of the minima showed a different trend and they displayed a higher attack response for the worker pheromone Fig. 6 Kruskal Wallis p value = 0.0042, df = 3. The same analyses were performed considering only the worker and soldier c lasses. The total response of the workers was significantly greater for the minor pheromone Fig. 7 Kruskal Wallis p value = 0.0099, df = 3. No significant differences were found in the initial worker response or either of the soldier responses for th e 4 treatments; minor, worker, soldier, and control. Analysis showed that there was no significant difference between the initial responses elicited for any of the 5 treatments, using the data from the 4 treatments where the individual colony attack pher omones were used and along with the control. The total responses, however, indicated that the attack pheromones from the four colonies received more attention from the ants than the control during the 20 minute trials Fig. 8 Kruskal Wallis p value = 0.0 260, df= 4. This justifies the concept the crushing the abdomen releases a pheromone that initiates an attack response. Detecting differences between colony attack pheromones required that the analysis was carried out four times, using the data from onl y one colony each time. These tests indicated that there was no difference in either the total or initial responses to the different treatments for colonies A, B, and C. Colony D, on the other hand, showed a significantly lower initial response for both the control and the pheromone from colony C Fig. 9 Kruskal Wallis p value = 0.0030, df= 4. This trend continued for total response and colony D showed a significantly higher overall response to the pheromones from colonies A and B than it did for the control and the pheromone from colony C Fig.10 Kruskal Wallis p value = 0.0269, df= 4. The data from colony D are unique in that, the response to the pheromone of colony C is significantly different to that of the other two colonies. Both the initial response and the total response figures show that colony D attacked the control and the colony C pheromone equally. Furthermore, several times when the colony C pheromone was introduced into the foraging trails of colony D, it was observed that the ants carrying leaves would cache them and returns to the colony. DISCUSSION The purpose of this experiment was to determine how the attack response is coordinated in A. cephalotes and to attempt to describe some of the underlying mechanisms. The data indica te some interesting relationships, in addition to some more straightforward conclusions.
The first two analyses indicate that overall, the minors are the first responders for any foreign object introduced into the foraging trail; both the initial response and the total response are predominately made by the minors, with workers responding second most often and soldiers, a distant third. This relationship could be explained by the apparent abundance of each of the three classes. Casual observations of how prevalent each of the classes was indicate a similar relationship. There appeared to be more minors than workers and soldiers, and more workers than soldiers on the trail at any given time. Another possibility is that the task of identifying threats fall s to the minor class, as it does for identifying suitable substrate for the fungal gardens Stevens 1983. Figure 5 shows that the minors responded most aggressively to their own pheromone in the first minute. This coincides with the possibility that min ors identify threats because more minors were recruited to attack and further assess the danger posed by the foreign object. They responded less aggressively in the first minute when the attack pheromone was that of a worker or soldier. An object having one of these pheromones would indicate to a passing minor that the object was already being taken care of by larger individuals better suited to attack. It is apparent that though the minors responded most aggressively to their own pheromone initially, th eir total response was greater for the pheromone for the worker class. This greater total response may be due to the minors sensing the attack pheromone of a worker but not sensing a worker attacking. This might have caused the minors to attack and secre te their own pheromone, recruiting more minors, since the minors were not recruited right away, the whole cycle of conscription started in the middle of the observation period. Perhaps, those minors that first attacked were still present when more minors were recruited and the total response was greater as a result. This relationship warrants further research. The workers showed a preference for minor pheromone over the whole observation period. Perhaps this relationship could be explained because once the minor attack pheromone reaches a certain density; it acts as a trigger for the worker class to begin attacking as well. The positive feedback from the minor scent recruiting more minors causes the phenomenal density to reach a critical level where rec ruitment of the next larger class is initiated. Whitehouse and Jaffe 1996 found that small ants or minors may play a key role in marking territories during battles. This marking behavior almost always decides who wins the battle and is an example of the square law Franks and Partridge 1993. It might be possible that the two colonies, being in close proximity to one another, battled recently and that colony D had been forced to submit to colony C; when reacquainted to the attack pheromone of colony C, the ants foraging submitted and cached their leaves to return to the colony more quickly. The intraspecific attack response in A. cephalotes appears to be primarily following the square law. Minor workers were the dominant attackers, indicating that the response to attack pheromones is more based on sheer numbers than more effective fighters. The linear law may have a different signal based on vibrations or foreign smells. The majority of soldier ants observed were attempting the attack the experimenter and not the toothpicks. The presence of a large mammal next to foraging trails for an extended period seemed to initiate a fairly strong response. Perhaps a future study could involve different odors such as human sweat in addition to the attack pherom ones. The
study of attack pheromones in A. cephalotes would also benefit from experiments on the residual effects of a battle between colonies. It would also be interesting to see how the response changes both closer and further from the entrance. ACKN OWLEDGMENTS First, I want to thank Javier MÃ©ndez. He supported me through problem after problem and was implemental in my last minute transition between projects. To Maria Jost, thank you for helping to narrow the focus and making everything seem small er and more manageable. Thanks to Oliver Hymen for his guidance during the statistical analysis. I am forever grateful to Kathy Rockwell, Alan and Karen Master, and Javier, who helped me through some of the more difficult times of my life in these past m onths and still they managed to make this semester the most incredible and life altering period of my life. Thanks to all my fellow students who visited me in hospital, convinced me that Panama was boring, and greeted me with open arms when I returned fro m the states. Finally, I would like to thank Gommie, who gave all the support and love a grandchild could ever hope to receive. The world is a little worse off now that she is gone. Thank you all. LITERATURE CITED Holldobler, B. and Wilson E. O. 1990 . The Ants. Harvard University Press. Cambridge, MA. pp. 596 608, 245 Franks, N. R. and Partridge, L. W. 1993. Lanchester battles and the evolution of combat in ants. Animal Behavior, 45, pp. 197 199 Stevens, G.C. 1983. Atta cephalotes Zompopas, Leaf cutt ing Ants In: Costa Rican Natural History, Janzen, Daniel H. Ed. The University of Chicago Press. Chicago, IL. pp. 688 691 Miller, S. 1992. A HitchhikerÂ‚s Guide to the Colony: A Close up study of major and minor worker of the leaf cutter ants Atta Cephalot es . Tropical Biology and Conservation Council on International Education Exchange. Summer, 1992. pp. 124 134. Swartz, M. B. 1998. Attack on an Atta cephalotes Colony by an Army Ant, Nomamymex esenbeckii . Biotropica, 30, 4, pp. 682 685 Wetterer, J. K. 1994 . Attack by Paraponera clavata Prevents Herbivory by the Leaf cutting Ant, Atta cephalotes . Biotropica, 26, 4, pp. 462 465 Whitehouse, M. E. and Jaffe, K. 1996. Ant Wars: Combat strategies, territory and nest defense in the leaf cutting ant Atta laevigata , Animal Behavior, 51, pp.1207 1217
Figure 1. Map of the Monteverde area. The Ecological Sanctuary black oval is located in Cerro Plano, Puntarenas, Costa Rica. Four colonies of Atta cephalotes were studied to test for differences in attack phe romones between colonies and classes. Data were collected between 0800 and 1700 hours from April 25 to May 8, 2006.
Figure 2. Map of the Ecological Sanctuary. Four colonies were labeled A D. Each colony was tested 3 times to check for a difference i n attack pheromones between colonies and the 3 classes. A control was also tested. Toothpicks were covered with the attack pheromones from the different colonies and classes.
Figure 3. The difference in the type of ants to respond in the first minute to any of the 8 treatments tested for Atta cephalotes . The x axis represents the type of ant and the y axis represents the sum rank produced by the Kruskal Wallis test. Significant differences were found between the numbers of ants that responded in the first minute. Kruskal Wallis p value < 0.0001, df= 2. Figure 4. The differences in responses in Atta cephalotes by the three different ant types over the 20 minute trial are shown above. The x axis indicates ant class type and the y a xis show the sum rank output produced by the Kruskal Wallis test. Significant differences were found between the numbers of ants that responded in over the full 20 minute period. Kruskal Wallis p value < 0.0001, df= 2 0 1000 2000 3000 4000 5000 6000 7000 8000 9000 Minor Worker Soldier Ant Type Sum Rank 0 1000 2000 3000 4000 5000 6000 7000 8000 9000 Minor Worker Soldier Ant Type Sum Rank
Figure 5. Only data for mi nors in the Atta colonies are used in this graph. The x axis indicates treatment type and the y axis show the sum rank output produced by the Kruskal Wallis test. Minors showed a preference for their own attack pheromone within the first minute of observa tion. Kruskal Wallis p value = 0.0036, df= 3 Figure 6. Only data for minors in the Atta colonies are used in this graph. Minors show a preference for the worker attack pheromone over the whole 20 minute trial period. The x axis indicates treatm ent type and the y axis show the sum rank output produced by the Kruskal Wallis test. Kruskal Wallis p value =0.0042, df= 3 0 50 100 150 200 250 300 350 400 450 Control Minor Worker Soldier Treatment Sum Rank 0 50 100 150 200 250 300 350 400 450 500 Control Minor Worker Soldier Treatment Sum Rank
Figure 7. Only data for workers in the Atta colonies are used in this graph. Workers showed a preference for the mi nor attack pheromone over the whole 20 minute trial period. The x axis indicates treatment type and the y axis show the sum rank output produced by the Kruskal Wallis test. Kruskal Wallis p value = 0.0099, df= 3 Figure 8. All coloniesÂ‚ response s to the scents of other colonies were tested against the control. Control is the only one to show a significant difference. This indicates that crushing the abdomen was sufficient to tract the attack pheromone from Atta cephalotes. The x axis indicates tre atment type and the y axis show the sum rank output produced by the Kruskal Wallis test. Kruskal Wallis p value = 0.0260, df= 4 0 50 100 150 200 250 300 350 400 450 Control Minor Worker Soldier Treatment Sum Rank 0 500 1000 1500 2000 2500 3000 3500 4000 Control Scent A Scent B Scent C Scent D Treatment Sum Rank
Figure 9. Using only the data from colony D for initial response, all four scents were tested. Control and Sc ent C were found to be significantly different. Colony DÂ‚s response to colony C was different from that of colonies A and B. The x axis indicates treatment type and the y axis show the sum rank output produced by the Kruskal Wallis test. Kruskal Wallis p value = .0260, df= 4 Figure 10. Using only the data from colony D for total response, all four scents were tested. Control and Scent C were found to be significantly different. Colony DÂ‚s response to colony C was different from that of col onies A and B. The x axis indicates treatment type and the y axis show the sum rank output produced by the Kruskal Wallis test. Kruskal Wallis p value = .0269, df= 4. 0 50 100 150 200 250 300 Control Scent A Scent B Scent C Scent D Treatment Sum Rank 0 50 100 150 200 250 300 Control Scent A Scent B Scent C Scent D Treatment Sum Rank